The Aircraft Engineer 1934

January 25, 1934
Supplement to FLIGHT
FLIGHT
ENGINEERING SECTION
Edited by C. M. POULSEN January 25, 1934 CONTENTS
Ethyl. By F. R. Banks, O.B.E.. P.E.Ae.S.. M.I.A.E.. M.Inst.P.T.. M.S.A.E In the Drawing Office Laying out Lines and Plating. By It. Haley Hiduminium E.R.53 B Power curves of De Uavilland "Gipsy Six" engine
Page 1 5 8 8
ETHYL
By F. R. BANKS, O.B.E., F . R . A E . S . , M . I N S T . P . T . , M.S.A.E. M.I.A.E.,
The paper on " Ethyl," read by Mr. Banks to the Royal Aeronautical Society on January 18, presents some difficulty to a paper like FLIGHT, in that its importance, in view of the Air Ministry's decision to use leaded fuel in the future, is such that it ought to be published in full. Space does not, however, permit of this course being tahen, and we have decided to publish a summary of the first part of Mr. Banks' paper in the FLIGHT pages this week. The second, and more tecltnical, part of the paper is summarised below, but will take one or two more, instalments to complete.ED. ENGINE DEVELOPMENT Some information indicating the lines to follow in order to make full use of fuels containing lead THE manner in which leaded fuels have been viewed, in some quarters, is extraordinary. If the same difficulties had been put in the way of the development of superchargers in this country, one is certain that they would not yet be in use except in a very experimental way, and in consequence we would not have been in the position to produce some of the world's best aviation engines, as we are doing to-day. With regard to the employment of high Octane fuels in America, while a large amount of test bench work was, and still is, being done, they appreciated the necessity of obtaining practical flight experience in the early stages of development and took the view that whatever the results of bench tests, they would probably come up against sundry difficulties in actual service, so the sooner experience was gained, the better. We, in this country, have admittedly obtained valuable data from the test bed during the past two years or so, particularly on leaded fuels, but little data has been obtained on fuels of high Octane value and the operation of high duty engines in actual service. The problem of the operation of aviation engines on any fuel, because of their relatively high specific power output, is quite peculiar to their class and cannot be compared with the normal operation of automobile
engines, with which little or no trouble is now experienced over long periods of use. The aviation engine has to deal continuously with a far greater bulk of heat per unit volume of cylinder than practically any other petrol engine. I t is, therefore, more difficult to obtain reliability ovei long periods of service without paying greater attention to it between times. Such items as the pistons and valves are very much larger than those in normal automobile engines, and it is, therefore, difficult to arrange the design of these parts so that they are able to dispose of the heat satisfactorily and still retain their form and material structure. Modern engine development, therefore, demands what may be called at the present time " super fuels," although in five or ten years' time these will probably be considered mediocre. The Effect of the Products of Combustion of a Leaded Fuel upon the Engine Parts The main products of the combustion of a leaded fuel with which we are concerned is lead bromide. Under particular circumstances, the presence of lead bromide may aggravate certain troubles which, previously, might have been latent in the engine, with the possible exception of one, namely, cold corrosion. The weak links in the chain connecting satisfactory engine performance with leaded fuel are as follows : (a) The tendency of the exhaust valve to scale and burn, due to " hot ' corrosion attack of the valve seat and insert, and/or " mechanical " attack by the pieces of detached scale. (b) " Cold " corrosion attack of the exhaust valve stems, when the engine is standing idle. (c) " Cold " corrosion of the cylinder bores under the same conditions as in (b). (d) Corrosion attack of the exhaust pipes and collector rings. (e) Deposition of lead salts on the sparking plugs. This list appears formidable, but with the materials available to-day, coupled with suitable design, there is no need for these troubles to persist. Exhaust Valves and Valve Seat Inserts, General Considerations The exhaust valve is undoubtedly the most vulnerable part of an engine, since it must perform well at elevated temperatures (around 800 deg. C.) when working without undue scaling or becoming mechanically weakened and, in addition, remaining gas-tight. In the case of an exhaust valve which works in the presence of leaded fuel, a hard, black, polished and
78 a
c2
SUPPLEMENT TO
PLIGHT
JANUARY 25, 1934
THE AIRCRAFT ENGINEER

and, in fact, one or two manufacturers have done a considerable amount of running with steel inserts and intend to standardise them in all their future models. The materials generally favoured are those of the semiaustenitic or austenitic variety, and in some cases are similar to those used for the valves themselves. These steels generally have a good resistance to corrosion and scaling at high temperatures, and there are some particular brands which are exceedingly tough and " work harden " to a high degree. Notable among these is N.M.C. (nickel, manganese, chromium), a high-expansion steel produced by Firth's. This steel has a coefficient of expansion which lies between that of aluminium bronze and the aluminium alloys generally used for cylinder head material, being 0.0000223 between 200 and 300 deg. C. It is extraordinary how a change in insert material afffictB the valve. Many cases have come to one's notice where valves have burned when operating against aluminium bronze inserts and have been completely cured on going over to steel inserts. The following combinations could, in the light of recent experience, be tried with advantage: (a) A valve of austenitio steel working against an insert of similar material or N.M.C., preferably the latter. (b) A similar type of valve as in (a) " Stellited " on the seat and working against insert material which is not " Stellited." (c) Both the valve and insert " Stellited," using the same materials as in (a). (d) An un--( Stellited " valve against a " Stellited " insert, similar materials used as in (a). In the case of (a), K.E. 965 valves have been tried with success on prolonged tests against inserts of N.M.C., both on complete air-cooled engines and single cylinders of the same type. Some Notes and General Information on the Valve and Insert Design, Construction and Fitting The following deals with the sodium-cooled valve and steel insert, because it is felt that this combination is a direct line of development to pursue immediately. Valves In Fig. 8 sketches are shown of the sections of two sodium-cooled valves differing in design. (a) shows the latest type of sodium-cooled valve developed in America. The hollow forged head should be noted, together with the swaged-in hollow plug or thimble at the stem end. The latter is to prevent excess heat from travelling up to the stem end upon which the valve rockers operate, particularly in the case of the
MAT at useDiHPVKC or cup/mo
M.3Q TO Ft* PLUS.
adherent skin generally appears to form on the seat surface in the early stages of running. It is not yet clear whether this skin is a lead product deposited on the seat surface, or is a result of corrosion attack of the material itself. One is, however, inclined to the former theory, that it is in the nature of a glaze, since it has also been found on valves which have operated successfully for long periods. The thickness of the glaze or skin varies somewhat, and is difficult to assess accurately, but from photomicrographs appears to be in the region of 0.0003 in. In cases where the skin has been found on valves which have operated successfully for long periods, it is of even thickness with an unbroken surface. In the case of a failure (valve burning) the trouble seems to start when the skin or glaze breaksdown. The degree by which the exhaust valve, when in operation, may or may not be attacked by lead bromide is mainly controlled by its working temperature and the material used in its construction. It is one's experience that low working temperatures for the valve with a given material may cause little or no trouble, but an increase in the former may give rise to valve troubles due to failure of the material to withstand, at increased temperature, the lead bromide attack, and it will then scale and perhaps burn. To eliminate these troubles at the outset, the broad principle is to ensure by suitable design that the working temperature of the exhaust valve will be kept to an absolute minimum and a valve material chosen which, in addition to high mechanical strength and durability at elevated temperatures, will have a good resistance to attack by the exhaust products. The experience we have in this country, and on the Continent, seems* to indicate that the valve seating* in the cylinders themselves exercise a considerable influence upon the conditions of the seat of the valve and therefore, the life and behaviour of the latter. In this connection it is interesting to note that the question of the conductivity factor of the neat insert material appears to be of secondary importance and, provided that the coefficient of expansion of the material is as near as possible to that of the cylinder head material in which it is fitted and the method of fitting satisfactory, i.e., good thermal contact maintained, its capacity for getting rid of heat is much greater than the amount of heat which the valve can give up to it. The important features which a valve seat insert should possess are : a high resistance to corrosion attack and good surface hardness, or, in any case, toughness. Corrosion resistance appears to be the principal answer to valve-burning troubles and, in addition, the provision of a hardened surface in one or both cases seems to prevent abrasion of the seat surfaces by any scale which would tend to spoil the thermal contact between the valve and its insert. The rate of development of the modern aviation engine is such that great difficulty is being experienced in keeping the exhaust valve temperature within reasonable limits. I t has been apparent during the* last year or so that however efficient the design of the valves and seat inserts, it is difficult, if not impossible, to get rid of the heat adequately in this manner. Obviously, the hotter the exhaust valve the more limited are the capabilities of the engine with fuels of given Octane value. This has led to the development of the internallycooled valve, which is one having a hollow stem, and sometimes head, partially filled with a medium, usually metallic sodium, which will efficiently transfer some of the heat from the valve head to the cooling medium of the engine (air or liquid) via the valve stem and guide. Suitable Valve and Insert Materials and Combinations The material most generally used up to the present time, for inserts, is aluminium bronze. This is still used,, but it will, undoubtedly, be superseded by the special alloy steels. These steels are already being tried,
rputto xm/o/r j t i w aauewa

T*/>ER PLUS
SOBiUM
Fig 8.Typical examples of sodium-filled valves
78 b
JANUARY 25,
1934
SUPPLEMENT TO

partially lubricated valve gears of air-cooled engines, where there is a risk of excessive heat getting to the valve springs and softening them. (b) gives a sketch of a similar valve with a hollow stem only, and it shows a different method of plugging which is also in use. The taper plug is pushed in, and the material peaned over it at the stem end. A loose, hardened cap is then fitted over the stem end, or a hardened tungsten-steel button welded on. Quite a good idea is to " Stellite " over the stem end, and this will ensure that the plugging remains tight and provides something of suitable hardness upon which the rockers can operate directly. The valve is of the modified tulip type. The full tulip head is not desirable since, due to its greater exposed surface, it usually runs hotter than the other types, and in any case the sodium is too far
FLIGHT
properly and cause valve burning. Distortion is very difficult to detect or control, but it valve burning is experienced, this possibility should be the first to be investigated. The rim of the valve, and the insert, should bo regarded as two concentric rings, and both designed to achieve this effect in practice. The design of the valve guide for a sodium-cooled valve is very important, because the stem has to deal with extra heat, and unless it can get rid of it to the cooling medium in an efficient manner, a high degree of guide wear will result. Fig. 9 gives rough sketches of different types of inserts and the methods of fitting them. The shrunk screwed insert is in general use in this country, whereas in America the plain shrunk type is practically universal. Further details will be found in the main text.
SHRUNK
NO
SCREWED
m INSERTS
Fig. 9Various methods of fitting valve seat inserts. The screwed type is in general use in England while the plain type is almost universal in America removed from effective contact with most of the head material. The most effective valve seat angle is, in my opinion, one of 30 deg. This gives a better average valve opening characteristic than, say, one of 45 deg., but it is more difficult to ensure maintaining a satisfactory seating with this than with one to the latter angle, and, perhaps, in the case of an exhaust valve, particularly, it is more important to ensure that the valve seats itself efficiently rather than to benefit by the small increase in porting efficiency offered by the former angle. The question regarding the most satisfactory width of valve seat to employ is a vexed one. For some time, the American engine firms have favoured a relatively wide seat, and in some cases the seat widths have been somewhat excessive. The tendency in the States now appears to be towards a narrower seat, particularly for engines of high specific power output using fuels containing lead. The seat widths of some typical American valves range from ^ in. to 1 in. for a large valve having a seat diameter of about 3 in. and from | in. to & in. for valves of 1.75 in. to 2 in. seat diameter. The British engines generally keep to the lower limits of the widths mentioned, and in some cases are only half the width, in proportion. One hesitates to speak of " wide " or " narrow " seats as such, and considers that the question of width is really a compromise which is controlled by the considerations of individual engine design. Satisfactory unit loading, to achieve good thermal contact, should be considered of primary importance. There is a risk, if the seat is very narrow, that any " blow past " which may occur will cause " guttering " right across the seat, whereas with a wider seat* under similar conditions, it usually takes longer for this to occur, and it may only show up in the form of pitting. I t will be appreciated, therefore, that it is unwise to dogmatise on such matters. Many valve troubles are occasioned by distortion of the cylinder head. This is particularly liable in the case of the monobloc arrangement used for the large modern liquid-cooled engines, and it has also been the cause of a large amount of trouble in some ordinary motor-car engines. Distortion, so far as it may affect the valve, generally causes misalignment between the valve guides and inserts, or actual " ovalising " of the latter, both of which prevent the valves from seating " Cold" Corrosion The cause of " cold " corrosion is generally accepted to be as follows: When an engine comes to rest, a certain amount of condensation takes place, particularly in the case of a cylinder in which the exhaust valve has stopped in the open position. Any lead bromide present will also condense, and this, with the moisture present, gives rise to corrosive action, particularly on steel parts. The principal points attacked by " cold " corrosion are the exhaust valve stems and the cylinder barrels. The extent to which they are attacked is influenced by the materials used, and also the working conditions. The general treatment, to avoid attack, is to ensure that the parts concerned are well covered by a film of mineral lubricant. Further details concerning particular treatment will be found in the main text. The use of " nitriding " in this connection has been quoted, but our experience here and in Europe seems to indicate that, so long as the parts made from these materials are kept well lubricated, little or no trouble is encountered. Adequate lubrication is naturally quite possible with, say, cylinder barrels, but the use of completely nitrided exhaust valves is not recommended, although the valve stems may be treated in this manner and are quite satisfactory so long as the nitriding is not carried too far down to the hot part of the stem. Exhaust Pipes and. Collector Rings The usual material employed in the construction of exhaust manifolds is mild steel, which after forming is welded and riveted. Mild steel, although fairly satisfactory, is not the best all-round material to use, and is very prone to split, due to a combination of temperature variation and vibration, which cause fatigue of the material. Steels of the stainless and /or austenitic variety are gradually coming into use, particularly in America. These are corrosion resistant - * 1 j a high degree and do not scale easily. Exhaust collector rings of stainless steel are, one understands, very generally used by the military machines in the U.S.A. The use of mild steel rings has been general with the civil aircraft operating concerns in the States, and although no great trouble was experienced from their continued use in conjunction with that of leaded fuels, it is understood that they are being replaced with stainless steel rings as occasion permits.
78 c
SUPPLEMENT TO
JANUARY 25,
FLIGHT
Sparking Plugs There 13 no more difficulty in obtaining a suitable plug tor operation with leaded fuel than choosing one for any engine whatever the fuel used. In either case a satisfactory choice largely resolves itself into a matter of practical' test. The three troubles which may possibly occur are (1) an excessive rate of build up of deposit on the insulator; (2) possible breakdown of the
(/HMODIfiED PLUQ
Fig. 11This diagram shows the various stages of stelliting a valve seat. cent. The amounts of the constituents vary according to the particular grade and degree of hardness required. Both at low and elevated temperatures its resistance to wear and oxidation is excellent. It is particularly effective in resisting, at high temperatures, corrosion attack by lead bromide or the products of its decomposition. The treatment of valves and inserts consists, briefly, of applying a layer of " Stellite " to the seat surfaces, this being done by means of an oxy-acetylene torch. " Stelliting " is not, actually, a welding process, since fusion of the Stellite and the material of the part being treated, is not desired. It might be described as a high-temperature brazing process. A description of the technique of applying Stellite will be found in the main text. The figures 11 to 15, however, show the essential points in the process. Fig. 12 shows a typical example of an exhaust valve of a large American radial engine. The valve has a hollow forged head, sodium filled, and both the seating and valve head are Stellited. This valve measures about 3 in. across the outer diameter of the seating and weighs approximately one pound in its completed form. Fig. 13 gives a sectional view of the same valve. Note the hollow swaged in plug at the stem end, and also the thickening up of the stem section at the neck. This valve is a remarkable example of accurate forging. The illustration in Fig. 14 shows Stellite deposited on a rough machined valve of martensitic steel. The segment cut from the valve head shows up the section. Fig. 15 shows a valve of the same type as Fig. 14, but it is of austenitic steel similar to K.E. 965, and is the product of another manufacturer. Note the difference in the Stelliting. (To be continued.)
Of HOT 6AS. So*ie Gerosir mu. ftHTM He#E
trs
. v 6000
aezc.I BUILO IM Of O**osrr\
TMEVE M*Y SM //y This Oue Tb /rsS*LL
- O>t/s//iG Ore* He*Tt*e
Fig. 10This sketch shows moditications made to mica sparking plugs to suit air-cooled engines using a "leaded" fuel. insulation; and (3) erosion or corrosion of the electrode tip. These are not necessarily related, and are differently influenced by temperature. More complete details will be found in the main text, but let it be said that any troubles experienced with sparking plugs are not critical, and the cures are relatively straightforward. Some Notes on " Stelliting " Although in use for some time, the application of Satellite to the valve and seat inserts of aviation engines is only a relatively recent development, consequently the following information may be of some assistance. " Stellite " is a material composed largely of cobalt and chromium, between 50-65 per cent, of the former and 30 per cent, of the latter, together with tungsten, which may vary, in proportion, from about 4 to 20 per
Figs. 12 to 15Show further details of different valve types and methods of " stelliting." 78 A
;' ""
JANUARY 25, 1934
SUl'PLEMFWT TO

IN THE DRAWING OFFICE
LAYING OUT LINES AND PLATING BY K, HALEY.
FLIGHT
IN getting out a set of lines for either an ordinary " fairing," drawing, or plating up a monocoque fuselage, it is impossible to get them correct unless the}7 are faired in three views; putting a line in " to eye " is hopeless. Only a few years ago, when 90 per cent, of the fuselage fairings were done in wood, it was quite a common sight to see a skeleton fairing in position, with pieces glued on here, and pieces cut out there, to allow the stringers to lie fair, proving that the lines were far from accurate. Imagine plating a ship with f-in. plating and the lines not being fair. No doubt this procedure was satisfactory to a degree, as when the fabric was put on the general appearance was passable, but now that we are familiar with metal moiiocoque fuselages the writer is of the opinion that it is worth the trouble to fair the lines properly in the first place, and eliminate any error, or endeavour to do so. In the following article the writer will try to describe the method employed in laying out lines, half-block model, and shell expansion, for a metal fuselage. The section on lines will, of course, apply equally to a set of " fairing lines " for an ordinary fairing drawing. Lines To begin with, endeavour to lay out the drawing paper for the lines to as big a scale as possible, and the whole bench can, and should be, utilised, unless the machine is extremely short. A steel straight-edge, 6 ft. long, is a useful tool when laying out lines and a set of splines or battens, with at least a dozen lead weights are essential. If time will permit, the cartridge paper should be pinned down the night before and allowed to stretch. Set down the " datum line," or " thrust line," as in Fig. 1, also centre line in plan and " body plan." Draw in outline in profile and plan, the plan being only drawn in the bottom half of the lower drawing. Subdivide the total length between the nose and the sternpost, into an equal number of " stations " (if possible, these stations should be a multiple of the frames or fairing formers). Number these stations 0, 1, 2, 3, etc., from the left-hand side. The number of stations will depend on the total length, and whether the fuselage has a lot of curvature in its length. At the ends, especially the nose, half-stations can be inserted for fairing purposes only. Having laid out the outline in profile and half-plan, draw in the shape of the widest station (usually at half-length) on the
" body plan," to the left hand of the centre line, station 5, fig. 1. Next, draw lines enclosing the width and total depth on the body plan, as A, B, C, D, Fig. 1, and divide this figure into equal parts from " 0," both vertical and horizontal, the horizontal lines being numbered from the base upwards, and the vertical lines numbered from the centre line to the right and left. The horizontal planes will be known as such, and the vertical lines will be called " buttocks." Mark off the intersection of these horizontal lines with the profile and the intersection of the buttocks with the plan outline, as at " XX " and " YY," Fig. 1. Now we are ready to transfer the height above and depth below the datum of each station to the body plan, also the half-width. Having chosen the middle station and marked same previously, mark off other stations on the body plan thus:All stations forward of the middle station mark off to the left of the centre line of the body plan, and all stations ait of the middle station mark off to the right of the centre line. On the body plan we now have the height and depth of each station on its centre line and the half-width along the datum about " 0." These positions are " lifted " off the profile and plan with strips of paper about in. wide. Next proceed to lightly outline the shape of each station on the body plan after having noted an} necessary flats that may occur, such as a bomber's window on the bottom surface, etc The shape of the sections will otherwise be largely governed by the shape of the middle section. After all the stations and half-stations (if any) have been drawn in, proceed to draw in the diagonals as OB, OC, 0A, 0D, Fig. 1. Fairing The real " fairing " has not yet been carried out, and we now proceed to fair our " body plan." Taking a strip of paper place same along the horizontal plane No. 1, and mark off the centre line on the strip and all the intersection points of all stations that cut this line. Transfer these points on the strip to the bottom half of the " plan " and mark off the half-width of each station. Place a batten through these points with weights thereon, and lightly draw a line through them. Proceed as above for all the horizontal planes. Do not attempt to correct shape unless there is an obvious bump, as it may be due to quite another cause. The alteration to another line later on may bring any irregular lines in the early stages to their correct spots. Now, with another strip of paper placed on Buttock No. 1, mark off the intersection of all the stations fore and aft, i.e., each side of the centre line on the " bodyplan." Transfer these points to the profile and mark off these points at each station. The ends; of these
/****
FIG.I.
.M*
SUPPLEMENT TO
JANUARY25, 1934
FLIGHT

FIRM'S NAME
STEEL ORDER SHEET NO.
To MESSRS No REQUMD
LENGTH BREADTX THICKNESS MARK.
CAlULCTED WEIGHT
r"
i ^
FIG.H
buttocks will be projected from the plan as at YY, Fig. I. Carry out this procedure for all the buttocks, having lightly drawn in a line with the batten through all the points. The third and last operation is to mark off on a fresh strip of paper the intersection of all " stations " with the diagonals. Note here that the stations forward of the middle station are taken on the line OA and the stations aft of the middle station oix the line OB, Fig. I, which is No. 1 Diag. The lines OC, OD is No. 2 Diag. Transfer these points on to the stations on the plan and mark off above the centre line. Other diagonals may be marked on the body plan if desired, but usually two will suffice. We now have lines in the profile and plan representing the shape of our fuselage were it made in wood and cut through at the various horizontal planes, buttocks and diagonals. It will have been observed by now where the different lines are not fair, and the worst cases shoiild be rectified first. Having smoothed out a particular horizontal plane in the plan, take a strip of paper, lift off the new spots for each station and transfer back to the body plan. On altering the shape in the body plan, it will be found that several other horizontal planes, buttocks and diagonals may be slightly altered and must be corrected. The whole procedure is gone over again until all lines in each view are " fair." With a little experience one is soon able to tell which line is badly out, and which to alter to avoid a lot of laborious work. Having " faired " the lines, we now mark in the position of all frames or formers with dotted lines. A table of " offsets " can now be made up for either the stations or formers, for issue to the shops. Shell Expansion and Half Block Model Suppose the machine we are working on is a monocoque type, the fuselage being built up of dural plates,
on frames, the writer is of opinion that a half-block model is essential, if wastage of plate is to be avoided. To construct a half-block model there are one or two methods of procedure. Assuming the lines have been drawn to a scale of, say, in. = 1 ft., and a print issued to the model maker. One method of making the model is to trace out the shape of each horizontal plane on tracing paper, place same on a suitable piece of soft wood of the same thickness as the distance apart of these planes and prick the outline through on to the wood. Shape each layer of wood to the marks, and when all are finished glue same together and cramp up until glue has set. Another method is to glue all the slabs of wood together first, and then when set mark off the position of the stations, and having made templates as in Fig. Ill, cut grooves in the block until the template nearly beds down on the block at its respective position. Pare away all surplus wood and carefully rub down with glass paper until templates fit nicely at each station. On completion of model give two or three coats of flat white paint and mount on a baseboard. This baseboard should project beyond the model on all sides by at least 1 in., and should have its bottom edge dead parallel to the datum line. Work on Model On arrival of the model in the D.O. the draughtsman will now proceed to mark off the datum line at each end of the model. The bottom edge of the baseboard being a fixed distance below the datum, this enables him to locate the datum. Also along the bottom edge of the baseboard he will mark off all the station intervals, or, if possible, the position of the frames or formers for preference, as shown on the lines drawing. A special tool is now required like a pkmchette shown in Fig. V to carry out the necessary work on the model. The pencil fixed in the " planchette " is adjusted until it meets the surface of the model, and having brought same in line with a frame position proceed to mark the frame on the model from top to bottom, holding the " planchette " hard against the guide, Fig. IV, and sliding same along, the model, of course, being placed with the baseboard flat on the bench. Having marked in all the frames, girth round a few frames on the lines Drg. that coincide with a " station " by taking a strip of paper and, from the bottom carry the strip round until you reach the intersection of the datum. Having marked the bottom and the datum on thie strip for, say, half-a-dozen frames at intervals in the length of the fuselage, transfer these points to the model, then pin a light pinewood batten through these points and mark in the datum on the model. This, of course, should be a straight line. We are now ready
PLANCHETTE: FOR MARKING LINES ON
MODEL
JANUARY 25,
1934
SUPPLEMENT TO

FIG.YI
FLIGHT
-f
LAP JOINT -SINGLE RlVfS
H/KLF BLOCK MOCtEL
# #
OOUBlt MVETS TREBLE RIVETS
to mark off the plates on the model, which is the sole purpose of making same. Although the gauge of the plates will vary from front to rear, the length from the manufacturers is generally the same. The usual method of plating a fuselage is to '' break joint," i.e., never to have two joints opposite one another, but to have at least two strakes between each joint measuring round the girth. Cases do occur where the butts are arranged all to finish on a convenient frame to enable certain lengths of plate to be utilised. When this is done the strength of fuselage is made up by a considerable number of longitudinal stringers, hence an increase in weight. As this article is not dealing with the pros and cons of design, but the practical side of the work, we will proceed to the next step in the work. Knowing the size and gauge of plates to be used, starting at the middle and working towards the ends, mark in the plates using as long a length as possible. With a strip of paper having the width of the plate marked thereon, place one mark on the top centre line of the model and bend the strip round at each frame, marking the width of the plate on the model. We now have the first line of plating marked which will be the top plate on the finished fuselage. Of course, the joint ma}7 not commence on the centre line, but equally on each side when, of course, the strip of paper will only have half the plate width marked on it. Next, one must decide the width of lap required to join the plates longitudinally, and mark same on our strip before proceeding to mark off the second line of plating. These lines of plating are called " strakes." The first one is " A " strake, and when marking off B, C, D strakes make sure your mark for the lap is inside the strake above. All longitudinal seams must overlap with the sight edge downwards (.nee section, Fig. VI). After having marked off all the width of plates on the frames from top to bottom, procure a pinewood batten and pin same on model through each successive strake and pencil in. On a second strip of paper mark off the length of the plate to be used on " A " strake, and using as long a. length as possible start at the middle with an equal length of plate each side of the middle frame and work fore and aft until you have all the lengths marked on
this strake. When carrying out this operation do not omit to allow for the width of the lap joint at each plate after the first one. The plate in front of each adjacent one should lap over, that is, all laps should point aft. Of course, one may be making all joints and seams butt jointed with a narrow plate behind to make the joint, when the allowance for overlap will IK- omitted. On marking off the lengths of plates on " B " strake start about one-third of the length from the plate above and carry on fore and aft. Start " C " strake in the opposite direction, and work as before. This insures that no two laps are opposite one another foi" at least two strakes. One must arrange this " shift of laps " throughout the whole fuselage and towards the ends it will not be possible to use the total length of plate in, and achieve this end, see Fig. VI at E l and K5. At the fore and aft end the plates become congested if one tries to keep the width constant. To avoid this cut the plate back in length at a convenient position to enable one standard width plate to occupy two strakes. By now it will be obvious that the fuselage has been plated with a minimum of waste, and the haphazard method of trying a plate on the job in the shop and the risk of scrapping has been overcome. Assuming that all the " shell " plates have been marked on the model, collect all information re openings, such as side windows, doors, hatches, etc;., and mark them on the side of the model. A complete half-block model should have all attachments to the " shell " dotted thereon, but the extent of detail will remain at the discretion of the draughtsman and the shop requirements. The primary object of the model is to obtain the exact siae of the plating. When everything has been marked on the model that is required, ink in, after final correction, and check, then give the whole surface a coat of clear varnish to preserve the marks thereon. Shell Expansion The next step is to prepare the " Shell Expansion " drawing for the half-block model, Fig. VII. This drawing is best described by imagining the model to be made of tin and flattened out from the round to a flat
Fig. VIINote, all openings, external fittings, doublings and shell drilling should be shown on this drawing
SUPPLEMENT TO
JANUARY 25, 1934
FLIGHT
surface. The " girth "' becomes flat, but the length remains the same, and is purely a guide to the plating squad in the shop to locate their plates. Proceed to mark off on the paper for this drawing the datum and all the frames or formers to the same scale as the model, then with a strip of paper girth round the frames on the model and mark off all plate edges above and below the datum. Transfer these marks to the " Shell Expansion Drg." and when all the plates are drawn in we have a view showing the position of all plates on one side of the fuselage. Note that any opening, bracket, etc., only occurring on one side of the fuselage can still be indicated on the " Half-Block " model and " Shell Expansion " by labelling same " Stb. only " or " Port only " as the case may he. Having transferred all the plates from the half-block model to the shell expansion and drawn in same, proceed to label all the strakes as Al, A2, Bl, B2, etc., and when finished this drawing should be a complete copy of' the " model " and is used hy the foreman in charge of the plating as a guide for his work. Plate Ordering Last, but not least, the Plate Order List is prepared from the " model." (See Fig. II.) Most of the plates amidships can be scaled off direct, as there is little double curvature, and entered on the order sheet with a J in. extra in length and J in. extra in width. Where a plate has double curvature, as at the nose, place a piece of tracing paper over same and mark the outline of the plate on the paper. When this paper is laid out flat we have the true shape of the flat plate for ordering. This process is called " lifting off.' When these plates come in from the makers or are cut from stock, they are labelled as per list, and when all work is complete thereon they are plated in store, and as they are marked according to the Shell Expansion Drg., the foreman in charge of the fuselage plating knows exactly where each plate goes. It is not suggested here that separate departments should handle the plates, but one squad does all the rolling and shaping, drilling edges for rivets, etc. With a reasonable amount of good progress work beforehand, there should he no waiting one department for the other, and when the plates are required for the final riveting up on fuselage, they should be ready in store. Here, perhaps, we should have mentioned that the " Order List " is prepared first from the model, and delivered to the ordering department, who, presumably, order same without delay, while the rest of the work is proceeding. Incidentally, the material for the frames, etc., is also " lifted off " the model, and a list made out similar to the plate order list. I have purposely refrained from mentioning the frames previously as they may he built-up plates with angles, or " Z " bars, etc., and require a separate drawing for each frame or batch of frames, but it is obvious that any information re the contour of the frames can be obtained from model. In conclusion, the above procedure for lines, plating, etc., is applicable to either a " Monocoque Fuselage " or a flying-boat hull. it HIDUMINIUM R.R.53 B T Under the name " Hiduminium R.R.53 B," High Duty Alloys, Ltd., of Slough, have introduced a new light aluminium alloy. This is the result of many months of research work in the laboratories of the firm, with the object of altering the standard alloys to suit a particular purpose. The new alloy is a slight modification of the well-known Hiduminium R.R.53, which is one of the series of allovs introduced and patented by Rolls-Royce, Ltd. The new alloy is finding wide application for fast-moving levers, treadles and brackets in the textile and electrical industries, and should also
Some parts cast in Hiduminium R.R.53 B. be suitable for the smaller forms of castings in the aircraft industry. The analysis of Hiduminium R.R. 58 B is as follows : Copper ... ... 2.5 per cent. Nickel ... ... 1.5 ,, ,, Magnesium ... 0.8 ,, ,, Iron 1.2 ,, ,, Silicon ... ... 1.2 ,, ,, The physical properties in the various conditions are as follows :
VhM Cast. 3 per cent. Maximum
TnL
in.
T^L
in. Tons
Elongation TJrinell
'
3 6 3 75 103 129
Tons 8 14 As cast ' 12 22 Solution'treated Solution'treated and artificially 23 aged Sand Cast Text Burs 1 per cent. Maximum
XJ 4J
80 As cast 110 Solution treated Solution treated and artificially li) 21 1 120 aged The specific gravity is much th-e same as fc)i '' Hiduminium ' ' R.R. 53 alloy
Tons/ sq. Tons/sq " in. in. Tons Tons 8 12 12 17
.0,
^t.nn.
^nt. naraiie.- ^
Power and throttle curves of De Havilland " Gipsy Six " engine. (See pages 84-86.)
78 h
February 22, 1934
FLIGHT
ENGINEERING SECTION
Edited by C. M. POULSEN
No. 97 ( V e 2 I X ) 9th Year February 22, 1934
CONTENTS
Engine Cowling. By J. D. North, F.R.Ae.S., M.I.Ae.E Ethyl. By F. E. Backs. O.B.E.. F.K.Ae.s.. M.I.A.E.. M.Inst.P.T., M.S.A.E Technical Literature Summaries of Aeronautical Research Committee Report s... ... Page 9 14 16
ENGINE COWLING By J. D. NORTH, F.R.Ae.S., M.I.Ae.E. In FLIGHT of February 8, 1934, we published a summary of the first part of the paper under above title which Mr. J. D. North, Chief Engineer of Boulton & Paul, Ltd., read before the Boyal Aeronautical Society on February 1. A brief report of the discussion was published in FLIGHT last week. Below we publish extracts from the concluding part of Mr. North's very interesting paper. It should be pointed out that we have retained Mr. NorWs numbering of the illustrations, which has resulted in certain cases in gaps in the, numbering.ED. The Townend Ring Inasmuch as my company have proprietary interests in the patents covering the Townend Ring, and have devoted much attention to the development of this device, I am naturally in a position to give more detailed information concerning this particular form of low-drag cowling. I hope that these details will be of general interest. Although the Townend Ring itself is a simple device, the factors which may influence its performance are many and various, and time will permit only a very general consideration of some of the more important of these factors. Ring Sections The section of a Townend Ring is an aerofoil section in so far as it is required to produce a radial outwardly directed lift with its consequent downwash. The effectiveness of a ring for given conditions is determined by the intensity of downwash per unit of circumference. Hence it is an advantage to use a section which develops a high lift coefficient in order that the chord length required may be a minimum. Experience shows that for rings of the usual singlesurfaced (plate) type a camber of about 10 per cent, of the chord length should be used. Double-surfaced sections usually employed for wings, of the same camber on their upper surface, are less effective than the plain plate type, presumably because the mean curvature is
reduced to one midway between that of upper and lower surfaces. There is some evidence that an increase of camber to more than 10 per cent, may be advantageous under certain conditions. The addition to a plate type ring of a bulbous nose similar in form to the leading edge of a moderately thick aerofoil section has been found usually to decrease the drag of the complete ring installation quite appreciably. Such a bulbous nose has been used by Boulton & Paul as an exhaust collector and provides a method of cooling and silencing the exhaust with no increase, and normally a decrease, in total resistance. Fig. 9 shows ring sections which have satisfactorily been used for Town. end Rings. Angle of Ring Chord The best angle between the chord line of the ring section and the thrust axis depends on many factors, such as the exact form of the engine, of the body behind the engine, of the section actually used for the ring itself and the relative fore and aft position of the ring relative to the engine. As a very rough guide,
mr
to
FIG. 9
40
*'
EZ WTW CCHNJ3T fOSt
174 a
SUPPLEMENT TO
FEBRUARY 22,
1934
FLIGHT

Chord Length The chord length required to produce a given degree of constraint on the tendency of the airflow behind the engine to break away from the body depends mainly on the lift coefficient which is developed by the ring section. Sections of the types previously illustrated which have been found satisfactory apparently develop when used as Townend Rings lift coefficients of the order of 0.5 to 0.6, and with these sections a chord length of approximately 0.5 of the engine diameter is found to give the maximum reduction in drag. The flow over the nose of a body engine combination is necessarily curved and change in chord length of a ring alters the effective angle of incidence between the ring and the airflow, therefore the effect of change of chord is not a simple effect. Fig. 11 shows the measured drag of a nacelle and the engine mounted on a wing when fitted with three different rings, (a) is a ring having a chord of approximately 0.33 engine diameter, (b) is a very similar ring with a chord length of 0.52 engine diameter, while (c) has a slightly increased chord 0.525 engine diameter and is fitted with a bulbous nose exhaust collector. The difference between (b) and (c) has little to do with chord length, but that between (o) and (b) indicates the kind of difference which attends on change of chord length.
useful for determining the range within which experiment may usefully be conducted, the lines of the body may be extra-polated forward past the engine to complete a reasonable streamline, and the chord of the ring should then be set at an angle between parallel to a tangent to, and at about 4 deg., converging rearwardly to that tangent, the tangent being taken in a plane corresponding to 50 per cent, of the chord length.
120
no
COST WITH NO TOWN END Rll*5
100
90
Fig. 10 shows the variation in drag of an engine and streamline nacelle fitted with a Townend Ring for various chord angles of the ring itself. The curves marked 1 relate to the ring when the midpoint of the chord lies in the plane of the cylinder centre line. Curves 2 and 3 relate to the same Townend Ring moved forward by successive steps each of about 20 per cent. of the chord length, this chord length in the particular case being about 48 per cent, of the engine diameter. For the position 1 it will be seen that there is a range of chord angle from 2 deg. to 6 deg. over which the drag is practically constant, and that at 8 deg. the drag starts to rise very rapidly. For positions 2 and 3 the minimum drag has increased appreciably and the flat portion of the curve has disappeared, No. 3 showing a sharply marked minimum value of drag, the curve rising steeply on either side of this minimum. The general characteristics shown for the lower set of curves without slipstream are retained by the upper set which represents the conditions with an appropriate airscrew running at conditions corresponding roughly to climbing airspeed. With very few exceptions, it has been found that the presence of slipstream does not affect the relative merits of different ring arrangements. Fore and Aft Position of Ring The curves of Fig. 10 also indicate thn nature of the effect of changing fore and aft position of a Townend Ring relatively to the body. The results relate only to one particular type of ring used on a particular form of body, and even for that case do not extend sufficiently far to prove that position 1 is the best possible. Experience, however, indicates that in nearly every case a ring which is placed with its chord extending equally ahead of and behind the cylinder centre lines will give better results than one placed in any widely different position.
80
FIG.I!
60
50
50
WINS KL
This figure shows also the increase in drag due to the engine nacelle when no ring is fitted, and it will be seen that with no ring this drag increases very rapidly with increasing wing lift. With any of the rings this increase in cost of engine as wing lift increases is very greatly reduced and within the range covered by the figures has completed disappeared for the best of the three rings, i.e., (c). It may be remarked that it is a general characteristic of a good Townend Ring that it maintains its effectiveness over a considerable range of conditions. Polygonal " Rings " Fig. 12 shows two Townend Rings of identical section, chord length, and chord angle, made for use w*ith the same nine-cylinder engine. Tested on a streamline nacelle, the measured drag using the polygonal ring was
11
FEBRUARY 22, 1934
SUPPLEMENT TO
FLIGHT
FIG. 12.
SECTION OHMTUW UNg f T P f t ^ - I
found to be slightly less than that with the circular ring. For a nine-cylinder engine 54 in. in overall diameter, the equivalent full-scale drags were 27 lb. for the polygon and 31 lb. for the circular ring, which is to be compared with 96 lb. with no Townend Ring, at an airspeed of 100 ft./sec. A very large number of tests have now been made in our wind channel giving a direct comparison between the performance of circular and polygonal rings of otherwise identical form and used on identical engine-body combinations. In no such case has the polygonal ring given results inferior to the circular one, and in the majority of cases the polygon has shown a definite advantage. Tesits made with the airscrew running show that slipstream does not adversely affect the superiority of the polygon. Body Shapes So many practical considerations govern the design of body shapes that it is quite impossible to give any hard-and-fast rules as to the shape of body which should be employed with the Townend Ring. Fortunately, however, the performance of the Townend Ring itself appears to be influenced mainly by the form
of the body for a short distance behind the ring only. Experience to date shows that for current type of radial engines the maximum reduction in drag with a Townend Ring can be secured if the body immediately behind the engine has a diameter of about 0.75 of the engine diameter and if the body lines over a distance of about one engine diameter behind the engine plate are reasonably fair and do not diverge or converge with abnormal rapidity. What the body form further from the engine may be may greatly affect the total resistance of the body and engine combination, but will not greatly affect the saving in drag caused by the Townend Ring. Fig. 13 shows three models of engines and nacelles which have been tested. The nacelles themselves are solids of revolution having the outline of the standard 3:1 streamline strut section and differ only in the position of the engine on the bodies relative to the maximum ordinate of the basic streamline. The diameter of this maximum ordinate was 0.78 of the maximum diameter of the engine. Of the three models tested the intermediate is slightly the best, both with and without the Townend Ring, but the differences between
- (mi-
FIG. 13
7.-&Z,'
to
's
j
- '
c H*aii_e SHAPE
174 c
SUPPLEMENT TO
FEBRUARY 22,
1934
FLIGHT

(g) in this figure shows a case where interference between wing and ring was greatly reduced by cutting away a segment of the ring in way of the wing. Generally speaking, cutting of gaps in the ring circumference causes the ring to become almost completely ineffective. If, however, a member is provided which will serve to carry the general ring circulation across the gap, the effect of the interruption becomes unimportant. The case illustrated is one in which the wing serves to bridge the gap.
the three arrangements are not large. This intermediate arrangement is that which, with the polygonal ring, gave a total full-scale drag for the 54-in. diameter radial engine of 27 lb. at 100 ft./sec., which is so far about the minimum resistance which it has been found possible to obtain for a radial engine of these dimensions with any form of Townend Ring. Although the tests relate directly to engines mounted on streamline nacelles in free air, experience shows that if the form of body immediately behind the engine
corresponds reasonably closely to that of any of the three nacelles shown over a length equal to the engine diameter, the drag reduction caused by a given Townend Ring will be of the same order as that which the same ring would cause on the streamline nacelle. Normally for best results the body section over the region above mentioned should be circular. If a polygon ring is used, a polygon body, sides parallel to those of the ring, is as good as, or slightly better than, a circular one. Townend Rings and Interference The effect of the Townend Ring on the drag of an engine is of the type of phenomenon usually described as " interference." Fig. 11 already shown indicates clearly how a Townend Ring may reduce interference between a wing engine installation and the wing itself, and in cases where, as is usual, such wing engine interference is appreciable, a satisfactory Townend Ring will almost invariably greatly reduce the interference drag. The Townend Ring itself is very sensitive to certain types of interference. If the flow over the outer surface of the ring is disturbed, it may be caused to break away and local stalling of the ring section provoked. More than a very limited degree of such local disturbance is found to produce an effect which spreads round the ring circumference and very rapidly destroys the effectiveness of the ring. The most difficult cases of interference with the Townend Ring yet encountered are those in which interference between the ring and a closely approaching wing occur. Fig. 14 shows at (a) and (b) conditions which almost inevitably lead to serious interference of this type and should be avoided; (c) and (d), which differ from (a) and (b) only in a relative vertical displacement of rang and wing so that the leading edge of the wing definitely cuts the ring periphery instead of being nearly tangent to it, are free from this trouble and give satisfactory results. A variant of (a) in which the engine is dropped below the wing and the nacelle is separated therefrom instead of being built on to it may be worse than (a) itself, and is only satisfactory when the engine is dropped far enough to give a large vertical gap between the ring and the leading edge of the wing. The arrangement (e) gives excellent results, but it is important not to move the ring and engine so far back that the leading edge must be mutilated to clear the engine itself.
The total resistance of a ring-cowled installation is very little affected by bodies which are within the wing itself. Circular struts for supporting the ring do not increase the drag as compared to streamline struts. Exhaust collectors within the ring have but a very small effect, which is often unmeasurable, and the effect of fitting inter-cylinder baffles, various types of air intakes, or the like, which do not protrude beyond the ring is invariably small, and usually negligible. Engine Cooling Comparison of a typical Townend Ring with any other form of low drag cowling for use on a similar engine indicates the probability that the ring will interfere less with cooling than will any of its present-day competitors. The necessary development of a lift force by the ring sections and the circulation round the ring which this implies, involves some reduction in the velocity through the ring as compared to that outside it. This reduction in velocity cannot be of any great magnitude and would not be expected to account for any serious effect on cooling. Practical experience ^has shown that the Townend Ring does give cooling superior to that of other available types of cowling for radial engines capable of substantially reducing the drag of those engines. Mr. Fedden has published results which show that satisfactory cooling can be obtained with a Townend Ring on a particular engine and aircraft combination to which the application of a complete cowling of the N.A.C.A. type was impracticable because it caused overheating. There is a considerable fund of experience in America indicating that whereas this complete type of cowling can normally be employed successfully on ungeared engines, it leads frequently to difficulties with cooling when it is applied to geared and particularly to geared and supercharged engines. The restricted frontal opening of the complete N.A.C.A. cowling is in the region mainly affected by airscrew boss shielding, and the area so affected is greatly increased, as has already been pointed out, when a large diameter slow running airscrew is employed. The Townend Ring, with its wider frontal aperture, is less affected in this way. Direct measurements of the air velocity close to the sparking plugs of a nine-cylinder radial engine have been made in flight (Ref. B. and P. tests 2161 and 2173), both with and without Townend Ring, using
174
FEBRUARY 22,
1934
U
TO

pitot heads and hot wire anemometers. The pitot heads show a small rise in velocity of the order of 5 per cent, when the ning was fitted, the hot wire anemometer, on the other hand, showed a reduction of the same order. Many explanations for this discrepancy are possible, the most probable being that the Townend Ring had substantially changed the direction of air flow, to which the hot wire anemometer would be insensitive. The hot wire anemometer being a direct method of measuring cooling and only an indirect measure of air speed, the results indicate a small loss in effective cooling. Fig. 16 shows temperatures measured on the rear face of the cylinder heads of a nine-cylinder radial engine
FLIGHT
Frig. 17 shows cylinder temperatures measured on an engine of the same type during climb, with and without Townend Rings on the same aircraft. The average temperature when the Townend Ring is fitted has obviously increased appreciably. The irregularity of temperature distribution around the engine is more marked both with and without ring than is shown in the test-bed case. Repeated tests under similar conditions, using the same engines and aircraft, have shown that this irregularity, though always present, varied from flight to flight, and it was quite usual to find that a cylinder which had developed an abnormally high temperature on one flight remained abnormally cool on the next. Fig. 17 relates to a type of engine which had had its power output boosted to about the limit of its cooling capacity when used without a Townend Ring, and even in this condition at the relatively slow climbing speed used in this particular case the cooling could not be regarded as completely satisfactory. Fig. 18 shows in a slightly different form the results of similar tests on an engine of generally similar type. In the bottom curves the temperature of No. 3 cylinder with the Townend Ring is definitely unsatisfactory. Tests were accordingly made with inter-cylindeir baffles around cylinder No. 3, which also half encircled cylinders
FtG.I8
300' tso"
running on the test bed with the standard fan and wind tunnel cooling arrangement. Two curves, which are very nearly identical, relate to the temperatures with and without a Townend Ring. The absence of any appreciable difference in temperature under the two conditions may be explained by the artificial cooling conditions. Attention is directed, however, to the irregularity of temperature distribution round the engines, the maximum variation between individual cylinder temperatures being about 50 deg. C.
1
soo_
cruNoea NO.
is
K
7
x
M~~~^
\.
w
- ^ \
FIG. 1 7
Nos. 2 and 4. As the second set of curves show, the temperature of No. 3 cylinder dropped from 285 deg. C. to 225 deg. C. No. l2 cylinder only half baffled showed practically the same temperature as No. 3 with the complete baffling. It may be noted that No. 1, complete unbaffled, dropped in temperature between the two tests by nearly 70 deg. and No. 5, also unbaffled, by 50 deg. C. Following this test all cylinders except No. 6 were fitted with half baffles of the type which had apparently effectively cooled No. 2. On a third test, indicated by the upper curve, the temperature of No. 3 had returned to practically its original high figure, and it will be obvious that no sort of connection can be established between the presence of baffles and cylinder temperatures. Further, the irregularity of temperatures round the engine is very obvious, and is found to a less degree even where cooling is considered satisfactory. Wherever the fitting of Townend Rings has led to actual overheating, investigation indicates that this irregularity of cylinder temperatures becomes very marked indeed. Increase in general temperature marked by such violent irregularities in temperature distribution can obviously scarcely be attributed to any direct effect of the Townend Ring on the effective cooling velocity over the engine, and many explanations of the effect have been considered. Of these, the one which seems to have the best foundation is that the very considerable change
174 e
U
SUPPLEMENT TO FEBRUARY 22, 1934
FLIGHT

as pure aluminium, etc. The sump would be deep enough to prevent any water reaching the joint between dt and the tank, in order to avoid the possibility of electrolytic action. Engine Tests and the Influence of Increasing Concentrations of Lead When considering the duration of engine tests in order to ascertain the effect of leaded fuel upon the engine parts, one is of the opinion that no tests of less than 100 hours' duration are of value. In order to promote rapid engine development on leaded fuels, the knock rating of the finished fuel should be decided upon in the first place, after which a basic petrol chosen, having an initial anti-knock value which demands a fairly large amount of lead in order to attain the required final value. This will ensure that the engine is capable of giving satisfactory operation with any concentration met with in service, even although it may eventually be provided with a fuel, the basis of which only requires a very small amount of lead in order to reach the desired anti-knock value. There are many contentions regarding the influence of increasing lead concentrations on engine condition, and in general the consensus of opinion appears to be that an increase in lead concentration gives the engine parts concerned a harder time by increasing the rate of deposition of the products of combustion. One is not substantially at variance with this view, and has always maintained that tests should be carried out on the lines suggested in the previous paragraphs of this section. It is quite feasible to suppose that an increase in the amount of lead must generally show up in the form of greater rate of deposit build up. However, the following points are put forward as a matter of interest. Firstly, the American view, backed by six or eight years of intensive experiment and use of leaded fuels, is that increasing concentrations of lead tend to increase the rate of attack and deposit build up, which may lead to troubles previously dormant. Secondly, tests have been carried out by the Air Ministry, at the works of the aviation engine firms in this country, over the last two years. The tests, of 100 hours' duration, were made on representative types of engines in service, and the results did not completely bear out American experience. The interesting point about these tests is that valve failure, due to burning, was experienced in some cases and occurred in about 50 to 70 hours of running. Further tests of 100 hours' duration were then made after completely reconditioning the engines concerned, but a fuel having only 1 c.c. of lead per gallon was tried, where previously a " 4 c.c." concentration was employed. The same class of petrol was used as the basis of the fuels, and a similar knock rating to the " 4 c.c." fuel was obtained by the use of added aromatics. However, in a directly comparative test with this and the " 4 c.c." fuel, precisely the same degree of valve failure in practically similar periods was experienced with both fuels. Therefore, from the experience in this country and in Europe, one would say that increasing lead concentrations do not necessarily give rise to trouble or to the same ratio increase of deposits in the engine. With regard to the apparent variation between the results obtained here and in Europe, to those indicated by American experience, a satisfactory explanation might be that the modern American engine has been developed over the same period as that of leaded fuels; consequently a certain amount of technique has been evolved to deal particularly with their use. A great deal more flying, with engines using such fuels, has also been done in America, while little or none has been done in this country, and practically all our leaded
in the direction and general turbulence of the airflow past the engine caused by the ring may disturb air intake and carburetter conditions and lead to variations in mixture strength and/or to irregular gas distribution. Serious cooling difficulties attendant upon the fitting of the Townend Ring rarely arise, except in the rapidlydiminishing number of cases in which the cooling margin of the engine, even without low-drag cowling, is small. The designer of radial engines has realised the importance of modern low-drag cowlings and of providing his engines with cooling capacity which will be adequate when such cowlings are fitted, consequently such difficulties are steadily growing rarer. Where irregular temperature distribution occurs, and there is strong evidence that it occurs to some extent in all air-cooled engines, the cooling which has to be provided is that which will keep the hottest cylinder down to permissible limits; and the analysis of the reasons for such irregular temperatures and methods for their cure should be of the utimost interest to the engine maker himself, since they are one method by which effective cooling can be appreciably increased. (To be concluded.)
ETHYL
BY P. R. BANKS, O.B.E., F . R . A E . S . , M.I.A.E., M.INST.P.T., M.S.A.E.
(Concluded from, p. 4.) General Notes on Engine Operation with Leaded Fuels THE internal appearance of an engine which has run on a fuel containing lead differs somewhat from that usually associated with the more ordinary fuels. The deposit from the use of the former fuel is harder in nature and perhaps more adherent than that of the latter. Its colouration is also different, being white to greyish white on the cooler parts of the combustion chamber and reddish brown on the hotter parts. This is due to the presence of lead bromide. There is, sometimes a yellowish tinge to the deposit, which may be accounted for by some lead sulphate present in the deposit. Where a part, such as an exhaust valve, has been running unduly hot, the deposit is generally exceedingly adherent to the valve head and has a dark " steel " grey appearance. The dye which is present in all leaded fuels is particularly useful for the relatively complicated fuel systems used in aviation engine installations, since it shows up, almost immediately, any leakB which may be present. Some queries have arisen regarding the effect of leaded fuels on ths materials used for aircraft fuel tanks. No trouble has been experienced in the; oase of tanks manufactured from the usual aluminium alloys, but with regard to those particular alloys which contain a large percentage of magnesium, such as Elektron, there seems to be some doubt as to the advisability of employing them for fuel tanks at all. One's personal experience is that corrosion trouble is manifest with high magnesium alloys when water is present in the fuel, whether the latter contains lead or not. If it contains lead, then the corrosion attack appears to be somewhat accelerated. From this one deduces that the presence of water is really the deciding factor, but it is almost impossible to avoid a certain amount of water collecting in fuel tanks. It is suggested, however, that magnesium alloy tanks could be designed with provision for a sump of some material which does not suffer from this corrosion attack, such
174/
FEBRUARY 22, 1934
16
SUPPLEMENT TO

fuel development has been restricted to relatively severe experimental running and type tests which, one submits, are more critical than flight conditions. One is of the opinion that any troubles experienced in America with leaded fuels of high concentration are not so much due to this feature, but rather to the increased " power per litre " of cylinder capacity at which American engines are now running under normal cruising conditions in flight. Knock Testing and Assessing Fuels in Relation to Engine Performance The question of testing fuels for anti-knock value is, admittedly, a subject in itself and hardly comes within the scope of this paper, but it has such direct bearing upon the successful development and satisfactory operation of aviation engines that perhaps little excuse is needed to mention it. The whole essence of knock-testing technique is the ability to correlate the results obtained on the fueltesting unit, with the performance of the fuels in the engine and to be able to assess the knock ratings of the various fuels in their order of merit. It is exeeedingly difficult to arrange a complete set of conditions for the fuel-testing unit which will imitate, accurately, those met with in the engine. Some time ago, the Institution of Petroleum Technologists appointed a sub-committee to formulate a suitable programme in order that experiments could be carried out and the data obtained therefrom used to enable a satisfactory technique to be evolved for the correlation of laboratory knock test results with actual engine performance. Tribute should be paid to the I.P.T., which is the first body to formulate a method of knock testing and correlating aviation fuels, to be accepted nationally. This should, however, only be regarded as a preliminary step. These test results are very clearly and completely described by Mr. Pye, who was chairman of the sub-committee, in a paper read before the World Petroleum Congress last year. The running tests were carried out at the R.A.E., Farnborough, on air- and water-cooled units, and at the engine works of the Bristol Aeroplane Co. Singlecylinder units of representative service engines were used, and in no case was a complete engine employed. Due, according to the report, to the " extreme difficulty of accurately detecting the onset of detonation in a complete aero engine and to the large quantity of expensive sub-standard fuel which would be required." One contends that, outside the expense, and particularly in the case of the air-cooled engine, the question of audible detonation is not necessarily important, but that the temperature effect on the cylinder head, due to the detonative characteristics of the particular fuel used, is more a measure of that fuel's ability to operate satisfactorily in the engine. With these large engines of comparatively high specific power output, a fuel can cause a dangerous rise in the operating temperature of the cylinder head, which on further increase will result in pre-ignition rather than detonation. This is markedly so when the nonknocking fuels are used. This is largely the reason why all the recent fuel tests formulated by the U.S. Army Air Corps specify that the sample tested shall not show a higher reading of the temperature plug than the reference fuel; rather than taking average bouncing pin readings. On would say that one or two fuel tests carried out on a complete engine, particularly of the air-cooled type, heavily thermocoupled at suitable points, would have yielded more valuable information than the singlecylinder tests. A further important point is that the tests at Farnborough were run at varying speeds, whereas general experience would indicate that constant speed is necessary when matching fuels. ..:--'.X-...v...v^.;:- . . " 174
FLIGHT
The report also gives an explanation for the test methods finally adopted as a result of the work done, i.e., C.F.R. Motor Method, modified to use a mixture temperature of 260 deg. F., instead of 300 deg. F., and why the method is less severe than that employed for correlating automobile fuels (Motor Method). The explanation offered is: " that when a read vehicle engine in pulling at low speed on full throttle the conditions ara not only severe by reason of the low speed, but that in many types there is also provision for a large amount of mixture heating which may vary and exceed even the heating provided by a supercharger." A road vehicle engine is certainly severe on the fuel under the conditions described, but it must be remembered that the B.M.E.P. and moan temperatures, when compared with those of an aviation engine, are not so high, and also, the cylinders are not so large with, generally, a greater ratio of surface to volume than the aviation engine. Consequently, audible knock may be severe for the comparatively short time that the vehicle is running under these conditions, but due to the inherent design of the automobile engine it will take a considerable time before detonation becomes great enough to build up excessive heat and/or bring about actual damage, although, admittedly, the performance of the engine, and consequently the vehicle, is impaired. It is not easy to see how these conditions can be more severe than, say, those of an air-cooled engine having large cylinders by comparison and running at full rated load in an aircraft under climbing conditions, where the rate of air flow over the cylinders is at a minimum and as altitude is increased, with a corresponding decrease in air density, the difficulty of dissipating heat from the cylinders is greater. Actually, the operating speed of an automobile engine, under which detonation usually occurs, is not always so low as might at first be supposed, when compared with the normal speed of an aviation engine, although it may be lower in proportion to the maximum speed in the former case. In fact, the C.F.R. road tests showed that maximum knock occurred at road speeds varying from about 15 to 40 in.p.h. corresponding to crankshaft speeds of approximately 900 to 2,500 r.p.m. respectively. The contentions put forward here are not in any way intended as a criticism of the accuracy of the I.P.T. investigation, since the results of the tests have proved their own accuracy and have shown that it is possible to obtain good correlation with a given set of engine conditions. Criticism might be made that the engine conditions specified did not accurately represent those met with in actual service, and in addition only fuels of comparatively moderate knock ratings were tested, whereas, for future engine development, the Air Ministry has already brought out a fuel specification, D.T.D. 230, in which an Octane value of 87 is called for when tested under the modified Motor Method and, also, the use of lead is allowed. None of the fuels in the I.P.T. tests contained lead. This fuel will be used for new types and these engines will undoubtedly produce higher specific power outputs than the previous models, upon which the I.P.T. investigation was made, and will probably give the fuel a harder time in comparison. It remains to be seen therefore, whether the correlation as carried out, applies in this particular case. Some correlation tests on complete engines are already in progress in America, and it will be interesting to study the results and learn the conclusions arrived at, when they are published. The latest fuel specification (No. Y-3557-G) evolved by the U.S. Army Air Corps, calls for a nominal Octane value of 92. A. C.F.R. engine is used as the basis, but is considerably modified to conform to Air Corps requirements. The temperature rise method of assessing the anti-detonation value is retained. The engine speed of
SUPPLEMENT TO
FEBRUARY 22,
1934
FLIGHT

Air Ministry. R. & M. No. 1551. 12 diagrams.) March 22, 1933. (28 pages and
the unit has been increased to 1,200 r.p.m. and a jacket temperature of 165 deg. C. (330 deg. F.) is used. The C.F.R. Committee, however, has tentatively adopted the C.F.R. Motor Method, unmodified, for testing aviation fuels, pending the results of the investigations now being made. The work of the C.F.R. Committee must not be confused with that of the U.S. Army Air Corps, which latter are an entirely separate and governmental body. The C.F.R. Committee represents the oil concerns, also automobile and aviation engine interests. Conclusion It is submitted that a fairly comprehensive exposition of the leaded fuel situation was needed in view of the lack of comprehensive and practical information which was available to the engine manufacturers and operators. It is not too much to say that air supremacy, whether considered from the civil or military standpoint, will eventually go to the nation which develops engines making the fullest use of fuels of really high anti-knock value.
The investigation represents an attempt to reduce the considerable labour invol ved in computing the stresses in the rings of rigid airships. The analysis is developed in terms of the radially-braced ring, but gives the essentials for an unbraced ring. The account is divisible into three sections: Section 1.It is shown that calculations of stresses, etc., in a symmetrical ring, loaded in any possible way In its own plane may be made to depend on the addition of results of calculations for two simple standard -types. The considerable reduction of labour which follows for all calculations save the first is independent of the particular system of stress analysis adopted. Section 2.An approximate solution to reduce the labour connected with the standard types; it is probable that the approximation could be carried further, if desired. Section 3.An illustration of the use of Sections 1 and 2 in the case of an airship ring. The illustration deals \* 1th the problems which arise when wires become slack under load. SUMMARY OF THE PRESENT STATE OF KNOWLEDGE REGARDING SHEET-METAL CONSTRUCTION. By H. L. Cox,
B.A. R. & M. No. 1553. (20 pages.) August 3, 1933. Price Is. net.
Investigation into the strength of constructions in thin sheet metal divides itself naturally into two parts, firstly, consideration of the phenomenon of buckling and the determination of buckling load ; secondly, the investigation of conditions after buckling has commenced. In the majority of aircraft constructions, buckling will normally occur at so low a load that the occurrence of buckling will not in itself be of great importance. Nevertheless there are important exceptions to this statement, of which built-up spars and constructions in corrugated sheet may be cited. In this summary attention is mainly confined to the effect of buckling under shear or under uniaxial compressive loading. The problem of buckling under combined shear and compressive loading has not been considered by any investigator. THE EFFECT OF AILERONS BRISTOL FIGHTER AEROPLANE. ON THE SPINNING OF A By A. V. Stephens, B.A.
TECHNICAL LITERATURE
SUMMARIES OF AERONAUTICAL RESEARCH COMMITTEE REPORTS
These Reports are published by His Majesty's Stationery Office, London, and may be purchased directly from H.M. Stationery Office at the following addresses: Adastral House, Kingsway, W.C.2; 120, George Street, Edinburgh; York Street, Manchester; 1, St. Andrew's Crescent, Cardiff; 15, Donegall Square West, Belfast; or through any Bookseller.
AN APPLICATION OF PRANDTL THEORY TO AN AIRSCREW.
Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1555. (7 pages and 5 diagrams.) June 10, 1933. Price 6d. net.
I t has long been accepted* that ordinary ailerons, although of little value as a control for recovering from spins, may exert a large influence upon the character of the steady spin. Conventional ailerons have been found to maintain their power of producing a rolling moment (chord axes) up to large angles of incidence, but they also tend to induce a yawing couple of the opposite sign. Moreover, it has been shown theoreticallyt that the effects of applying a pure rolling couple against a spin, whether dynamically or otherwise, are not in general such as to accelerate recovery; and also that even a small applied yawing couple may cause a radical change in the nature of the spin. Accordingly, the influence of ailerons is in general governed by the yawing moment (chord axes) due to them. Complete measurements of a series of spine, covering the available range of aileron movement were obtained (a) with the aeroplane loaded normally; (6) with 40 lb. of shot in each wing tip. The effect of applying aileron in either direction during the recovery from a normal spin was also investigated. Ailerons were found to exert a large influence upon the character of the steady spin ; crossing the ailerons rendered the spin flatter, whereas setting them " with " the spin had the opposite effect. The total range of incidence obtainable in this way was from 33 to 60 for the aircraft under normal load. " The Spinning of Aeroplanes." Gates and Bryant, October, 1926. H, & M. 1001. t " Note on Recovery from a Spin." Bryant and Jones, October, 1930. R. & M. 1426.
By C. N. H. Lock, M.A. R. & M. No. 1521. (41 pages and 13 diagrams.) August 30, 1932. Price 2s. 3d. net.
The velocity field of the helical trailing vortices of an airscrew is obtained on the basis of Prandtl's artifice of replacing the helical vortex sheets by plane vortex sheets, thus reducing the problem to one of two-dimensional flow of a perfect fluid. Prandtl's original treatment was confined to the special case in which the trailing vortex sheets are equivalent to rigid lamina: (the airscrew with minimum energy loss); the present paper solves the problem for a general distribution of vorticity with radius by means of a Fourier expansion, the method being analogous to that used by Trefftz for the monoplane aerofoil. In its simplest form the method neglects squares and higher powers of the blade incidence and is subject to errors arising from the replacement of helical vortices by straight vortices ; these limitations are afteiwards removed by various artifices which, although they do not represent a rigorous solution of the problem, should give results which are sufficiently accurate in all practical cases. The method is compared (a) with the result of assuming the number of blades to be infinite (Vortex theory) and (b) with the result of determining the performance of each blade element separately on the assumption that the rest of the wake has the special distribution (minimum energy loss) for which the numerical solution was given bv Goldstein (approximate Goldstein method of R. & M. 1377). Numerical comparisons are confined to the case of an airscrew of constant pitch with " square tipped " blades. A formula for an approximate overall correction, given by Prandtl, to results calculated by method (a) is discussed and it is shown that If the formula iB applied to calculations for an infinite number of blades with square tips (as in Glauert's " Airscrews for high speed aeroplanes "*) the corrected values will apply approximately to two- and four-bladers of normal plan form. Alternative formulae are given which can be used to convert from an infinite number of blades of normal plan form to two- and four-bladers of the same plan form. E. &M. 1342.
ABSTRACT. '" ''*'- ''""
A. Bailey, M.Sc, Assoc.M.Inst.C.E., and W. F. Cope, B.A. Work performed for the Department of Scientific and Industrial Research. R. & M. No. 1560. (11 pages and 12 diagrams.) May 24, 1933.
The present state of our knowledge of heat transmission has been summarised by M. Fishenden and O. A. Saunders,* and a study of the chapter of their book dealing with convective flow in pipes shows that although the results of many ad hoc researches have been published, there have been few systematic investigations of the problem. In particular, little work appears to have been done on pipes of other than circular cross section. At the suggestion Of the Aerodynamics Sub-Committee Of the Aeronautical Research Committee the present investigation was undertaken at the N.P.L. using an apparatus which would enable the friction and heat transmission to a stream of fluid flowing through a pipe of given cross section to be measured simultaneously over the same portion of the pipe. For simplicity, it was decided to start with water as the working fluid and to use drawn copper pipes, as these were easily obtainable in a variety of cross sections. A technique has been worked out and apparatus constructed for the simultaneous measurement of the fluid friction and the heat transfer from either tube to fluid or fluid to tube over the central portion of a copper pipe of circular or rectangular section. Tests have been carried out on pipes having internal dimensions 1 71 cm. diameter, 1 40 cm. square, 1 90 cm. x 0-90 cm., 2-22 cm. x 0-63 cm., and 2-53 cm. X 0-32 cm. rectangles over a range of Reynolds numbers from 3,000 to 25,000. The observations taken of both friction and heat transmission have been reduced by the method of dimensional analysis. The following conclusions have beea drawn : (1) That the hydraulic diameter is the correct length parameter to use in correlating heat-transmission results. (2) That an increase in surface friction does not necessarily lead to a corresponding increase in heat transmission, (3) That ths heat transmission of a narrow rectangular pipe through which water is flowing and being heated is greater than that of a circular pipe under the same circumstances. If the water is being cooled, the heat transmission of the flat pipe is less than that of the circular pipe. It is proposed to extend the range of the observations with water and to carry out similar tests with other fluids. ' The Calculation of Heat Transmission." H.M.S.O. 1932.
HEAT TRANSMISSION THROUGH AND RECTANGULAR PIPES. By
CIRCULAR,
SQUARE
THE FLOW PAST CIRCULAR CYLINDERS AT LOW SPEEDS.
By A. Thorn, D.Sc, Ph.D. R. & M. No. 1539. (2 pages and 3 diagrams.) June, 1932. Price 3d. net.
Abstract only of paper published in Royal Society Proe.
ABSTRACT. THE CONVECTION OF HEAT FROM ISOLATED PLATES AND CYLINDERS IN AN INVISCID STREAM. By N. A. V. Piercy,
D.Sc, and H. F. Winny, Ph.D. R. & M. No. 1540. (2 pages.) September 22, 1932. Price 2d. net.
Abstract only of paper published in the Phil. Mag
THE RADIALLY-BRACED AIRSHIP RING.
Bairfltow, C.B.E., F.R.S.
By Prof. L. Communicated by D.S.R.,
174 h
March 29, 1934
lili
FLIGHT"
ENGINEERING SECTION
No. 98 Year March 39, 1934 CONTENTS Some Developments in Aircraft Construction Engine Cowling Technical Literature Summaries of Aeronautical Research Committee Reports ... Page 17 22
SOME DEVELOPMENTS IN AIRCRAFT CONSTRUCTION THE paper under above title written by Mr. H. J. Pollard, of the Bristol Aeroplane Co., Ltd., and of which he gave a spoken summary at a meeting of the R.Ae.S. on March 15, was of particular interest to aircraft engineers. In FLIGHT last week we published a summary of the paper, and some of the photographs which illustrated it. For the benefit of our more technicallyminded readers we give below the appendices in which Mr. Pollard gives details of various formulae, etc., from certain publications to which he referred, but which are not readily available to British aircraft engineers. Before giving Mr. Pollard's appendices, there is a section of his paper which had to be dealt with somewhat sketchily in FLIGHT last week, but which should be mentioned here. That is the section on multi-spar wings. Concerning the problems of economic use of
materials in monoplane and biplane wings, Mr. Pollard said : " As to structure weights all that can be said at present is that when we have chosen the most favourable material and placed it most favourably for giving strength and stiffness, then we have done all that can be done to obtain the minimum possible weight. The principles I have outlined certainly show the way. An idea of what is to be gained as regards spar weights is shown in Fig. 11, the two aircraft concerned being very nearly the same weight and designed to the same load factors. The variations in sectional area along the length of a monoplane front spar are shown in curve o, while the stresses along the boom for one condition of flight are shown in curve b. Curve c shows the variation in stress along the length of the top front spar of a biplane wing, while curve d gives the average sectional area of the front top and bottom wing spars taken together as a single spar. The variation of sectional area of this virtual combination of spars cannot be shown conveniently in the diagram, but the variations in area of the top front spar in regions of high stress can be clearly distinguished at points z, z, etc., on the stress curve b. Due to the fixity at the point p, where the centre section and outer wing spars join, together
Fig 11. Sectional area and stress curves for monoplane and biplane spars, plotted against span, a, area of monoplane front spar. b. stress in monoplane front spar, c, stress in biplane top front spar, d, average area of all spar sections in front truss of biplane. 306a D2
IS
SfJPPLRUZNT TO MARCH 29, 1934
PLIGHT Fig. 14. The mechanical properties of 4S were given in a table last week. The relationship between fatigue shear stress and endurance for various materials is shown herewith. The consistency of the 4S alloy is most notable.

N
OOOC
2 j
i
i I I & *
z
\ Vs.
=1=
.
KaOO
*o
to
IIXIO
with the lamination, the stress distribution shown is good for a single bay biplane wing spar, but even if improvement could be obtained by modification to the detail design, the construction of a biplane spar member having the uniformity of stress indicated in curve b is not a feasible proposition. " Only in really large biplane constructions would such refinements in design be possible, whereas no difficulty exists in getting the uniformity of stress shown in curve b, in monoplane constructions as described. At the moment we cannot proceed further than a simple monoplane wing; the word simple is perhaps a misnomer here, the calculations may, in a casual manner, be regarded as simple in comparison with what is required for a biplane structure. In the case of the monoplane, however, flutter and allied investigations become ef prime importance and the additional calculations connected therewith make the matter more involved and complicated than it might at first sight appear. " Another form of wing construction in which the whole skin or a large part of it is reinforced against considerable longitudinal forces is worth attention; in this the bending is resisted by a reinforcing combination and not by the booms of separate and distinct spars. Corrugated internal plate attached to a plain outer skin would appear to be the best arrangement, graduation in area against the variation in load could well be executed through the use of lamination in the flat skin.
/
There are certain manufacturing difficulties, however, which so far have precluded an experimental investigation of this construction. The main difficulty, which has been excessive riveting, is, however, likely to be overcome in the near future." Appendix I Critical Stresses of Thin Curved and Flat Plates The formula derived by Redshaw for the critical buckling stress for a curved panel axially loaded and simply supported at axial and circumferential edges is p = {E/6 (1 - a2)} {V [12 (1 - <r2) <7r2 + ir^/fc*] + */&} ... (1) a = Poisson's ratio, b = length of arc, the other symbols having the meaning allocated in other parts of the paper. If (bit)2 is large is comparison with rjt the formula reduces to that obtained for the complete tube (see references in table below) p = V {1/3(1 - o2)} E(tjr) ' (2) so that if the angle subtended by the panel is fairly large the critical stress is dependent only upon the curvature and thickness and not upon the length b. If the angle is small, the above formula should be amended to p = V (1/3(1 - o*)} E(/r)Fx" (3)
/
/
= i
\
^y
g , a S
- ,
,
1
fc ^
T fiULCC
k " >" - T V I t C K H I AT!O
"
<
"*
's
Fig. 1, Appendix I : Curves showing multiplying factors to be used in connection with the critical stresses for curved panels in compression. Based on R. & M.1565.
306
MARCH 29, 1934
19
SUPPLEMENT TO

where F 1 = 1 + ~D(tfd) D = 7^/8^/ {12(1 - <?)) = 0-368 d = r ^ / 8 = " the bulge "'
FLIGHT
<j> = angle subtended by are of panel. If (fc/f) is small in comparison with rjt the formula reduces to that for a flat panel p=[w2/{3(l-a')}]E(/6)* (4) For slightly curved surfaces equation (4) must be modified
2
to
where F 2 = [1 + y1 {1 -+- ""* W)2}]/2. For slightly curved panels it is advisable to use formula (3), although the critical stress will be the same if formula (5) is used. The graph shows the relationship between the multiplying factors F x and F 2 and the djt ratios. It is stated in the text of the paper that the theoretical critical stress given by formula (3) above was in many cases 80-90 per cent, of the collapsing load of some actual tests ; in this connection, it must be remembered that the critical stress is the stress at which instability will occur, but that it is possible for a test specimen to take additional loads after the critical stress has been reached. There is generally a discrepancy between the theoretical and actual critical stresses for most cases of instability; this is mainly due to the difficulty in reproducing the exact theoretical boundary conditions on the specimens. Another cause which is often overlooked is that in most cases the formulae are only valid providing the limit of proportionality
Fig. 1, Appendix II
Method of Test, in shear.
Square sheets
is not exceeded. The question of plastic instability is of importance in certain phases of structural design and has, as yet, received little consideration.
Fig. 2, Appendix II : Sequence of sheets tested. 306 c
Types of failure.
SUPPLEMENT TO
FLIGHT
MARCH 29,
1934

The sudden fall in stiffness indicated at points x, x, etc., on the curves cannot be associated with any observable phenomenon on the panel while being loaded, e.g., appearance of tension waves or their permanence. From the aspect of restoring the greatest percentage of the original ultimate strength, the most effective type of stiffening weight for weight is laminated plates. As might be expected, the use of sectioned stiffeners (panel five, Fig. 2) results in greater stiffness over the first range of load, and this type of stiffening also gave the greatest elastic strength. The amount of elastic strength for these panels is not apparent from the curves shown in Fig. 3. This is an extremely important design consideration, the determination of the true elastic strength necessitating the use of elaborate strain measuring apparatus. Other forms of stiffening are being tested, and subsequently the investigation is to be extended to holes of various diameters and forms, and then to the case of lightened plates under combined end thrust and shear force. Appendix III Theory of Webs of Girders The following notes and formulae should be read in conjunction with the relevant part of the paper : / = Tension in the web in stress units. H, = Load in tension boom in force units. H r = Load in compression boom in force units. a = Angle of waves to horizontal. V = Load in vertical members in force units. Other symbols shown in Fig. 1. / = 2P/M sin 2a = 2Y/ht (for a = 45) (1) H ( , r = Px/k - (P/2) cot a = Vxjh - ()P (for a=45) (2) V = -F(d/h) tan a = -V(djh) (for a = 45) (3) I t is stated (Ref. 10, page 1) that when the load acts in one direction only or to a much greater amount in one than in the opposite direction, it pays to put the stiffeners at about 30 to the vertical; it is proved that the weight of the sheet web is lowered 15 per cent, and the stiffness is raised 55 per cent, by such means. The formulae for these cases follow, although it frequently happens that the external forces act for different cases in opposite directions, the total force in one direction not being greatly different from the total force in the opposite direction, thus the angle /S (Fig. 2) for general use should be taken as 90. The following formulae apply, however, for any angle , and cover the case of the stiffeners or struts (the words are used interchangeably though stiffener is probably the more appropriate) secured to the web. / = (2H/ht) {l/sin2a(l - t a n a cot j8)} (4) H T C = Mfh - S/2 (cot a + cot p) (5) V 1 = - [(Si+ SR)/2] (djh) (tan a/sin /S) {1/(1 - tan a cot /3)} Pw/sin ,8 (6) V 2 = - l ( S t + SE)/2](d/7i) (tan a/sin 0) {1/(1 - tan a cot )} (7) S = M = SL = SR = Pw=
Q-O
In addition to the foregoing case, the table below shows, with references, the problems relating to critical instability that are known to me to have been solved. Reference (1) Plane panels shear KE(/b)2 17 (2) Plane panels compression ... KE(t/b)2 18 (3) Curved panels shear KE(/r 19 (4) Tubes, torsion KEfi/r)1-35 (Empiric) 20 (5) Tubes, compression KE</V 21 (6) Segment of tube compression KEtf/r ... R. &M.1565 (7) Corrugated panels shear ... KE(</6)2 22 (8) Corrugated panels compression KE(/fe)2 23 Appendix II Effect of Holes on Strength of Panel Bracing The necessity for the investigation of the case of lightened and reinforced plates subjected to shear forces is, at present, of more urgency than the case of similar plates subjected to end load. Unlike the former case, an approximate estimate of the strength of the latter can be made. The purpose of this investigation is to find by how much strength and stiffness of panels in shear are reduced by cutting hcles in them, and to find what forms of stiffening will restore the maximum elastic strength and stiffness for a minimum expenditure of weight. The first part of the investigation was confined to square panels with circular holes cut in them, and the method of test is as shown in Fig. 1. A square panel is secured to rigid bars freely pivoted at their ends. The pin centres are 22 in. apart, and the material is Alclad 0-022-in. thick. Panels tested to date are as follows : (1) Plane sheet. (2) Sheet with 6-in. dia. hole (unbeaded edge).
(^)
,, ,,
"-in.
,,
,,
,,
,,
,, (weight of beading equal to weight of metal removed for hole). (4) ,, ,, ,, ,, ,, vertical stiffeners. (5) ,, ,, ,, ,, ,, vertical and horizontal stiffeners. (6) ,, ,, ,, two laminated rings each side. (7) ,, ,, ,, ,, four laminated rings each side. All the panels except No. 7 are shown in Fig. 2. The results shown plotted against shear load in the graph, Fig. 3, are deflections measured at and parallel to the direction of loading. The investigation has not proceeded far enough to allow of the drawing of final conclusions. Two matters, however, arc worthy of comment.
(3)
,, 9-in.
,,
Mt*.
! /
\t
o
Total shear force at section considered. Moment at section considered. Shear to left of the strut under consideration. Shear to right of the strut under consideration. External load at the strut under consideration.
OM>
o-ao
9
QCPLieTlON IN*
Fig. 3 : Graphs showing relation between load and relative displacement of scales shown in Fig. 1, App. II. No. 1, complete panel. No. 2, panel 6 in. hole. No. 2a, panel 9 in. hole. No. 3, as 2a, edge stiffened. No. 4, as 2a, vertical stiffeners. No. 5, as 2a, vertical and horizontal stiffeners. No. 6, as 2a, laminated rings.
Other symbols have the meaning previously allocated or as shown in Fig. 2. Providing the spacing of the struts is not outside the limits one-half to one-sixth of the beam depth, the angle a is approximately equal to /3/2. Since the angle will, in general, be 90, the formulae reduce to : / = 2S/ht (8) HTC= M / A - S / 2 (9) V B = - [ ( S t + S)/2] (d/h) Pw/2 (10) P H = the external force applied at the strut V,,.
MARCH 29.1934
21
SUPPLEMENT TO

In the relevant references, formulae covering the case of beams having struts flexible in the plane of the web sheets are quoted. In general, however, it is preferable to secure the struts to the sheet, thus securing the greatest rigidity, the proportionate increase in the strength of the struts obtained by this means is indicated below (Fig. 6 and equation 18) while the whole stiffness of the girder is raised by such fixing ; the formulae for flexible struts are therefore not given. In addition to the stresses arising from the force calculated from equations 1, 2 and 3, bending stresses are also set up in the flanges due to the pull of the web sheet on these parts.
FLIGHT
The stress at the point A is given as (Ref. 10, page 5)

/ A = / , , . ( V / V ) = /1/(1 - cot a tan 8) (15)
hL,(KV) //( cotatan 8)s where/,,, = (Sw/kxt) 1/sin a cos a
(16) (17)
In the above, the boom angles 8 and y are assumed equal. The strength of the vertical secured to the skin is derived by means of Fig. 6. The elastic support given to the verticals, results in what is, in effect, a reduction of the slenderness ratios or a virtual reduction of lengths of those members
Vk I.
c.
\1\ \
z
e
to >*
Appendix H I : Figs 1 to 6. The bending moment is a maximum at the points of attachment of the web struts and is M F = SdÛh . tan a (11) If the resistance of the flanges to bending is small then the pull of the central parts of the web becomes ineffective (Fig. 3), it is then necessary to evaluate the expression wd= l-25isin ^ [l/(lt+Ic)h] (12) where 1/ and I,, are the moments of inertia of the tension and compression flanges about their own centroidal axes, and from Fig. 5 to obtain a value for the constant CV Then a new bending moment M*F= CiMp can be derived (13) In the case of flexible flanges, the equation (1) gives only the average stress in the web sheet, the max. stress is
/max = /average X 1/^2
and the virtual length I1 is given by =C3Z
(18)
C3 being obtained from Fig. 6. Using this effective length and the loads V or V the strength of the member as a strut can be obtained from the " strut curve," allowance being made for the " effective width " of attached material.
17 P.B.S., Vot. 105. On the stability under shearing forces of a flat elastic strip. B . V. Southwell. 18 Uuckling of thin plates in compression. R. & M. No. 1554. H. L. Cox. 19 Strength in shear of t h i n curved sheets of Alclad. N.A.C.A. Technical Note No. 343. O. M. Smith. 20 Strength tests of thin-walled duralumin cylinders in torsion. Technical Note 427. E. E . Lundquist. 21 P.E.S., Vol. 121. The strength of t u b u l a r s t r u t s . A. Robertson. 22 Z.F.M., 28/9/29. The buckling of corrugated sheet in shear. S. Bergmann and H . Reissner. 25 D.V.i,., 6/3/31. Buckling tests on panels of corrugated sheet. E . Seydel.
the value of C2 also being obtained from Fig. 5. In most wing constructions the spars taper, thus part of the shear is taken on the flanges direct, the amount taken by the web being the difference (Siu) of the total shear and the vertical components of the flange forces (Fig. 4). Wagner gives this case special treatment as indicated in Ref. 10, page 1, et seq. The main fact established is that the web stress vanes over the depth of the girder. At the same time the stress is constant along the length of any wave, thus the stress at the point A on section AB (Fig. 4) is the same as at A1 and the stress at point B is the same as at B 1 .
PIETZKER, NOT PESCA In the account of the discussion of Mr. Pollard's paper published in FLIGHT last week, Mr. Langley was made to refer to Pesca's theory. Some of our readers may have been puzzled by this unfamiliar name. The author to whom Mr. Langley referred was Henr Pietzker. Reference to Pietzker's theory is made in A.It.C It. & M. No. 1553, but there is a misprint in the reference. The analysis of Sir John Biles' experiments on H.M.S. Wolf in the light of Pietzker's theory was published in the Proc. Inst. Naval Architects in 1925, and not, as printed in R. & M. 1553, in 1905. 306 e
SUPPLEMENT TO
MARC H 29, 1934
FLIGHT

ENGINE COWLING estimated drag of the unringed machine, no check on the real drag being available from a full-scale test with the ring omitted, the maximum speed obtained has frequently fallen short of the designer's hopes, but this characteristic is not confined to aircraft with Townend Rings, and should not be taken as evidence of their failure to produce, full scale, the saving in drag shown by model tests. In a figure shown earlier, the model results of varying the chord angle of incidence of the Townend Ring were shown. In order to obtain if possible full-scale confirmation of these results, a variable angle Townend Rang, precisely similar to the model, was tested full scale.
By J. D. NORTH, F.R.Ae.S., M.I.Ae.E. (Concluded from page 14) Townend Rings on Pusher Engines A Townend Ring will effectively reduce the resistance caused by an engine fitted at the rear of an aircraft body. For use under these conditions the ring chord angle requires to converge fairly rapidly to the rear, whereas for the more usual type of installation the convergence is almost invariably in the opposite sense. 4 ring used on such an installation is subject to a largft down-wind force, whereas on the normal installation there is a large up-wind force. A reduction in drag in this case must be associated with large increases in the pressure on the rear end of the body within the ring, but the essential characteristic of the Townend Ringthat of producing a downwash which prevents the flow from breaking awayremains unaltered. In tandem installations Townend Rings have been used on both front and rear engines. It appears that in these cases the reduction in drag which is possible' is mainly that due to the front ring. In certain cases the use of a Townend Ring on the front engine of a tandem pair is stated to have very considerably improved tl.ie cooling of the rear enginea point of very considerable importance. In view of the effect of the ring in preventing a breakaway of flow behind the engine, this effect is not altogether surprising. Model Tests and Full-Scale Results on TownendjRings The Townend Ring owes its origin to investigations carried on in the wind channel with small-scale models. There is inevitably some doubt as to the direct applicability of the results of such tests to the prediction of performance for the full-scale aeroplane. Full-scale tests of Townend Rings have now been made in sufficient numbers to place it beyond doubt that the effect shown by small-scale models also occur in the fullsize aeroplane, and there is enough evidence to show that the magnitude of the saving in resistance shown by model tests is of the same order as that indicated by full-scale performance tests. Models used for Townend Ring investigations require to be a fairly large scale. Our own experience indicates that a scale of one-fifth with a wind channel speed of 60 ft./sec. is sufficient for most purposes, but that it is dangerous to go to a much smaller scale. I t is very common practice to represent the engine on small-scale models by a conventionalised dummy engine which is arranged to give the known or assumed drag of the real engine. Such conventionalised models cannot safely be used with a Townend Ring on any scale, because the change in drag caused by a Townend Ring depends on the exact pattern of the air flow caused by the engine, and not by the engine's absolute resistance. The representation of engine installations by conventionalised models is usually extended to cover cowling details, and air passages through cowlings are either not represented at all or are replaced by a few passages drilled through the body block. Tests which hare been carried out comparing the resistance of a representative model of a complete engine with the normal type of cowling accurately represented, invariably shows very much higher drags for the normally-installed engine than are given by the usual t3pe of conventional representation. Many comparisons between the results of model tests made by my firm and of full-scale tests showing the effect of Townend Rings are available. Where the fullscale tests show directly the change in performance due to the fitting of the Townend Ring, the correspondence between model result and full-scale test has generally been very satisfactory. Where estimates of the effect of fitting Townend Rings have had to be made from the
Fig. 20 shows the estimated variation in performance with ring angle, and that measured full scale. I t should be noted that the full-scale results plotted are the mean of six speed readings for each ring chord angle. The total estimated change in speed between no Townend Ring and Townend Ring at its most effective setting is 15.5 m.p.h. The variation between the various measured full-scale speeds at each individual ring angle vary between 2 and 7 m.p.h. I do not regard these particular tests as showing any abnormal degree of variability. Total Power Plant Drag with the Townend Ring It may be interesting to compare estimates of the drag necessary for cooling with results which have actually been obtained with the Townend Ring. The estimate made by Pye, which has been earlier mentioned, that at 150 m.p.h.. about 1.5 per cent, of the total engine output must necessarily be used in providing engine cooling when the temperature difference available for heat transfer is 180 deg. C , relates only to ideal conditions which we can scarcely hope to approach in the air-cooled engine, even if the suggestion which has once been made of using streamline sections instead of circular ones for engine cylinders were found to be practicable. McKinnon Wood, in a paper to which reference has also been made, has suggested a method of cooling which is certainly practicable, involving the enclosure of the cylinders in ducts which confine the cooling air flow to a path closely following the cylinder contour and embodying a fan so that the cooling is not dependent on the air speed of the aircraft. The use of the fan is indicated as preferable on account of the claimed selfregulating characteristics of this cooling system, and if aircraft of much higher speeds than are at present common are to be attained with engines having characteristics not greatly different from those of existing
306/
MARCH 29,
1934
23
SUPPLEMENT TO

type*, fan cooling may become necessary and does not appear to be inconsistent with the use of ring cowline of the Townend type. Assuming a mean temperature difference of 250 deg. C. between cylinders and cooling air, McKinnon Wood estimates that, allowing for fan losses, 4 pea- cent, of the b.h.p. will be used in providing cylinder cooling alone. A mean temperature of 250 deg. C. between cylinders and air is certainly not available in the majority of engines to-day in service. The maximum permitted temperature at a position on the cylinder heads, which is certainly above the mean temperature of the cylinders as a whole, is normally between 215 deg. and 235 deg. C. For a mean temperature difference of 200 deg. C, McKinnon Wood gives the power absorbed in cooling cylinders as 10 per cent, of the b.h.p. Even 200 deg. C. is almost certainly higher than the mean difference which can be permitted with engines of existing type, but we may take this figure of 10 per cent, as representing about the figure which might be achieved in practice with a cooling system of the type suggested. The best result within my knowledge which has so far been obtained, using a Townend Ring, is that given by 54-dn. diameter nine-cylinder radial engine mounted on a streamline nacelle of 42-in. maximum diameter, fitted with a polygonal Townend Ring, and having a total drag at 100 ft./sec. of 27 lb. This result has already been referred to (Fig. 12). With an engine developing the maximum b.h.p. of 480 at 4,000 ft., the speed of the aircraft fitted with this installation was 140 m.p.h. true (132 i.a.s. at 4,000 ft.), and the power absorbed by a resistance of 27 lb. at 100 ft./sec. at this speed is 43.5 h.p., or 9 per cent, of the engine b.h.p. When an engine of the same overall dimensions and type, but supercharged to give 600 h.p. at 5,500 ft., was fitted in place of the lower-powered engine, the speed of the aircraft increased to 160 m.p.h. true (148 m.p.h. indicated at 5,500 ft.), the engine installation drag being unaltered, and the power absorbed in overcoming engine drag became 54 h.p.still 9 per cent, of the total. Allowing for airscrew efficiency, the drag of the engine installation in this case at top speed is about 11.5 per cent, of the total drag of the aircraft. I understand that Messrs. Armstrong Siddeley Motors, Ltd., have measured a total drag of 35 lb. at 100 ft./ sec. for the fuselage, engine and Townend Ring of a military machine with open cockpit, wind screen and pilot. The engine in this case was a " J a g u a r " capable of developing about 500 h.p. at sea level. This is 8 lb. in excess of the nacelle figure above quoted, and it is certainly not to be expected that the resistance of the fuselage alone, engine removed and a faired nose substituted, could reach so small a figure as this. The drag to be attributed to the engine in this case must therefore have been less than in the examples which I have considered.
FLIGHT
it
Igo
'*O
-^ -I I*A s ' U / \ u \* * .I 7 /
>
v>
2e
752
& It" \ 17a \
*4f> /SO
'to
'70
/3O
/Bo
SLOO 2/0
Sio
\tATCR TEMP T. Fig. 1 : Professor Gibson's Experiment. A practical method has been developed for increasing the capacity of a cooling fluid for heat dissipation by using the latent heat of vaporisation. The greatest difficulty to be overcome on any type of I.C. engine is to keep the working temperature level. Temperatures should be kept high, but within the limits of effective lubrication. The most desirable condition is a uniform high temperature, in all parts of the engine. That is to say, an important feature of the cooling system is to keep the engine hot. In an aircraft installation, fitted with evaporative cooling, the control of temperature, external to the system, is of course impossible, due to varying conditions of flight, but some form of control over the cooling system fitted is necessary, and the method adopted will be described later. In this country the cooling system is vented to the atmosphere, but in America they are using successfully an unvented system, i.e., a pressure gauge is fitted, set to 5 lb. per sq. in. Before describing the layout of an evaporative cooled system it will be interesting to note the results of certain experiments carried out by Pirof. Gibson in 1910 and reported to the Institute of Engineers and Shipbuilders. The tests consisted of measuring the surface temperature of an iron vessel in which water was heated by a gas flame. The water was vigorously stirred during the experiment, and it was found that the temperature of the surface actually decreased as boiling point was reached. Fig. 1 shows graphically, on a base of water temperatures, the results he obtained. As the temperature of the water increased, that of the vessel increased
EVAPORATIVE COOLING
By R. HALEY* EVAPORATIVE cooling in aircraft engines has undoubtedly made great progress in the past five years, and the attempt to improve the efficiency of the air-cooled engines by the addition of the Townend Ring and others has no doubt hastened the efforts of the watercooled engine manufacturers to improve their system of cooling.
Mr. Haley is on the Technical Staff of the Gloster Aircraft Co., Ltd.
Fig. 2 : Diagram of the Antoinette cooling system.
306 g
24
SUPPLEMENT TO MARCH 29, 1934
FLIGHT

ninth web. Torsiograph tests revealed that torsional vibration was an important contributory cause of these failures and yielded data for an in vestigation of damping. The torsiograph investigations are described in this report in conjunction with frequency calculations and consideration of minor criticals. Also, a general analytical method of investigating damping in multi-crank engines is set out and applied to the Liberty engine to obtain damping factors for use in assessing the severity of important critical? in proposed designs for new engines. The crankshaft-airscrew system has been reduced to an equivalent dvnamical system and the natural frequency of single node torsional vibration calculated. The relative importance of the several orders of vibration, for the same forcing torque and damping, has been estimated, and the effects on the magnitude of the vibrations that would result from increasing the Vee angle from 45 deg. to 60 deg. have been examined. Using the B.A.E. torsiograph, series of torsiograms have been obtained from a Liberty engine driving an airscrew at various speeds up to 2,000 r.p.m. The actual torque variations have been determined from static calibration of the torsiograph as fitted. The specific conclusions relate mainly to the Liberty engine, but they serve to indicate the characteristics of similar engines. The most important torsional vibration in the operating speed range occurs at 1,300 r.p.m. and is of 4J order with a single node near the airscrew. Torsional vibration can be considered as producing additional stress variations which increase in range towards the airscrew end of the shaft, where the single node occurs. Thus, failure at the fifth crank from the free end is consistent with pronounced but not extreme single-node torsional vibration. The severity of any critical is greatly dependent upon the Vee angle. For example, if this angle were 60 deg. (instead of 45 deg.), the forcing harmonic of the 4 order critical would be 28 per cent, less and that of the 6th order critical would be 41 per cent, greater. With an 80 deg. Vee the 4 order critical would be zero, assuming uniform cylinder performance. The damping appears to be proportional to the third power of the stress amplitude instead of to the second power as for viscous friction damping. The total mechanical hysteresis in the crankshaft material, as derived from data given by Dorey for pure stress reversal (Reference 10), accounts for only a small fraction of the damping present. Aerodynamic damping for a rigid airscrew is readily calculated and in this instance it is negligibly small. METHODS OF VISUALISING A I R FLOW WITH OBSERVATIONS ON SEVERAL AEROFOILS IN THE W I N D TUNNEL. By
Fig. 3 : Diagrammatic representation of present British system also. This continued only until the water had reached 180 deg. F . ; after that there was a gradual decrease in the temperature of the vessel. There was, as boiling poant was approached, a corresponding increase in the rate of transmission of heat. This indicates that one need not fear overheating of an I.C. engine when using the evaporative system of cooling. Fig 2 shows a diagram of the early Antoinette cooling system, designed by the well-known French designer Levavasseur. Fig. 3 shows diagrammatdcally the present system of evaporative cooling in this country. The steam from the cylinder jacket passes into a " header tank," wherein the moisture from the steam is separated and adds to the water already in the tank. The dry steam then passes up to the wing condensers, the condensate being returned to the cylinder jacket via the circulating pump. Due to the fact that the condensers could not cope with the varying rise and fall of temperature under varying conditions of flight, it was found necessary to have some means of controlling the amount of cooling within reasonable limits. To do this a honeycomb radiator was mounted under the fuselage and a mechanical means of raising and lowering same by the pilot was installed. Readers will be familiar with a typical example in the Hawker " H a r t " published in FLIGHT. The condensate and uncondensed steam is taken from the wing condensers and passes through the bottom radiator, where it is finally condensed and cooled before returning to the cylinder water jacket.
1
K. W. Clark, B.Sc., D.I.C. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1552. (10 pages and 22 diagrams.) May 8, 1933. Price Is. net.
During tests and research work carried out in the wind tunnels it is often desirable to be able to see the air flow over the body, and to know its direction and whether it is turbulent or not. The following is a brief discussion of some of the methods in use, considered primarily as an aid to general work at the higher speeds. They include (1) attaching fine tufts of silk to the surface ; (2) the use of a pencil of dense smoke; (3) heating the air stream locally by electric sparks or an electrically-heated fine wire ; and (4) changing the colour of the surface in the wake of the flow. Tufts provide most information and are simplest and quickest to use. The use is for flow near the surface and is complementary to smoke or heated air methods of visualising the flow above the surface. The observations show that whereas the initial stages of the breakdown in flow and the final stalled condition show marked similarities for all the aerofoils there is an intermediate stage in which the differences between the aerofoils are very noticeable. THE BUCKLING OF T H I N PLATES IN COMPRESSION. By
H. L. Cox, B.A. Work performed for the Aeronautical Research Committee. R. & M. No. 1554. (21 pages and 8 diagrams.) August, 1933. Price Is. 3d. net.
The strength of constructions in thin sheet metal that are now being used to a large extent by aircraft manufacturers, is often very considerably affected by instability of parts of the structure. The present paper describes an investigation by both theoretical and experimental methods into the influence of buckling on the strength of rectangular plane panels of thin sheet metal subjected to compression parallel to one pair of edges. Theoretical analysis of the behaviour of panels both before and after buckling is developed and the results are compared with the results of tests on panels with two different types of edge support. The results of the experiments carried out to check the theoretical conclusions provide reasonable confirmation of the relation between stress and strain as determined theoretically. The collapsing load of a panel of width 2d, thickness 2A and height greater than id is shown theoretically to be equal to the crushing load ofa similar panel of width 2rf'where d' = LA + M<2, the values of L and M depending upon the crushing strength of the material and upon the edge fixing conditions, and the term LA being normally considerably greater than the term Md. The collapsing loads recorded in actual tests are shown to be in reasonably good agreement with this theoretical formula.
ABSTRACT.
(To be
continued.)
SUMMARIES OF AERONAUTICAL RESEARCH COMMITTEE REPORTS These Reports are published by His Majesty's Stationery Office, London, and may be purchased directly from H.M. Stationery Office a t the following addresses: Adastral House, Kingsway, W.C.2; 120, George Street, Edinburgh; York Street, Manchester; 1, St. Andrew's Crescent, Cardiff; 15, Donegall Square West. Belfast; or through any Bookseller.
TORSIONAL RESONANCE CHARACTERISTICS OF A TWELVEOYLINDER VEE AERO ENGINE. By B. C. Carter,
A SURVEY or THE A I R CURRENTS IN THE BAY OF GIBRALTAR IN 1929-30. By J . H. Field, C.S.I., M.A.,
and R. Warden, Ph.D., M.Eng. R. & M. No. 1563. (3 pages and 1 diagram.) November, 1933. Price 6d. net.
Occidents to naval aircraft in lee of the Rock of Gibraltar led to a survey in 1929-30 of the winds causing them, and the survey included a preliminary examination with a model of the Bock on the scale 1/5,000 in a wind tunnel at the N.P.L., and full-scale work in Gibraltar Bay with pilot balloon* and theodolites, and with kites carrying a new form of recording instrument. It was found that the measurements with the model closely forecast what occurred in nature at Gibraltar in regard to wind directions and the distribution of vortices and vertical currents. The inference follows that model measurements would in future cases be a good preliminary, when the suitability of a proposed aerodrome or sea base is to be considered. The present paper has been published in full as No. 59 of the Geophysical Memoirs of the Meteorological Office.
F.R.Ae.S., M.I.Mech.E., and N. S. Muir, B.Sc. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1304. (39 pages and 37 diagrams.) February 2, 1933. Price 3s. net.
In a number of Service Liberty engines, fatigue (lacks devcicred at t i c fillet between the fifth crankpin from the tail end of the crankshaft and the
306A
April 26, 1934
FLIGHT" ENGINEERING SECTION

NO. 99
Year
April 36, 1934
CONTENTS
Evaporative Cooling. By R. Haley An Investigation into the Variation of Engine Power with Height Oil-cooled Exhaust Valves Page 85 28 31
siderable portion of its chord, a condenser of low volume surface ratio could be obtained by using a double-skin construction. Where, however, the area available is definitely limited, a retractable honeycomb radiator could be introduced into the circuit to assist under conditions of peak load." Cooling Surface The capacity to dissipate heat from the system described depends on several factors, and it is extremejy difficult to hit on a happy medium. It is well known that the amount of cooling surface required is not constant for any particular engine, but varies with the atmospheric conditions and the climb characteristics of
EVAPORATIVE COOLING
By R. HALEY
(Oonchuied from page 24.)
Experimental _AIR VENT Experiments were carried out at the R.A.E. in 1927 n a Bristol Fighter to ascertain the advantages of STEAM HONEYCOMB RADIATOR Evaporative Cooling. The aircraft was fitted with a triangular honeycomb radiator, placed above the top FILLER centre section. Fig. IV is a diagram of the system emGAUGE ployed which had no steam separator but relied on the bottom of the condenser to act in this capacity. It was found in this test that the steam separated from -THERMO POCKET the circulating water too late in the system and it was suggested that a separator fitted close to the exit from the water jacket would have prevented a large percentage of the water being carried up with the steam. WATER In wing surface condensers a large proportion, of water CIRCULATING PUMP in the steam becomes a positive danger as there would be an excess of weight in the condensers as well as retaining a large quantity of water, thereby depleting the amount required round the cylinder jackets. Fig. IV. : Diagram of experimental evaporative cooling To complete the experiments the same machine was system fitted on a Bristol Fighter in 1927 used but with wing surface condensers. These condensers were subjected to a test of 1 lb. per sq. in. both internal and external; relief valves were fitted to the aircraft. Also the heat supplied to the water jacket in the naturally aspirated engines falls with, but not at blow off at 1 lb. per sq. in. The first flight tests were the same rate as, the power, as the altitude increases. made with all four condensers in operation, i.e., top and The capacity to dissipate heat from the condensers, bottom wings, and a loss of water slightly under 1 pint for a climb of 5,000 ft. and a cruising flight of two however, tends to increase with the altitude when climbing at a constant I.A.S. Allowing, therefore, for the hours' duration was registered. heat reservoir capacity of the water jacket, oil, and In a paper read before the I.A.E. in 1930 by Capt. Andrew Swan, B.Sc, some of the concluding remarks metal parts of the engine, the time of climb from G.L. to a given altitude will have a material effect on the might be of interest: " The leading edge condenser occupying the wing in size of the condensers, due to the fact that the machine leaves the ground with the cooling system well below front of tho main span would appear to be restricted in its application to wings of comparatively small size boiling point. Other factors in estimating the surface and of large surface per horse-power. In large wings required are:(1) Change in boiling point of water with increased altitude; (2) variation in air temperathe condensers would have a high volume surface ratio, and cooling might be adversely affected. As the upper ture throughout the year; (3) change in cooling air temsurface of a wing has good cooling properties for a con- perature and density at various altitudes. 410 a D2
26
SUPFLSMBMT TO
PLIGHT
APRIL 26.
1934

Then C = '- (1 WkL* + 12 27k% 4- ] 7 54) (vad)-u ^^ Example:An aeroplane with a wing section of 8 ft. chord, giving approximately a perimeter of 16.32 ft., is required for tropical summer conditions with a mean steam pressure of ^ lb./sq. in. within the wing condenser. The total length of condenser is 40 ft. and cooling surface is of " standard properties," i.e., 25-1 per cent, of top surface and 15.2 per cent, bottom surface. What h.p. can be dissipated at an altitude of 9,000 ft aond an air speed of 150 ft./sec. with a /cL of 0.3P
4Q C=67'6X (21-34) 106 106
The condensers, of course, have to get rid of the " latent heat " of the steam and the quantity of steam generated is a big factor in the design. If the rise in temperature of water round the cylinder jacket is 20 deg. Fahr. the quantity of heat carried by 1 lb. of water is 20 B.T.TL, whereas the quantity of heat carried by 1 lb. of steam is 966 B.T.TT. I t is a wellknown fact that no steam is produced until the temperature of evaporation is reached, and this is controlled by the pressure. Whilst evaporation is taking place the temperature remains the same until all the water is evaporated. The heat supplied to the water after raising it to evaporative temperature until it is all converted to steam is termed the " Latent Heat " of steam. As the heat transferred from the surface of the wing condensers to the atmosphere must pass through the boundary layer, the condition of the layer becomes of great importance and at some point between the L.E. and T.E. the laminar layer becomes turbulent and is effected by the steadiness of the air advancing on the L.E. of the wing. Hence the importance of keeping the surface of the wing condensers as clean as possible. Having obtained the quantity of heat to be dissipated in terms of h.p. from the engine manufacturers, make
-692 X 16-32) + (1-022)-26 = 142-3 b.p. see Tables I and II R & M 1481 On climb there is an additional amount of heat to be dissipated, depending upon the rate at which the boiling point, corresponding to the pressure, is falling as the altitude increases. At 5,000 ft. the boiling point of water falls about 1.05 deg. C. in a 1,000 ft. of pressure altitude, so that with a steam cooling system a 1,000 ft. ascent has the same effect as a supply of
200 190 180 RO 160 150 KO 150 H O1 1 0 I0&90 80 70260 50^40 30 20 10 0 10 20 30 40 5 $ 60 70,? 80 30 100 1 1 0 120 130 140 150 160 170 BO
L /\ 1 \
if A"
I/O
V> \
r,V
/
/
A
s KL- 125
\ / / / K -32 | /
L
+ 1
"I
"I
30^-
IOOA-
\ - 1
L \ \
10024 OX
--
\
\
50^-
.g $ 8
\KL-I25
6 )% \ 2.
\ \
32
t<
/
/
t
/I
f
FROM LEA
7~
/
\ \ \ \ /
\ \
70,Z- UJ
z 5
<D
8 LOWER SURFACE 90^-
/ f
\ / /
UPPER SURFACE
uJ
N \ \
\
\ \
ft HALEY
Fig. VI. : On left of datum line, cooling from lower surface compared with total cooling of standard surface at the same kL. On right of datum line, cooling from upper surface compared with total cooling of standard surface at same kL. a layout of the wing condenser using about 25 per cent. sensible heat sufficient to raise the cooling water and its associated metal 1.05 deg. C. If We is the water equiof the top surface and 15 per cent, of the bottom surface, making the area of the surface approximately valent of the system which must be maintained at the 0.2015 dh, where d = total profile of aerofoil surface boiling point corresponding to the pressure, and " R ' ; in feet and L = length of L.E. occupied in feet. These the rate of climb in feet per minute, then this addiproportions will be governed largely by the design of . 1-05 X Well TZ. horse-power. the wing and wing tips, and must be tried out. To tional cooling is about 1,000 23-6 obtain the amount of cooling from the condenser let Two distributions of emissivity taken from results C = total cooling in h,.p., d = total profile of aerofoil studied in R. & M. 1163 are illustrated in Fig. VI. section in ft., i- = length of L.E. occupied in feet, v = speed of oncoming air in feet per sec.; o- and p. are the one referring to a fcL value of 0.125 taken as typical density and viscosity of the air, expressed relative to for level flight and the other with a 7cL of 0.32 as typical The abscissa; are distances I.C.A.N. ground level values, M = the mean tempera- of climb conditions. measured from the L.E. over the profile towards the ture difference in deg. C , 7 f L = a function of the lift T.E. expressed as percentage of the profile distance coefficient. 410 6
27
APRIL 26, 1934 SUPFLEMNT TO

between these edges. The ordinates at any point gives the total cooling obtainable out of a surface extending from the L.E. up to that point. The scale is such that the cooling obtainable from the standard surface is represented by a total ordinate of 100. Thus, for the climb case, from 25.1 per cent, of the upper surface the coaling is 70.8 units, and from 15.2 per cent, of the lower surface it is 29.2 units. Such a scale makes the calculation of the actual surface from the standard surface an easy matter. Suppose the standard surface required in a particular condition of climb to be 110 sq. ft., while the standard area available on the aircraft is only 88 sq. ft., then the cooling required on the wing surface per unit length of the L.E. will be 125 pea' cent, that of the standard. From the curve, one combination which will meet this requirement is 41 per cent, of the upper surface and 15.2 per cent, of the lower surface measured from the L.E. The actual area then used will be: 41 + 15-2 X 88 or 122-8 sq. ft. 2 5 - 1 + 15-2 Of course, there ace infinite numbers of positions of the profile which will combine to give the required effect, and the position chosen will be largely governed by the position of the front spar relative to the L.E. and the angle of climb which governs the angle of the drain pipe from the T.E. of the condenser. Retractable Condenser Experiments carried out at R.A.E. show that the triangular honeycomb radiator is more efficient than the square type. Readers who are familiar with marine condensers will have noticed that the steam is introduced into a large opening at the top of the condenser, and also at the hottest point of the condenser. We may see something very similar in aircraft in the near future, complete with air pump, etc. To return to the present model, if steam is supplied above the heat dissipating capacity of the radiator some must flow out of the air vent and be lost, see Fig. I I I . The efficiency of the radiator partly depends on its capacity to adjust its heat dissipation by drawing in air or expelling air as the heat supply is altered. The design of the radiator should be such that no air pockets can be formed, preventing the easy flow of steam. The steam should be introduced at as low a point as possible with the air vent at the top of the radiator, when the full efficiency of the cooling surface of the tubes will be made use of. Here it would be as well to mention that the tubes are of the standard hexagonal ended type either 360 mm. or 400 mm. long by 7 mm. or 10 mm. dia., according to the design in hand. The working weight of the condenser must include the film of water round the tubes and is in the range of 0.38 lb. to 0.43 lb. Tilting the radiator relative to the horizontal plane of flight has no decided advantage, owing to increased drag, but tapering the tubes has a decided advantage. Arrangements can be made in the design for the condensate to be collected at the lowest point of the sump. There is a definite increase in efficiency when the steam is introduced at the front end of the radiator. The problem of selecting a criterion for the cooling efficiency of a radiator is an intricate one. I t is necessary to compare, in given temperature conditions, the rate of heat dissipation for the radiator with the maximum rate of heat dissipation it is capable of, i.e., at the boiling point of the liquid, when the maximum demand is made of it. Maximum demand implies that the engine is at full throttle and that the air velocity through the radiator is a minimum at maximum radiator exposure. A practical example to maximum demand is obtained in flight at maximum rate of climb. This condition makes it difficult to measure steady heat conditions, since the atmospheric conditions a n l the
FLIGHT
factors governing the flow of heat are continually changing. Assuming that the lieat entering the radiator per second is a constant fraction of the b.h.p. (1 b.h.p. = 2,545 B.T.U./hr.) at a height "h" (in a standard atmosphere) ; let p and < r be the relative pressure and density respectively, A the air temperature, B the foiling point of the liquid in the radiator, M the mean radiator temperature, V the forward velocity. The heat to be dissipated per second is proportional to p, and the heat discharge from the radiator per second is proportional to (M-A)o-V; and in a climb " V " is proportional to a power of a- between o and , so that the rate of heat dissipation is -proportional to (M-A)cri. The maximum rate oi heat dissipation possible is (B-AVi. To enable the pilot to know when the radiator is too far within the fairing, i.e., when the radiator is not condensing all the steam received from the wing condensers, a warning device is fitted to the bottom of the air vent, which, when steam passes to atmosphere, makes electrical contact with a small lamp fitted on the instrument board, giving the pilot warning that steam is being lost. The writer can foresee the day when some electrical device will operate the radiator in and out, functioning by the emission of steam through the air vent. This air vent must protrude clear of the fairing when the radiator is in the full-up position. In this position also, the sump must be in the slipstream. Header Tanks The design of header tank is largely in the hands of the engine manufacturers, who should be consulted before any drawings are issued to the shops. As this unit
T
CRANKSHAFT
oc = ANGLE OF TAIL DOWN
Fig. V. is still in its experimental stages, it would be useless to describe any particular type in this article. The material used in construction can be either aluminium with welded ends, or Dural with riveted ends. As most of the internal piping is of intricate Bhape, they can be made up in aluminium and welded. Certain features are common to all header tanks, viz., the sump must have baffles to prevent the water swirling, a pressure gauge, temperature thermometer, and filler neck must be fitted; also a fitting to accommodate the low water level warning device. This is operated on the same principle as the steam loss device fitted to the retractable radiator. On certain type of aircraft it is necessary to fit a valve for inverted flight to prevent the water from passing up the steam pipe to the wing condensers. To fix the position of the filler neck it is necessary to find out the cylinder jacket volume, plus the volume of all the piping containing cooling water, i.e., the main feed pipe from the tank sump to engine pump, the pipe leading from the cylinder head to tank, and any internal piping. About 50 per cent, of the tank's total capacity
410
28
SUPPLEMENT TO APRIL 26, 1934
FLIGHT

In the case of an aircraft the temperature variation will be much more than that covered by the N.A.C.A. tests. The standard height corresponding to 20 is just under 17,500 ft., but in winter this temperature might easily be recorded at 10,000 ft. A careful examination of the curves of I.M.E.P. at constant pressure from the N.A.C.A. report shows that almost in every case where full throttle was used the I.M.E.P. at 20 was less than that indicated by the curve of the reciprocal of the relative absolute temperature. With the engine in the throttled condition the fall off is not so apparent, but since we are only concerned with the full throttle performance of the engine this fall off in power at low temperatures should be taken into account. A similar fall off in power at the highest temperatures will also be noticed on these curves, but not quite so consistently as at the lowest temperatures. The inference to be drawn from these results is that there is a linear variation of I.M.E.P. with temperature at constant pressure. The alternative relationships for the variation of I.M.E.P. with temperature and pressure can be expressed as : I.M.E.P. P
is air space, and the volume of internal steam pipes must be deducted from the tank capacity before fixing the total water capacity of the tank. A typical example of the method of arriving at the water level in the header tank is shown in Fig. V. The position of this level naturally governs the position of the lip of the filler neck when the aircraft is tail down. From a graph provided by the engine manufacturers one can plot the intersection of the water level line with the centre line of engine, which, when produced through the header tank, fixes the level relative to the vertical.
AN
INVESTIGATION INTO THE VARIATION ENGINE POWER WITH HEIGHT
OF
By W. R.
DURING
ANDREWS,
A.F.R.Ae.S.*
the last few years much investigation has been done regarding the best basis of reduction to standard of aircraft performance. The outcome has been the adoption of a basis half-way between the pure density and the pure pressure. The question of the actual variation of engine power with height has become of secondary consideration. Messrs. Coales & Lingard (Ref. 1) have discussed the question at great length from the analysis of a number of full scale results. The final presentation of their results does not make extrapolation possible, and over the range of heights considered gives no indication of the variation of engine power at other than standard temperatures. Any empirical relationship for the power factor must cover extrapolation to great heights, as well as tropical and arctic conditions. In an endeavour to produce a rational basis of reduction Diehl (Ref. 2) introduced a power factor based on the results of bench tests of an aero engine in the N.A.C.A. altitude chamber. These tests (Ref. 3) were made at varying temperatures and constant pressure, separate tests being made at three or four pressures. Indicated mean pressures only are recorded in the report but from data, probably not published. Diehl deduced the following relationship :
I.M.E.P.,, I.M.E.P. = P (1-5 - 0 - 5 T) or I.M.E.P.,,
(5)
B.H.P. where P
pi 15 T - .
. Relative pressure compared with standard G.L. pressure d 29-92 in. Hg. T = Relative absolute temperature compared with G.L. standard of 288 C. absolute. Such a formula as 1 gives excellent results at low height and can be expressed in the form Power factor = /(P 5 6 r > < r 0"435) (2) The power of an engine is rarely proportional to R.P.M., but assuming this to be the case, then equation (2) expresses the basis of reduction of flight test data to standard conditions. This is in close agreement with the findings of Hutchinson and Finn (Ref. 4). who found the weighed mean law for the correction for N\/o- to be Power foctor = / ( P 5 * <r-) (3) It must be remembered, however, that the above comparison is only possible on the assumption that B.H.P. is proportional to R.P.M. The conclusions arrived at in the tests carried out by the N.A.C.A. in the altitude test chamber on an engine at different temperatures and pressures were that the I.M.E.P. was proportional to (a) the absolute pressure and (b) the reciprocal of the square root of the absolute temperature. The tests were only intended to supply information as to the correction necessary to engine bench test data for reduction to standard conditions, which in general would be covered by temperature variations of 20 C. The tests, however, were carried out for a range of + 30 C , so that all practical variations of temperature near the ground are well covered.
" Ml-. Andrews is on the Technical Staff of A. V. Boe < f c Co., Ltd.
where I.M.E.P. is in non-standard conditions I.M.E.P.,, is in standard conditions of temperature (15 C.) and pressure (29-92" Hg.). Both these alternatives will be used in conjunction with the expressions which follow in an attempt to determine the best form of the power factor. An attempt will also be made to illustrate the difference these alternatives make in the reduction of flight test data to standard. As stated previously, equation (1) will only be an approximation at low heights as its very nature suggests a power output at infinitesimal pressures. Obviously, at some fairly low pressure, the whole of the energy of the engine is absorbed in overcoming the internal friction of the moving parts. So far as the standard atmosphere goes it will be found that equation (1) is almost identical with P1-05 at heights at any rate up to 20,000 ft. It is obvious that if a power factor is to be capable of extrapolation in all directions it must include some factor to cover the internal losses of the engine and also make some allowance for the variation of these losses themselves with varying atmospheric conditions. In this connection there is a scarcity of data. It is only by comparisons between indicated and brake power under varying pressures and temperatures that reliable information can be obtained for use in the determination of the power factor. The nearest approach to the required data is supplied by tests made to determine the motoring losses of a single cylinder Benz engine (Ref. 5) under varying atmospheric pressures. The results are expressed in terms of the density of the air in the cylinder at the B.D.C., i.e., at the commencement of the compression stroke. It is pointed out in the report that carbonisation of the lubricating oil can materially modify the internal losses to as much as 40% of those determined by motoring tests. The results do seem to indicate that at constant R.P.M. the motoring losses can be expressed as a constant plus a linear function of the charge weight at B.D.C. Unfortunately, these tests also indicate that the loss in M.E.P. due to charge density is also proportional to R.P.M. This fall off in M.E.P. can then be expressed as I.M.E.P. - B.M.E.P. = e + i i N o I (6) where N and [ = R.P.M. = Density at B.D.C.
Assuming that the I.M.E.P. will be proportional to the charge weight it follows that ax will bear the same relationship to the outside air as does the I.M.E.P., so that 6 can be rewritten in the form -,.r .: I.M.E.P. - B.M.E.P. = a -j- b N - ; . . -: ^
410
2'.)
APRIL 26 1934 SUPPLEMENT TO

or its alternative form I.M.E.P. - B.M.E.P. = a + b N P (1 5 - 0 5 T) where and b are constants. From which it follows
FLIGHT
If now the assumption is made that the power factor given in Fig. 6 of R. & M. 1141 is correct at about 12,500 ft., then / = 0-0731 and the following Table shows the comparison between the values calculated by 10 and those from R. & M. 1141. :
TABLE 1
J H.P. = N | a, + b, N - ^ | d > \/Tl \/Tl
(8)
OveT quite a large range of R.P.M. the I.M.E.P. may be taken as constant so that the power factor can be expressed as : p
B.H.P. X.I.M.E.P. A - N
P Standard Height, f feet. = P'lPo
U
bL N) 0 5,000 10,000 15.000 20.000 25,000 1-0 0-8320 0-6876 0-5642 0-4594 0-3709
Power factor = p, T From Fig. 6. T" 1 0731 =.-00731 R. & M.I 141 To (mean values) 10 0-8353 0-6912 0-5659 0-4575 0-3639 1-0 0-838 0-693 0-567 0-454 0-335
Since
N.I.M.E.P.,,* - N I.M.E.P. P
I.M.E.P" ~ -y/f
(9)
then
1-0 0-9656 0-9312 0-8968 0-6824 0-828
Since the power factor is a function of R.P.M. it follows that the factor obtained from climb data must be different from that obtained from the level speeds. However, the
AIRSCREW CHARACTERISTICS
The agreement between the two values is really no better than would be expected considering that the value o f / w a s
POWER CURVE MAXIMUM R P M - 10 MAXIMUM B.H P 10 P/,! AT MAX RPM 10
K\
9 8
WOODEN AIRSCREWS Pf/o - 0-7 Pe/D - 0 -964
20
1-9
1-0 -
X \
\
\
/
\
7
\
/
FIG.i. \
/ / /
FIG.2
\ \
\
s
\ \
fi
more important quantity is the climb and by a lucky coincidence it is unusual for the R.P.M. on climb for an unsupercharged engine to vary more than 1 per cent, per 5,000 ft., so th at there will be no very great loss of accuracy in omitting the R.P.M. from the equation and re-writing in the form.
l - d - g d-g
Since Pe must equal 1 -0 when P / \ / T
then K = / + fe - 1 so that finally

(10) or in the alternative form
chosen to make equation (9) agree with the observed values at roughly 12,500 ft. Any form of power law must give a method of reduction in keeping with that obtained from the analysis of many tests. It is suggested that to test this a hypothetical engine and airscrew be taken and the R.P.M. at different heights and temperatures calculated. The results so obtained can then be compared with those of Flight Trials. "*' It has been shown (Ref. 4) that the best method of reduction of flight test results to standard is nearly the half pressure half density method. If it is possible to obtain a power factor which agrees with the findings of reference 1 and gives a basis of reduction of approximately (pi ai), then by virtue of the allowance for internal losses extrapolation can be made with greater confidence than by any of the existing methods which make no allowance for engine friction. For the purpose of determining the basis of reduction an airscrew of the following characteristics has been chosen : P//D = Face pitch/diameter =0-7 Pe/D = Experimental mean pitch/ = 0-964 dia.
410<J
30 SUPPLBMBNT TO APRIL 26, 1934
FLIGHT

V * / n- =
v
A.S.I. = where J =
0-725
(11)
TABLE 2
P = 10731 r- -
"0731
N = R.P.M. ratio n = R.P.S. ratio = N P/3 = 0-739 p, 36900 P Pw3D5 1 = ^ - a t J = 0-5 (12) Standard Temperature
0 ... 1 0 10 1-0 1*0 1-0
where P = G.L. Power at " n " r.p.s. The values of the generalised torque coefficient KgQc are plotted in Fig. 1. Since we are assuming that the power factor does not vary with R.P.M., it is only necessary to determine the values
rV
5,000 . . . 0-8616 10,000 07384 15,000 . 0-6291 20,000 0-5327 25,000 0-4480 30,000 0-374
0-8353 0-6912 0-5659 0-4575 0-3639 0-2841
0-9282 0-8593 0-7932 0-7298 0-6693 0-6115
: 1031 1-068 ! 1112 1-165 j 1-231 ! 1-317
0-7755 0-594 0-4485 0-334 0-2436 0-1738
of N v ' a for a constant value of or what is the same thing n at constant J. Choosing J = 0-6 then Kq Q, = 0-901 from which it follows by substitution in 12 that (13) = 1-219 P/ B a always corresponding to ground level conditions. By plotting P/n s against N we can read off the value of N corresponding to any value of P/ n 3 and so calculate the appropriate value of N \/a. See Fig. 2. Since is constant for any value of a by virtue of
P,
Standard Temperature + 30 0 5,000 10,000 15,000 20,000 25,000 30,000

0-9055 0-7775 0-6640 0-5635 0-4754 0-3980 0-3306 0-9474 0-7896 0-6518 0-5321 0-4281 0-3390 0-2628 9516 8818 8149 7507 6895 6311 5750
0-956 0-985 1-019 1-059 1-110 1174 1-258
0-9015 0-6960 0-5310 0-3995 0-2954 0-2140 0-1454
Standard Temperature 30 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 ... ... ... ... ... ... ... ... 1-116 0-9655 0-8315 0-712 0-6060 0-5125 0-4305 0-3590 10611 | 10565 0-8884 0-9826 0-7382 0-9129 0-6067 0-8438 0-4924 0-7785 0-3943 0-7159 0-3102 0-6561 0-2386 0-5992 1-052 1-0870 127 174 1-231
300 388
equation (11), it follows that by plotting the N \/a for each ft eight in the standard atmosphere at the corresponding height a curve is obtained from which the corresponding standard height can be read for any value of N \/a determined from non-standard conditions. The atmospheric conditions chosen for the investigation are (a) Standard Temperature, (b) Standard Temperature 30. (c) Standard Temperature 30. The calculated values for the three atmospheres are given in Table 2.
1-505
1-1115 0-8575 0-6740 0-5120 0-3833 0-2822 0-2035 0 1430
The calculations of the values of N v a n ( i the corresponding standard heights are given in Table 3 and Fig. 3.
TABLE 3
= 1-0731;^= -
0-0731 Equivalent Standard Height.

HE
Height
H. feet
/
a
P/ra3
HE -
Pe
JN V
Hs
AH
; 1) Standard Temperature
0
5,000 10,000 15,000 20,000 25,000 30.000
1-0
10
0-9282 0-8593 0-7932 0-7298 0-6693 0-6115
1-031 1-068 1-112 1-165 1-231 1-317
1-219 1-257 1-303 1-357 1-421 1-501 1-606
0-915 0-837 0-7628 0-691 0-6225 0-5567 0-493
"
(2) Standard Temperature + 30'

0 5,000 10,000 15,000 20,000 ... 25,000 30,000 0-9516 0-8818 0-8149 0-7507 0-6895 0-6311 0-5750 0-956 0-985 1019 1059 1-110 1174 1-258 1165 1-201 1-243 1-292 1-354 1-432 1-5345 0-889 0-812 0-739 0-669 0-601 0-5365 0-4735 1,700 6,700 11,650 16,650 21,600 26,600 31,550
+ 1,700 + 1,700
1.650 1.650 1.600 1.600 1,550
410/
31
APRIL 26, 1934
SUPPLEMENT TO
THE AIRCRAFT
ENGINEER
FLIGHT
TABLE 3(continued)
Height H. feet
P/
P,
Equivalent Standard Height
1 1
1!
= AH
i; "s
(3) Standard Temperature 30
0
5,000 10,000 15,000 20,000 25,000 30,000 35,000
1-0565 0-9826 0-9129 0-8439 0-7785 0-7159 0-6561 0-5992
1-052 10870 1127 1174 1 231 1-300 1-388 1 505
1-282 1-325 1-375 1-432 1-501 1-586 1-693 1-835
0-945 0-865 0-7905 0-718 0-6473 0-580 0-516 0-454
3,200 8,150 13,150 18,100 23,200 28,200 33,200
1,800 1.850 1,850 1.900 1,800 1,800 8,000
From the tables it will be seen that there is a small tendency for the change in equivalent height A H to decrease as the standard height increases,for constant temperature differences.
VARIATION! OF N J f f WITH HEIGHT !
The average value for the 30 is 1.750 ft, per 30, or 58-3 ft. per degree change from standard. This corresponds to a basifi of correction of flight test data to standard of: Power factor = /(i>"'rpl ' 49 ) as compared with the weighed mean law from_R. & M. 1532 Of/ (p0-58 ff0.42).
KKFERENCKR
fe 1-0731 - -0731
1.K. & M. 1141. The determination of the horsepower height factor of engines results of type trials of aircraft. 3. I). Coaleg and A. L- Ltagard. 2.N.A.C.A. Report No. 297. The reduction of observed airplane performance to standard conditions. Walter 8. Diehl. 3.N.A.C.A. Report No. 190. Correcting horsepower measurements to a standard temperature. Stanwood W. Sparrow. 4.R. < f c M\ 1">:)2. The l"Rt basis of aircraft performance reduction. Hutohinson, B.A.. and E. Finn, B.8c. .1. ]..
75
-85
90
95
(To be
continued.)
OIL-COOLED EXHAUST VALVES Salt-filled exhaust valves have been used extensively in aero engines, more particularly in the United States of America, but oil-cooled valves are something of a novelty as far as aero engine design is concerned. A British inventor, Mr. It. C. Cross, of Bath, Somerset, has been experimenting with oil-cooled valves on a car engine, and has obtained promising results. I t goes without saying that the conditions to be faced by the valves of an aero engine are a good deal more severe in many ways than those to which the average car engine valves are subjected, but the inventor does not expect any serious difficulty in applying the system to aero engines. I t might be mentioned incidentally that Mr. Cross is also the inventor and patentee of the Cross rotary valve, which has been fitted on motor-cycle engines, one such engine, using a 60 per cent, petrol, 40 per cent, benzol fuel mixture, having run satisfactorily at a compression pressure of approximately 300 Ib./sq. in. without detonation. The B.M.E.P. reached 157 Ib./sq. in. for an unsupercharged engine. The suggested arrangement of an overhead oil-cooled valve is shown in the diagram. In the main, the arrangement is obvious. A gallery tube Ji is connected to the oil supply and has attached to it a small tube C, which passes through the hollow stem of the valve into the hollow valve head. Between the tube C and the walls of the hollow valve stem is a small annular space Z), through which the oil returns. A is a double or forked valve rocker. I t will be obvious that the Croes oil-cooled valve
CROSS 0IL-C00LED VALVE : A, is a double or forked rocker ; B, a gallery oil tube ; C, a small tube connecting B with valve bead ; D is an annular space between tube C and inside of hollow valve stem
.rf
SUPPLEMENT TO
FLIGHT
APRIL 26,
1934

BENCH AND FLIGHT TESTS OF A ROOTS TYPE AIRCRAFT ENGINE SUPERCHARGER. By the Staff of the Engine
arrangement in the form shown in the diagram is applicable only to engines in which the overhead valve mechanism is totally enclosed. With the open type of valve gear found on many engines the oil returning from the valve head and stem would be flung out and, apart from the loss of oil, would make the aircraft behind the engine very dirty. The oil supply is, of course, intended to be taken from the general supply, and in some cases it might be necessary to fit an oil pump of slightly greater capacity than normally provided. Apart from arranging for the gallery oil pipe, it would be necessary, in converting an existing aero engine to take oil-cooled valves, to bore out the valve guides to take the larger valve stems. I t might have been feared that carbonisation of the oil inside the valve might occur and might block the passages. Experience with a car engine indicates that this does not take place, the heat transference via the quickly-flowing oil stream being so rapid that the valve never gets really hot. The temperature of the oil after passing through the valve is raised by a few degrees only.
Experimental Department, R.A.E. Communicated by the D.S.R., Air Ministry. R. & M. No. 1558. (23 pages and 16 diagrams.) September, 1933.
The purpose of the investigation was to investigate the application of the Boots class of displacement supercharger to an aircraft engine, bench tests of a unit of this type, built to N.A.C.A. specification, were carried out to examine its performance characteristics, and subsequently flight tests were undertaken to determine the capabilities of a Lion V.S. engine fitted with this supercharger. The bench calibration tests covered a range of rotor speeds from 2,750 r.p.m. to 3,900 r.p.m., and the necessary quantities were measured to enable the power absorbed by the supercharger and the volumetric and adiabatic efficiencies to be calculated. From this information the probable rated altitude in combination with the Lion engine was deduced. The flight tests, which were carried out in a D.H. 9A aircraft, consisted of determinations of (1) the rated altitude on climb and in level flight; (2) the ceiling ; and (3) level speeds at the rated boost pressure of the engine below the rated altitude and at the maximum obtainable boost pressure above it. During the several flights, additional observations were recorded relating to the individual performances of the aircraft, engine and supercharger. The Roots supercharger wag found to be especially suitable for engines with rated heights in the vicinity of 20,000 ft. At pressure ratios corresponding to altitudes of this order a gatlsfaetory volumetric efficiency is maintained, and the adiabatic efficiency compares favourably with that of centrifugal superchargers. The power wasted below the rated altitude by permitting a proportion of the compressed air to escape to atmosphere is not excessive, and at ground level the total power absorbed by the supercharger can be reduced to a negligible amount. NOTE ON THE USE OP NETWORKS TO INTRODUCE TURBULENCE INTO A W I N D TUNNEL. By E. Ower, B.Sc..
A.C.G.I., and R. Warden, Ph.D., M.Eng. R. & M. No. 1559. (8 pages and 4 diagrams.) July 31, 1933.
Variations of turbulence in the wind-tunnel stream are known to have a marked effect on the results of certain types of experiments, and in attempts to study this effect wire or cord networks have been used on various occasions to introduce artificial turbulence into the tunnel. Experience at the N.P.L. suggested that this method might be unsuitable on account of the fluctuations in the time average of the velocity across a section introduced as a result of the "shadows" of the cords. Some measurements of the distribution of mean velocity and static dsestyse at different distances downstream of two networks were therefore made and the results confirmed the impression previously formed. While the transverse static pressure distribution was uniform at a distance of only 10 in. behind a network of J-in. cords with li-in. mesh, a violently-disturbed distribution of mean velocity was observed at this distance, and even at 3 ft. behind the network the time mean of the velocity at different points in a section was found to vary by over i per cent. Similar explorations were made with a streamline body in the tunnel, and it was found that the network seriously modified the velocity fleld near the body. It is concluded that networks are not suitable agents to employ for any quantitative investigations of the effect of turbulence on wind-tunnel data. THE FLOW NEAR A WING WHICH STARTS SUDDENLY FROM REST AND THEN STALLS. Aeronautics Laboratory,
SUMMARIES OP AERONAUTICAL RESEARCH COMMITTEE REPORTS
These Reports are published by His Majesty's Stationery Office, London, and may be purchased directly from H.M. Stationery Office at the following addresses: Adastral House, Kingsway, W.C.2; ISO, George Street, Edinburgh; York Street, Manchester; 1, St. Andrew's Crescent, Cardiff; 15, Donegall Square West, Belfast; or through any Bookseller.
THE EXPERIMENTAL DETERMINATION OF PITCHING MOMENT OP AN AEROPLANE DUE TO ROTATION IN PITCH.
Cambridge. R. & M. No. 1561. grams.) August 8, 1933.
(9 pages and 12 dia-
By A. S. Halliday, B.Sc., Ph.D., D.I.C., L. W. Bryant, B.Sc., A.R.C.Sc, and C. H. Burge. R. & M. No. 1556. (27 pages and 28 diagrams.) March 30, 1933.
All previous measurements of the damping moment of an aeroplane in pitch have been made by the method of oscillation in the wind tunnel. Unfortunately, as pointed out in R. & M. 718, this method does not determine the true value of M^ directly, but is subject to a correction due to the fact that the incidence changes during the experiment. In consequence of the appreciable time taken by the downwash from the wings to reach the tail, the effective incidence of the tail does not at a given instant correspond to the incidence of the winss at the same instant. It has been shown in R. & St. 826 that at the incidences of normal flight the correction for this effect can be applied with some confidence, particularly if the rate of change of downwash at the tail is actually measured on the model. But the simple theory of R. & M. 718 can hardly be accepted without question for application to attitudes at the stall and above, and any method of experiment which avoids the necessity for making the correction is of great value in determining M^ at large incidences. The whirling arm provides uch a method, since the incidence remains constant throughout the motion There are large differences in the values of the derivative K B m- at angles. of incidence above the critical angle obtained by the two methods. Below the critical angle they give the isame result to an average accuracy of about 10 per cent, provided the oscillation measurements are corrected for rate of change of wing downwash in accordance with the principle laid down in R. & M. 718.* In view of the uncertainty of this downwash correction it is considered that the whirling arm measurements are to be preferred for the normal flying range of incidences. For ordinary calculations of longitudinal stability it is probably sufficiently accurate to compute KB- from the tail forces alone making no allowance for wings and body. The derivative Z ? for the wings may equal or considerably exceed Z for the tail. The experimental measurements of Z ? and KBj for a sing aerofoil agree reasonably with the theoretical values.
18 f downwash n 718. The effect ofJthe lag thejongitudinal * *v.,K,i * M. J1- 7 - T ,he effe S* Of ? 6 la F of J 0 ";? 8 1 1 on the longitudinal stability of an aeroplane and on the rotary derivative M ? . - W . L. Cowley
When an aerofoil is started suddenly from rest into steady motion, the fluid flow in its neighbourhood passes through a series of different forms which approach asymptotically to the steady form which occurs in the wind tunnel and In the steady flight of aeroplanes. During this transition period a vortex sheet, or surface across which the velocity Of the flow is discontinuous, is shed from the trailing edge of the aerofoil and remains behind in the fluid, the strength of the discontinuity at the trailing edge falling asymptotically to zero as the steady state is approached. H. Wagner* has advanced the hypothesis that, if the wake is infinitely thin and the trailing edge sharp, the form of the flow and the magnitude of the lift during this transition stage should be approximately calgulable on the assumption that the flow everywhere outside the wake is irrotational, and that the strength of the discontinuity across the wake at the trailing edge has that unique value which is necessary to avoid infinite velocity at the edge. On this hypothesis, he calculated the distribution of vorticity in the wake and the lift at various distances from an impulsive start of an aerofoil idealised to the form of a thin flat plate of infinite span, set at an infinitesimal angle of incidence to the direction of motion. P. B. Walker, t using the same apparatus as was used for the experiments now to be discussed, photographed the two-dimensional fields of flow at various distances from a sudden start of an aerofoil of R.A.P. 30 profile set at 7'5 incidence. These photographs showed the flow changes to be substantially of the form supposed by Wagner, and measurements of the velocity fields near the aerofoil gave a curve of circulation around the profile against distance from the Btart, which was in remarkably close agreement with Wagner's theoretical curve, provided that, in each instance, the circulations were expressed as fractions of the circulation ultimately reached when the motion had become steady. The first signs of the stall were observed when the distance travelled was 2-8 times the chord of the aerofoil and, at that instant, it is deduced from the measured circulations that the lift coefficient was more than one and a half times the maximum which can act steadily upon this profile at the Reynolds number (1-3 x 10 ) of tho experiment. Photographs taken after the stall had begun show the shedding of vorticity which must occur as the lift falls to the steady value.
fa
and H. Glauert
Angewand'teltathematik und Mechanik. Vol. 5, No. 1,"February I92T Growth about a Wing for Measuring Growth of of Circulation Circ g and an Apparatus pp Fluid Motion." P. B. Walker. Reports and Memoranda N. 1402.
ff
410 k
May 31, 1934

No. 100 ( V o l ^ e s I X ) 9th year
iNHi
May 31, 1934
CONTENTS
The Effect of Swirl on Petrol Engine Combustion Investigation into the Variation of Engine Power with Height Technical Literature Summaries of Aeronautical Research Committee Reports Page 33 37 39
THE EFFECT OF SWIRL ON PETROL ENGINE COMBUSTION

B Y J. F. ALCOOK*
SINGLE-SLEEVE-VALVE ENGINE: Diagram showing oblique inflow during early part of induction stroke. itself. At this time, however, the flow in the port nppi'oaches may be utilised to increase or reduce the ' natural " swirl previously produced by the port. Thus, in Fig. 2, which shows a section through the ports of the engine used for the tests described herein, the sleeve is moving anti-clockwise when the inlet ports open, and the natural swirl is therefore clockwise. If
" S W I R L , " or rotation of the charge within the combustion chamber, is known to have a profound effect on combustion in Diesel engines, but its effect on petrol engines has not received much attention. The reason for this is that in the Diesel it is a valuable asset, easing as it does the critical problem of fuel distribution, while in the petrol engine it is generally a nuisance. Until recently petrol engine swirl has been of no great importance, for in most engines of normal design the amount of swirl, if any, is so small that its effect is negligible. There are, however, a few known exceptions to this rule, and perhaps a fair number of unsuspected cases. Recently, however, the single sleeve valve engine has entered the aircraft field, and in this type of engine swirl is of far greater importance. As will be seen later, the method of inlet port control in these engines automatically produces a high rate of swirl, and painful experience soon showed that it was a factor to be reckoned with. Method of Swirl Production In the single-sleeve engine the inlet ports are opened by the circumferential motion of the sleeve and closed by its upward motion. During the early opening period the orifice is faired on one side only by the edge of the cylinder port, and the mixture therefore enters obliquely, causing the charge to rotate in a direction opposite to that in which the sleeve is moving. Fig. 1 illustrates this effect. In the later part of the suction stroke the sleeve is moving vertically and no swirl is produced by the port
Mr. Alcock is on the Technical Staff of Ricardo & Co., Ltd., of Shoreham.
EXPERIMENTAL SINGLE-SLEEVE-VALVE ENGINE : tion through Ports.
540 a
the mixture is admitted to the belt by the left-hand inlet, the flow in the belt is clockwise and reinforces the natural swirl, but if the right-hand inlet is used, the total swirl is decreased. These effects are clearly shown in Fig. 3, which shows the results of a " blowing " test on the above engine. In this type of test the engine is stationary, and air is blown through the inlet ports (the cylinder head being removed) and the rate of swirl is measured by a freely rotating vane within the cylinder. The test shows that the flow in D2
34
SUPPLEMENT TO
FLIGHT
MAY
31, 1934

1
' 1
- FIG.3
/
\
> \ \ \
sV
1 8 0 / 200
given engine if it is to give its best results, is proportional to the engine speed. Thus the most logical measure of the swirl intensity is the " swirl ratio " or ratio of swirl r.p.m. t o engine r.p.m. Thus a swirl meter speed of 2,000 r.p.m. at am engine speed of 1,000 r.p.m. is expressed as a " swirl ratio " of 2. In what follows this notation is used throughout. EXPERIMENTAL WORK The work to be described was performed on an experimental single sleeve valve engine with one water-cooled, cylinder of 5.5-in. bore and 7-in. stroke, with a normal running speed of 1,300 ir.p.m. There are three inlet and two exhaust ports, giving nominal gas velocities at 1,000 r.p.m. of 125 ft./sec. for the inlet and 169 ft./sec. for the exhaust. The port timing is:
10 |
40 I 20
60 I 40
80 100 l?0 \ HC I TIMING CRANK O E O R H S \ I 60 80 IOO Njzi
\
WO 1 6 0
BLOWING SWIRL TEST. Pressure 10 0 in. of Water. Intake P (pro-swirl). Curve (neutral). Curve C, air to swirl).
E.O. 52-5 E. I.O. 10 L. ENGINE STATIONARY : Air I.C. 22 L. E.C. 22 J L. Curve A, air to Inlet Belt from B, air to Inlet Belt from Intake N The inlet ports are connected by a belt, shown in Inlet Belt from Intake A (antiPig. 2, which has three intakes. By fitting the carSee Fig. 6. buretter on one or other of these intakes, the normal charge swirl caused by the asymmetrical port opening the inlet belt affects the swirl mainly during the later part of the suction stroke, and has no influence at all can be increased or decreased by varying the tangential at the beginning of the opening period, owing to the flow in the belt, and the swirl ratio thus varied over a considerable range. This range can be extended on low velocity of the air in the belt at that time. either side by the use of pro-swirl or anti-swirl baffles on the outside of the ports. - Swirl Measurement " Blowing " tests such as those just described are aids to the understanding of the mechanism of swirl production, but give no measure even relatively of the irate of swirl wfei the engine is running. To obtain this a " swirl meter " is employed, consisting of a freely rotating vane in the combustion chamber, with a spindle
\\W\\\\\\\\\\\\\
EXPERIMENTAL SINGLE-SLEEVE-VALVE ENGINE : Section through Combustion Chamber. The combustion chamber is of the lenticular form shown in Fig. 5, giving a compression ratio of 5.88. There are two alternative sparking plug positions, one Swirl Meter for Motoring Swirl Tests. central and the other offset by l-,7e in. (0.261 of the sleeve bore). passing through the head to external counting mechanThe carburetter is a Claudel-Hobson type, 62-mm. ism. Fig. 4 shows such a swirl meter (fitted t o a body, 48-mm. choke, with variable main jet. Ignition Diesel combustion chamber). The engine is then was by coil to a single plug. ./-,-,r. motored, aitid the vane speed observed. Such a test is called a " motoring swirl " test. I t does not, of course, Test Equipment, etc. .--..:. give the actual rate of swirl during the combustion period, which is the swirl that matters, but appears to The engine output was measured by a swinging-field be consistently related thereto. electrical dynamometer, the r.p.m. by a tachometer which was frequently checked against a counter, and Swirl Notation fuel consumption by calibrated measure and stop-watch. The indicator diagrams were taken with a " FarnAs will be shown later, the swirl r.p.m., as measured boro " indicator. by a motoring swirl test, which is required by any 540
MAY
31, 1934
SUPPLEMENT TO

Water flow in the jacket-heat tests was measured by weight, and water temperatures by the usual mercury thermometers. In these jacket-heat tests no allowance
I S O
FLIGHT
FIG.8.
140
8 130
iJ 120
flour
p \
A
M O
MO/WH
FIG. 6.
2 3 SWIRL RATIO
Diagram of Inlet Manifold and Ports.
EFFECT OF SWIRL ON POWER OUTPUT AT VARIOUS SPEEDS : Central Sparking Plug.
is made for convection heat losses from the engine, etc.; this is obviously independent of the swirl rate, and thus will not affect the differences due to swirl variation.
Details of Tests. Swirl Calibration
Motoring swirl tests were made over the speed range with eight intake conditions, as follows:
Table 1 Table of Intake Arrangements and Swirl Ratios at 1,300 r.p.m. (see Fig. 6)
State
Carburettor on Intake
Port Baffles in ports (see Fig. 5)
Swirl ratio N/n at 1,300

r.p.m.
1 2 3 4 5 6
A {anti-swirl) P (pro-swirl) N (neutral)

P N P
L
A-S
P-S A-S
A-S A-S A-S P-S
P-S A-S
P-S
0 85
4-6
3-35
0 5-6
-l-o (reverse swirl)

6-8
7
8
N N
-2
P-S
P-S Pro-swirl baffle. A-S Anti-swirl baffle. For swirl rates of above states over the speed range, see Fig. 7. Engine Performance
EFFECT OF SWIRL ON FUEL CONSUMPTION (MINIMUM), AVAILABLE MIXTURE STRENGTH RANGE AND IGNITION ADVANCE REQUIREMENT: Full lines refer to central sparking plug and 1,300 r.p.m. Dotted lines refer to offset sparking plug.
Measurements were made at a number of swirl rates, of the following quantities:
FIG.10
65000
5-W000
-50000
Szoooo
STATE 4 CM/lON IM MIT ~ ANTI-SWIM. BAFFLES IN CNTRE FOKT3
uJIOi
-NORMAL SWM OUtttnvt

CADI* ON t.H. mi STATt 2. WITH BAFFUS ON AOI FVXT
I 2 SWIRL RATIO
3 %
I J
MOTORING SWIRL TESTS : Effect of Engine Speed on Swirl Ratio.
EFFECT OF SWIRL ON HEAT TO COOLING WATER: 1,300 r.p.m., constant total fuel consumption (17 pts. per hr.). Full lines refer to central sparking plug, dotted lines to offset plug.
: .: ..:---:.:
540C
SUPPLEMENT TO
MAY
31, 1934
FLIGHT

FIG 12.
700
(A) B.M.P. at various speeds. (B) Minimum fuel consumption and available mixture range a t 1,300 r.p.m. (c) Heat to cylinder and head water jackets at 1,300 r.p.m. (D) Detonation tendency at 1,300 r.p.m. (E) Maximum cylinder pressure and rate of combustion pressure rise (1,300 r.p.m.). Except in the detonation tests, a non-detonating fuel was used throughout.
IN
P
500
MAX/MUM
PKCSSV/I&
20|
FI6.II.
t20 z - ^
.50
S^40
KliO
a:
< a:
50
si
ISO
z
10 *
t-
B.M.e.p.
is
OCTANC -75 1 NUMSER
y
150
-.
REVERSE SWtRL
g
K0
-/fcae/rst
2 3 * SWIRL RATIOS (AT 1500R.P.M)
-2
-1
SWIRL RATIO % 2 I
EFFECT OF SWIRL ON DETONATION : 1,300 r.p.m., central sparking plug. The letters F.S.-O.M. indicate " frequent, slight, occasional, moderate detonation." The results of these tests are shown in Figs. 8, 9, 10, 11 and 12 respectively, while the figures obtained at 1,300 r.p.m. are summarised in the following table: Table 2.
Figures obtained at 1,300 r.p.m. tio
EFFECT OF SWIRL ON MAXIMUM CYLINDER PRESSURE AND RATE OF PRESSURE RISE: 1,300 r.p.m. Max. power mixture strength. Full line refers to central sparking plug, dotted line to offset plug. In the second series of tests the engine was run at 1,300 r.p.m., full throttle, and benzole added to the petrol (D.T.D. 134) to reduce the detonation to standard intensity. The " octane number " of this mixture was then measured in a C.F.R. engine by the " Aircraft Method " (900 r.p.m., shrouded inlet valve, mixture temperature 260 deg. P.). This process was repeated for various swirl rates, giving the results shown in the lower curve of Fig. 11. COMMENTS ON TEST RESULTS Series A. B.M.P./Swirl (Fig. 8).The highest output is given, at all speeds, by a swirl ratio (swirl-meter speed/engine speed) of about 2-2.5. The curves are fairly flat, on the low-swirl side of the maximum, giving an available swirl range, without serious loss of performance, of 0.5-3.5. The motoring M.E.P. was found to be independent of swirl, at any rate within the limits of engine variation (1.0 F.M.P.) so that the above conclusions apply to I.M.P. as well as B.M.P. Scries B. Fuel Consumption and Mixture Range at 1,300 r.p.m (Fig. 9).In these respects also a swirl ratio of about 2.0 gives the best results. There is little difference between the central and offset plug positions, except that the latter appears to mitigate the increase in consumption due to excessive swirl. Why this improvement should exist is not clear, especially as the mixture range and the ignition advance requirements at these high swirls are the same as with the central plug. In all these mixture range tests the ignition advance was adjusted to suit each mixture strength. The ignition advance requirements decrease steadily with increasing swirl. Series C. Jacket Losses at 1,300 r.p.m. (Fig. 10). The minimum gross loss is at zero swirl, but the minimum loss per b.h.p. is at a swirl ratio of 1, that at the optimum power swirl ratio of 2 being some 6 per cent, higher. ' Incidentally, the ratio h.p. in water/ b.h.p. is very low, being only 0.6 at a swirl ratio of 1.
5. P.
fc >'
s
cS
Max. M.E.P.
Si
QQ ^ "
lb./in.t Brake lnd.
jjl li "f , gx"

c
is
2|
-1-6
0 B
4
"ft
0-605 0-60 0-625 0-65 0-695 0-745
O '" 75-7 75-7
4-6 5-6 6-8
0-85 1-75 3-35
1 7 3 2 5
g
6A 4A 1A 7A 3A 2A 5A 8A
15!) 141-2 157-2 l i t 8 160-8 14*-5 160-5 144 160 140 156 137 153 129 145 156 155 144-8 160-8 145-5 161-5 143 159 139 155 137 153 129 145 140 139
143
Central Plug 40 0 465 20 37 0-456 16 36 0-456 18 37 0-444 18 29 0-450 17-5 28 0-458 15 13 0-470 12-5 8 0-506 12-5 37 34 32 41 29 24 10 9
lire
^s
per
620 640 615 605 585 635 660 630 600 600
40 47
75-5 75-9 76-5
40 38 34
-1-5
0
0-85 1-75 3-35

4-6 5-6 6-8
Offset Plug 0-450 2 0-464 20 0-63 0-450 25 0-625 22 0-450 0-635 0-450 20 0-66 0-452 15 0-76 0-455 12-5 0-805 0-485 12-5 _.
to
38 35 39
i
"o
37-5 35-5
All readings (except fuel consumption) at maximum power mixture strength. For maximum power mixture strength.
Detonation Tests. Series D. These were of two kinds. In the first series the engine was run throttled on a fuel of relatively low octane number and the throttle opened to give a standard intensity of detonation. This was done with various swirl conditions and the B.M.P. giving standard detonation taken as an (inverse) index to the detonation tendency. These B.M.P. figures are shown in the upper curve of Fig. 11..
540^
MAY
31, 1934
37
SUPPLEMENT TO
THE AIRCRAFT
The jacket losses are increased by offsetting the plug, but in the optimum swirl region the difference is negligible. The head losses are about 25 per cent, of the cylinder losses and vary with swirl in the same manner. This ratio does not, however, necessarily represent the proportion in which the head and cylinder receive heat from the gases, for experience indicates that there is a good heat-flow path from head to cylinder via the head-rings and sleeve. iSeries D. Detonation Tests at 1,300 r.p.rn. (Fig. 11). As stated above, tests were made by two methods, the first that of throttling, and the second that of fuel mixtures. Both methods show minimum detonation with a swirl of 1.5-2.0, or practically the same value as the optimum for power and efficiency. The total range of requirements in fuel quality is, however, not great,
ENGINEER
FLIGHT
the extreme difference being only one " octane number." Incidentally, it appears that this engine is very sensitive to the fuel quality as far as audible detonation is concerned, a difference of 0.1 octane number being distinguishable by ear, but, on the other hand, it does not suffer seriously from the secondary effects which normally render detonation dangerous. Thus the jacket heat loss is not appreciably increased by detonation, and there is no tendency to pre-ignition with any reasonable degree of detonation. These characteristics were, incidentally, also shown by a similar air-cooled sleeve-valve unit, which could run continuously with a really alarming intensity of detonation. The path of the detonation noise is affected by the swirl varying from the usual " pink " at low swirl ratios to a dull thud with high swirl. (To be concluded)
INVESTIGATION INTO THE VARIA TION OF ENGINE POWER WITH HEIGHT

By W. R.
ANDREWS,
A.F.R.Ae.S.*
(Concluded from p. 31) A similar treatment is given to the alternative form of the power law of equation 10, when / a = 0-0674, so that . . . (14) Pe = 1-0674 P (1-5 - 0 - 5 T) - 0-0674 Tables 4 and 5 give the characteristics and calculation of N \/<7 corresponding to Tables 2 and 3. The plotted value corresponding to Fig. 3 are given in Fig. 4. It is at once apparent from Table 5 that the correction for temperature quickly increases with height, and does not
Mr. Andrews is on the Technical Staff of A. V. Roe & Co., Ltd. TABLE 4
0-0674 P;. = 1-0674 P ( 1 . 5 _ o-5 T) -
Height Hfeet
a a P P.:
V!
In
Equivalent Standard Height HE
TT
IT
Xlfi xlg
= AH
Standard Temperature 0 ... 1-0 5,000 ... 0-8616 10,000 ... 0-7384 15,000 ... 0-6291 20,000 ... 0-5327 25,000 ... 0-4480 30,000 ... 0-374
10 0-836 0-6921 0-566 0-4567 0-3625 0-2821
10 1031 1-067 1-111 1167 1-236 1-325
1-219 1-257 1-301 1-355 1-423 1-507 1-615
1-0 0-9282 0-8593 07932 0-7298 0-6693 0-6115
0-915 0-838 0-763 0-6913 0-622 0-5555 0-4920
Standard Temperature + 30 0 5,000 10,000 15,000 20,000 25,000 30.000 ... ... ... ... ... ... ... 0-9055 0-7775 0-6640 0-5635 0-4754 0-3980 0-3306 0-9444 0-7804 0-6365 0-5104 0-4011 0-3065 0-2265 0-959 0-996 1-0435 1-104 11855 1-299 1-460 1-169 1-214 1-272 1-347 1-445 1-585 1-780 0-9516 0-8818 0-8149 0-7507 0-6895 0-6311 0-5750 0-887 0-8075 0-731 0-656 0-5835 0-512 0-442 1,800 7,050 12,250 17,650 22,900 28,400 34,000 1,800 2,050 2,250 2,650 2,900 3,400 4,000
Standard Temperature 30 0 5,000 10,000 15,000 20,000 25,000 30,000 ... ... ... ... ... ... ... 1116 0-9655 0-8315 0-712 0-606 0-5125 0-4305 1-0556 0-8916 0-7477 0-6216 0-5123 0-4181 0-3377 1-056 1-083 1-1125 1146 1-183 1-226 1-275 1-288 1-321 1-357 1-397 1-4425 1-495 1-555 540 e 1-0565 0-943 0-9876 0-866 0-9129 0-7955 0-8438 0-7255 0-7785 0-6595 0-7159 0-5965 0-6561 0-537 3,100 7,850 12,600 17,400 21,950 26,500 . 1,900 2,150 2,400 2.600 3#50 3,500
SUPPLEMENT TO
MAY
31, 1934
FLIGHT

TABLE 5
p, = 1-126 -y= - 0 1 2 6
Height H. feet
V~
Equivalent Standard Height HE
HR
Hg
Standard Temperature 0 5,000 10,000 15,000 20,000 25,000 30,000 1-0 0-9282 0-8593 0-7932 0-7298 0-6693 0-6115
1-0 10
0-8272 0-6759 0-5445 0-4308 0-3325 0-2488
1-042 1092 1156 1-237 1-347 1-503
1-219 1-271 1-332 1-410 1-509 1-643 1-833
0-915 0-833 0-755 0-6785 0-605 0-534 0-4635
Standard Temperature + 30 0 5,000 10,000 15,000 20,000 25,000 30,000 ... 0-9516 0-8818 0-8149 0-7507 0-6895 0-6311 0-5750 0-9448 0-7793 0-6346 0-5091 0-4000 0-3064 0-2264 0-959 0-998 1047 1107 1-188 1-299 1-460 1-170 1-218 1-276 1-350 1-449 1-584 1-780 0-887 0-8065 0-730 0-6555 0-583 0-512 0-442 1,800 6,800 11,700 16,600 21,450 26,450 31,400 1,800 1,800 1,700 1,600 1,450 1,450 1,400
Standard Temperature 30 0 5,000 10,000 15,000 20,000 25,000 30,000 10565 0-9826 0-9129 0-8438 0-7785 0-7159 0-6561 10642 0-8829 0-7253 0-5873 0-4674 0-3645 0-2762 1048 1093 1147 1-212 1-297 1-407 1-559 1-278 1-333 1-399 1-478 1-582 1-714 1-901 0-946 0-863 0-784 0-707 0-632 0-5595 0-489 3,200 8,200 13,200 18,100 23,100 28,100 1,800 1,800 1,800 1,900 1,900 1,900
TABLE 6
Column
Pe
pi 05
p i ir> T - O 5
1-126^-0-126 10731^-00731
VT
Height ' feet 40,000 50,000 60,000
VT
0 1852 0-7569 01149 0-7569 0-07125 0-7569
0-1702 0-1032 0-0624
0 1564 0 0955 0 0551
. 01136 0-0227 - 0034
0-1552 0-0677 0-0148
for N \/ a and from p to a for Vc \/ a and the mean represented by the power law given above. In Ei. & M. 1532, it was stated that to obtain the power factor it was assumed that />,. had a linear variation with density. There is no indication given as to whether there is a systeP. = 1-0731 - 00731 (15) matic variation from this linear relationship, so that no which gives as close an agreement with test results as can be comparison can be made. Fig. 5 gives the values of p, (as expected and also agrees well with the power factors deter- determined by (15)) against a. For values up to 25,000 ft., a straight line can be drawn mined from flight trials. Any modification to coveT the variation of the power which is well within 1 per cent, of the calculated values. factor with r.p.m. must be a complication which is unjustified Without a knowledge of the way in which the departure since the observed bases of reduction vary from p to p0'2 a0"8 from the linear occurs in R. & M. 1532, it is impossible to
seem at all in agreement with the results of flight test data. The most noticeable thing is that the ground level correction is practically the same for both the methods. The conclusion drawn from this part of the investigation is that the most satisfactory power law is of the form
540/
MAY
31, 1934
39
SUPPLEMENT TO

ysoooo'
'60000' 1 1 000' soooo' 000'
FLIGHT
+ n i I 11 i ii 11 1111111 ii 11! [ r~n
i i 111 i i I 11 ii 11 i: i 1111
i I M i L
VARIATION OF P e WITH <r IN A STANOARO VTMOSPHE.RE Pc -10731 ?- - 0 7 : 1 ^,00
.JL-
35 VARIATION Of N/ir WITH HEIGHT f>e -1-176 J ^ - 126
50
a 25
^
oo'
X)'
20 -
FIG.5
draw any conclusions from this curve, except that in so far as p,, is almost a linear function of a it agrees with the tests upon which R. & M'. 1532 are based. There is some evidence to show that the variation of power with height varies according to the relationship.
SUMMARIES OF AERONAUTICAL RESEARCH COMMITTEE REPORTS
These Reports are published by His Majesty's Stationery Office, London, and may be purchased directly from H.M. Stationery Office at the following addresses: Adastral House, Kingsway, W.C.2; 120, George Street, Edinburgh; York Street, Manchester; 1, St. Andrew's Crescent, Cardiff; 15, Donegall Square West, Belfast; or through any Bookseller.
ON THE EFFECTS OF VISCOUS AND SOLID FRICTION IN AIRSCREW DRIVES IN DAMPING TORSIONAL VIBRATION.
which as pointed out previously agrees closely with Diehl's power factor
p,
= p i - i s cT-o-5
In order to investigate the effect of this smaller power factor a value of / in equation (10) was chosen to agree with these power factors at about 12,500 ft. The new power factor takes the form = 1-126 ^ - 0-126 (16)
By B. C. Carter, F.R.Ae.S., M.I.Mech.E. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1557. (25 pages and 6 diagrams.) February, 1931. Price Is. 3d. net.
Torsiograph investigations made with a single-throw radiul engine fitted with a spring hub have shown that the damping inherent in the vibrating system is quite small when the hub damper is not fitted. The combination (if spring hub and damper constitutes a positive nieana of introducing damping into the system, in any required degree. In K. A.M. 1053 the author analysed the functioning of the viscous friction Lanehester damper, and made some deductions as regards the operation with solid friction. It was shown that the effectiveness of the damper, with optimum setting, is limited by the size and weight of damper flywheel that can be adopted, and by the strength of the drive. These particular limitations do not apply to the damped spring hub. What the limitations are can be estimated for any particular case by applying the results of the present analysis. Range of Investigaluytw.This report contains three main sections. Section I comprises a mathematical analysis of forced torsional vibration for a simple engine system incorporating a spring hub fitted with a viscous friction damper. Viscous frictif/n damping within the crankshaft system is taken into account. Section 11 comprises a mathematical analysis of forced torsional vibration for a simple engine system incorporating a spring hub fitted with a solid friction damper, the crankshaft being taken to be so much stiffer than the spring hub as to be treated as being rigid. Viscous friction damping within the crankshaft system is taken into account. The investigation is limited to conditions under which there is no pause at the extremes of motion before the return movement occurs. Such conditions apply for light and fairly heavy damping but not for very heavy damping. The limiting conditions beyond which pauses occur are revealed by the analysis. Section III comprises applications to some typical problems of the results of the investigations given in Section I and II. A special application of the equations derived in Section I is given whereby the action of a continuously slipping clutch in damping torsional vibrations is examined. The formula and graphs derived may be applied in making design calculations for dampers to be used in conjunction with spring hubs and couplings for aircraft engines. In general where a solid friction damper is adopted, the slipping torque should be about 80 per cent, to 100 per cent, of the amplitude of the main harmonic forcing torque to be subdued under operating conditions. TORSIOGRAPH INVESTIGATIONS ON A RADIAL ENGINE WITH AND WITHOUT A SPRING H U B , WITH SOME REFERENCE TO DAMPING. By B. C. Carter, F.R.Ae.S., N. S.
The values of N y/c have been calculated for this case. See Table 5 and Fig. 6. The values of A H show definite differences of + 30 as compared with 30 from standard. The former shows a definite drop off with height while for the latter A H is almost constant for any height. The average value of A H is 57 ft. per 1 C , which is roughly the same as that found for p,, = 1-0731 P / \ / T - 0-0731. It is concluded that the basis of reduction is dependent only upon the form of the power law and not upon the numerical value of the constants. The best form of the power law seems to be of that given in equation (10) with / = 0-0731. Table 6 on p. 38 has been compiled to show the effect of extrapolation into the Isothermal atmosphere of all the forms discussed. It is noticed that the power factor in columns 3 and 4 (of the form of equation (10) is much below that given by columns 1 and 2 which assume a power output down to zero pressure. In fact, column 3 (equation (16) ) shows that at 53,800 ft. a normally aspirated engine will not supply any external power. With column 4 (equation (15)), the height at which the power output is zero is some 10,000 ft. higher. Over the range of heights likely to be covered by unsupercharged engines the power factor of equation (15) (/ = 0-0731) seems to be the most justifiable. Obviously the question of the power factor for supercharged engines requires separate treatment, as that determined for unsupercharged units suggests that flight is impossible above 60,000 ft. for any engine. It is also conceivable that the form of the power factor will vary with compression ratio, but these questions must be left until further data are available
Muir, B.Sc, and H. Constant, M.A. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1562. (14 pages and 18 diagrams.) Price Is. 6d. net.
The spring hubs used in these investigations were made as an outcome of previous torsional vibration research. Torsiograph tests were taken partly to examine the effects of the torque fluctuation in the airscrew drive of fitting & spring hub, both with and without a damper incorporated, and partly to make use of the spring hub in determining the amount of damping present in the engine airscrew system.
540
SUPPLEMENT TO
FLIGHT
MAY
31, 1934

FLEXTJRAL AND SHEAK DEFLECTIONS OF METAL SPARS.
The observed main critical speeds with spring hubs fitted agree reasonably Closely with the calculated values. The minor criticals occur at speeds which agree closely with the critical speeds calculated from the main critical speeds. The effect of fitting an engine with a spring hub incorporating a suitable damper is to remove practically all trace of torque fluctuation in the airscrew shaft at all speeds. The provision of stops does not serve nearly so well as the provision of a damper, and it is doubtful whether stops are needed when a damper is fitted. If fitted, they should not be arranged to come into contact with spring arms, but should operate between rigid portions on the driving and driven sides of the springs, and the clearance given should be such as to prevent the springs from being over stressed through accidental circumstances. The degree of damping indicated by theory as being requisite is effective and It can be obtained with a solid friction cone damper of moderate size, loaded by airscrew thrust. FURTHER EXPERIMENTS ON A MODEL FAIRET III.F SEAPLANE. By A. S. Batson, B.Sc, and A. G. Gadd.
(Part I) By I. J. Gerard, M.Sc., A.M.I.C.E., and H. Boden, B.Sc. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1567. (11 pages and 20 diagrams.) September, 1933. Price Is. 6d. net.
The research of which this is the preliminary stage aims at producing a simple procedure for estimating the ultimate strength of metal spars. The method at present in use is to deduce from a test to destruction of a relatively short sample an apparent failing stress. This stress is then assumed to be characteristic of that spar under loading conditions not confined to those under which the sample spar was tested. This method can give rise to considerable errors and a more satisfactory procedure is urgently required. Five typical spars, supplied by aircraft firms have been tested within the elastic limit as simply supported beams carrying one or more concentrated lateral loads. No axial loads have been included in any of the tests recorded in this report. The slope and deflection corresponding to a range of lateral loads have been measured. Some tests made in 1928 on the shear deflection of metal spars have also been included in an appendix to the present report. The apparent values of El deduced from the measured slopes and deflections over portions of the spar loaded in pure bending (i.e., with no shear) have been found to be independent of the applied load, indicating that within the limited range considered the simple theory of bending is applicable.
ON THE CALCULATION or THE CRITICAL REVERSAL
R. & M. No. 1564. (6 pages and 13 diagrams.) August 9, 1932. Price 6d. net.
The experiments herein described follow on those given in R.& M. 1356,* which provides data on the spinning properties of a typical float seaplane. The same l/15th scale model of the Fairey IIIF. seaplane was used. The range of incidence over which rolling and yawing moments due to rolling on model, were measured in R. & M. 1356* was extended to 69-8. This included a determination of fin and rudder (0) rolling moment and control due to rudder set to + 32-5. As at 60-9 incidence, fin and rudder moment was again positive (i.e., helping rotation). Curves of rolling and yawing momenta due to rolling measured on wing alone (incidence 22-4" to 60-9), showed marked similarity to those for a R.A.F. 15 biplane, gap Chord, stagger 0. Elevator hinge moments were measured at 0 and 30 yaw over a range Of Incidence approximately 10 to 60 for elevator setting 35 to + 20. Hinge moment was not altered greatly by angle of yaw of the model and with elevators set to trim, was such as to exert a small force on the control column towards the pilots, which decreased as incidence increased. R. & M. 1356. Spinning of a model of the Fairey IIIF. Seaplane. H. B. Irving and A. S. Batson. THE ELASTIC INSTABILITY OF A THIN CURVED PANEL SUBJECTED TO AN AXIAL THRUST, ITS AXIAL AND CLRCUMMKENTIAL EDGES BEING SIMPLY SUPPORTED. By S. C.
SPEEDS OF WINGS.
By D. M. Hirst, M.A.
W I T H AN
by
APPENDIX ON ROLLING MOMENT INDUCTION FACTORS
G. R. Brooke, Nat. Dipl., and D. M. Hirst, M.A. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1568. (24 pages and 4 diagrams.) September 20, 1933. Price Is. 3d. net.
In R. & M. 1506* and 1490,t Roxbee Cox and Pugsley have developed the general theory of the loss of lateral control due to wing twisting for a scmirigidj wing, and experiments and calculation on three elastic wings of rectangular plan form have given results in good agreement with this theory. The present report describes investigations into the influence of wing and aileron shape and twist distribution on the critical reversal speed, on the basis of this theory. In addition, certain simplifying assumptions v hic-h were introduced into the general theory of the semi-rigid wing are here discussed in detail, and a further verification of the usefulness of the semi-rigid theory provided by a calculation on an elastic wing differing in shape and elastic qualities from those previously examined. Finally, a procedure is given for rapidly estimating the reversal speed of a normal wing, and is suitable for design use. It appears that the most important of the four variables in the analysis, is wing plan form ; increase of wing taper considerably increases the reversal coefficient. The prediction of the semi-rigid theory is shown to be reasonably accurate, and this result, following similar comparisons for rectangular wings may be considered to justify the use of that theory for any normal wing. " Theory of loss of lateral control due to wing twisting."H. Roxbee Cox and A. G. Pugsley (R. & M. 1506, 1932). t "The aerodynamic characteristics of a semi-rigid wing relevant to the problem of loss of lateral control due to wing twisting."A. G. Pugsley (R. & M. 1490,1932). t A semi-rigid wing is denned as one having arbitrarily fixed and unrelated distributions of wing twist and aileron angle, both distribution* being independent of aerodynamic loading. THE N.P.L. OPEN-JET WIND TUNNEL. By A. R. Collar, B.A., B.Sc. R. & M. No. 1569. (17 pages and 10 diagrams.) October 19, 1933. Price Is. net. The present report deals mainly with preliminary experiments on models, which were carried out to ensure that the projected tunnels should give the beBt possible aerodynamic performance. The design finally reached can be briefly described as follows : The tunnel is of the open-jet type, with an elliptical nozzle, of which the major axiB is horizontal; it has two return ducts, passing one above and one below the Jet. At each of the right-angle corners of the duets there is a cascade of guide vanes. Immediately before reaching the open working section of the tunnel, the air passes through a contracting channel, which has a square section at the larger end and an elliptical one at the jet, the ratio of the areas being 3-9:1. After passing across the open working section, the air is received In a collector channel of approximately conical form, whose section varies from elliptical to circular. The airscrew is four-bladed, and is situated at the enlarged circular end of the collector cone. Radial blades cf aerofoil section are provided in order to neutralise the rotation which would be imparted to the air by the screw. The power factor of the model was subject to a large scale effect; at the jet speed mainly used during the experimental work, namely, 50 ft./sec, the power factor was 1-8. The variation with Reynolds number indicated that a full-scale power factor of about 2-6 might be expected. The full-scale tunnel, which has been completed, has equalled expectations. The distribution of velocity in the jet is as good as those of most other tunnels of this type, and the power factor has the predicted value of 2-6. The elliptical nozzle of the tunnel has a horizontal major axis measuring 9 ft. 1 in. and a minor axis of 7 ft. 0 in., and an input of 375 b.h.p. at the airscrew yields an air speed of about 210 ft./sec. in the jet.
AN IMPROVED MULTITUBE TILTING MANOMETER. By
Redshaw, M.Sc, A.F.R.Ae.S. R. & M. No. 1565. (15 pages.) May 1, 1933. Price 9d. net.
The stability of a thin circular tube has been dealt with by numerous investigators, and it is of interest to observe that the results obtained by Professors Timoshenko and Prescott, who applied the Rajleigh Ritz Minimum Energy Method, agree very closely with the more rigorous treatment of Professor Southwell. In the analysis Prescott's notation is adopted. Some experimental investigations are analysed and compared with the analytical results obtained by the application of the theory. A reasonable agreement was obtained for most of the experimental results examined. In each case the predicted critical stress was lower than the actual failing stress which would appear to be reasonable, as a panel will generally support an increase of load beyond the critical load. No test data is available for panels which approximate to flat plates, and further test data is deemed desirable. WIND TUNNEL INTERFERENCE ON WINGS, BODIES AND AIRSCREWS. By H. Glauert, F.R.S. Communicated by
the Director of Scientific Research, Air Ministry. R. & M. No. 1566. (75 pages and 34 diagrams.) September 13th, 1933. Price 4s. 6d. net.
This report provides a comprehensive survey of the subject of wind tunnel interference on wings, bodies and airscrews. The limited extent of the artificial stream of a wind tunnel, bounded either by the rigid walls of a closed type of wind tunnel or by the free surface of an open jet, inevitably leads to some constraint of the flow and to some interference on the behaviour of a model tested in the wind tunnel. This interference could be minimised by using very small models but it is desirable for many reasons that the model should be as large as possible. The study of wind tunnel interference Is therefore of great importance since some interference is inevitable, and since an accurate knowledge of this interference will justify the use of larger models than would otherwise be permissible. The general nature of the interference can be appreciated most readily by considering the conditions in a closed tunnel. If a large body is placed in the stream, the first and most obvious constraint imposed by the rigid walls of the tunnel is that the stream is unable to expand laterally as freely as it would in an unlimited fluid, and in consequence that the velocity of flow past the body is increased, leading to an intensification of the forces experienced by the body. Another choking constraint of a different character arises if there is a wake of reduced or increased velocity behind the body, as occurs respectively with a blunt body or an airscrew. The necessity of maintaining continuity of flow in the tunnel then implies that the velocity and pressure of the stream surrounding the wake will differ from the undisturbed values far in front of the body, and this change of pressure reacts back to cause a change in the force experienced by the body. The interference experienced by a lifting body, such as a wing, is of a different character. The lift of a wing is associated with a general downward movement of the air behind the wing, and the constraint of the tunnel walls on this downwash modifies the behaviour and aerodynamic characteristics of the wing. Finally, a fourth type of interference occurs if there is a gradient of static pressure down the stream of the wind tunnel. This pressure gradient arises owing to the development of the factional boundary layer of reduced velocity along the walls of the tunnel, which leads to an increase of velocity and a decrease of pressure along the axis of the tunnel. Any body is therefore tested in a slightly convergent stream, and experiences an increased drag owing to the drop of static pressure from nose to tail. These various interference effects in a closed tunnel or in a free jet are discussed in detail for different types of body, which are grouped conveniently under the headings of wings, symmetrical bodies and airscrews. The basis of the theoretical treatment of the subject is examined critically and the method of analysing particular problems is explained in detail. Experimental results are quoted to justify the theoretical formulae or to derive empirical values to complete the theoretical analysis, and the results required for the practical application of the correction formulae are given in suitable tables and figures. A full list of references Is appended to the report.
R . W a r d e n , P h . D . , M . E n g . , A . M . I . M e c h . E . R . & M. No. 1572. (6 pages and 3 diagrams.) November 23, 1933. P r i c e 6d. n e t . A new model multitube tilting manometer designed to overcome the disadvantages of the older models is described. Precautions were taken in the design and manufacture of the instrument to eliminate any danger of trouble due to distortion. The table assemblage rotates about a pin and Is roughly balanced about its axis of rotation. The measuring tubes are matched and calibration factors for individual tubes are not required. A method of setting the table is described whereby the actual observations are the values of the standard coefficients (P/j>V2). A system of " indirect " lighting which gives a bright meniscus and scale against a dark background and eliminates parallax errors and eyestrain is incorporated. A sensitivity and accuracy equivalent to that of a 26-inch Chattoek gauge has been achieved.
June 21, 1934

No. 101 ( V O l ^ e , , I X ) 9th Year June 21, 1934
CONTENTS
Atomic Hydrogen Welding The Effect of Swirl on Petrol Engine Combustion The Polar Diagram Technical Literature In the Drawing Office Page 41 43 44 48
ATOMIC HYDROGEN WELDING

BY C. R. DEGLON, B . S C *
THE atomic hydrogen process of welding is probably one which is not very well known in that it is a comparatively new addition to the many processes already available to the welding side of the Engineering Industry. In view of this fact it is proposed to outline briefly the history of the process and to give a description of the principles involved. The origin of the atomic hydrogen process is due to Langmuir, whose early experimental work showed that when tungsten was heated in molecular hydrogen the heat loss was roughly proportional to the square of the absolute temperature. This was found to apply for comparatively low temperature only, and the heat loss increased at higher temperatures until it was proportional to the fifth power when temperatures approaching 3,000 deg. C. were reached. The reason for the sudden increase in the heat loss from the tungsten was found to be due to the dissociation of molecular hydrogen into its atomic state. In order to perform this dissocia Mr Dcglon is'onTthe Technica^Staff of Metropolitan-Vickers Electrical Co.. Ltd., of Trafford Park, Manchester.
Fig. 2. a considerable amount of energy is required, and the most convenient way of obtaining this energy is by means of the electric arc.
Fig. 1.
Fig. 3.
608 a
42
SlJPPLEMBNT TO
FLIGHT
JUNE 21,
1934
Fig. 4, Atomic hydrogen welding consists in maintaining a continuous independent A.C. arc between two tungsten electrodes, around which is blown a stream of hydrogen. Due to the catalytic action of the tungsten the energy of the arc changes the hydrogen from its molecular to its atomic state. The atomic hydrogen rapidly diffuses from the region of the arc and recombines some little distance from it into the molecular hydrogen again. These conditions can be seen in Fig. 1, which shows the atomic flame. In changing from the atomic to the molecuar state the energy originally supplied is given up in the form of heat. This atomic flame is surrounded by ordinary hydrogen burning in the usual way, and thus welding takes pace in a hot flame of a single gas, which means that the weld will be taking place in an actively reducing atmosphere which is free from the contamination of atmospheric nitrogen and oxygen, the adverse effects of which are well known. The constituents of the material being welded are not, however, immune from attack because the hydrogen
Fig. 5. has a great affinity for carbon, and consequently there is always a reduction in the carbon content of the material being welded. This loss of carbon can, of course, always be made good by utilising a filler wire containing about 50 per cent, more carbon than the material being welded. By this means the deposit will be of approximately the same analysis as the parent metal, because no other constituents of the steel are attacked. Welding by the atomic hydrogen process (see Fig. 2) is similar to most flame welding processes, with the exception that the metal is considerably more fluid and the atomic flame is not so rigid as other flame methods. To the inexperienced this may at first present difficulties due to the operator not being able to control the metal, but proficiency can easily be obtained with a little practice. In Fig. 3 is shown a 35-ampere atomic hydrogen welding set, whilst the welding torch itself can be seen in Fig. 4. With a set of this type it is possible to
43
JUNE 21, 1934 SUPPLEMENT TO

weld most of the non-ferrous metals and alloys, and exceptionally good results can he obtained when welding rustless steels. The advantage of the process will br readily appreciated when it is pointed out that th:deposit is of precisely the same composition as the parent metal, so that no deterioration of the stainless properties will be apparent at the weld. The extreme ductility of the metal deposited by the atomic hydrogen process renders it eminently suitable for all types of welding, and the fusion of thin material is no exception. All classes of joints can be made, but perhaps the easiest method of welding very thin plate is to use the edge weld. By this means a considerable amount of the inevitable distortion which is produced in thin plates can be eliminated, and the finished product will have a very pleasing appearance. The atomic hydrogen process has not only been developed for hand welding, but is also suitable for automatic operation. Fig. 5 shows a completely automatic atomic hydrogen welding machine utilised for fabricating transformer radiators. This machine is used for edge-welding two pieces of .031 steel strip together. The clamping mechanism can be seen in Pig. 6, and in this connection it should be noted that in the welding of low carbon steels it is essential to keep the material hot sufficiently long to allow the dissolved hydrogen to get away, otherwise blow-holes will be formed, and that is why the clamping jaws are placed some little distance from the seam. If they are too near, chillingwith consequent blow-holesresults. In the welding of high carbon steels this effect is not so apparent because the hydrogen is attacked by the carbon in the steel, and is thus not absorbed to the same extent.
FLIGHT
Thus, in State 4, the running is much rougher than in State 1, although the pressure rise rate is lower. The bamd widths, however, agree well with the changes in engine behaviour, and this suggests that the roughuess is due to inequality between successive cycles, though the fact that the dynamometer readings were steady shows that the variations must average out over quite short periods. The audible roughness was observed to be uneven, in contrast to the regular " thudding " heard with an over-turbulent side valve head (which usually gives a narrow rising-pressure band). More difficult to account for is the combination, in the high-swirl condition, of a very regular diagram, which presumably means a constant I.M.P. with the observed irregularity of torque. It is possible that the high heat flow in this condition caused a slight " drying-up " with consequent variations in friction (which would not, of course, be revealed by a motoring test). Swirl Control The moral of these experiments is that in a petrol engine the swirl ratio should not exceed a value of about 3, but that, if it is below this figure, the exact value does not matter much. Since the permissible range of swirl ratio is so wide, an accurate quantitative basis for design is unnecessary, and it will be sufficient to discuss qualitatively the factors on which the swirl ratio depends. Engine Size and Speed Dimensional reasoning shows that, if the effects of air viscosity and elasticity are ignored, engines of exactly similar proportions will have the same swirl ratio, whatever their size wr speed. This exact similarity, however, cannot apply except in engines desigmed to run at the same piston speed, for of two engines, one designed for high piston speed and the other for low, the high-speed unit must have the larger inlet ports, and thus, at any given speed, the inlet velocity, and the swirl produced thereby, will be less. For otherwise similar engines designed to give the same inlet gas velocity at their working speeds, the swirl ratio will, roughly, be inversely proportional to the piston speed. For the experimental engine above described the torque curve peaks at 1,300 r.p.m., the corresponding piston speed being 1,520 ft. per min., and the inlet gas velocity 162 ft./sec. At this speed the swirl ratio in the " neutral " condition (State 3), that is, without any pro-swirl or anti-swirl arrangements, is 3.3, a figure which is slightly beyond the desirable limit. Thus one may say that for engines designed to give their peak torque at piston speeds below 1,500 ft. per min., anti-swirl precautions will probably be necessary, and they are generally desirable where the peak-torque piston speed is less than 2,000 ft. per min. to give a margin of safety to cover errors in design or manufacture. Viscosity and elasticity effects are very complicated, as can be seen by the variations of swirl ratio with speed shown in Fig. 7, but generally the swirl ratio, if high enough to matter, is near its maximum at the peak-torque speed, and for the present purpose one can, therefore, neglect these secondary effects. Bore-Stroke Ratio For a given inlet velocity the swirl is roughly in inverse ratio to the bore, and a short-stroke engine will therefore have a lower swirl than the long-stroke type. The effect is, however, small within the limits of stroke-bore ratio which are practicable for aero engines. Combustion-Chamber Diameter When a rotating mass of gas is reduced in diameter, its rotational velocity is. increased. Theoretically, the swirl should vary inversely as the square of the radius
THE EFFECT OF SWIRL ON PETROL ENGINE COMBUSTION

BY J. F. ALCOCK
(Concluded from p. 31) Series E. Maximum Cylinder Pressure and Hate, of Combustion Pressure Rise (Fig. 12).The maximum pressure is highest at a swirl of about 1.3, but the highest and lowest figures are only about 50 lb. per sq. in. apart, which is not a serious matter. The pressure is slightly higher with the offset plug. With the central plug the rate of pressure rise varies considerably with the swirl, being greatest at a swirl rate of about 1.0, but with the offset plug it remains practically constant over the swirl range. It was observed, however, that the variations in smoothness of running did not correspond with those of the pressure rise, a better indication being the width of the rising-pressure band on the diaphragm, a wide " fuzzy " band corresponding to rough running. The following table illustrates this point: Width of rising Bate Diagram P.R. 2 band X (see lb./in. degr. Fig. 13) X degr. crank angle
A 40 47 40 36 112 61 7-2 4
State
Swirl
Engine running
4 1 3 5
0-85 3-35
5-6
B
C
Very rough Moderately rough Smooth (but irregular)
All with central plug. The above diagrams are given in Fig. 13. 608 c
44
SUPPLEMENT TO
FLIGHT
JUNE 21,
1934

TO C TDC
A
. .
"
,;
: ' . '
roc
"';. ;'?
'':
* '.'''.
* v
'
'
'
FIG.13
State 4. # / = 0. State 1, # / = 0-85. State 3, A7/ = 3-35. State 5, J\T/ = 5-6 SWIRL INVESTIGATION: Full-throttle Indicator Diagrams with different Swirl Rates. Speed 1300 rpm. Large dots give T.D.C. and atmospheric lines. The vertical spacing is one dot per 100 lb/in2, the horizontal spacing one dot per 10 deg. crank. of gyration of the combustion chamber cross-section, since the moment of momentum is constant, but in practice the variation is somewhat less than this owing to friction and other factors. I t is desirable, therefore, for the combustion chamber to be the full diameter of the cylinder and spherical. And conical sections and others of small gyration radius should be avoided. Inlet Port Timing A late inlet opening increases the swirl, since the inflow velocity is increased during the early part of the suction stroke, when the obliquity of inflow is greatest. The inlet ports, therefore, should open as early as possible. Inlet Port Depth As explained above, the swirl is produced by the one-sided fairing of the inlet opening by the cylinder port wall. The deeper the port, therefore, the greater the swirl, as shown in Fig. 14, and it follows that the inlet ports should be made as shallow as possible. This cases " blowing " tests on a model form useful, if rough, guides to design. Swirl in Poppet Valve Engines There is no inherent swirl tendency in poppet-valve engines, but a certain amount is produced in overheadvalve engines with inlet bends directed tangentially to the cylinder axis. This is usually too slight to do any harm, but in a few cases excessive swirl has been traced to this cause. The remedy is obviously to modify the inlet porting so as to give a more nearly radial flow. Summary (1) Excessive swirl rarely occurs in poppet-valve engines, but is liable to do so in single-sleeve valve engines, especially low-speed engines. (2) The swirl ratio should not exceed a value of about 3, but, if below this, its exact value matters little. (3) The symptoms of " over-swirl " are loss of power and thermal efficiency, low-pitched detonation noise, small ignition advance requirement and smooth running. The combination of the last two features distinguishes over-swirl from over-turbulence, in which quick ignition is combined with rough running. (4) Factors tending to increase swirl and therefore to be avoided, are: Pro-swirl induction-belt flow, small diameter of combustion chamber, late inlet opening, deep inlet ports. (5) If avoidance of the above pro-swirl factors still leaves the swirl too high, the remedy is anti-swirl induction-belt flow or anti-swirl baffles. Conclusion The writer would like to thank the Air Ministry, on whose behalf these experiments were made, and Ricardo & Co. Engineers (1927), Ltd., a t whose experimental works they were carried out, for permission to publish the information given herein.
FIG 14.
bort depth ratio
SHALLOW PORT .APPROACH Air approaches or idee from both sides inflow obliquity sitohi
OCEP PORT APPROACH Air approaches orifice from one side only, inflow
Effect of Inlet Port Depth on Swirl Production. is particularly difficult to ensure in multi-cylinder blocks, owing to the possibility of core displacements, and in such cases, therefore, anti-swirl precautions should always be taken. Shape of Induction Passage As shown in Fig. 14, a considerable degree of swirl control can be effected by varying the direction of flow in the induction belt. Where possible the flow in the belt should be given an anti-swirl bias by suitably locating the inlet to the belt. In multi-cylinder blocks, however, this is often difficult for mechanical reasons, particularly where the induction belt is common to several cylinders, and in such cases one can with advantage use short baffles near one or more inlet ports in each cylinder, giving the air an anti-swirl bias. In such
THE POLAR DIAGRAM

THE ELEMENTS OF A METHOD OF STRESSING BEAMS UNDER COMPRESSIVE END LOAD.
By EDGAR H. ATKIN.*
AMONG the methods devised for calculating the strength of beams under compressive end load, one of the most elegant, and certainly one of the simplest, is the method of the polar diagram. This method was first described by its discoverer, Mr. Howard, in the Aeronautical Research Committee's Report and Memorandum No. 1233. As there expounded, however, the presentation is more suitable for the mathematician than for the practical engineer, hence there is a
Mr. Atkin is on the Technical Staff of Boulton & Paul, Ltd.
608 d
45
JUNE 21, 1934 SUPPLEMENT TO

need for a more practical account of the method. This paper is intended to be such an outline. Although the chief applications of the method are to be found in the design of Aircraft Structures, a sufficient number of cases arise in general engineering to make it worthy of study by others who are not aeronautical men. Tke great advantage of the polar diagram is that it replaces complex and laborious mathematical analysis by simple graphical constructions. During research on beams under combined lateral and compressive end load, it was recognised that if the bending moment diagrams previously plotted on a straight base were plotted as polar diagrams about a pole, great simplifications in the nature of the curves involved would result. The complicated functions of sines and cosines represented graphically on a straight base by sinuous curves could be represented in a polar diagram by arcs of circles amenable to elementary geometrical methods. In the first place, we must show how to define a point on a beam by an angle. This is done quite simply as follows : Take some arbitrary point on the beam, the midpoint for example, and assume some constant fi. To start with, /J. will be assumed to be the same for every point on the beam. Then, if x is the distance from the assumed point to any other point which has to be specified, fix is the angle to this point, and if 2a is the length of the beam the angle 2 \M represents the length 2a. Fig. 1 explains this. the bay a must always be less than (or 90).
FLIGHT
In the case of single bays where I is constant throughout We can proceed now to the various cases. CASE I.
Beam with end moments MA, MB and end load P ; no lateral load.
First calculate p and a. Set out the angle AOB equal to 2a. (See Fig. 2a). Along OA and OB measure off to some scale MA and MB respectively. The sign convention for moments is indicated in Fig. 1. In Fig. 2a, both moments are assumed positive. Through A and B draw perpendiculars to OA and OB respectively to meet in Y. Join OY. On OY as diameter draw the circle OAYB.
<
A
Y
j i
Polar Diagram for Beam subjected to End Load and End Moments only.
7A
III
FIG.I
1/ A >\
1
/
The point Y at the end of the diameter OY will be referred to as the vertex of the circle. The bisector OX of the angle AOB may be inserted to represent the mid-point of the beam. From this line, angles can be measured to any other point on the spar. The radius vector at any given angle to OX drawn from O to the circle OAYB equals, to the scale of the diagram, the bending moment at the corresponding point of the beam. Fig. 2b shows the form of the diagram when MA is positive and MB negative.
CASE II.
MOMENTS POSITIVE IN THIS DIRECTION
Sign Convention. Take any line OX with 0 as pole. This represents the arbitrary origin {e.g., mid-point of beam). To each side of OX draw the angles AOX and BOX. OA and OB represent the ends A and B of the beam, and any other radius vector OY at an angle fix with OX, represents the point Y distant x from X on the beam. Distances and angles to the right of OX are considered positive ; those to the left negative. The following notation will be used : 2a = length of beam. I = moment of inertia of beam. P = compressive end load. w = distributed load per unit run. W = concentrated load. E = Young's Modulus.
/
Beam, with or without end moments MA and MB, with end load P, and uniform distributed load w. This case must be introduced by a more complete sign convention than was required for Case I. Sign Convention.Distance x along beam positive to the right. Deflection y of any point on beam positive upwards. Slope i of beam positive if upwards to the right.
POSITIVE EXTERNAL MOMENT
POSITIVE EXTERNM MOMENT
FIG.3
MB
tt t t m t t 1 1 tiT
' UT/unil- run
a = ix a (called angle of beam). MA, MB = end moments.
Sign Convention.
46
SUPPLEMENT TO
JUSE21, 1934
FLIGHT

CASE III. Beam subjected to compressive end load, a concentrated lateral load W, and with or without end moments. W Determine JX, a, and the quantity . Commence the diagram as before, and put in the radius vector OC corresponding to the point at which W is applied. (See Fig. 6.) Measure off the overhang moments MA and MB and insert the perpendiculars AXi and T3X2.
External Moment M positive when tending to bend beam concave upwards. Distributed load w positive if directed upwards. Concentrated Loads W positive when directed upwards. Sut note that is negative when w is positive, and positive when w is negative. Fig. 3 shows the system of signs diagrammatically. We may now continue with our problem. Assume that MA, MB and w are positive. First calculate w H, a, and the quantity. Let out the angle AOB as before, and with 0 as centre w strike an arc of radius ; as this quantity is negative, the
M 6 -673O IBS.INS.
MA-5500 LBS.INS.
P 7500 LBS.
TTttttTTTTTTt
uT - 8 5 LBS/lNS. RUN
P-7500 LBS.
90 ins.-
eve moment
- v e momertf
FIG.49
FIG.4b..
Beam with End Moments, Distributed Lateral Load and End Load. arc will line in the angle equal but opposite to AOB (see Fig. 4). The arc KK, will be the base line from which to measure the bending moments. MA and MB are positive, therefore we set off KA, and K,B/ along OA and OB equal to MA and MB in the positive direction from K and K,. This is shown clearly in Fig. 4. Through A' and B ' draw perpendiculars to OA and OB meeting in X. On OX as diameter describe the circle OA'XB'; the shaded area is then the bending moment diagram. Fig. 4b shows the same diagram drawn for negative moments. In this case the bending moment is everywhere negative. Example (1). A steel beam 90 in. long, moment of inertia 0-3313 in.4, is subjected to a compressive end load of 7,500 lb. and positive bending moments of 5,500 lb.-in. and 6,730 lb.-in. at each end. In addition, there is positive lateral load of 8-5 lb.-in. run. 'Determine the maximum bending moment. The loading system is as sketched in Fig. 5.
B.M. MAX.4-530 LBS INS.
Diagram for Example No. 1. Through any point D on AX t draw a line perpendicular W to the dividing radius vector OC. being positive, measure off to the left DE equal to this quantity, and through E parallel to AXX draw EX 2 to cut BX 2 in X 2 . On OX! and OXs as diameters draw the arcs BF and FA. The shaded area is then the polar bending moment diagram. It will be well here to introduce the reader to the method of obtaining the " true shear " at any point on the beam, and also to define positive and negative shear. true bending moment\ The " true shear," is defined as d
da-
2 _ Z. _
* *~ = H = 2a =
7 500
'
El ~ 0-3313 x 3 x 10' 0-000756 in. 2 0-0275 in. x :. 0-0275 X 90 X 57-3 = 141 -T 8-5 = 11,240 lb.-in. 0-000756
The resulting diagram is shown in Fig. 5 and should be self explanatory. Max. Bending Moment = 4,530 lb.-in. and occurs at a point corresponding to 72-8 from the end A on the diagram. The actual distance x is given by :
o 72-8
57-3 X /* 57-3 x 0-0275 = 46-2 in. from A. We proceed now to a slightly more difficult problem.
while the " apparent shear" is positive if the resultant external lateral load to the right of a section is downwards (upwards to the left of a section). Take the diagram of Fig. 6, and imagine that the shear is required at Y. If then we measure YX 2 to the scale of the diagram and multiply by JX we shall have the " true shear " at Y. If, looking along the radius vector concerned (e.^., OY) in the positive direction, the distance to the corresponding vertex (e.g., X 2 ) is measured to the right the shear is positive ; if to the left it is negative. The direction of DE must always be decided by the sign convention for true shear. In addition it should be checked by a consideration of the general nature of the problem, from a common sense point of view. For instance, it can be seen at once in the present case whether W will increase or decrease the bending moment due to the end moments : the relative displacement of Xj and X 2 must therefore be such that the bay moment is decreased when W is positive.
608/
47
J U N E 21, 1934 SUPPLEMENT TO

M A - 5500 LBS. IN&
FLIGHT
M-6730 LBS. INS
P-7500LBS.A
P.7500 LBS.
Beam with Concentrated Load, End Moments and End Loads. End Moments and Lateral Load are shown in their positive Directions. The term " true shear " is used to distinguish it from the " apparent shear " which may be denned as the shear neglecting the end load, i.e., the shear due to the lateral loads and end moments only. Fig. 7 shows a small part of a beam at an exaggerated inclination to its initial position. The components of the apparent shear S1 and the end load P tangential to the cross section XX are S1 cos i and P sin i. The angle i is small, and therefore in accordance with the usual assumptions of the beam theory we may put Cos i 1 and sin i i. The true shear S is then given by the equation
Diagram for Example No. 2. From example No. 1 p = 0-0275 in. and . . ; - = 1,455 lb.-in. 0 0275 al> a for whole bay = 141-7 -. a for 30 in. - X 141-7 - 47-2
40
S1 = S + (P
Example (2). The beam of example (1) has the same end load and end moments, but instead of a distributed lateral load there is a single negative concentrated load of 40 lb. 30 in. from the end at which the moment is 5,500 lb.-in. Determine the maximum bending moment.
S'APPARENT SHEAR
Having obtained the requisite data, we can now proceed with the diagram according to the construction given above. (See Fig. 8a). Set out the bounding lines OA and OB to include 141-7 and draw OC at 47-2 to OA. Measure the end moments along OA and OB and draw perpendicular to these lines through the points EO obtained. Through any point D on the perpendicular to OA draw DE, to the right (W will evidently increase the bending W momentB in the bay) at right angles to OC, equal to and through E draw EX 2 parallel to AD to meet the perpendicular BXj to OB in X 8 . Through X 2 draw X,X 2 parallel to DE to meet AD in X t . Then X , and X 2 are the vertices required, and the arcs OAZ and OBZ of the circlet) on OX, and OX 2 respectively as diameters complete the diagram. It remains to draw the modified diagram for the case when the load W is in the opposite; (i.e., positive) direction. The construction proceeds on exactly the same lines but D E is drawn in the left instead of to the right. This is shown in Fig. 8 by the dotted circles on the diameters OX^ and OX, 1 . (To be continued.)
-i SLOPE
RELATION BETWEEN TRUE AND APPARENT SHEAR: True Shear on section xx = S' iP, or S' = S + iP.
48
SUPPLEMENT TO
FLIGHT
JUNE 21,
1934
IN THE DRAWING OFFICE

MONOPLANE WING LAYOUT
BY H. H. HUTCHISON.
THE ensuing method of laying out a monoplane wing is applicable particularly where it may be found necessary for reasons of expediency to vary the proportionality of the rib contours. In other words, the locus of the C.P. is made to depart from the conventional path, as in the case of a normal two-spar wing. In the case I am dealing with two spars are shown, but, of course, the method of laying out applies equally as well for a monospar. The wing dealt with is straight in front elevation from left-hand wing tip, along the tops of the ribs to right-hand wing tip. Fig. 1 is an outline in plan indicating the rib spacing. If the positions of the spars are already fixed up, dimensions for the nose and trailing edge can be given from these, otherwise a datum line is necessary.
Next mark off on the chord lines 0.0125, 0.025, 0.05, 0.075, 0.10, 0.15, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90 and 0.95 of the respective chords and draw perpendicular lines, cutting the contour. Scale these vertical dimensions, enter them in a table, and prepare a series of foreshortened viewsone of which is Fig. 4in order to use the dimensions thus foimd.
-1 -075
05
025 0125
H pzJ :

J
1
- - _ _-
A
X
e >
0125
_-
'
*^= fc=2
.i
iO
"
1\
025
j-
as i-^g 075 t = ? =
FI6.4
1 . 1
f T .,
1 j 1
The dimensions for the chord, nose to centre line of front spar and trailing edge to centre line of rear spar can be found readily from Fig. 1 by calculation, or by the following graphical method. For simplicity sake it is assumed that the rib spacing is 12 in., and that the spar axes are parallel in plan view. Draw a horizontal line representing the spar centre lines coincident. Draw parallel lines to represent the ribs perpendicular to this line at 1 in- apart.
At line 1 measure up, full size, dimension " b " from Fig. 1 above the horizontal line, this being the amount "the nose is ahead of the front spar at rib 1. Below the horizontal line measure off dimension " C," again at line 1, i.e., the amount the trailing edge is behind the rear spar at rib 1. Proceed in the same way at line 15, using dimension " a " above, and " e " below, Fig. 1. By joining these points all the other rib lengths + the constant can be measured off full size, due allowance to be made at the end ribs when detailing for the wingtip curve. Having decided the wing section, superpose in end elevation the rib contour of the smallest rib, neglecting end ribs, on the rib contour of the largest rib, placing the chord lines at that distance apart required to give the correct dihedral, dimension " x , " Figs. 3 and 4.
Start by drawing twelve vertical lines 1 in. apart, Fig. 4, and a horizontal line across them. From the horizontal line measure up line " 12 " full scale the amount the small rib chord line is above the large rib chord line, dimension " x," Figs. 3 and 4. Join this point to the point of intersection of line 1 and the horizontal lime. The intersections of this sloped line with the vertical lines are therefore the chord lines of all the other ribs between "1" and " 1 2 . " Above and below points " A " and " B " measure off full size the dimensions of the ordinates of the contour of ribs " 1 " and " 12 " at the various stations, i.e., 0.0125, 0.025, etc., taken from Fig. 3, and join up. Thus the ordinates of the contour of all the other ribs may be scaled off full soale each from its respective chord line, i.e., the point at which the line A-B intersects the rib lines. A similar diagram to that of Fig. 4 can be further used to lay in the axes of the spars in front elevation, for by this means the relative position of the rib chord line and spar centre line at any rib can be measured off straight away full size. And so with the hinge centre line of the aileron. Select a rib, say 9, and fix the centre line of the inner hinge upon it. From this point draw a straight line to the extreme tip. Again the relative positions of the hinge centre line, chord line, and spar centre at each rib is obtainable to full size scale, no matter how big the wing span. Mention, of the extreme wing tip calls for a brief word upon fairing off the wing tip from rib 12 outwards. A compromise is necessary as a rule. The chord lengths of ribs 13 and 14 are definitely fixed in Fig. 1 in order to obtain the good end curve. Thus a fair curve in end elevation, Fig. 3, allied with a reasonable curve in Fig. 4 (extended), using the rib lengths of 13 and 14 chords, correctly spaced as regards other ribs in plan, but not so in end elevation necessarily, is a practical solution of the problem.
608 h
July 26th, 1934
Supplement to
7660
'FLIGHT ENGINEERING SECTION

Edited by C M . POULSEN No. 102 ( V e 7 IX ) 9th Year
July 26, 1934
THE AUTOMATIC
CONTROL
OF AIRCRAFT
By G. R. M. GARRATT, M.A. (Cantab.)* The Automatic Pilot has been in continuous use in the Royal Air Force for five years, and forms part of the standard equipment of large machines such as night bombers. Recently the instrument has been approved for use in civil aircraft, and there is little doubt that in certain classes of civil work it will be found as useful as it already has proved in military flying. Smith's Aircraft Instruments control the Patent Rights throughout the world. In the article of which the first instalment is published this week, Mr. Garratt, of the Technical Development Staff, Farnborough, describes the principle and details of the two-axes Automatic Pilot, which has been developed by the Scientific Research and Technical Development Staffs of the Air Ministry, notably by Messrs. Meredith, Cooke and Kerr, of the R.A.E.
HE general problem of the automatic control or stabilisation of aircraft has received consideration in many different countries, and although many inventions have been recorded, there is little data available of the results which have been achieved. The instrumental problems and the aerodynamic requirements are not easily satisfied, and the credit for the practical development of the automatic pilot in this country belongs very largely to Mr. F . W. Meredith, Mr. P . A. Cookc and Mr. P. S. Kerr. The principal objects of automatically-controlled flight may be briefly summarised as follows : 1. Reduced fatigue for the pilot, who is able, when necessary, to leave his seat or to attend to other duties, such as navigation or the operation of his radio equipment. 2. Greater accuracy of control and course-keeping. 3. Reduced risk of loss of control in bad weather, or during night-flying. 4. Improved stability of the aircraft when accuracy of flying is of importance, e.g., air survey and photography. There are a large number of possible forms of automatic control, and while some of them might be practical, others would be impossible of satisfactory achievement. For ex-
ample, one might consider an azimuth control depending fundamentally upon a magnetic compass or upon some detector of the sun's rays. One might equally consider a pitch control depending on air speed, altitude, acceleration, the rate of change of altitude or on almost any function which varies with the pitch attitude of the aircraft. The main disadvantage of such controls, however, would be the fact that they would be useless unless they were applied to a completely stable aircraft on account of the phase and time lag which would be involved. An aircraft in steady flight frequently receives disturbances, and it depends on the stability characteristics of the aircraft what form of motion will result from the initial disturbance. Assuming a stable aircraft, a complex oscillation will generally be set up, anr! when the oscillation is eventually damped out the aircraft will usually be flying on a new course. The first function of an automatic control should, therefore, be to counteract the effect of the initial disturbance and to ensure that the aircraft returns to .its original course. Some datum is necessary from which it is possible to detect any deviation and t o which the course may be reset. In preference to such unreliable data as the earth's magnetic field or the sun's rays, the gyroscope is supreme. A carefully balanced spinning gyroscope has the property of keeping the direction of its axis fixed in space unless * The author desires to acknowledge the courtesy of Messrs. Smith's Aircraft Instruments in providing the facilities for the acted upon by some external force, and this characteristic stability is used in a few of the largest liners t o reduce t h ' production of the illustrations in this paper.
SUPPLEMENT TO
PLIGHT
7666
m
JULY 26,
1934
rolling of the vessel. The direct use of a gyroscope to provide stability in an aircraft would, however, be quite impossible on account of the prohibitive weight. The gyroscope used for controlling an aircraft provides only an accurate datum from which deviations can be detected, and to which the craft is directed when a disturbance has occurred. Since the gyroscope itself does not provide stability, the aerodynamic characteristics of the aircraft must receive some consideration. It must be remembered that while the degree of stability of an aircraft in pitch and in roll is largely under the control of the designer by suitable disposition of the wings and tail surfaces in relation to the centre of gravity, it is not possible to design an aeroplane to have inherent directional stability.- In other words, no aeroplane could be designed to maintain its flight upon a predetermined course without some form of control. The first essential feature of any automatic control is, therefore, that it shall be capable of directing the aircraft along a predetermined course, and it is only essential to control the pitching motion of the craft if it is unstable in pitch, or if a rapid damping of a pitch oscillation is required. In practice, however, a single gyroscope is capable of providing control about two axes, and it is usual to control the aircraft both in azimuth (i.e., directionally) and also in pitch. Stability in roll is very largely determined by the dihedral angle, and control of the ailerons is not usually necessary except for certain special purposes. In this paper we are only concerned with the control of rudder and elevator movements.
ailerons in proportion to the roll displacement. This, however, involves the use of a second gyroscope, and is therefore a serious complication. The second method, which is due to Mr. F. W. Meredith, consists in applying rudder angle in proportion to the roll displacement. The total application of rudder angle is therefore the sum of two amounts, one being determined by the angle through which the aeroplane has turned off its course, and the other being determined by the angle to which the aeroplane is banked. It was predicted by mathematic investigation, and confirmed by practical experiment, that the application of rudder angle when the aircraft is banked tends to increase the lateral stability of the craft in much the same way as if the ailerons were used to correct the bank. The application of rudder angle in response to a roll displacement is accomplished in a very ingenious but simple manner by tilting the axis of the gyre- rotor upwards in the forward direction. As will be seen later, from the detailed description of the gear, the effect of inclining the forward end of the gyro axis upv/ards is to cause a relative movement between the gyro axis and the aircraft when the machine banks, and this relative movement causes the application of rudder angle, even though the aircraft has not rotated about the azimuth axis. By this means, adequate damping of the disturbed motion is achieved and lateral stability assured without the added weight and complication of a second gyroscope, which, for all ordinary purposes, is quite unnecessary. ..
The Automatic Pilot

Having thus briefly outlined the essential features of an automatic control and the aerodynamic requirements to be satisfied, we may proceed to a more detailed description of the Automatic Pilot, familiarly referred to as " George," which has been developed at the Royal Aircraft Establishment, and the patent rights of which are now controlled by Smith's Aircraft Instruments throughout the world. Unlike some other systems, the Automatic Pilot described below depends on pneumatic means for its operation. The use of compressed air has so many advantages
The Importance of Stability

Since the gyroscope is only the datum, to which the aircraft is controlled, it is of importance to consider the effect of applying corrections when a deviation is detected. Considering the rudder movements, for example, it would be feasible to apply a correction of fixed amount as soon as an error in course is detected, and to reverse the direction of the correction as soon as a reversal in the sign of the error is indicated. Such a control would, however, cause the rudder to move in a "bang-bang" fashion, and as a result the motion of the aircraft would be continually disturbed. A great improvement is effected by arranging that the amount of rudder angle applied varies directly with the displacement of the aircraft from the datum line. The motion resulting from this form of control will generally consist of a complex oscillation, the damping of which is of considerable importance. Negative damping is, in fact, by no means impossible with such a control. By negative damping we mean that the amplitude of the oscillations would continue to increase instead of dying away. Disaster would obviously be the consequence of -iiv, uninterrupted negatively damped oscillation. While a full explanation of the reasons for the setting up of an oscillation would require mathematical treatment and would be out of place in these pages, a brief explanation is required. It must be remembered that a movement of any one of the control surfaces not only exerts a primary effect about its own axis, but, in addition, it usually has important secondary effects about the other axes. "Aileron drag" is, perhaps, the most familiar example of this secondary effect, when the depression of an aileron so increases the drag on that side of the aircraft that the machine yaws to one side in consequence, in addition to the primary rolling movement. Similarly, a movement of the rudder not only causes a yawing movement, but also introduces roll and sideslip. It is these secondary effects which complicate the damping of an oscillation. Fortunately, there are two methods by which satisfactory damping of the disturbed motion may be produced. The first method -was originated by the Sperry Gyroscope Company, and consists in applying a movement of the
FIG. 1.Air Compressors, showing Interior Details. over electrical operation, particularly in that it results in an exact and almost instant response of the -control surfaces, that it is most improbable that it will be superseded. So far as the actual gyroscope is concerned, there is little to chopse between electrical or pneumatic operation. It is in the means whereby a relative movement between the aircraft and the gyrosocope is translated into a movement of the control surfaces that the pneumatic system demonstrates its superiority. In an electric system a relative movement between the gyroscope and the aircraft has to complete a circuit which brings some form, of reversing motor into operation. No matter what form this motor may take, it must possess considerable inertia, and the resulting lag of the control surfaces seriously affects the damping of an oscillation. The small amount of power available from an electric motor without reduction gear-
JULY 26, 1934
51
SUPPLEMENT TO
FLIGHT
FIG. 2.Component Parts of Compressor System, and Control Levers, a, Compressor and Windmill, b, Oil Reservoir, c, Oil Separator, d, Pressure Relief Valve, e, Test Cock, f, Main Control Cock, g, Course-change Cock, h, Air Expansion Chamber, j , Pitch Control Lever, k, "Pins " Lever. ing is a further argument against the use of an electric system, and the introduction of reduction gearing further increases the time lag with consequent loss of damping. The quantity of air required is of the order of four cubic feet per minute, compressed to a pressure of 30-35 lb. per sq. in. While such a supply is not large, it is beyond the capacity of compressed air bottles, and a compressor system is a necessity.
Compressor System
The compressor system has been designed to operate for long periods without attention, and possesses several unique features. A rotary type of compressor is used, which may be driven from the engine, but which is more usually mounted on a wing and driven by a small windmill by which it is driven at a speed of 1,200-2,400 r.p.m. The interior construction of the compressor is shown in Fig. 1. As may be seen, an eccentrically placed rotor has slots cut in it, in which hardened steel blades can slide under the action of centrifugal force. Air is drawn in through the inlet port and compressed by the rotation of the blades to the outlet port at the side. Unless the clearances between the ends of the blades and the case were very small, such a compressor would be very inefficient if some satisfactory seal were not provided. It is found, however, that if a copious supply of oil is continually fed to the compressor a very satisfactory performance results, besides being accompanied by the minimum of wear. Referring to the diagram of Fig. 3, the oil is supplied to the compressor through the jet A, and is ejected from the compressor with the compressed air, in the form of an emulsion, to an oil reservoir. The emulsion is thrown against the walls of the reservoir through a jet, and this action causes the oil to flow to the
bottom while leaving the compressed air nearly free of oil. An outlet is provided at the bottom of the reservoir, to which a filter is fitted through which the oil passes again to the inlet jet on the top of the compressor. The size of the inlet jet controls the quantity of oil supplied. The compressed air, as shown in the diagram, passes away from the top of the reservoir to an oil separator interposed between the compressor system and the automatic gear. The function of the separator is to extract the finer globules of oil, which it does by means of a jet through which the air impinges on a small plate. The oil separator is fitted with a drain cock at the bottom through which the oil collected may be periodically removed, while the compressed air is supplied to the gyroscopic apparatus and associated mechanisms from the outlet pipe at the top of the separator.
Simple Qyroscopic Theory

In order to convey a true understanding of the operation of the Automatic Pilot, it is necessary to describe very briefly the simple properties of a gyroscope, but for a more complete account of the principles underlying gyroscopic action, the reader must refer to any of the standard textbooks on dynamics. Any rotating body has the property of tending to maintain the direction of its axis fixed in space unless acted upon by some external force. The greater the angular momentum (or mass, radius, and angular velocity) of the spinning body, the greater is the tendency to maintain the fixed direction of its axis. The second property of a gyroscope which must be described is the apparently peculiar manner in which it responds to the application of an external force. In Fig. 4 a spinning gyroscope is shown mounted in gimbal rings
SUPPLEMENT TO
JULY 26, 1934
FLIGHT
766^
THE AIRCRAFT
RELEASE VALVE
ENGINEER
DRAIN COCK
FIG-3
\
OIL RESERVOIR
FILTER
OIL SEPARATOR
FIG. 3.Diagram of Compressor System, and FIG. 4.Diagram of simple Gyroscope.
which are free to rotate about their respective axes as shown. Let it be supposed that a torque is applied to the inner ring about the axis A-A by the addition of a weight W to one side of the ring. A casual observer unacquainted with gyroscopic action would assume that the weight W would cause the ring and the gyroscope to rotate or tilt about the axis A-A, allowing the weight to fall. He would, however, be wrong. In actual fact, the gyroscope and the gimbal rings would rotate (or " precess ") in azimuth, i.e., about the vertical axis B as shown by the arrow.
An explanation of this property would be out of place and too lengthy for these pages, and the reader must refer to a text-book for a detailed account. Throughout the following description the reader is asked to bear in mind that if a gyroscope is spinning about any given axis and a torque is applied about an axis at right angles to the axis of spin, then the gyroscope will not rotate about the axis of the applied torque but will precess about an axis at right angles to both the axis of spin and to the axis of the applied torque.
(To be continued).
THE POLAR DIAGRAM

The Elements of a Method of Stressing Beams under Compressive End Load
BY EDGAR H. ATKIN
{Continued from p. 47) ~ T H E next case of importance is as follows : CASE IV. Beam subjected to compressive end load, a distributed load changing in magnitude at one or more points along the span, , and with or .without end moments.
OC, then the sense of PQ is the same as the sense

of RS. \
To fix our ideas let us assume that one end of the beam has a positive distributed load w, and the remainder a positive distributed load w2. Having started as in Case II, calculate both
W,
and -$-
Through Q draw QX2 to meet BjY in X2, and through X 2 draw X 2 Xi parallel to PQ to meet AjX in Xj. On OX, and OX2 as diameters draw the arcs AXD and BjE respectively. The bending moment diagram is the part shown shaded in Fig. 9. The direction in which PQ is drawn can be decided quite simply if the reader remembers that there can be no sudden change of bending moment at the dividing line OC.
---:-''. 4 . *"., _ E x a m p l e (3) '- --:
As wx and w2 are positive - and f are negative and, therefore, the corresponding arcs are placed in the negative direction, as shown in Fig. 9. The end moments are measured off and the perpendiculars to OA and OB drawn as in previous cases. It will be noted that these end moments are positive. Let these perpendiculars be AjX, and BjY. Take any point P on A t X and draw PQ parallel to the dividing line OC corresponding to the change in w. If R and S are the points
W* Wa
in which the arcs of radius -j- and ; respectively meet
A steel beam 80 in. long, with a moment of inertia of .35 in.4, is subjected to a compressive end load of 7,800 lb., a positive end moment at A of 4,000 lb. in. and a positive end moment at B of 5,890 lb. in. A distributed load of 6.3 lb./in. run over AC, and a distributed load of 8.1 lb./in. run over CB. AC = 30 in. CB = 50 in. The diagram for this example is shown in all its constructional details in Fig. 10. It should be unnecessary to repeat in terms of numerical quantities the construction already described. The follow-
53
JULY 26, 1934
SUPPLEMENT TO
p
A!
UJ,
FLIGHT
L,
FIG. 10.
4 ' C 1 L l *. *
cJz
P Ma
vjmnmmtmm
tlfffiiti
C
-V
\
/
/ \h
Y
X
/s A
'.0O0L8S/INS
ttttrnnttttt t
U/-6-3LBVIMS uJ-811-BVlNS
FIG 9
OX, at the correct angles and put in the arcs of radii

iV
Beam under compressive End Load and a distributed Lateral Load, changing in Magnitude in the Bay. ing figures will enable the reader to follow the drawing of the diagram through for himself. 7,800 V .000743 V .35 x 3 >Tio 7 = .02725 30 X 57.3 X .02725 =. 46.8 deg. oAC = nCB = 50 X 57.3 X .02725 = 78.1 de Total Angle a = aAC + aCB = 124.9 deg. K'AC 6.3 = 8,470 lb. in.
.000743
,ancl ..set off the end moments AA. and BB. and draw
Hz'
7i!'
Mi"
normals A,R, BjT to OA and OB through A, and B , respectively. AjR intersects OX in Q,. Through any point R on AjR draw RX perpendicular to the dividing line OX and divide RX at S, so that RX . fi2 and join RQ. RS = ^ As an aid to memory, it should be noted that the angle RQX which the locus (>f 1 lie vertex for AX makes with
DISTRIBUTED LOAD UJ
wCB
8.1 .000743
= 10,890 lb. in.
1 2 over
LENGTH lj
ttttttitt
I, over LENGTH L,
The distance X! X2 (or PQ) parallel to OC is, therefore, i^| ' *"
'
^ = 10,890 8,470 = 2,420 lb. in. " CASE V.
Beam subjected to compressive end load, a uniform distributed lateral load, end moments and one or more changes of moment of inertia in the bay. All the necessary quantities, /xx ju2 : "1 "2 . ^L^. . . are calculated as usual.
Hi Hz
It will be
noticed that this time-^- -^differ.because of the differences
in the moments of inertia ; in Case IV these quantities differed, because of the differences in the lateral distributed To simplify the description of the method, it will be assumed that w is constant over the whole span, and tttat there is only one change of I in the. bay. Let-the moments of inertia over AX and XB be Ij and I 2 respectively, and lor definiteness assume that I1 >I 2 . It is useful to note that because I 2 > I 2 thereHw*u, <uo, and, other things being equal 2>Cj> * The wethod of constructing the diagram is a? follows (see Fig. n ) : Having "set off the various lines OA, OB, Beanoinder compressive End Load and Lateral Load with a Change in Moment of Inertia in the Bay.
84
SUPPLEMENT TO
JULY 26, 1934
FLIGHT
766/

,_ 138-3
OX is greater than the angle SQX which the adjusted locus RQ makes with OX, and that AX is stiffer than XB. In other words : in passing from a part of the beam to another part less stiff the angle through which the locus is turned reduces the angle which the adjusted locus makes with the dividing line. Along OX set off in the positive direction a distance / w w\ QY equal to ( 2 5 1 and through V draw UV parallel to SQ. UV is then the final adjusted locus and cuts BjT in X2> the vertex for the length XB. Since X 2 is now known, the construction can be repeated from the end B and X s found. To do this draw X2W X2W perpendicular to OX and divide X2W, so that Y T W = and join X21Y.
w
HI
7
/X / /
x \
A
\
zr
\
-ve \
Along OX measure off from Y distance YZ in the positive direction equal t o 2 5 and through Z draw ZXX M2 Mi" parallel to X21Y to meet A ^ in X,, then X t is the vertex for the length AX, and circles on OX, and OX2 as diameters complete the diagram. In this case the number of degrees in the sector AOX corresponding to a unit of length of AX is not the same as the number of degrees corresponding to a unit of length of XB. This obtains generally : in more complicated cases a different angular scale must be used for each part of the beam into which the " jumps" of moment of inertia divide it. It is also to be borne in mind in calculating the true shear that the appropriate vertex and value of p for the segment under consideration must be used. The actual position of a vertex is no indication that it applies to a given segment. It is only by following the construction rigorously that diagrams of unusual shape can be exactly drawn. Example (4) A steel beam AB, 95 in. long, is so constructed that 45 in. of its length from A to X has a moment of inertia of .32 in.4 while the remaining length has a moment of inertia of .4 in.4 There is a uniform distributed upward loading of 7.5 lb./in. over the entire length and a compressive end load of 7,000 lb. The end moment at A is 3,500 lb./in. clockwise externally to the left, there is a moment at Bj of 3,000 lb./in. anticlockwise externally to the right. The necessary quantities are as follows : .
f
\l
V
1/ /u
ll/
/
-ve
1
r Mr
z
3S00 Les INS.
F1G.I2
Examp/eNo.*
I Kins*
I.-4I11S*
"ttttttJtttt B "Mtttttttt U-7iLBs/lBK,
i *
3000
LBS INS. P700OLBS.
It may also arise because of a load offset on its point of attachment to the beam. As this interesting case is of rather infrequent occurrence the details of the constrviction will not be given here : they have already been described in the pages of Flight: the reader is referred for these particulars to Flight Aircraft Engineering Supplement, August 26, 1932, where he will find all details. Having mastered the six cases already described he may now, in order to deal with any combination of the various types of loading, superimpose the constructions.
V
uXB =
.32 x 3 x 10
7,000
= \f.OOO729 = .027
-7 = v 000583 = .02415 ,4 X 3 X 10 = 69.7 deg. aAX = .027 X 45 X 57.3 6 aXB = = 9 - 2 deg. .02415 x 50 x 57-3 IV 7-5 = 10,270 lb./in. .000729 w 7-5 = 12,860 lb./in. .000583 ' The diagram is shown in Fig. 12. The lettering is the same as in Fig. 11, but in this case the moment of inertia increases as we pass through X.
-."-.; .:--:/ v; " / ; ' C A S E VI ' .--._,
Change of Moment in the Bay This case may occur because of a change in the cros> section of beam over a part of the spar which results in a shift of the neutral plane above or below the neutral plane of the remainder.
FIG. 13
Example No. S.
Example No. 5 : Beam under complex Loading Conditions. -
65
JULY 26, 1934 SUPPLEMENT TO

This does not mean, however, that a change of end load in the bay can be dealt with graphically by the polar diagram. Such a change cannot be allowed for by pure graphics ; if it is necessary to take account of it recourse must be had to a combination of analytical and graphical methods outside the scope of this article. When the end loads in the various parts of the beam do not differ greatly it is usually sufficient to use the mean end load throughout the bay. It is, of course, evident that in no case is the method applicable to a tensile end load. The angles a would, for negative values of P become imaginary, and therefore could not be drawn. A sudden change in material along the span can be dealt with quite easily by exactly the same construction as for a change in moment of inertia. In fact all the observations which have been made, or will be made about the latter, apply equally to a change in the modulus of elasticity. With the preceding proviso we proceed to an example. ' Example (5) A beam is subjected to the system of loading shown in Fig. 13. It will be noticed that there is a change in the '. moment of inertia 40 in. from the end A, and 25 in. from the end B. In addition there is a change in the distributed loading and a concentrated load. It will be found convenient in this and all complicated cases to tabulate the necessary quantities as follows : A X Y Z W B
1 in. 4 .. 1.2 9 9,500 .000352 .01875
FLIGHT 766^
9
9,500 .000352 .01875 10.5 29,850
1.2 9,500 .000264 .01625 10.5 39,800
Plb. . .
9,5
.000264
9.5
.000352 .01875 10.5 29,850
fi . . W
. .
.01625
If this is done, there will be no appreciable change in the diagram, and the ordinary methods previously described can be used. It is much better, however, not to introduce more different points of change than are necessary : a large number of changes calls for much more careful drawing and sometimes repeat diagrams to improve the accuracy of the first attempt. Consider in particular the case in which a concentrated lateral load W occurs at X where the moment of inertia of the beam changes from I, to I 2 and, consequently, ix changes from /xt to it2. It is easily seen that a change in the magnitude of the distributed loading has no influence on the solution to the problem. The difficulty is this : are we to use fj.l or fi,2 in calculating w for use in the construction at the point X ? /" Since the solution in the general case is unique, it is evident that there can be no ambiguity in the solution for this special case, because it is simply the limiting case arrived at by reducing the distance, originally finite, between W and the change in I until it is zero. Assuming that I always changes at X there are two ways of looking at this apparent anomaly. The case may lie assumed to be either the result of moving a concentrated load W initially a small distance to the left of X up to X r or, the result of moving a concentrated load W initially a small distance to the right of X back to X,. Looking at the problem from this point of view, we infer that it is the sequence of the construction for the load \V and the change in moment of inertia which decides the value of p to be taken. If in our construction we are crossing a dividing line from a part A where n = 11, to a part H where n = ii2 we should expect the following rule to hold. If the construction for W is performed first, the value of (i for the part A should be used (viz., p,) to calculate the w value of : if the construction for W is performed after the construction for the change in moment of inertia, the value of y. for the part B should be used (viz., ii2) to w. calculate the value of These rules may be proved if the reader so desires by considering all the sequences which can be obtained by combining the different values of /x with the alternative sequences oi the constructions. ' It is then easily proved that the rule gives the only sequences compatible with a unique and consistent solution when the constructions have been performed from both ends of the beam as described above. It appears unnecessary to include a proof in the text of this article The descriptions will enable anyone to use the method for any single bay beam without an understanding of the principles underlying it. Once the reader is conversant with the method he will gradually appreciate its mechanism and will realise that every construction has for its basis a known change in shear at each discontinuity in the bending moment diagram. In the case of a concentrated load W there is a change W in the shear : in all other cases there is no change in shear By stating whichever condition is applicable in any case the reader should have no difficult}' in proving by simple geometry the construction given.
lb./in.
13
49,250
13
36.95O
w
a
. 37.25 deg. 43 deg. 215 deg. 26.85 deg. 23.3 deg.

1 <
60
The various vertices for the parts AX, XY, YZ, ZW and W B will be denoted by X t , X 2 , X 3 , X 4 and X 5 . Having drawn all the radial dividing and bounding lines and arcs of radius j the end moments may be set orf. This gives the points A! and B r On the normal to OA at Aj any assumed vertex X , for AX taken, from which by the construction for change in moment of inertia, we arrive at the vertex X 8 ' for XY, X 3 [ is displaced from X2> parallel to OY by an amount ~ in the positive direction.
l 1
While X 4 is displaced from X., normal to OZ in the positive direction. Having arrived at X 4 J the construction for moment of inertia is again performed, and, by drawing a line through X 6 ] parallel to the final adjusted locus EF meeting the normal to OB at B1 in X6; we obtain the correct vertex X5 for WB. Starting now from OB with the correct X5 the construction is repeated in the reverse direction, and the remaining correct vertices found. A very interesting special case arises if a change of shear, due to a concentrated lateral load, occurs at the same place as the change in moment of inertia, or more generally the change in the quantity ft. It becomes, in this cxse impossible to say from the general construction which is the correct value of / x to take at the discontinuity. An obvious way of overcoming the difficulty, is to separate the two changes by a small distance along the spar.
THE INSTITUTE
OF METALS
This year's Annual Autumn Meeting ot the Institute oi Metals will be held in Manchester from September 3 to September 6. The thirteenth Autumn Leclure will be delivered by Dr. J. L. Haughton on September 3, the subject being "The Work of Walter Kosenbain." Of particular interest to aircraft engineers should be the paper on " The Influence ol Pickling on the Fatigue Strength of Duralumin," by H. Suttos: and W. J. Taylor. Those wishing to take part in the proceedings should communicate with the Secretary of the Institute of Metals, 36, Victoria Street, London, S.W.I.
SUPPLEMENT TO
' JULY 26,
1934
FLIGHT 766A
SUMMARIES OF AERONAUTICAL RESEARCH COMMITTEE REPORTS These Reports are published by His Majesty's Stationery Office, London, and may be purchased directly from H.M. Stationery Office at the following addresses: Adastral House, Kingsway, W.C.2. ; 120, George Street, Edinburgh ; York Street, Manchester; 1, St. Andrew's Crescent, Cardiff; 15, Donegall Square West, Belfast; or through any bookseller.
MODEL SPINNING TESTS OF AN INTERCEPTOR FIGHTER. By A. V. THE FRICTIONAL DRAG OF FLAT PLATES BELOW THE CRITICAL REY-
NOLDS NUMBER.
By A. Fage, A.R.C.Sc.
R. & M. No. 1580.
(7
pages and 3 diagrams.) April 21, 1933. Price 6d. net.

Measurements of the frictional drag of flat plates have been made over the wide range of Reynolds Number. At the low values the measurements were made on two small plates of length 0.125 and 1 in. respectively, mounted in a small water tunnel, and at the high values on a 5 in. plate in a 1 ft. open-jet wind tunnel. The drag of each plate was predicted from changes of momentum in the fluid flowing along its surface. The velocity measurements needed for these predictions were obtained for the small plates by a technique involving the use of an ultra-microscope, and for the 5 in. plate by small pressure tubes. An analysis of the results and those obtained by other experiments leads to the conclusion that over the wide range, the drag is given fairly closely by the Blasius relation. Over a lowest range the measured drag is higher than the Blasius value.
TESTS OF THE EFFECT OF F U E L EVAPORATION ON THE PERFORMANCE
This Interceptor Fighter is a heavily loaded low-wing monoplane of R.A.F.34 wing section : A-B is negative and of considerable magnitude and the fin and rudder although unusually large are situated entirely above the tailplane. The fuselage is of approximately circular cross section. It has been shown theoretically that an aeroplane combining these qualities would be exceptionally liable to spin flat. The aeroplane was spun at Martlesham with C.G. 10.5 in. behind the leading edge of the mainplaue, and on the first occasion developed a flat spin to the right; the pilot, after attempting to recover by every artifice at his disposal, was finally compelled to abandon the aircraft. It appeared at first sight that, since the spin had been unquestionably flat and the controls correctly applied for recovery, the Interceptor Fighter was a clear case of a dangerous flat spinning aeroplane. A 1/22 scale dynamical mode' of the aircraft was spun in the R.A.E. Free Spinning Tunnel, and the influence of the following factors investigated :(a) elevator angle ; {b) aileron angle ; (c) C.G. position ; (d) moments of inertia ; (<;) direction of spin. Recovery tests were carried out with several control movements, embracing as far as possible the various stages by which recovery from the flat spin at Martlesham was attempted. The experiments show a close agreement between model and full scale and provide a plausible explanation of the flat spin at Martlesham, compatible with the satisfactory behaviour of the aeroplane in the earlier tests. It was found that the model spun steeply with elevators hard up and recovery on reversal of the controls was rapid; but if the elevators were moved downwards through as little as 10 dog. from this position a flat spin was induced from which recovery was sometimes very prolonged. Several variations of the original design were later tested and the experiments on them are described in Part II of this report.
A CONTINUOUS ROTATION BALANCE FOR THE MEASUREMENT OF YAWING AND ROLLING MOMENTS IN A COMPLETELY REPRESENTED S P I N . By
OF A CENTRIFUGAL SUPERCHARGER. By G. V. Brooke, B.Sc.Tech. Stephens, B.A., and R. H. Francis, M.Sc. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1578. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1574. (15 pages and 7 diagrams.) November, 1933. (17 pages and 2 diagrams.) May, 1933. Price is. net. Price is. net. The ratio of the absolute pressures at delivery and intake of a centrifugal air compressor operating under conditions of constant adiabatic efficiency is a function of the ratio of the square of the impeller speed to the absolute intake temperature. As the pressure ratio is known to increase with increase of impeller speed at constant intake temperature, the form of the function is necessarily such that an increase of pressure ratio with reduction of intake temperature at constant impeller speed would be expected. Complete confirmation of this effect when the fluid compressed was air was provided by the experiments described in R. & M. No. 133(i.* Subsequent tests of a Kestrel U.S. engine (twelve cylinders, Vee construction, watercooled) under artificial altitude conditions showed the existence of a similar effect when air and petrol mixture was passing through the supercharger, although decrease of intake temperature was found to produce slightly less than the anticipated increase of pressure ratio. In comparable tests of a " Jupiter VII " engine (nine radial cylinders, air-cooled) there appeared to be an entire absence of any influence of intake temperature upon supercharger pressure ratio. Tests were undertaken to ascertain whether these differences in the behaviour of the several superchargers could be attributed to the action of petrol evaporation upon the performance. The investigation comprised three groups of experiments, in which the fluid compressed was (1) air, (2) a mixture of air with petrol to specification D.T.D.134, and (3) an air and white spirit mixture. The third series was included in order to examine the performance under conditions of greatly reduced evaporation without the necessity for tests at inlet temperatures sufficiently low to involve risk of freezing at the supercharger intake. At initial air temperatures between 15 deg. and 20 deg. C. the performance when compressing air was only slightly affected by the admixture of white spirit. Under conditions of the maximum air temperature of 60 deg. C. at the carburetter intake, the admission of petrol produced an increase in the pressure ratio of the supercharger agreeing approximately with the estimated theoretical effect of complete vaporisation before entry to the, impeller. The magnitude of the increase of pressure ratio due to vaporisation became greater with increase of impeller rotation rate. At the highest speed reached in the tests (865 ft. per second at the impeller tip), the pressure ratio was increased almost 6 per cent, by the admission of petrol to the air at an intake temperature of 60 deg. C. This increase of pressure ratio represents a gain in rated altitude of approximately 1,500 ft. * R. & M. 1336. The application of dimensional relationships to air compressors, with special reference to the variation-of performance with inlet cotidiLions. R. S. Capon and G. V. Brooke. (June, 1930.)
P. H. Allwork, of the Aero-Dynamics Department, N.P.L. With an Appendix on the Experience Gained in the Use of the Apparatus. By H. B. Irving, B.Sc., und A. S. Batson, B.Sc, of the Aerodynamics Department, N.P.L. R. & M. No. 1579. (6 pages and 2 diagrams.) November 2, 1933. Price qd. net.
In R. & M. 828* and R. & M. 936t descriptions were given of two earlier continuous rotation balances in which the radius of spin was fixed at zero and incidence only was variable The present balance was designed to obtain test conditions more nea.rly representative of a true spin, and by its means a complete aeroplane model can be tested at any presentation to the wind, within limits, and at any radius of spin up to 12 in., the maximum rate of spin being three revolutions per second. The model can be set at any presentation to the wind within limits A continuous rotation balance for the measurement of l.p at small rates of roll.Relf and Lavender, R. & M. 828, August, 1922. t A continuous rotation balance for the measurement of pitching and yawing moments due to angular velocity of roll.Lavender, R. & M. 936, February, 1925.
COMPARATIVE TESTS WITH PETROL AND BUTANE ON A I R AND WATER COOLED AIRCRAFT E N G I N E S . B y P . H . Stokes, B.Sc., a n d F . G.
AMERICAN N.A.C.A. REPORTS

TECHNICAL NOTE N O . 491 : TANK TESTS OF A FAMILY OF FLYING-
Code Holland, M.Sc., A.I.Mech.E. Communicated by the Director of Scientific Research. Air Ministry. R. & M. No. 1570. (66 pages and 28 diagrams.) July 14, 1933. Price 4s. net.
The experimental work, dealing respectively with an air-cooled and a watercooled engine, was planned as part of a concerted effort towards an improvement in the fuel economy of engines in normal service. At the time when the work was started, the fuel consumptions per brake horse-power-hour commonly met with were substantially higher than the best obtainable upon a single cylinder test engine, and an element of uncertainty which it was often suggested must be held at least partly responsible was inequality of distribution of the fuel between the cylinders on a multi-cylinder engine. It was suggested that this might be the cause of a difference of some 15 per cent, between the highest thermal efficiency obtainable on a multi-cylinder engine and that reached under properly controlled conditions in a test engine. It was decided to compare the fuel consumption per hour and the thermal efficiency of two typical engines, running on petrol in the normal way, with the results when the same two engines were run using a gaseous fuel, which could be thoroughly mixed %vith the air at its entry to the induction manifold. Under these conditions, although a perfectly uniform dispersion of the fuel within the cylinders could not be taken as certain, inequalities of fuel distribution as ordinarily understood would at any rate be completely eliminated. A good many preliminary trials were made using coal-gas as fuel, but the fundamental difference of chemical composition renders comparisons with petrol on a thermal efficiency basis uncertain, and butane ((-4Hj0) was therefore decided upon as providing a gaseous fuel with molecules of the same general type as those of petrol. The results achieved in the experimental work, if regarded from the point of view from which it started, are largely negative in character ; for they appear to prove that losses through inequality of fuel distribution cannot be greater than about 5 per cent, and are therefore of secondary importance. Apart from this answer to a particular question, there is to be found in these reports a great deal of accurate data upon the efficiency-power relationship for aero-engines both supercharged and normally aspirated, and at altitudes up to 23,000 ft., more complete, perhaps than any hitherto published.
BOAT HULLS. By James M. Shoemaker and John B. Parkinson. This report presents towing tests made in the K.A.C.A. tank of a parent form and five variations of a flying-boat hull. The beams of two of the derived f'^rrê were made the same as that of the parent and the lengths changed by irrrdu- '..g and decreasing the spacing of stations. The lengths of two others of the derived forms were made the same as that of the parent while the beams were changed by increasing and decreasing the spacing of buttocks, all other widths being changed in proportion. The remaining derived form has the same length and beam as the parent, but the lines of the forebody were altered to give a planing bottom with no longitudinal curvature forward of the step. The test data were analysed to determine the minimum resistance and the angle at which it occurs for all speeds and loads. The results of this analysis are given in the form of nondimensional curves for each model. The effect of variation in overall size, as indicated by a " complete " test on any given hull, is pointed out. The effect of changing length alone by the spacing of stations, of changing beam alone by the spacing of buttocks, as well as the effects of the changes in length-beam ratio and longitudinal curvature that result from these operations are discussed.
TECHNICAL NOTE N O . 49S: WIND-TUNNEL MEASUREMENTS OF A I R
LOADS ON SPLIT FLAPS. By Carl J. Wenzinger. Tests were made in the N.A.C.A. 7 by 10 foot wind tunnel, to determine the control forces and air loads acting on split flaps. Clark Y wing models were used with two different sizes of full-span split flaps ; one having a medium chord (0.2.r; C), and the other a narrow chord (0.15 c). Hinge moments of the flaps were measured and also the division of load between the flaps and the wing. The investigation showed that, at the angles of attack and flap deflections for maximum lift, the lift loads on the split flaps were only 5 per cent, and 8 per cent, of the total lift for the narrow and medium-chord flaps respectively. The ratio of drag on the flaps to total drag increased greatly with decreasing angle of attack, reaching a value of approximately unity at small negative angles of attack with the flaps fully deflected. The normal force on the split flaps increases both with angle of attack and with flap deflection for angles of attack below the stall. The value of the normal-force coefficient is about 1.40 at the angle of attack and flap deflection for maximum lift with either of the flaps tested. The centre of pressure of the load on the split flaps in general moves forward with decreasing flap deflection and with increasing angle of attack from small negative angles up to the stall. The hinge moments of the narrow-chord split flap were about 42 per cent, those of the medium-chord flap when deflected to give approximately the same maximum lift, but they are considered to be still too large for rapid and easy flap operation.
August 30, 1934
Supplement to
898a

No. 103 ( V ^ e 8 I X ) 9th Year August 30, 1934
THE AUTOMATIC CONTROL OF AIRCRAFT

.. . By G. R. M.
GARRATT,
M.A., (Cantab.)*
(Continued from page 52)
Qeneral Construction of Automatic Pilot

The general construction of the Mark 1 Rudder and Elevator Control is shown in Figs. 5, 6 and 7. Fig. 5 shows the actual gyro rotor A mounted on its ball bearings in the inner gimbal ring B. The inner gimbal ring is mounted in pivots in an outer ring C, which is carried by vertical pivots in the supporting framework, which may be seen in Figs. 6 and 7. On the rotor spindle may be seen a pair of nuts D. These nuts, which will be referred to later, are for the purpose of balancing the rotor in " azimuth.'' The air supply for driving the rotor, which revolves at a speed of 11,000 r.p.m., is led to the bottom pivot of the outer gimbal ring. This pivot is hollow and from it the air passes through a pair of pipes, which are embedded in the vertical ring, to the two jets located close to the pivots of the inner ring. Attached to the inner and outer gimbal rings are various valves and links, which may be seen in Figs. 6 and 7. These associated mechanisms are of a somewhat intricate nature and will be described in detail in their correct sequence later in this paper. The construction and assembly of the gyro rotor in its bearings, and of the two gimbal rings with their respective pivots is of the very utmost importance, and only the highest degree of workmanship may be permitted in order that bearing and pivot friction may be reduced to the absolute minimum. Any friction about either the horizontal or the vertical pivots causes inaccurate behaviour of the controls. It will be seen that the gyroscope is given three degrees of freedom, i.e., it spins about its own axis, and it is entirely unrestrained about both the pitch and azimuth axes. Bearing in mind, therefore, that the dynamic stability of the gyroscope tends to maintain the direction of its axis fixed in space, it will be realised that when any deviation of the aircraft occurs, a relative movement takes place between the supporting framework and the gimbal rings. The gyroscope is installed with its ax-is approximately in the fore and aft line of the aircraft, and thus a pitching of the aircraft causes a relative movement
between the supporting framework and the inner gimbal or pitch ring, while a yawing movement causes a relative movement between the supporting framework and the outer gimbal or azimuth ring. It is solely the relative movement between the outer gimbal or azimuth ring and the aircraft itself which controls the application of rudder angle, and the operation of the rudder control will therefore be considered first.
Rudder Control
When the aircraft deviates from i s course, the relative movement between the azimuth ring of the gyroscope and the supporting framework brings a small air valve which
FIG. 5.View of Gyroscope dismounted, snowing inner and outer Gimbal Rings.
-58
SUPPLEMENT TO
AUGUST 30,
1934
FLIGHT 898b
trol the movements of the servo-motor, the piston would travel to the extremity of its stroke and thus apply full rudder angle irrespective of the amount by which the aircraft had deviated from its course. Such a control would cause the motion of the aircraft to be continually disturbed and would be quite impracticable. To achieve smooth operation of the control, it is necessary to arrange the mechanism so that the application of rudder angle varies proportionately with the angle through which the aircraft has departed from its course, and thus, as the aircraft returns to its course, the angle is progressively reduced. This requirement is satisfied by the incorporation of a "follow-up" system, by the action of which the movement of the servo-motor piston causes the rudder valve E to " follow " the movement FIG. 6.Rudder and Elevator Control from right-hand side, showing Rudder of its piston and so limit the Control Mechanism, Course-change Cylinder, etc. travel of the servo-motor may be seen at E, Fig. 6, into operation. The piston of piston to an amount determined by the actual displacethis valve is connected by a link to the azimuth ring (see ment of the rudder valve piston. This will be more clearly photo.), while the valve body is carried on the framework. understood by the following description. The supporting framework L is not rigidly attached to Movement of the piston in one direction or the other admits compressed air to one or other of the flexible the baseplate, but is carried on a bearing which permits pipes F, G, which lead to a double-acting servo-motor H. a small amount of rotation about a vertical axis. The The piston of the servo-motor is connected by its rod to framework is connected by a pair of levers (which may be the dummy rudder bar J, which is connected by cables seen at M, Fig. 9) to the rudder bar. When the rudder bar moves, the gyro framework, together with the body ol to the rudder of the aircraft. It will be noticed that the servo-motor piston rod is valve E, rotates slightly on its bearing in such a sense as actually- connected to a lever which is pivoted about the to follow the travel of the piston and close the valve. Suppose that the aircraft has received a disturbance. same centre as the dummy rudder bar, and which can move independently between the top and bottom members of the rudder bar. This may be more clearly seen in Fig. 7. When the automatic control is in use, these two levers are locked together by means of the spring-loaded plunger K, which is operated by a Bowden cable from a lever (k, Fig. 2) in the pilot's cockpit. The purpose of this arrangement is to provide a complete mechanical cut-out between the automatic gear and the control surfaces for use in case of emergency. When the aircraft is disturbed from its course, the valve E operates on account of the relative movement between the gyroscope and the aircraft, and admits compressed air to the servomotor, which, in turn, controls the rudder so as to correct the course of the aircraft. It will be clear, however, that unless some FIG. 7.Rudder and Elevator Control-from Front End, showing Centraliser Unit, means were adopted to conand Rudder and Elevator Valves.
.AUGUST 30,
1934
S9
SUPPLEMENT TO
FLIGHT
HqSc
FIG.
trolled by the position of the gyroscope, it automatically changes course as the gyroscope precesses. From the brief description of gyroscope theory earlier in this paper, it will be recalled that in order to precess a gyroscope about the vertical axis, a torque must be applied about a horizontal axis at right angles to the axis of the rotor. This torque is actually applied to the inner ring of the gyroscope by means of a small double-acting air cylinder, which may be seen at N, Fig. 6, just beneath the rudder valve. The piston of the air cylinder is connected to the inner gimbal ring by means of a link and the perforated arm O. When air is admitted to one side or the other of the course-change cylinder, the thrust of the piston applies a torque about the horizontal gimbal axis and so causes the gyroscope to precess in azimuth. The admission of air to the course-change cylinder is controlled by the "Course-Change Cock" (g. Fig. 2) which is under the control of the pilot.
Balancing for Azimuth Accuracy

The accuracy with which the Automatic Pilot keeps the aircraft on the predetermined course is adjustable by means of the rotor nuts shown at D, Fig. 5. By means of these nuts the position of the centre of gravity of the gyro rotor and inner gimbal ring may be adjusted so that a negligible torque is exerted about the horizontal pivots of the inner gimbal ring. It is possible to adjust these nuts so that the aircraft will not deviate from its course by more than four or five degrees per hour.
FIG. 8.Diagram showing Operation of Relay Valve.
Pitch Control
and in consequence has departed from its course. The gyroscope and the gimbal rings will remain "fixed," and relative movement will occur between the gyroscope and the supporting framework. This will cause the rudder valve to operate and, in consequence, compressed air will be admitted to one side of the rudder servo-motor. The servo-motor piston will therefore operate the rudder bar to apply a correcting movement to the rudder. As the rudder bar moves, however, the ratio levers M,M, cause the supporting framework of the gyroscope, together with the casing of the rudder valve, to rotate slightly in the same direction as the apparent displacement of the rudder valve piston. The rudder valve is thus closed, and, in consequence, no further movement of the servo-motor piston and no further application of rudder angle takes place. By adjusting the effective lengths of the levers M,M, it is possible to regulate the magnitude of the rudder correction applied for a given departure of the aircraft from its course. As the aircraft returns to its course, the rudder valve causes the servo-motor to reduce progressively the applied correction until the aircraft is again on its true course, when the rudder will be in the neutral position. In practice the aircraft usually '' overshoots'' the true course slightly, to which it returns by a diminishing oscillation. The accuracy with which the Automatic Pilot maintains the aircraft on its course is in no way effected by any inherent tendency of the aircraft to yaw off its course, since the mean position of the rudder is automatically kept at the requisite position for straight flight in much the same manner as a pilot frequently has to keep the rudder applied if he wishes to fly straight. As will be realised from the earlier description of gyroscopic theory, a frictional torque about the horizontal axis of the inner gimbal ring will cause a precession of the gyroscope in azimuth. In other words, any friction about the horizontal pivots will be detrimental to the accuracy with which the Automatic Pilot can maintain the aircraft on its course. For this reason it is undesirable to connect the piston of the elevator valve direct to the pitch or inner gimbal ring, as, no matter how carefully the valve might be made, the friction between the piston and the casing would result in poor course-keeping qualities. In the case of the rudder valve this friction is not of serious consequence, since friction in the rudder valve is
Change of Course
There are two possible methods of altering course on a gyroscopically-controlled aircraft. The first is to change the relative positions of the gyroscope and the aircraft. While this method has the advantage of accurately measuring the amount of the turn, it possesses several serious practical disadvantages. The second method has been adopted in the Automatic Pilot, and consists in precessing the gyroscope itself, and since the aircraft is conVALVE
FIG. 9.Rudder and Elevator Control from left-hand side, showing Connection between Pitch Ring of Gyroscope, Relay Valve and Main Elevator Valve.
SUPPLEMENT TO
AUGUST 30,
1934
FLIGHT 898*2

admission of compressed air to the elevator servo-motoi? which is shown on the illustration of the elevator plate, Fig. 10. The servo-motor is connected to the elevator cables in a very similar manner to that already described above for the rudder control. As in the case of the, rudder control, it is necessary to use a follow-up system in order to ensure that the movements of the elevators are proportional to the amount by which the aircraft has been disturbed from its mean pitch attitude. The operation of the elevator follow-up system is controlled by a Bowden cable which connects the movements of the elevator servo-motor to the casing of the main elevator valve. When the piston of the elevator valve is displaced by the operation of the relay valve, compressed air is admitted to one side of the servomotor, causing a movement of the servo-motor piston, and thus producing the application of elevator angle. As the piston moves, however, the Bowden cable mechanism causes the casing of the elevator valve to follow the displacement of its piston, thus closing the valve and preventing the further application of elevator angle. The angle through which the elevators are moved, therefore, depends on the actual pitch displacement of the aircraft. It will be clear from the above description of the pitch control that the constancy of the pitch attitude maintained must depend solely on the accuracy with which the gyroscope maintains fixed the direction of its axis, and, since the constancy with which the gyroscope can define a pitch angle is affected by the rotation of the earth and by the frictional torques which must be present about the vertical axis of the gyroscope, it is essential to provide means for ensuring that the gyroscope will be substantially unaffected by such torques, and will be capable of maintaining a constant pitch attitude, in spite of a possible change of latitude or the rotation of the earth.
equivalent to a torque about the vertical axis. This will cause certain errors in the pitch attitude, but, since the pitch attitude is accurately determined by gravitational means, as will be described later, small frictional torques about the azimuth axis can be tolerated. In order to reduce the frictional torques about the horizontal axis, the inner gimbal ring is connected to the main elevator valve by means of a pneumatic relay valve, the construction of which will be made clear by reference to the diagram, Fig. 8. The small inner piston of the valve is connected to the inner gimbal ring by means of the rod which may be seen at P, Figs. 8 and 9. Referring to Fig. 8, which is diagrammatic only and inaccurate in detail, it will be seen that the inner piston moves inside a sleeve which forms part of the main piston of the relay. Small air passages are arranged in the main piston, as shown, one of which connects the top chamber of the cylinder with an exhaust port at the bottom end of the normal position of the " fairy " piston, and the other of which connects the bottom chamber with an exhaust port at the top of the fairy piston. Air is supplied to the relay from an expansion chamber via the inlet R, and it then passes to the top and bottom chambers of the relay through the tapered needle throttles as shown. In the neutral position, the fairy piston just covers the two exhaust ports of the main piston, and, since there is only a negligible leak through the ports, the air pressures on either side of the main piston are equal. When the
Adjustment of Pitch Attitude

In order to maintain a constant pitch attitude under all conditions, it is necessary so to control the gyroscope that its pitch angle remains constant with respect to the vertical. This is achieved by gravitational means. The outer gimbal ring, which, it will be recalled, is carried by vertical pivots, is put out of balance by the addition of a weight T, Fig. 6. When the pivots are exactly vertical, it is clear that this lack of balance will have no effect whatever, but consider the effect of the weight if the line of the pivots should be tilted either forwards or backwards. The weight will then cause a torque about the line of the pivots, and this torque produces a precession of the gyroscope about the horizontal pivots of the inner gimbal ring. In other words, the gravitational torque about the outer gimbal axis when that axis is tilted results in a precession of the gyroscope in pitch. By means of the relay valve and the main elevator valve, the aircraft is constrained to follow the datum set by the gyroscope, as has already been explained. When the gyroscope is precessed in pitch, therefore, by the gravitational torque about the outer gimbal axis, the aircraft changes its pitch trim, and in doing so the axis of the outer gimbal ring is once again returned to the vertical, and the gravitational torque disappears. The pitch attitude of the aircraft is thus governed by the relation between the machine and the axis of the outer gimbal ring, and it would be possible to adjust the pitch attitude by varying this relationship. There are, however, certain disadvantages to this method of adjusting the pitch attitude, particularly when the ailerons are uncontrolled, and another method is therefore adopted. Countering or balancing the torque about the outer gimbal axis due to the weight T is a second torque which is applied to the outer gimbal ring through the lever U, Fig. 11. This lever is connected to one end of a coiled spring V, the other end of which can be rotated through
FIG. 10.Elevator Servo-motor and Mounting. fairy piston is displaced, however, in either direction, it uncovers one of the exhaust ports, and consequently allows the air pressure on one side of the main piston to escape. The pressure of air on the other side then causes the main piston to move in the same direction as the fairy piston until both the exhaust ports are again closed. In this manner a movement of the fairy piston, which is practically unrestrained by friction, causes the main piston to move by the same amount but with sufficient power to overcome any friction which may exist in the main elevator valve. The main piston of the relay valve is connected by a lever to the piston of the elevator valve shown at S, Figs. 8 and 9. It will thus be realised that if the pitch attitude of the aircraft should be changed by some disturbance then relative movement will occur between the pitch ring of the gyroscope and the aircraft. As a result, the fairy piston will be displaced, and, in consequence of the relay action of the fairy valve, the main elevator valve will be operated. The main elevator valve controls the
AUGUST 30,
1934
SUPPLEMENT TO
FLIGHT link, the whole arrangement forming a "Watt's linkage." This arrangement operates in exactly the same manner as the single weight described above, so far as fore and aft accelerations or pitch angle changes are concerned, but it has the advantage of insensitivity to lateral accelerations.
Tilt of Qyroscope Axis

When discussing the matter of stability earlier in this paper, it was stated that it had been shown both mathematically and practically that satisfactory damping of a disturbed motion could be obtained by the application of rudder angle in proportion to the roll displacement of the aircraft. A full explanation of the reasons for this would require mathematical treatment, and would be out of place in these pages. It will suffice here to state that th? :i application of rudder angle when the aircraft is banked Rear End. tends to increase the lateral stability in much the saint: way as if the ailerons were used to correct the bank. Separate control of the ailerons would require the use of a second gyroscope with a consequent increase in the complexity and cost of the installation, and, except for certain special purposes, or in an aircraft having little lateral stability, control of the ailerons is not usually necessary. ,'.. , Throughout the previous description of the Automatic Pilot it has been assumed that the axis of the gyroscope lies parallel to the fore and aft axis of the aircraft. It was shown by Meredith, however, that if the forward end of the axis is tilted slightly upwards, so that the axis of the gyroscope is no longer parallel to that of the aircraft, an application of rudder angle takes place when the aircraft is banked. When the axis of the gyroscope is parallel to the fore and aft axis of the aircraft, it is clear that no relative motion can occur between the gyroscope and the aircraft,
FIG. 11.Rudder and Elevator Control from
a small angle by means of a Bowden cable from the pilot's cockpit. The pilot is thus able to control the magnitude of the counteracting torque. Since the gyroscope must continue to precess in pitch so long as the two torques are unequal, it is clear that the normal pitch attitude of the aircraft is defined by the angle at which the gravitational torque exactly opposes the spring torque. When the pilot wishes to change his pitch attitude, he makes an adjustment to the spring torque, thus causing the gyroscope to precess and control the aircraft into a new attitude, at which the gravitational torque again balances the spring torque.
TKe Effect of Lateral
Accelerations
It has been stated above that the outer gimbal ring is put out of balance by the addition of the weight T, Figs. 6 and 11, and while the addition of this weight alone would permit the gear to function as described, it would be sensitive to lateral accelerations, which would cause small inaccuracies in the pitch control when the axis of the gyroFIG.IZ scope was displaced laterally from the fore and aft line of the aircraft, as will occur during a turn, or if a continual application of rudder is required to maintain straight AIR flight. In order to make the unbalanced mass, consisting of the outer gimbal ring and the weight T, insensitive to lateral accelerations, the weight is divided into two HAINJCONTROL parts, one of which is COCK attached to the outer gimbal ring, as shown at T, and the other of which is shown at W, Fig. 6. This second weight is carried by a pivoted arm which is connected to FIG. 12.Diagram of Pipe Connections. The Detail Connections on the Gyro the outer gimbal ring by a Mechanism have been shown diagrammatically for the sake of clarity.
62
SUPPLEMENT TO AUGUST 30, 1934
FLIGHT 898/
if the latter is banked without either turning or pitching. If, however, the gyroscope axis is tilted the relative motion which must occur between the gyroscope and the aircraft when the latter is banked will clearly have a component about the azimuth axis. This component of the relative movement serves to apply rudder angle by means of the rudder valve in exactly the same way as has been described for the rudder control. The total application of rudder angle is thus seen to be the sum of two parts, one being determined by the azimuth deviation of the aircraft and the other being determined by the roll displacement. Provided that the aircraft is reasonably stable in roll the damping of the disturbed motion of an aircraft fitted with gyroscopic control of rudder and elevators only is greatly improved by the tilting of the gyroscope axis. In such an aircraft it would only be necessary to control the movements of the ailerons if an improvement of the roll stability in rough weather were desirable, or if a highly accurate degree of stabilisation were required.
Preliminary Operation of Controls

While the foregoing description gives a complete picture of the operation of the gear while the controls are in use, there remain the preliminary operations and the apparatus connected therewith before the controls can be put into use. Referring to Fig. 12, it will be seen that the compressed air, at a pressure of 30-35 lb. per sq. in., first passes a " Test Cock" (e, Fig. 2), and thence communicates with the "Main Control Cock" (f, Fig. 2). The object of the Test Cock is to provide an easy connection for a compressed air supply for ground testing of the gear when the ordinary compressor system can obviously not be used. The Main Control Cock has three positions, in the first of which the compressed air is completely cut off from the Automatic Pilot. In the second position the compressed air is connected to the spinning jets which play on the periphery of the gyro rotor. The gyroscope runs at a speed of 11,000 r.p.m., and an interval of at least five minutes must be allowed to elapse to allow the gyroscope to reach its normal speed before the controls are put into operation by moving the Main Control Cock to the third position.
FIG. 13.View of " Tuning Table " on which the various Adjustments and Balancing Operations are carried out. The Table can be oscillated in Pitch and in Azimuth by1 means of the Electric Motor. Complete Tuning of a Rudder and Elevator Control usually occupies between three and six days. drawal of the cone and piston simultaneously seals the four communicating pipes from the servo-motors and supplies the compressed air to the various valves which thereafter control the air pressures in the respective servo-motors as has been previously described.
Qyroscope Centraliser
In Fig. 7 may be seen the exterior portion of the Centraliser unit. This unit consists of a spring-loaded piston and cone. The cone may be seen at Y, Fig. 6. In the normal unoperated position this cone is pressed towards the gyroscope and engages with a pin on the inner gimbal ring, which is accurately located by the apex of the cone. The gyroscope and gimbal rings are therefore locked when out of use, and the risk of accidental damage to the bearings is minimised. The centralising cone remains in the locked position while the gyroscope is being run up to speed, and so sets the initial position of the gyroscope when the controls are put into operation. At the back of the centraliser (Figure 7) several pipes may be seen. These communicate respectively with the two ends of the rudder and elevator servo-motors and when the controls are not actually in use, all these four pipes are in intercommunication. The pistons of the servo-motors are thus free to move in their cylinders and offer no appreciable restraint to the free movement of the control column and rudder bar by the pilot. When the main control cock is turned to the third or " In " position, the compressed air is connected to the pressure side of the centraliser piston which then compresses the powerful spring contained therein and withdraws the cone from the inner gimbal ring. The with-
Conclusion
It is inevitable that with the extension of long distance air transport the use of automatic control will become universal. The relief which is afforded to the pilot by the use of the Automatic Pilot permits him to devote the greater part of his time to navigation, map reading, and the operation of his radio installation, whereas with manual control, his whole time is occupied with the actual flying of the aircraft. Perhaps the value of the Automatic Pilot is most appreciated by a pilot who is compelled to fly tinder rough weather conditions, when visibility is bad and the control of the aircraft rendered difficult by the violence and frequency of the disturbances. Under such conditions the safety of the aircraft and the lives of the occupants depend solely on the skill and endurance of the pilot. The navigation of an aircraft at night is never an easy matter, and the strain on the pilot whose duties compel night flying under all weather conditions often demands the utmost endurance. Assisted by the Automatic Pilot, his task is rendered far lighter, and provided there is sufficient visibility for taking off and landing, the dangers which are otherwise inevitably associated with night flying are reduced to a negligible quantity. The Automatic Pilot can control the movements of an
63
AUGUST 30, 1934 SUPPLEMENT TO

aircraft far more accurately than the most skilled human pilot. It is important to remember that while a disturbance must actually have occurred before it can be appreciated and corrected by a human pilot, the Automatic Pilot detects a disturbance instantly with its growth and almost simultaneously applies the means for its correction. In consequence, the somewhat unpleasant motions of an aircraft in rough weather are very much reduced, and it is therefore probable that public opinion may, in the future, demand the universal use of automatic control in passenger carrying aircraft. The accuracy with which the Automatic Pilot can maintain the aircraft on its true course must inevitably result in a reduction of the fuel consumption. In practice, the course steered will not deviate by more than four or five
FLIGHT
degrees per hour, and thus only a very occasional correction is required by the pilot. The additional weight of the equipment is more than balanced on a long flight by the reduced fuel consumption and not infrequently the carrying of a relief pilot or wireless operator is rendered unnecessary. In comparison with the extra safety assured, the economies and the extra comfort afforded, the initial cost of installation is not high. The considerable experience of the working of the Automatic Pilot which has been gained during recent years by its use in several squadrons of the Royal Air Force has demonstrated the safety and tht> value of the installation, and it is safe to predict that within a few years, no passenger or freight carrying aircraft will remain unequipped with gyroscopic control.
NICKEL WIRE FOR SPARKING PLUGS

By A. G, AREND
ICKEL wire has been adopted as representing the most suitable economical metal for the sparking plugs of internal combustion engines. The intensely hot spark which is formed heats the zone around the gap to a very high temperature. Accordingly, nickel has been adopted as the most suitable metal for the central electrode points, whilst in many cases the side electrodes are stamped from nickel sheet to the required template. The successful manufacture of the nickel wire for this purpose has been the outcome of much research work, because nickel is one of the most impure metals on the market. There are, of course, many classes of the metal, but to the present day it is quite customary to find nickel which contains several per cent, of impurities. Such a metal as this is of little value for drawing purposes, and would only result in brittle and weak wire. It is thus recognised that only the purest form of nickel be used for wire drawing. When it is known that the wire has to be ultimately used in the making of sparking plugs, an addition of about 2 per cent, of manganese is made. This is better suited to withstand the effects of current and heat, but sometimes adds to the difficulties of successful drawing. The best nickel is obtained from the electrolytic process. This is first cut up into long strips which resemble ingots of small section. These are first hot-rolled, and later cold-rolled, until the section has been reduced to a thin rod. Heat treatment has to be performed between the different rolling operations to remove the strains which exist in the crystalline structure of the metal. Much more difficulty is experienced in this heat-treatment, than in the
annealing of copper, brass or similar metals for wire production. Brittleness appears during the reduction of the billet, before it has even reached the rod stage. This is due to the ready solubility of nickel oxide in the nickel. The result is that to-day the pure metal from the electrolytic extraction, is not directly rolled, but first melted, and a minute quantity of a deoxidising agent added, which ensures the removal of this undesirable oxygen. The metal is then poured into rod-ingot moulds, which thus economises in part of the rolling operations which would otherwise have been necessary. The final size of the rod is about Jin. diameter, after which it is reduced to the form of wire by drawing through dies. The drawing process does not differ materially from that adopted for the making of copper wire, since the relative ductility of copper and nickel are very similar. (These conditions, however, do not obtain in the rolling of the two metals as copper is much more malleable than nickel.) Ductility simply signifies the power of the metal being permanently extended by traction, and although ductility and malleability are closely associated in most cases, an exception to this exists in the drawing of nickel and copper wire. The plant used for the making of the nickel wire is the same as that used for copper wire. The wire required for sparking plugs is expected to possess a fair amount of rigidity, since although the shape must be capable of being altered from time to time, it must remain in such a position. The melting point of nickel is only surpassed by three
FIVE PHOTO-MICROGRAPHS t A specially-drawn Nickel Wire under diflEerent magnifications.
SUPPLEMENT TO
AUGUST 30, 1934
FLIGHT
SqSh

fallen so far as 5 per cent, of the original tensile strength, whilst in the cold condition. Some plugs are probably better insulated than others, whilst heating-up and vibration varies somewhat widely with different engines. In practically all cases the breakages of the wire are due to the temperature, and the carbon monoxide and acetylene only act as secondary measures. With most broken electrodes, the breakage took the form of intercrystalline fracture. The conclusions which may be drawn from this are that the assembling of the sparking plug parts is not so likely to be responsible for a,breakdown of the electrodes, as the method of production and preliminary heat-treatment of the wire. Further, nickel wire cannot recover in the ordinary way from the effects of overheating in a badworkfng" engine, and the strains are present in the wire and will develop to a fracture sooner or later.
metals of ordinary commercial importance, which are cobalt, chromium and platinum, all of which are more expensive than nickel. There is thus little likelihood of a substitute being used in place of nickel at the present time.
Examination of the Electrode Points

The small section which the electrode point offers does not assist the examinations which take place. When troubles are known to exist in the wire, a number of different nickel wires are removed from their respective sparking plugs. These are accurately ground and polished by a small machine specially adapted for the purpose. The small section of wire is then etched by immersing in dilute hydrochloric acid for a few minutes. It is then washed in water, dried, and carefully fitted in a mounting device with the assistance of some soft modelling clay and the etched surface examined under the microscope. Different minute cracks are sometimes found to exist in the electrode points. This was only ascertained after a number of sparking plug wires had fractured after prolonged use. The microscopical examination revealed that these cracks were due to intercrystalline complication's. They usually commenced from the outer skin of the wire and travelled inwards to the core of the wire. When substituting nitric acid for hydrochloric acid during the etching process, it was found that the former penetrated so rapidly that the crystal boundaries were impaired for examination, and that some of the grains fell out completely. Some central electrodes taken from old plugs showed a similar surface change, but in the ordinary way transverse cracks were practically never experienced. The wire of the sparking plug is known to be exposed to a certain amount of stress, due to differential expansion of different parts of the plug, or to vibration caused by the ordinary working of the engine. The same transverse cracks were practically never apparent in that class of plug in which bent side electrodes terminate in a point. It has been asserted that the chemical composition of the wire is unimportant, and that commercial wire will behave as well as pure nickel wire. This is scarcely accurate as it only alludes to those classes of wire to which manganese has been added. Many sparking plug electrode wires contain upwards of 3 per cent, of manganese, which neither impairs the electrical conductivity or heat-resisting properties, but this should scarcely be regarded in the light of impure nickel. The heating of this wire in air or oxygen differs from the conditions which obtain in the combustion chamber. When tests were first made on the wires, it was found by microscopical examination that the oxygen penetrated along the crystal boundaries. This formed a eutectic of nickel oxide which resulted in brittleness in the electrode wire. In actual working practice of the plugs, this is not so important, since the conditions inside the cylinder are different. Carbon monoxide and acetylene are present in the latter, which are more effective, than air itself. When the electrode points have been raised to a high temperature in these gases, intercrystalline brittleness is readily produced, and a fracture of the wire is always liable to result. Overheating caused by the local action of a strong electric current, can result in similar circumstances, although this is quite independent of the oxidation. Irrespective of what the conditions are, nickel wire is always liable to fracture, even when exposed to a small load, at high temperatures. A large series of tests were made in order to ascertain the relative strength of the wire at different degrees of heat. That is, the tensile strength was determined before being heated and then whilst the wire had been raised to the different pre-determined temperatures. The strength showed very clearly that it rapidly diminished, as the temperature rose, and in one extreme instance, it had
SUMMARIES OF AERONAUTICAL RESEARCH COMMITTEE REPORTS These Reports are published by His Majesty's Stationery Office, London, and may be purchased directly from H.M. Stationery Office at the following addresses: Adastral House, Kingsway, W.C.2; 120, George Street, Edinburgh; York Street, Manchester; 1, St. Andrew's Crescent, Cardiff; 15, Donegal! Square West, Belfast; or through any bookseller.
AILERON ANGLES IN HIGH SPEED MANCEUVKES WITH SINGLE SEATER
FIGHTERS.
By B. V. Williams, M.Sc, and J. H. Hartley, B.Sc.
Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1582. (8 pages and 6 diagrams.) September 11, 1933. Price 6d. net.
It was required to investigate by full scale experiments the movements of the ailerons on single-seater fighters during high-speed manoeuvres, in order to provide data for the revision of strength requirements. Synchronised records were obtained of aileron movement and air-speed in those manoeuvres which required large aileron angles at high speeds. On Aeroplane " A " the R.A.E. and Service Unit pilots applied their controls with the maximum force possible, considerably in excess of the force used in normal service manoeuvres. In the case of Aeroplane " B " the R.A.E. pilots applied their controls in a manner which they thought would cover most service contingencies, whilst a Sen-ice Unit pilot was asked to reproduce these manoeuvres in the normal service manner. The ailerons of Aeroplane " A " are so heavy that full aileron could not be applied at top level speed, whilst at high speed in the dive with engine throttled only a small movement of aileron can be obtained. It is possible to apply full aileron in an upward roll from a dive at 200 m.p.h., but this is because the speed has decreased below top level speed when the ailerons are applied. On Aeroplane " B " full aileron would appear to be applied seldom at top level speed or higher speeds, but it might be possible to apply it at such a speed by the use of abnormal force. A considerable amount of aileron can, however, be used, at speeds well in excess of top level speed, when diving with engine throttled. In similar manoeuvres at the same speed the Service pilot applied less aileron than the R.A.E. pilot.
EXPERIMENTS ON THE WESTLAND-HILL PTERODACTYL, MARK I V . PART
I. EXPERIMENTS ON A 1/5 SCALE MODEL. By A. S. Batson, B.Sc. PART II. FULL SCALE TESTS. By J. E. Serby, B.A. R. & M. No.
1577net.
(J9 pages and 21 diagrams.) January 5, 1931. Price is. 6d.
The object of the experiments was to provide a comparison of results on a larger scale model (one-fifth) with results of experiments made previously on a one-sixteenth scale model at Westland Aircraft Works, and to co-ordinate the results with those on the full scale, which are given in Part II of this Report. Lift, drag, pitching moment and controller hinge moment were measured at 60 ft. per second originally over a range of incidence from 2 deg. to + 33 deg. approximately and with controller or flap settings from -f 15 deg. to 45 deg. approximately. Incidence range was afterwards extended to 70 deg. for hinge moment measurements. The results agreed reasonably well with those on the smaller model. The slope of the curves, pitching moment against incidence, indicated that the machine should be longitudinally stable. Rudders of two different sizes were fitted to the wing tips and their effect on drag estimated over a range of incidence from 2 deg. to + 10 deg. Presence of the rudders increased minimum drag by 5.7 per cent, for the small and by 6.7 percent, for the large rudders. Various modifications to the model were also made and their effect on minimum drag was found to be small, the greatest reduction in drag, about 7 per cent, being that due to fairing the undercarriage. Variation of the wind speed showed little scale effect, but the results indicated a slight reduction in minimum drag and a slight increase in maximum lift with increase in wind speed. Lift and drag controller angles to trim have also been measured; minimum drag was carefully measured to compare with the exceptionally low value obtained in the model tests. The Pterodactyl design as developed in the Mark IV is, from a performance point of view, good, the very low minimum drag achieved by its clean body and undercarriage lines giving a good top speed and climb. The stalling speed is high, due to its rather poor maximum lift coefficient. The latter is difficult to avoid in this design, as the controllers have to be held at a large negative incidence to stall it and therefore cannot contribute to the Hit. Its geDeral stability and control is fairly good and comparable to that of a normal aeroplane, although the anticipated lateral stability at the stall has not been realised. The bad take-off involving a run of 250 yards, is the worst feature of the Pterodactyl Mark IV.
September 27, 1934.
Supplement to
1002a
FLIGHT ENGINEERING SECTION

Edited by C. M. POULSEN No. 104 ( V O 1 N V X ) 9th Year
September 27, 1934
STRENGTH CALCULATIONS FOR A TUBULAR SINGLE-SPAR CANTILEVER WING

BY F. DUNCANSON
From a description published in " Flight " of August 9, 1934, readers will be familiar with the Blackburn-Duncanson single-spar wing. The process of stressing such a wing is explained in the following article
HE methods of strength calculation and weightestimating for a single-spar cantilever wingfirst outlined in THE AIRCRAFT ENGINEER (Supplement to Flight) in March, 1929have now been put to practical test and their validity established by direct comparison with the usual handbook methods. The following article is illustrated by a worked-out example in order to lend reality to what would otherwise be rather an academic discourse. For purposes of illustration, the case adopted is that of a hypothetical twin-engined low wing cantilever monoplane (Fig. 1). The wing is of the straight tapered untwisted type.
The approximate formula used ior the load grading is now generally accepted as giving an equivalent result to that obtainable from A.P. 970. When it is considered that the usual stressing convention postulates that the load grading suddenly drops off to zero across the body (Fig. 2), instead of the more probable condition shown in Fig. 3, it will be seen that the crudity of this assumption rather swamps any slight differences due to refinements in load grading estimates. The following notes and calculations show briefly how it is possible to arrive at the design and weight estimate of a single-spar wing having a tapered circularor nearly circularsection spar situated at or near the maximum ordinate of the wing section.
Wing Structure Strength Calculations and Detailed Weight Estimate

Tapered Tubular Single-Spar Wing Span 70 ft. o in. Maximum chord, 15ft. Nett area, A = 735 sq. ft. The plan form of the wing and the load diagram may be represented, to a close degree of accuracy, by the equation y = kx$ When x 32.75', y = 15' /. k = 1 = 4.685 32.75* * D Allup weight, W = 10,886 lb., wing weight, w = 1,236 lb. Central load, W = 9,650 lb. Load factor, N = 7.5 NW x9650 ,, ,, J '.-:-.;* A 735 = 98.5 lb. per sq. ft. Equation to load diagram : y = 4.685 x 98.5*3 461*3
SUPPLEMENT
TO
SEPTEMBER 27,
1934
FLIGHT
1O0 2&

^
((^685x4
PIAN FORM & LOADING
-kxi \
-~-""
i 5'
LOAD FT/LBSjCHOKD 0
FIG.5.
15 20 25 SEMI SPAN FT SHEAR FOt
I
30
35
/
y
A
<
/
End ronjue
10000 LBS . 10" 15 TO SEMI SRftN Ft
FIG.6.
10 25 30 40 15 20 25 SEMI SPAN FT. 50 35
FIG.4.
Equation to shear force diagram: y = $bifxzdx = | x Eq uation to bending moment : y = ^6fx^dx = f x 346*3
= 148.2*3
The plan form, loading, shear force, and B.M. curves are shown by Figs. 5, 6 and 7. Relief loads due to engines, etc. : lb. Weight of engine unit .. . . 1,400 Weight of half undercarriage . . 240
1,640
For the purpose of calculation, each corrugated arc may be replaced by a plain arc of radius equal to the mean radius of the corrugated arc and of a thickness such that the cross sections of the corrugated and plain arcs are equal. 1.To find the height of the neutral axis above the spar centre line we have : Distance of the e.g. of each arc from x x, H chord
= rx arc
. a
sin= 2f . 2
A
1405 B 1405 /BODY FRAME
Factored load = 7.5 X 1,640 = 12,300 lb. On the starboard side the engine torque is subtracting from the relief B.M. due to the above weights. Engines give 640 h.p. at 1,200 r.p.m. : Engine torque, T = ''
^ 2iX
\ 39' -
= 2,810 1b. ft.

I,2OO 2 8lO
- -
<*
24"-
6150
6150
FIG.8
Reactions at A and B =
2
= 1,405 lb.
T J
Relief B.M. at root due to petrol and oil weight = 7.5 X 16.1 x 392
2 .
= 92,000 lb. in.
The relief bending moments are shown plotted on Fig. 7.
Method of Estimating the Moment of Inertia and Bending Modulus of the Spar Section
The following method enables a rapid and accurate estimate to be made of the moment of inertia of the spar section about its neutral axis.
HEIGHTS OF CGs OF CIRCULAR ARCS ABOVE THEIR CENTRES
CROSS SeCTiONAL AREAS OF ARCS H
FIG 10
1 3
12
II 1 0
in U J
\ \
\
1
H
s^
la
z
. >v
r
\ \ \ \ s\
t
4
/ / /
^-> ^^
s\ \
\
V
\
N \
6 5 4 2 } -i 5 AI'.C IN RADIANS 1-5

2-5
FIG 12
s\
3 ARC IN RADIANS
SEPTEMBER 27, 1934
67
SUPPLEMENT TO
FLIGHT
IOO2C
2.Height of NA above x x = 2
Zart
3.Moment of Inertia of an arc about x x .. .

.
Diameter = = 28in. / = .o8in. Developed width 1.7 P. 1st top arc, 2 3 radians Equivalent thickness 1.7 X .048 = .0817m. 2nd top arc 1.2 radians Equivalent thickness 1.7 X .064 = .io88in. Bottom arc 1.0 radian. Equivalent thickness, .io88in.
Height above XX. in.
0 0
I., =/
J
trddr* sin20
*i /
rH]
sin2 Odd
j_2
jBl
s i n 2 0 , s i n 2 0 2 ~|
~^+ ~A
92 61
Element.
Cross Sectional
J
Shell 1st top arc 2nd top arc Bottom arc
Area. sq. in. 7-5

2.70 1.87 i-55 -
Moment of Area about XX.
Since the arc is symmetrical about YY we have :

sin 2 0j sin 2 02 4 ~ j sin 2 0j 2
11.42
3 3-52
1381
3O-85 25-3 - 21.4
and
sin 2 (?j = cos (a .'. I** = a+ C O S (a - -
Total area 13.17 sq. in. Total moment 34.75 in.

h =
3475
n*7
Element. Shell 1st top arc 2nd top arc Bottom arc Total I about N.A. Moment of Inertia about XX. 690 373 346 300 Moment of Inertia about N.A. 739-5 228.5 225.7 4239
1,617.6 i n . 4
Moment of Inertia of arc about NA : I N A = lxx rcc*(H - h) 2 It has been found convenient, in order to expedite the work, to plot curves of cross sectional areas, heights of cg's and moments of inertia of circular arcs of various radii. These are shown in Figs. 11, 12 and 13 (Figs. 11 and 13 are for t = .iin.)
MOMENTS OF INERTIA OF CIRCULAR ARCS
D< > K
16
A* 17
y.not
1617.6 3 Zc = ' = 137.5m. o / J " 11.76 7 1617.6 . Z, = = 95.0m. 3
g,-6 S 15
y
FIG.15
IK
1-3 12 1
Zc-KD t
1-1 1-4 1 5 1 6 17 1 8 19 20 2-1 2-2 2 1 BENDIN6 MODULUS COEFFICIENT K
The Zc of this particular section may be expressed in terms of a coefficient x D 2 x /. 137-5 The coefficient for this section is 28 2 X .08 = 2.19 The section considered has a cross sectional area of 13.17 sq. in., is 28in. diameter and the shell thickness is .o8in. A plain tube of the same cross sectional area would be
n X 2 8
The following example is worked out to illustrate the application of the above method to a section of the proposed spar near the wing root. .
= .1495 thick.
.I495 The thickness ratio is therefore =- = 1.869

,Oo
SUPPLEMENT TO
SEPTEMBER 27,
1934
FLIGHT

(a) Wing span. (b) Wing area. (c) Thickness ratios at wingroot and wingtip. (d) The positions of wing engines, fuel tanks, etc. 2. The loads on the wing structure, comprising : (a) Load per unit area. * . (b) Load factor. (c) Weight of wing engines, fuel, etc., in relation to the central load. %^ 3. The strength/density ratio of the material used. A wing structural weight of 8 per cent, or even less may be achieved in quite a practical design having fairly low aspect ratio, large amount of taper in chord and wing thickness, small load factor, high wind loading and the use of a material of high strength/density ratio.
A number of similar cases have been computed and the results plotted on Fig. 15 which shows the thickness ratios required to obtain various Zc coefficients. The following procedure has been adopted in designing the spar and estimating its weight : At intervals along the length the gauges of the shell plating have been determined from considerations of torsional stiffness. The Z c coefficients have then been selected so that a proof stress of 16 tons per sq. in. in compression shall not be exceeded.* From the diagram shown in Fig. 10 the corresponding thickness coefficients have been read, thus enabling the complete weight of the shell plating and corrugated flanges to be calculated.
DETAILS OF THE WEIGHT ESTIMATE

Large Dia. D!ft. 2.42 2.42 2.42 2.12 1.89 1.61 1.36 1.082 .84 Small Dia. D 2 ft. 2.42 2.42 2.12 1.875 1.61 1.34 1.082 .825 .167 Plate length Plate Area including A butt (Di + Da)/ strap 2 ft. sq. ft. 6 5 8.25 8.25 9.5 8.5 8.5 8.5 10.5 45.65 38.1 58.9 51.8 52.3 39.4 32.6 25.5 16.6 Shell thickness t .08 .08 .072 .072 .064 .064 .048 .048 .028 Thickness Ratio. 1.89 1.83 1.83 1.76 1.76 1.55 1.55 1.4 1.2 Equivalent thickness T .151 .1464 .1318 .1268 .1125 .0993 .0745 .0673 .0326 Weight Ib. 144x.103 AT 102.5 82.9 115.3 97.5 87.4 58.0 36.1 25.5 8.1 613.3 58.3 15.0 47.0 36.4 7701b.
Item.
DJ+DJ
Mid portion, , 1st Section, .. ,, 2nd Section, > ,. 3rd Section, ,. Tip
inner outer inner outer inner outer inner outer
4.84 4.84 4.54 3.995 3.5 2.95 2.442 1.907 1.007
Weight of shell and corrugations Weight of diaphragms Rivet heads Frames and fittings at wing root and engagements Tank ends and petrol system fittings Total weight of spar
]\]
].]
.'."
... ...
Any material may be used, but Duralumin to Spec. L3 has been chosen for the present example.
ESTIMATED WING WEIGHT Item. Spar Ribs, stringers, etc., .33 lb./sq. ft. Wt. lb. Remarks. Calculated in detail after investigation of structural strength. In the case of the " Monospar " wing for the Fokker F.VII, this item weighed .26 lb./sq ft. (vide Aeronautical Journal, Oct., 1932). This is increased to .33 lb./sq. ft. to cover increased load factor requirements. In the case of the " Monospar " ailerons the weight was .5T lb./ sq. ft. This has been increased to .76 lb./sq. ft. for the same reason as in the case of the previous item. Proportioned from Segrave Meteor. Vide weight analysis of " Monospar " wing for Fokker F.VII.
77
242
Such a design would be desirable for a freight carrying aircraft to transport heavy loads at comparatively slow flying speed. Where high speed is of more importance, a wing of greater aerodynamic efficiencywhich must necessarily be heaviershould be used. Some day a formula may be discovered which takes account of all structural and aerodynamic conditions and produces the right answer to any given specification of requirements.
THE INSTITUTE OF METALS

The 26th annual meeting of the Institute of Metals was held in Manchester from September 3 to September 6. The thirteenth Autumn Lecture was delivered by Dr. J. L. Haughton in the Great Hall of the College of Technology, and dealt with "The Work of Walter Rosenhain." In the morning of September 4 there was a general meeting in the Reynolds Hall, where a civic welcome was given by the Lord Mayor of Manchester. In the afternoon visits were paid to the works of the Broughton Copper Co., Ltd., Metropolitan-Vickers Electrical Co., Ltd., W. T. Glover & Co., Ltd., Mather & Platt, Ltd., and the British Cotton Industry Research Laboratory. On September 5, at a general meeting in the College of Technology, papers were read and discussed during the morning, and in the afternoon visits were paid to the works of the British Copper Refiners, Ltd., British Insulated Cables, Ltd., Callender's Cable and Construction Co., Ltd., and the Chloride Electrical Storage Co., Ltd. In the evening there was a reception by the University of Manchester. The last day of the meeting was spent in making excursions by motor coach to Froghall, North Staffordshire, and to Buxton.
Skeleton ailerons, 72 sq. ft.
56
Controls in wings. . Fabric and dope, 735 sq. ft. at o-2 Ib./sq. ft.
21
i47
Total wing weight 1,236 lb. :, Wing area . . - * Weight per sq. ft... Wing structural weight
735 sq. ft. 1.69 lb. 11-3%
Astonishingly low structural weight percentages have been claimed for single spar construction, but these claims mean nothing unless the design conditions are stated at the same time. The most important of the design conditions are : 1. The geometry of the wing structure comprising particulars of:
SEPTEMBER 27, 1934
69
SUPPLEMENT TO
FLIGHT
I0O2
LONGITUDINAL STABILITY CALCULATIONS OF SEAPLANES ON WATER

In a letter which accompanied the article published below, the author points out that when, some time ago, he had to calculate the stability of a pair of floats for a ^seaplane, he had to evolve the following method. He is, of course, the first to admit that it is much easier to obtain the results with a model in the tank, but when no model is available the following method may be helpful
By H. PARKINSON
EAPLANE characteristics are readily obtainable when a tank and models are available, but in the absence of the latter, it becomes necessary to investigate the problem by the use of mathematics. The particular problem which these notes are intended to deal with is that of plotting a figure to show the range of longitudinal stability of the aircraft when at rest on the water. Having found that the metacentric height is suitable, the figure, or " Righting Moment-Trim " curve, may be proceeded with.
DISPLACEMENT 2870 LBS DATUM (DECK) -RIGHTING LEVER - C . G AIRCRAFT (287OLBS)
is then integrated in order to obtain the displacement to L.W.L. The result is added to Fig. 1, and the necessary calculations computed thereon. The draft position of the water line is a matter of trial, but an error of about x 2 per cent, in the estimated displacement should be considered satisfactory. In any case, a small error in the calculated displacement will have little effect on the later calculations for the position of the centre of buoyancy. If the L.W.L. is shown on the Lines drawing, it is possible to make a very close estimate of the new water line by inspection.
DISPIAUMCNT 1870 tSS. WATER LINE ' a ' 19 3 ' .HSj'l 2-46', S l \ 5-32* I B-tn', 8 - 5 ^ , 0-2^
~"l4> BY STERN L.W.L AT REST
STEP FOR* STEPAfT.
C.B AT REST

C.B. NOSE OF VftTER PLANE
FIG. 2.
?At this point it may be well to recall that independent float designs, which are numerically equal to Reserve The " Moment Curve " is now required for the purpose of Buoyancy, Total Displacement and Metacentric Height, of assessing the fore and aft position of the centre of are fairly certain to produce righting moments of different buoyancy. The required data are obtained directly from numerical values at the larger angles of trim. the Displacement curve in Fig. 3 and is obtained as indicated Apart from an artificial increase in righting moments by the appropriate heading on Table 2. The values shown by a corresponding increase in reserve of buoyancy, the in Table 2 are plotted on Fig. 3, and the resulting figure optimum values of the moments to cause instability, described as the " Moment Curve." The " Moment compatible with other requirements, is a factor requiring Curve " is then integrated to obtain the total moment of the displacement about the nose of the Water Plane. attention to the float shape in the early stages of design. The arithmetic in connection with the work is appre- Following this, the fore and aft position of the centre of ciably laborious, and one has found the graphical method, buoyancy is calculated as indicated on Table 2. A close used herein, to be a satisfactory substitute for the more approximation of the draft position of the centre of buoyancy may be made by eye. classical Simpson's Multipliers. The centre of buoyancy position may now be drawn on Referring to Fig. i, it is readily seen that at any angle Fig. 1 and the Righting Lever Z scaled directly from the of trim (f>, the righting moment is defined by : figure. Table 3 clearly indicates all the remaining calculaRM = W.Z. tions for obtaining the righting levers and righting moments where : RM = Righting moment. W = Total displacement to L.W.L. DISPLACEMENT ZS70 LBS Z = Righting lever. 3 0 NOSE OF WATER PLANE The foregoing condition defines the calculations as an MOMENT CURE DISPLACEMENT CURVE investigation into : (T) The positions of the centres of buoyancy for several arbitrarily chosen angles of trim over the range . of stability. (2) The righting levers at the chosen angles of trim. For the first stage, the Lines drawing. Fig. 2, will be necessary. Considering any water line " a " at an angle of 4> to the datum, the underwater areas at the various transverse sections are tabulated together with the appropriate distance from the Float bow This information is shown on Table i The information is then plotted on Fig. 3 under the title of " Displacement Curve," and the curve
SUPPLEMENT TO
SEPTEMBER 27,
1934
FLIGHT
1002 f

TABLE I Case : Displacement to water line " a . ' Trim io by stern.
Section
A B C D
The " Righting Moment-Trim " curve shown on Fig. 4, is obtained by plotting the appropriate information from Table 3. Should a cube law factor be used to modify the linear dimensions of the existing float to suit any other displacement, then the righting moments of the new flotation system follow immediately. In addition to the above, the percentage reserve of buoyancy must be numerically equal for both flotation systems.
DATUM (DtCK)
1
Area to W/L
0
0 0 .38
~^
E F G. For'd. G. Aft
H K
1.09
1.8
M O O TOO BOO FT. U MOO
N
\ \
\ \
J
.
BOWS AWA , ( r
DISHACEMEt*T
2S7O LBS.
LV,
^DATUM (DECK)
3"
L M Stern
2.34 1.99 2.17 2.17 1.98

1.5
Distances from Bow, Feet 1.19 2.46 3.81 6.32 6.7 8.52 10.25 10.25 11.48 13.08
34.Bo
.9 0
lfi.2 ' 17.85 19.3
SMO E000
1
TRIM YBO 5
s \
\
FIG.4-
OST AWASH STUNf
| m>
0
5" -4" J" 2'
\ \ AN01E4 srtm
3
ANGU Of L L./STR / V
TRIMI ^STER N
\
,
From Fig. 3 : Displacement to water line " a " in cu. ft. = 22.4 Displacement for two floats in cu. ft. = 22.4 x 2 = 44.8 Displacement in lb. = 44.8. x 64 = 2,870 lb. Sea water at 64 lb. per cu. ft. TABLE II Case : Centre of buoyancy position. Water line " a." Trim io bv stern.
Element Stern to M ,. .. L , K , H , G Volume cu. ft.
.89 2.9
\) 71
3" Irf
4"
5'
mo
BTBO us AteN E W VM
l'
|S
\
*1
sow
woo
000
s
/
Distance from Bow, Feet 17.85 16.2 14.65 13.08 11.48 10.25 .52 6.87 5.32 3.81
~.
, J
5.73
8.0
z =
RMl
RM W
=A ( z Vw
~
A
Stern to nose of water line
, F , F. .D
12.4 15.02 18.6 20.98 22.2 22.4
From Fig. 3 ; Moment of float about nose of water plane = 23.7 x 8 = 189.6ft.* C.B. aft of nose of water plane =
1
Where : RM = Known righting moment at W displacement. RMi = Required righting moment at A displacement. This also assumes that BG varies in the same ratio as the linear dimensions of the float. The slope of the " Moment Curve " in the region of the " at rest " position, is indicative of the metacentric height. As the slope of the " Moment Curve " is approximately constant for small angles of trim, it will be in order to assume an angle of i in the following : RM = W x Z Z GM Tan i RM Z ~~ W RM GM W tan Where GM = metacentric height. The metacentric height from Fig. 4 is : GM =
1400
= 8.46ft. 22.4
At Rest Trim by Stern
1 3 5 8 7 10
TABLE III
Trim by Bows Angle 6 Angled ... ...
9" 6" 3 3 0
11 11
.955 .143 .812
3
0 0 0
4
.5
" Y " f t . ... 3.972 3.218 1.854 .852 .508 .284 x 5.47 Sin 6 Z= Y - X 3.12 2.65 1.57 Righting Moment 2,870 x Z ft. lb. ...
1.271 1.805 2.167 .284 .475 .667 .987 1,330 1.500
.096 .404
8,940 7,600 4,500 2,330
1,160 2,830 3,820 4,300
2870 tan i c = 26.5 feet.
TO CONTRIBUTORS Contributors are reminded that articles submitted should, if at all possible, be typewritten, on one side of the paper only. Illustrations, particularly graphs and diagrams, should be clearly drawn, although it is not essential that they are drawn in ink. Pencil diagrams, if neatly drawn, are quite suitable, but they should not, as many of our contributors appear to think, be drawn in among the text. All illustrations should be on separate sheets and clearly numbered. It is also of assistance if an indication is given of where in the text it is desired to place each figure, although it is not always possible to place the illustrations exactly where the author wishes them to be, type metal being somewhat "inelastic." ._. ..... . _ . _
SEPTEMBER 27, 1934
71
SUPPLEMENT TO
FLIGHT
BOW-STRAIN
From the earliest days, the name T. W. K. Clarke has been associated with the technical aspects of aeronautics, and we welcome this contribution from an "Old Timer"
BY
TABLE
OF BOW STRAIN
Arc lengthchord chord
COEFFICIENTS.
Bow strain =
/d\' 'ly I for flat curve only
c bow strain coefl., d = camber and / = half length of chord.
Bow Strain Coeff. C. No, Curve and its Cartesian Co-ordinates. Formula. Numerical Value approx. Diagram.
T. W. K. CLARKE, B.A., F.R.Ae.S.
N connection with the bowing, due to end-load or to any other cause, of a structional member such as a strut, platework, etc., it is sometimes necessary to know the decrement of length due to this bowing. Whatever type of curve the bent member assumes, the ratio of this change in length to the original length, which
fd\
%
j
Circular arc.
O.GOT
Catenary or cosh curve V = a d a cosh a where a = 3-7

L'i
may be termed " bow-strain " lies close to 2/3 ( -j ) ; where d is the maximum bow deflection, and / is half the length of the bowed portion of the member. The appended table gives the numerical coefficients for various forms of curves ; these values are termed bowstrain coefficients. For example, suppose AB (see diagrams 1 and 2 of the table) represents a straight unstressed flexible wire, fixed to rigid anchorages at its ends A and B. Then, if this wire be stretched, within the elastic limit, by bending to the form ACB, the bow-strain of the curve will be identically the same as the mechanical strain set up in the wire, and if it were multiplied by E, Young's Modulus for the wire, it would give the stress in the wire. If its length were il and it were hanging freely with a sag = d it would assume the form of a catenary and have a " Bow-strain " coefficient (see No. 2 in the table) = - ; the stress in it would therefore be - ( j) .E If the anchorages A and B instead of being rigid were elastic, then it is easily seen that we must replace E by E' where -pr, = ^ + 7 where e is the " approach " of the h, b, I anchorages A and B per unit stress in the cable. NOTES ON THE TABLE The above table applies only to "flat curves," i.e., those in which the fourth and higher powers of the slope at any point, relative to the chord, can be neglected. In the case of the circular arc, bow-strain = the limiting value of R (9 sin 0)/R sin 6 = g02 = - j , where 2 0 is the angular length of the arc, and R its radius. In general, the bow-strain coefficient is calculated from the formula : bow-straincoefficient
1 2d
i
0.6(57
Parabolic arc. 0.667
Sine curve
y = o lii
0.C16
Half sine curve
y dcos
()
IB
0.616
Supported beam ccn trally loaded (or end loaded cantilever)
0.6
2 /d\'2
Supported beam formly leaded
unii'S 0.589
Clamped beam centrally loaded. This may be divided into four equal end loaded cantilevers, and is therefore the same as 6. Clamped beam formly loaded y= , An elliptic segment of camber dkx semi minor ax is. 10 ( Refer to its own axis. Two straight lines. An unsymme trical curve formed of two curves of the same type ; if the crown of the complete curve has an eccentricity e. In this case the coefficient is the same as that of each of the curves from which it is formed. But instead of the argument = we shall have Thus bow strain
0.0
0.61
5+A*
U.867-) 0.007
11
0.5
It will be seen that the value of the coefficient ranges from ^ for two straight lines (see No. 11 in the table) to slightly more than -, as for the case of an elliptic arc (see No. 10) or a curve whose curvature is greater near
2
SUPPLEMENT TO
n
SEPTEMBER 27, 1934
FLIGHT
1002A
its ends than in its central portion. As another example, let us consider the shortening of a bent strut of length il under an end-load P. If the strut is of uniform section, it will, under end load, tend to bend to a sine curve, if " tapered," to a curve intermediate to a sine curve and a circular arc. In the first case the bow strain coefficient is 0.616, while for a circular arc it is 2/3. An estimate then of say 0.64 will probably be quite near enough for practical purposes. If the load bends the strut to a known camber d, the shortening of the strut will be expressed by the equation :
. " ' " /dy d* (bow strain) x -a/ = 0.64 ( %-) 2/ = 1.28 -j If d is not known, but one knows 8 the " effective " Ve initial camber*, d = 8 approximately where P e is the Euler load for the strut. The substitution of this value d in the preceding equation gives the shortening of the strut for any end-load P. * Approx. = the initial camber -f 1.2 X eccentricity of the neutral axis due, for example, to want of uniformity in the thickness of a tubular strut.
SUMMARIES OF AERONAUTICAL RESEARCH COMMITTEE REPORTS These Reports are published by His Majesty's Stationery Office, London, and may be purchased directly from H.M. Stationery Office at the following addresses : Adastral House, Kingsway, W.C.2 ; 120, George Street, Edinburgh : York Street, Manchester; 1, St. Andrew's -Crescent, Cardiff; 15, Donegall Square West. Belfast; cr through any bookseller. Fisher, B.A. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1537. (26 pages and 16 diagrams.) February 2, 1933. Price is. gd. net.
NOTE ON A METHOD OF REPRESENTING SPAR TESTS. By H. R.
SOME CALCULATIONS ON THE STRESSES FUSELAGE OF A PARTICULAR AEROPLANE. INDUCED By H . BY G U S T S I N THE H. Fisher, B.A.
Communicated b y t h e Director of Scientific Research, Air Ministry. R. & M. No. 15S1. (18 pages a n d 7 diagrams.) J a n u a r y 31, 1933. Price i s . n e t . It has been suggested that gusts may sometimes give rise to serious stresses in an aeroplane fuselage and this report describes some calculations for a particular aeroplane which have been made with a view to determining the order of the stresses actually likely to occur. A method similar to that given in R. & M. 1463 has been applied to calculate the behaviour of the aeroplane when flying fully loaded at 116 m.p.h. into a gust perpendicular to the wing-chord whose velocity increases linearly along the flight path. Curves are given showing the history of the forces in various members of the fuselage. A gust of 30 f.p.s. is not capable in the conditions assumed of adding in any member examined a load of more than 0.23 of its estimated strength. Calculations for other cases, elasticity being considered, would be very unlikely to raise this figure to as ' much as 0.5. Sudden tail forces due to irregular air motions not amenable to calculation would have their efforts reduced by about 50 per cent, in the front bay of the rear fuselage, because of the inertia of the rear fuselage and tail. Fuselage stresses due to gusts thus appear in general unlikely to be of serious magnitude.
Professor Southwell has put forward a method* of graphical representation in which the points representing the observations iall approximately on a straight line whose slope is a measure of the load corresponding to elastic instability. It is thus possible to obtain an indication of the value of this critical load when elastic failure of the material prevents this load being nearly approached. A spar, which may have a small initial bowing, is1 supposed subject simultaneously R. & M. 1463. Acceleration of Aeroplanes in Vertical Air Currents. Part I, by to end load, which may be applied eccentrically, and to symmetrically distributed H. R. Fisher. lateral loading. Calculations of its behaviour, made on the usual assumptions, are put in a form showing the amount of deviation from linearity of the curve obtained by plotting deflection against deflection/load,* both when the lateral load WIND TUNNEL TESTS ON JUNKER TYPE AILERONS. By F. B. Bradis constant and when it is kept proportional to end load. The effect of zero errors in measurements of deflection and end load are con- field, Math, and Nat. Sci. Triposes, and W. E. Wood, B.Sc. Communicated by the Director of Scientific Research, Air Ministry. sidered. R. & M. No. 1583. (6 pages and 7 diagrams.) August 26, 1933. Experimental confirmation of the theory is obtained from tests on a steel bar Price 6d. net. within the elastic limit, and results are shown of applying the method to tests in which the elastic limit is exceeded. Wind tunnel tests have been carried out on ailerons of Junker type, to determine Below the elastic limit approximations of tlie Berry type make the method -of whether this is a suitable type for use when a very high degree of balance is required. Reference 1 applicable to spar tests, provided the deflection is appropriately defined. The aileron is an independent surface carried below and behind the trailing edge of The rough experimental material at present available suggests that it may also be the wing, giving in effect a slotted flap, but arranged so that the aileron is not shielded of use beyond the elastic limit. behind the main wing. This arrangement gives a larger rolling moment per unit aileron area than the balanced aileron forming part of the wing section. In the E.F.297. " On the analysis of experimental observations in problems of;elastic ^.following tests the Junker aileron has a chord width of 0.2c to give rolling moments stability." R. V. Southwell. Also Proc. R. Soc. (A), Vol.135,1932. JHJomparable with those of a Frise aileron of the same span and of chord 0.3c. The Junker type of aileron produces roughly 50 per cent, greater rolling moment EXPERIMENTS ON THE DISTORTION OF A STRIPPED TWO-SPAR METAL per unit aileron area than a Frise balance aileron of normal type, and the moment is less reduced at the stall. The yawing moments for a given rolling moment are WING UNDER TORSIONAL LOADING. By D. Williams, B . S c , A.M.I.Mech.E., and H. F. Vessey, B.Sc., A.F.R.Ae.S. Communi- somewhat larger on the Junker aileron except at the stall, when they are no worse cated by the Director of Scientific Research, Air Ministry. R. & M. than for the Frise aileron. The ailerons may be closely balanced.
No. 1571. (17 pages and 21 diagrams.)
April 19, 1933. Price is. net.

A N EXPERIMENTAL S T U D Y OF THE STALLING OF W I N G S . Aeronautics
In some recent work on the elastic deformation of two-spar aeroplane wings under torsional loading, theoretical results have been obtained on the convenient assumption that the rite connecting the spars are rigid in their own planes and so impose on the spars the same angular displacement as the wing at each point along the span. A subsequent investigation (4) suggests that actual ribs will behave very approximately according to this assumption, and the primary object of the experiments described in Part II of this Report was to discover whether ribs do behave in this manner, and if not, to find the effects of their flexibility. In other words, it was desired to check the accuracy of wing stiffness calculation on the basis of the " rigid rib " assumption, by actual experiment. Good agreement between theory and experiment was obtained and the basis of the theoretical work above referred to appears to be established.
Laboratory, Cambridge. R . & M. N o . 1588. (21 pages a n d 12 diagrams.) December 28, 1933. Price i s . 3d. n e t . The problems presented by the flight of aeroplanes near their minimum flying speeds are not yet completely understood. Predictions of behaviour in flight based on wind tunnel experiments have not hitherto been uniformly successful and do not explain why aeroplanes generally similar in design and having the same nominal wir.g profile may differ widely in the reputations which they acquire for behaviour, at low speeds. One cause of these difficulties may lie in variation in the mode of separation of the air stream from the upper* surfaces of the wings as incidence increases. For it is known that this separation may be either sudden or gradual and that sudden separation causes sudden changes in the air reactions, while the mode of separation mav depend upon small variations in wing form or surface roughness which may fall within the customary tolerances of full-scale construction. The experiments were accordingly undertaken to relate observations of turbulence above the surfaces of wings to the air reactions upon them, during the process of stalling. Measurements of lift, drag and centre of pressure, and of distribution of pressure and total head were made upon a series of wings about which theflowwas approximately two-dimensional. These measurements were correlated with observations of turbulence obtained by exploring the field of flow with very small silk threads attached to the points of thin wires. Observations of turbulence can give a good idea of the character of the forceincidence curve for a portion of a wing as it passes from the unstalled to the completely stalled state, and it is possible to recognise those changes of turbulence which lead to ambiguities and dicontinuities in the curves. The observations throw light on the process of the stall and the nature of scale-effect and suggest that, in certain circumstances, discontinuities which may affect aeroplane manoeuvre may not be revealed in the force-incidence curves obtained, either on the model or the full scale, by methods in use at the present time. * The words " upper " and " lower " will be used with reference to an aeroplane flying in the normal manner.
PART I. MODEL TESTS.
By A. V. Stephens, B.A., and J. Cohen,

J u l y 16,
B.A. Communicated by the Director of Scientific Research, Air

Ministry. R. & M. No. 1576. (10 pages a n d 2 diagrams.) 1932. Price gd. n e t . The Pterodactyl is controlled by large flaps on the wing tips, which function as elevators when moved together and as ailerons when operated differentially. There are also small rudders on either wing tip, so connected that the starboard one only is deflected when the right rudder is applied and vice versa. In the past ailerons and other wing tip controls have been found to be of little value in recovering from spins and it was thought possible that the controls of the Pterodactyl might prove altogether inadequate. The Pterodactyl exhibits unusual properties in that the ailerons are of paramount importance in establishing or recovering from a spin, whereas the rudders are ineffective. The tests show that rapid recovery from any spin should be obtained provided that the pilot is able to exert sufficient force to set the ailerons fully against the spin, or alternatively to move the control column fully^ forward. The existing rudders are inadequate and an appreciable improvement can* be effected by doubling their area.
November 1, 1934.
Supplement to
1152a

Editsd bv C. M. POULSRN
No. 105 November 1, 1934
FIGURES OF MERIT
From the earliest days of Flying, comparisons of the " Goodness " of one type of Aeroplane with another have been difficult. The " Ever ling Quantities," particularly the " High-speed Figure," have been used by " Flight " for several years. To them Mr. Rodger now adds another, the "Lift Ratio"
By R. RODGER
1 .Introduction
OR some years now it has been the custom oi FLIGHT, when offering technical descriptions of new types of aeroplanes, to assess the approximate relative efficiencies, both aerodynamic and structural, of such aeroplanes on a " figure of merit " basis. A bald statement of facts does not seem to create a clear-cut impression of relative values whereas the figure of merit, or index figure, possesses the distinct practical advantage of immediately emphasising one's ideas in a general sense. The statement of speed range as, say. 165-63 m.p.h. is less impressive than the corresponding ratio of 2.6, the single ratio at once providing a clue as to whether the aeroplane is good, bad, or indifferent in the particular characteristic under consideration. The example quoted is, of course, quite elementary, but the idea can, with advantage, be elaborated upon as, for instance, in the Everling Quantities. " On the other hand, however, the figure of merit should not be regarded as an absolute measure of efficiency. A machine may have special features inherent to the type which account for a poor figure of merit and apparently inferior design. One such example which comes readily to mind is the case of the large civil aeroplane. Fitted out as, say, a 15-20-seat transport, such an aeroplane would require internal cabin furnishings of not inconsiderable weight. Converting the same machine into a freighter of identical gross weight, most of the internal cabin furnishingsseating, upholstery, lavatory, etc.could be dispensed with, and would, in fact, be replaced by pay load, i.e., freight. As a transport such a machine might quite possibly"show a somewhat indifferent value for the ratio of gross weight to tare weight, whereas as a freighter the value of the same ratio might be passably good. Under these circumstances it is obvious that the figure of merit is a misnomer and conveys an entirely erroneous impression
because the true structural efficiency in both cases would be substantially the same. Iu comparing aeroplanes of more or less similar type, the figure of merit may reasonably be accepted as direct evidence. Where any appreciable divergence of type exists., however, it is desirable, probably essential, that more or less detailed information regarding the design should be available if the figure of merit is to assume its proper significance.
2.Symbols
For ready reference and in order to keep the subsequent text clear, the various symbols used and the significance attached thereto have been tabulated as under. W - gross weight (weight fully loaded), in lbs. \VT tare weight (weight empty), in lbs. V - speed level flight, in m.p.h. Vc --- climbing speed, in m.p.h. : P - engine(s) b.h.p. f(h) = factor for variation of engine power with' height. THP R thrust horse-power required. THPA -- thrust horse-power available. r) ~ airscrew efficiency. " 'i 1, =- lift, in lbs. =/cLpSV3 ; D = overall drag, in lbs. = kDpS\~* . k,. absolute lift coefficient. , kD --- absolute overall drag coefficient. ' p -- air density. . * S -- wing surface, in sq. ft. s = semi-span, in ft. ' D, = - induced drag, in lbs. c -- angle of induced downwash. to = induced downwash velocity, .....-, IT --- r e l a t i v e a i r d e n s i t y . ' p == a i r p r e s s u r e a t h e i g h t . . , ,;._. ,.;; p0 a i rp r e s s u r e a t s e a l e v e l . -..'., . : * * - ; ;
SUPPLEMENT TO
NOVEMBER
I,
1934
FLIGHT

3.The Weight Ratio
systemstwo or more superimposed aerofoilsthe induced drag is also a function of the gap/span ratio. " Span squared over weight" is, however, the preponderating factor and the influence of gap/span ratio is relatively small. Profile drag may be considered as included in the parasite drag group, the important ratio for the wings being the Hit/induced drag ratio.
INCIDENCE. AT INFINITE ASPECT RATIO
The simple proportion of the gross weight (weight fully loaded) to the tare weight (weight empty) has been termed the weight ratio and is assumed to be indicative of the structural efficiency, thus, W Weight ratio = .. . . (1) WT Mention has already been made of one precaution to be observed when comparing the weight ratios for dissimilar types. Another item which has a marked influence is, of course, the load factor, and as this is usually graded oft against gross weight the question of size is also involved. Tare weight is defined here as the gross weight less the disposable load, the latter item comprising crew, fuel, oil and paj- load or military load, as the case may be. The tare weight is sometimes inconsistently quoted in press descriptions and misleading weight ratios result. Attention will be drawn to this point again later on.
INDUCED DOWNWASH ANCLE
4.The Speed Range Ratio

The simple proportion of the maximum flying speed to the minimum flying speed has been termed the speed range ratio, thus, Vmax. Speed range ratio = \ . . . (2) Vmin. Here again ambiguity occurs in press descriptions as regards the minimum flying speed. Sometimes the stalling speed is quoted, sometimes the landing speed. The writer is of the opinion that the latter is the more important and should be standardised as the minimum speed. Referring to Fig. 1, D, = L tan but, hence, tan < = .-
D, =
(6)
The induced downwash velocity, to, has been correlated by Prandtl with several main features of the aerofoil, assuming a distribution of lift along the span according to a semi-ellipse, thus,
2L
5.Wing Power
This term was, the writer believes, first introduced by Prof. E. Everling in connection with the now well-known Everling Quantities. It is a combination of the wing loading and the power loading in which the gross weight term is eliminated, thus, Wing power = - ^ in h.p./sq. ft.
(3)
Substituting in (6). D, or. Substantiallv, therefore, L

~~ 1
2LS
L
2L
L =W
In FLIGHT the two principal loadingswing and power are nearly always quoted, even when the horse-power and/or wing area are omitted, and it is, therefore, sometimes useful to consider the wing power in terms of these two principal loadings, thus, P W P W /W Wing loading S = ~S~ X W = ~S / P = Power loading "' M .
' \v"'
y> ** '"
Lift Span3 *"" Induced drag """" ~ Weight ~ ' where, V = the indicated air speed, in m.p.h.
7.The Everling Quantities

These figures of merit were enunciated originally about eight years ago by Prof. E. Everling in the German periodical, Zeitschriftfur Flugtechnik und Motorluftschiffahrt, a translation of the article being presented in T H E AIRCRAFT ENGINEER, dated November 25, 1926. In this translation the original Continental units and symbols were retained, a rather unfortunate feature in the writer's opinion.
(The Continental units were retained because they enabled a direct comparison to he made between British and Foreign machines, of which many are described > "Flight." Otherwise one would have corresponding "High-speed Figures" of one nationality twice the value of those of another.Ed.)
6.Span Loading
In the opinion of the writer this quantity does not appear to be very well named as one might quite reasonably expect it to express load carried per unit run of span, say, lbs./ft. run. Actually its arithmetical significance is " span squared divided by gross weight," and as British nomenclature usually deals with the semi-span, the quantity is generally quoted as . Span loading ="7?7
(5)
The span loading is a qualitative indication of the induced drag, a high value for span loading being associated with low induced drag. Its significance will be better understood with the aid of a little simple arithmetical derivative analysis. The total drag of an aeroplane wing is equal to the sum of the profile drag and the induced drag. The profile drag is dependent on the wing section and its attitude, and is proportional to the wing area and the square of the air speed. It is unaffected by span, aspect ratio and truss arrangement. The value of the induced drag is proportional to the square of the lift and inversely proportional to the square of the span and the square of the air speed. It is independent of wing section and aspect ratio. This is true for all trusses, except that in the case of multiplane
Transformed into British units and symbols, the Everling Quantities reduce to quite simple expressions of a form appealing strongly to the practical reader. Their origin, however, becomes somewhat obscure in the process, and the various steps in their derivation are completely masked by the introduction of numerical coefficients into the final formulae. Using British units, and symbols the three Everling Quantities are generally defined as follows :
Vs v High sr>eed figure = r = 6 b zku 147,000 L V S = "/(*) P W
(8) (9)
Distance figure = - = .
75
NOVEMBER I , 1934 SUPPLEMENT TO
THE AIRCRAFT ENGINEER iv-2*.=.

W
FLIGHT 1152c
Altitude
figure
/w V" s
(10)
18.95
Each of these general formulae is derivable from the standard equations for level flight as explained in the subsequent notes. For the benefit of the non-technical reader very complete derivations have been given and the various steps have been kept as simple as possible.
(It would appear that Mr. Rodger is unaware of the fad that an article on the subject of comparison, of aircraft performances, by Mr. H. A. Me'aam, M.A., A.F.R.Ae.S., was published in ' The Aircraft. Engineer," of February 24, 1!)27, in which the Everling Quantities were derived in British symbols and units, and a comparison made between them and the " Coales Curves."Ed.)
In any case, the term P in (15) must be multiplied by f(h) to allow for height, and introducing the entire " height factor " we have, iku 147,000 P where Y = top speed level flight, in m.p.h., at height H ft. above sea level. P = maximum b.h.p. at sea level at r.p.m. appropriate to V. Note particularly the significance attached to P, which suffers a two-fold reduction due to (i) the airscrew characteristics which limit the r.p.m., and therefore the b.h.p. which the airscrew can absorb in torque, in relation to the forward speed V, and (ii) the deficient carburation due to the reduced barometric pressure and/or air density. Still further complications arise in the case of engines which are not normally aspirated, i.e., supercharged engines which develop their full power at some rated height, or ground boosted engines substantially maintaining their power up to some considerable altitude. It will be obvious that values for the High-speed Figure are determinate from any set of level flight performance data at any height, but only one of such results will be the one required, i.e., the optimum. For an accurate estimate of the High-speed Figure under all conditions it is, therefore, necessary to possess data for top speed level flight over a wide range of heights, variation of engine speed with aircraft speed, variation of b.h.p. with engine speed at sea level, and variation of b.h.p. with height. Such data are generally outside the scope of type technical descriptions appearing in FLIGHT, and the reader is usually confined to (15) for his comparison. This is fairly satisfactory if approximately similar types fitted with normally aspirated engines are being compared. When, however, other special features, such as those mentioned above, are characteristic to the aeroplane, one should use (8), and certain suggestions on this point will be made later.
8.The High Speed
Figure
For top speed level flight at any height, THP.A = THP. R But, THP. A = -nV and THP. R = 375 (12)
Equating ( u ) and (12), 375 By definition, D Substituting in (13), =
03)
(14) 375 Now, as mentioned above, the original Everling Ouantities were evolved in Continental units, and as it is often desirable to compare British aeroplanes directly with Continental types, it is advisable to render the formula? expressing the Everling Quantities in British units and symbols at once comparable with those in Continental units, thus eliminating the necessity for laborious and irritating conversions. The Continental coefficients for lift and drag, CA and Cw, are double the corresponding British coefficients kL and kn, and making allowance for this in (14) we have, 2ADSY3 P
V
or,
~~ ~375~ ^ 1 ? _ p/2_ SY;J
9.The Distance Figure

Yery simply, Distance (miles) = Duration (hrs.) x Speed (m.p.h.) but, Fuel supply (Jbs.) Duration = Consumption (lb./hr.) and Consumption = Specific consumption (lb./b.h.p./hr.) X b.h.p. whence, Fuel supplv (lb.) Speed (m.p.h.) Distance (miles) = 5 r^ ^-^ x r--x ' Specific consumption b.h.p. (lb./b.h.p./hr.) The specific consumption of the engine is here considered to be substantially constant over the entire range, the effects of height and throttling being ignored for present purposes; hence, for a given fuel supply, Distance o c p V and this ratio p is then itself a figure of merit for maximum range when the engine is throttled to give the maximum value of the ratio. At any speed level flight with the engine appropriately throttled, THP. 4 = THP. B DV whence, as before, r;P .. .. . . (13) 375 Y or, . D ' 375P W = L Substantially.
^ > ~~ 375 ' P For m.p.h. units, p --- 0.0051 at sea level, whence, 7 7 0.0051 S V 3 ikD ~ ~ " 750 ' P
JL _ ' (15) 2i, 147,000 This is the High-speed Figure a t sea level so familiar to readers of F L I G H T , where, V = t o p speed level flight, in m.p.h., at sea level. P = maximum b.h.p. at sea level. g In (15) t h e term p is obviously the reciprocal of the
wing power, whence, High-speed figure = 0.0000068 V3
a form which is sometimes handier than (15). Air density decreases with height above sea level, hence the value of p is proportionately less than 0.0051 at height
1
H feet above sea level, and the fraction - o o p is correspondingly reduced. The ratio of the air density at height H to the density at sea level is indicated by the symbol a, the relative density, by which value (15) must be multiplied. For normally aspirated engines the power output is reduced with height, but it is still a controversial point as to whether the variation follows more nearly the pressure or the density law. The favoured method in modern practice appears to be a fifty/fifty compromise between pressure and density, such that
76.
SUPPLEMENT TO
NOVEMBER I, 1934
FLIGHT

"V W For steady climbing flight, approximately, W= L hence, 1,-Pfjh) DV I_
(9)
whence,
or, introducing the " power factor," L V \V
^^Ws'WF)
where, V = any speed level flight, in m.p.h., at any height Hft. P/(/i) = b.h.p. with engine throttled to give THPA = THPR at speed V. Again note carefully the significance attached to the power term. In the case of the High-speed Figure we were dealing only with one specific point in the power curves, i.e. the intersection of the curves for THPA at full throttle and THPR as indicating the maximum speed level flight at the particular height considered. At some speed Vx substantially below the maximum speed there will be excess power available at full throttle, and the machine will climb. To maintain level flight at speed Vx it is thus necessary to throttle the engine so that THPA (throttled) equals THPR at speed Vx. This process can be repeated for all speeds and all heights within the flying range, so that obviously almost innumerable THPR and THPA (throttled) curves may be calculated. Only one pair of these, however, will determine the Distance Figure, the pair required being, as previously stated, that which gives the maximum ratio of speed level flight to power. This ratio is really analogous to the " miles per gallon " figure so familiar to motorists. The reader will now appreciate why the Distance Figure is seldom, if ever, referred to in type technical descriptions, an intimate knowledge of both machine and engine performance being essential if the Distance Figure estimate is to serve any useful purpose as a comparison. Such data is, of course, quite outside the scope of the usual technical descriptions.
whence, By definition,
L
1?f7=
V^ 375
. . (18)
W
V =
or, allowing for the difference between British and Continental units, V = Substituting in (18), KLP/2S_ W _ I
375
or.
tl
/W"
- - 3 7 5 P 2 Pf(h)
whence, writing p 0.0051 for m.p.h. units, 18.95 P/(A) ~J^
where, P = b.h.p. at full throttle at sea level at engine r.p.m. appropriate to the ceiling speed. o and f(h) have values appropriate to the absolute ceiling.
11.-The Lift Ratio

This is a figure of merit which the writer is suggesting here for, he believes, the first time. At least, he has never seen anything of a similar nature quoted previously in technical descriptions of new types of aeroplanes. The suggested figure of merit has the following significance, KLmax (from machine characteristics) Lift Ratio = K max (from aerofoil characteristics) L By definition. L = K,.pSYor, Substantially, Then, L =W
10,-The Altitude Figure

For steady climbing flight, Vc = ^ - ( T H P A - T H P K ) or, by previous reasoning,
.. (17)
If a climb be commenced at sea level with the aeroplane in the attitude corresponding to minimum power required, and this altitude be maintained throughout the climb right up to the absolute ceiling, the optimum rate of climb will not be realised at low altitudes, but the final altitude reached will remain unaffected. Assuming (17) to apply to such sea level conditions, then, V = speed level flight, in m.p.h., corresponding to minimum power required. P = b.h.p. at full throttle at engine r.p.m. appropriate to V. Since the altitude of the aeroplane is assumed to remain constant throughout the climb, the overall drag coefficient, kD, will also remain constant. But the mass density of the air will decrease with height, hence, for constant resistance, the speed level flight will increase with height. Therefore, at the ceiling the speed level flight, V, will have increased to V/\/a. Also, as previously, power will fall oft with height in accordance with f(h). Thus, at the absolute ceiling, (17) becomes.
At the ceiling, whence,

vPf(h) W
= O
. t:
DV 375 y/
Whence, from machine characteristics. Surface loading K L max.= p ( V landing)*' The other value, K,max. from aerofoil characteristics, is the full scale maximum lift coefficient for the basic aerofoil independent of such auxiliaries as slots, flaps, etc., and is determinate, with appropriate corrections, from tunnel data. The object of the Lift Ratio is to indicate to what extent the designer has cheated the basic laws of aerofoils by using flaps and/or slots, and by general layout of his wing truss to avoid loss of lift by interference and to take advantage of cushioning effects near the ground, short landing run, etc. It is suggested that the ratio might be useful in making preliminary estimates for a proposed new design by taking the landing speed as a datum and deriving the other quantities therefrom. In any case it would add to the value of type technical descriptions.
(To be concluded in our next issue.) ,;...!/'.
NOVEMBER I,
1934
77
SUPPLEMENT TO
FLIGHT
1152f
TESTING THE FILM STRENGTH OF LUBRICANTS

A Machine Which Also Measures Friction and Wear
SIMPLE machine has been designed which, it is claimed, gives accurate information on the loadcarrying capacity of lubricants. The machine also measures friction and calculates the wear-resisting properties of materials. It is manufactured by British Timken, Ltd., the makers of tapered roller bearings. Until recently reliable information of this character had been obtainable only through complicated and expensive laboratory research. The new machine has therefore proved a boon, not only to manufacturers and users of lubricants, but also to producers of all kinds of materials. The Timken lubricant and wear-testing machine was developed during research work in the company's laboratories. Originally it was used for obtaining measurements of the film strength of lubricants. Later it was improved and adapted to perform the three functions mentioned. The general appearance of the machine is as shown in Fig. 1. Fig. 2 shows the essential details of the load and testing lever system. The main features of the device are the lubricant container and return pump, the lever system, and the arrangements for direct or belt drive. The lubricant container holds about a gallon of lubricant, which can be raised to and maintained at any tempcraturo up to 210 deg. Fahr. by means of an electric hotplate. When the machine is in use the lubricant under test is constantly circulated by means of a pump in the base.
of the levers ensures that the test piece is always in the same position relative to the revolving cup and that the unit-loading over the length or surface of the two test pieces is always constant. Provision is made for either direct or belt drive by means of a 2 h.p. synchronous speed electric motor. An 8in. pulley with a 3m. face is mounted on an extension of the mandrel, and a key way is provided so that the motor may be coupled direct to the mandrel if desired. The oil pump is mounted in the base of the machine, and is driven from the testing mandrel. In operation the lubricant from the container flows over the test piece, the rate of flow being controlled by means of a valve in the pipe line. It then enters a sump, whence it is pumped back to the container. The mandrel is mounted on two Timken tapered roller bearings to preserve positive alignment, and is tapered at the end opposite the drive to receive the test cup.
Test Procedure for Lubricants

In using the machine for testing the film strength of lubricants, both test pieces, that is the cup and the block, i;re usually made of carburised steel, hardened to 60 deg. C. Rockwell and ground. Four tests may be carried out with one test block, but each new test requires a new cup or a reground surface of the cup. Newly ground faces on both pieces are necessary to ensure accurate and uniform results. Assuming both test pieces are in position, and the lubricant is heated to the required temperature, the driving motor is brought up to the desired speed and the lubricant is allowed to flow over the test block. The loading lever is then loaded by means of weights until the desired unit pressure is obtained. The full load should be applied at once rather than gradually, because the first few seconds of operation determine the possibility of sustaining loads when near the scoring point. Breakdown tests are usually run for a period of ten minutes, and the surface of the block after test is trufinal check. If the surface is very scratchy or scored, then the load-carrying capacity of the lubricant has been ex ceeded. Rubbing speeds vary according to the type of test, and as an example it may be observed that lubricants for automobile rear axles are usually tested at 400 r.p.m. (800 r.p.m. of the spindle). Light oils, such as free-wheel ing lubricants and motor oils, are tested at 200 r.p.m. Greases may be tested by using a special chamber which surrounds the test pieces. The chamber is filled with grease before the tests begin, and is kept filled by adding grease from a screw-down container, which is interchangeable with the oil container supplied with the machine. Care must be taken to add grease at a constant rate in order that tests may be truly comparative.
The Lever System

The lever system consists of two levers, one above the other. The upper lever, carrying the test block, is known as the "load lever" and is pivoted on a knife-edge mounted in the lower or "friction lever." The latter, which is also pivoted on a knife-edge, has a stop at the unloaded end, and is provided with a Vernier scale and a sliding weight for obtaining accurate measurements.
, ' 1^T*^?I
Use of a Chart
A chart provided with the machine shows the number of lb. avoirdupois placed on the loading lever in terms of 1,000 lb. of unit pressure on the test pieces. The chart also shows the turning speed necessary to produce any given rubbing speed, both by direct and belt drive. Standard formula- for use with the machine are given on the charts. Reference to Fig. 2 in conjunction with the following list of symbols used will make clear the method of taking readings and calculating results: A. Weight on the load lever. B. Weight on the friction lever.
1:
The Timken lubricant and wear-testing machine.
A tapered mandrel carries the testing cup, and a notch in the adapter carried on the load lever is provided to take the test piece Friction is measured between cup and test the Pû { r o m t h e con tainer flows regularly the two contacting surfaces. The arrangement
SUPPLEMENT TO
NOVEMBER I ,
1934
FLIGHT
i 15 h

Turning back to the question of eliminating the web in the ordinary I-section rod, this, of course, applies only to the bearing shell and not to the shank of the rod. So long as the bottom end of the web does not bear on the shell metal, the shank of the rod need not differ from existing design. Fig. 6 shows diagrammatically the ultimate design aimed for, and, if necessary, the flanges may be locally thickened at " A '' to compensate for the removal of web at this section. The lightness of this construction will readily be appreciated when compared with the usual type, which
over the whole area. The flanges, too, are strongly tied together at the centre, while having freedom to give under load at their outer edges. Fig. 4 indicates in an exaggerated manner the effect on the bearing of the central web and shows the unhappy load distribution. Fig. 5 shows the impro\-ed loading with removal of web from shell metal. Whilst the deflections indicated in Fig. 4 are of a fairly small order, they are sufficient to make themselves felt through the thin film of lubricant in the bearing with consequent ill-effects.
FIG.4.
UNIFORM LOADING ALONG ANY LONGITUDINAL SECTION THROUGH SHELL METAL WITH CONSEQUENTLY UNDITTORTED BEARING SURFACE
The author is of the opinion that many of the failures of big end bearings which occur from time to time, particularly in diesel engines, are due to the hammering of the bearing immediately under the web, and the inability of the shell metal to " wrap " itself round the journal because of the restricting action of the web. Thus an unequal distribution of load in the shell gives rise to an unequal stress and so an unequal deflection, which in turn means an unequal load on the journal, with dire results, especially at high peak loads.
tends towards complete solidity and rigidity, especially in the larger rods. At first sight it appears logical to apply the same treatment to the small end, but conditions are lebs severe, and the smaller overall dimensions make it awkward to deal with. However, if a substantial diameter is employed for the gudgeon pin the scheme may be used in certain cases with success. It is possible that considerable trouble may be met in practice through the development of fatigue stress at tb
It will be evident that the same principle is applicable to the cap half of the bearing, the inertia loads being equally distributed along the edges of the shell and so " wrapping " the bearing round the journal with uniform pressure. In present-day practice these conditions are much more nearly obtained than in the main half of the bearing because of the freedom from stiffening members ; and it is significant that by far the greater number of failures occur in the main half of the bearing in spite of the fact that the inertia loads may often exceed the gas loads. Reference to Fig. 2 will show that to obtain the maximum possible advantage the amount of stiffening given to the joint edges of the shells by bolt bosses, etc., should be a minimum consistent with safety. Otherwise, the arc of flexible shell metal may be seriously reduced.
radii " JB " B " (Fig. 6). Experiment would, no doubt, overcome this trouble, and it is not impossible to visualise a pin joint introduced at " B " " B " (see also Fig. 7) the pin being in multiple shear after the style of that used by Birkgit in the Hispano-Suiza engines.
The Paris Aero Show

In the November 29 issue of The Aircraft Engineer will be published an article by Mr. M. Langley, A.M.I.N.A., A.M.I.Ae.E., dealing with the subject of progress in metal construction of aircraft as exemplified by the machines exhibited at the forthcoming Paris Aero Show. Mr. Langley is an instructor in design at the de Havilland Technical School, and author of the book "Metal Aircraft Construction." He is thus thoroughly familiar with the subject, and will give readers of Flight the benefit of his views.
November 29, 1934.
Supplement to

No 106 9th Year November 29, 1934
METAL CONSTRUCTION AT THE PARIS

;., .. SALON -;,... i^;
By MARCUS LANGLEY, M.I.Ae.E., A.M.Inst.N.A.
Mr. Langley, who is Instructor in Design at the De Havilland Aeronautical Technical School, is also the Author of the book " Metal Aircraft Construction " : In the following article he records his impressions of this yearns Paris Aero Show
HE succession of Paris Shows every two years gives us a series of checking points from which we can gauge the development of the aircraft industry in Europe. My first object in visiting the exhibition this year was to see what had happened in structural design, and to judge or speculate on the effects that any changes might have on main designs and on production problems. In the Salon of 1932 there were evidently two schools of thought corresponding roughly to the English and French industries. The first, represented only by small military machines, had developed the wire-braced fabric-covered structure to perfection. It might be said that there was nothing in the layout of the machines as such which called for metal construction. The mere designs could have been executed in wood, though perhaps with more weight and a shorter life. The exhibits were very impressive in their simplicity and cleanliness, and there appeared to be little more development possible in the corrugated strip steel spar. The French, on the other hand, were developing metal-clad structures in duralumin, and it was giving them an opportunity to create types which would have been very difficult, if not impossible, to build satisfactorily in timber. Of course, one could see geographical and geological reasons for this in that aluminium and its alloys are much more a native product of France than of this country. In England, however, it is fair to claim that we know more of the higher tensile steels than anyone else. Each country, at that time, appeared to be developing its native products. This year there is distinct evidence of a breakaway amongst the British - firms. They, too, are now turning over to the monocoque fuselage and stressed skin wing.
If one considers what is also happening amongst the non-exhibiting countries, such as America and Holland, it becomes quite plain that the " metal-clad " is rising in popularity throughout the world. This is coincident with a rapid growth of commercial flying and also with the development of the low-wing cantilever monoplane. The fabric-covered wire-braced biplane still remains popular, however, in the limited class of small military aircraft. New materials were not very much to the fore. Elektron alloys have made surprisingly little advance for structural purposes, although they may be displacing aluminium for secondary uses and in engine construction, which it is not my purpose to discuss. Stainless steel sheet has become much more popular, and although it was particularly seen on the Hawker and Bristol stands in 1932, it is now found extensively in the French, Italian and Russian exhibits as well. There is, however, a particular reason behind this for which England can take no credit I refer to the increasing use of electric spot or " shot" welding, a process particularly suitable for the austenitic chromium nickelsteel alloys. A number of English constructors are experimenting with spot welding but their work has apparently gone little further than the making of specimens and of their mechanical testing. One firm, it was said last year, was spot welding a pair of seaplane floats in stainless steel. At the Paris Salon, however, there were already two very large floats made by different French manufacturers. There was a complete Italian flying boat and a Russian landplane in which spot welding was used throughout. It may be that we in England are not far behind in the use of this process, but if so we are keeping very quiet about it. The large Mureaux float made of stainless steel sheet was very impressive. The stiffeners were
SUPPLEMENT TO
NOVEMBER 29, 1934
FLIGHT

that is to say they are built up of corrugated strip tubes, but where the Bristol Company has used riveting or a wrapped joint, Poutiloff has used welding. The nodal joints in the structure are made up somewhat after the manner of the " Fairey spool," joined with bolts. The great criticism that I would make is of the workmanship, which in places would not have been passed by British Air Ministry Inspectors. One might also have doubts about the soundness of attaching wiring lugs to the spars by the method shown. . Spot-welding Duralumin
probably welded to. the shell plating on a fixed welding machine before erection. The final assembly would make use of the specially designed long-armed welding pincers. The Savoia-Marchetti which appears in the Salon is the second to be built with Budd shot welding. The first, which was built about three years ago, is now, I understand, back in America after its European tour of 1932. This machine has a two-spar fabric-covered wing structure which is conventional in its conception. But welding is used instead of riveting throughout. The material is thin stainless steel strip. The wing floats are built directly on to the underside of the lower plane.' The hull is completely welded together and made of stainless steel sheet and strip. This material has certain evident advantages on account of its corrosion-proof properties and also because it is extremely simple tc spot weld. The finished article is very smooth and clean and requires no painting. One might expect, however, that when one came to building a large flying boat it would be difficult to get sufficient rigidity with the thin sheet which would be necessary if the weight were to be kept within reasonable proportions. On the bottom of the hull, for instance, there are such loads when alighting in rough water that the extra bulk of a lighter alloy would give preater robustness. The Russians have made a particularly courageous effort in the use of electric welding on their Stal 2, which was shown on the stand of the U.S.S.R. : the machine was designed by Poutiloff. The structure is entirely of stainless steel but is covered with fabric. In layout it is a conventional high-wing semi-cantilever monoplane. The fuselage sides are vertical Warren girders, joined together across the top and bottom by struts ' with cross-wire bracing. The courage in this design lies in the use of welding in the fabri- A wiring lug on the Stal 2. cation of every part. The fuselage longerons and struts are of a section which is familiar to us in the work of the Bristol Company
On going up into the balcony I found yet more spot welding on a small stand belonging to the Sciaky Company. Possibly I should have known it before, but the name was new to me. What particularly took my attention was a fin or some.such small unit evidently built of duralumin and yet spot welded, which was leaning in the corner of the stand. I made further inquiries and found that this firm has supplied machines to practically all the French aircraft constructors, and that their English Company is responsible for the equipment of the Vauxhall car body works. Apparently Sciaky welding is used throughout in the Vauxhall pressed steel car-body and chassis. They claim to have developed a successful technique, in the spot welding of duralumin. The secret of this lies in the very short time given to the actual passage of the current, perhaps as small as one-tenth of a second, and associated with this a particular method of applying a variable mechanical pressure between the electrode points. The actual welding takes place at a temperature of about 5500, which is above the critical range of material. This overheating is very local, owing to the high conductivity of duralumin, and at the moment when the current is passing the mechanical pressure is relieved. It is exerted again as the current is cut off and this working is sufficient to retain the strength. The tabulated figures have been developed not in the laboratory but in the workshops of one of the leading French constructors. If rivets had been used of a size appropriate to the thickness of the sheet one rivet would have given about the same strength as one spot weld. This, however, is a very unsatisfactory comparison because welds can more easily be pitched closely than rivets. The possibilities of the method are enormous. One of the great objections which
RUSSIAN SPOT-WELDING : Wing and tail plane details on the Stal 2. Note the unusual spar booms. The numbers In the large drawing refer to the detail insets.
NOVEMBER 29,
1934
83
SUPPLEMENT 10

SHEARING TEST RESULTS AND TENSION OF WELDS IN ALLOYS AND STAINLESS STEEL. Thicknesses. in.
039 0315 039 0315 039 0315
FLIGHT
1 280c
LIGHT Strength. lb. 265

TQ9 132.5 no.2 199
1.LIGHT ALLOYS.
(a) Shearing Duralumin Aluminium Dural. + Alum. . . (b) Tension. Duralumin Aluminium Dural. + Alum. . .
2.STAINLESS STEEL.
+ +
+
+ +
.039 0315 039 0315 O39 O3I5
155
121.5
039 O39 + 039 039 0315 O3I5 039 O39
+ +
55-3 44-3 88.5
For Shearing strength, take 283 lb. for each o.oioin. of thickness, if a fixed machine is used. Take 173 lb. for each o.oioin. of thickness, if a portable machine is used. For example : .0315m. + .0315m. = 885 lb. on a fixed machine. When the two thicknesses are unequal, the strength of the weaker should be taken. .O2oin. + .0315m. == 5601b. on a fixed machine. h a s been raised t o t h e development of t h e tnonocoque fuselage and t h e metal-covered wing is in t h e expense of drilling t h e holes and closing t h e t h o u s a n d s of rivets which are required. I t would seem t h a t a more robust s t r u c t u r e could be m a d e in d u r a l u m i n t h a n in stainless steel, and t h a t none of t h e disadvantages of riveting would occur. T h e work would be quick and cheap, and t h e finish perfectly smooth, superior in fact t o t h a t obtained with countersunk riveting. F u r t h e r experiments are being m a d e b y t h e Sciaky Company b o t h in England and in France, and one m a y v e n t u r e t o forecast t h a t t h e spot welded duralumin monocoque and flying b o a t hull have a t r e m e n d o u s future. I t is t o be hoped t h a t no " snags " will be discovered, because t h e m e t h o d gives us all t h a t we could hope for. T h e Bristol Aeroplane Co. L t d . , was one of t h e first if not actually t h e first firm in t h e world t o build a metal
METAL MONOCOQUE
The Fuselage of an Avia 51
movocoque fuselage. T h a t was sixteen or seventeen years ago. After a period during which t h e y developed the steel strip s t r u c t u r e t o a high degree of perfection t h e y have returned t o t h e monocoqite. T h e low-wing cabin monoplane which t h e y show here in a half-finished s t a t e is probably t h e nearest approach t o t h e pure monocoqite in t h e whole Salon. The difference between t h e monocoque and t h e " metal-clad " is t h a t t h e first a t t e m p t s t o t a k e all t h e load in t h e skin, with a m i n i m u m of supporting structure inside, whereas t h e second h a s a s u b s t a n t i a l internal frame with t h e skin acting as little more t h a n shear bracing. Whilst it is relatively easy t o m a k e t h e rear end of a fuselage as a pure monocoque there are considerable difficulties a t t h e wing roots, owing t o t h e concentration of load. Most constructors t a k e these loads in a braced framework which gradually " dies down " towards t h e tail as t h e loads are dissipated. Mr. Pollard of t h e Bristol Co. h a s a t t e m p t e d t o keep t h e conception of t h e monocoque pure t h r o u g h o u t t h e length. T h e t w o deep main plane spars go t h r o u g h from side t o side a n d each is gripped between t h e webs of a double frame which might almost be considered as a vertical extension of t h e spar t o t h e t o p of t h e fuselage. Closely spaced stringers
BRITISH SIMPLICITY : Details of the wing roots on the Bristol 143. The letters refer to the enlarged detail v i e w s .
SUPPLEMENT TO
*4
NOVEMBER 29, 1934
FLIGHT
J2&Od
run across these two frames fore and aft along the fuselage'. At intervals there are transverse frames or hoops, which extend right to the skin. These are notched to allow the stringers to pass through and there are no connections between the two except that provided by the skin. Another extremely interesting nionocoque is that of the Avia 51. In this, however, the designer has not been able to forget his four longerons, which are considerably heavier than the other stringers. There are only seven main transverse frames throughout the length of the fuselage. One forms the engine bulkhead and there are two more double ones opposite to the front and rear wing spars. In between the main frames, however, there are smaller hoops, not attached to the skin but running round the inner flanges of the stringers. The main plane of this aeroplane is fabric covered and in one piece from tip to tip. It has two deep duralumin spars of N girder form and drawn duralumin strip ribs. The lines of the machines as a whole are pleasing, if one can get used to a rather smaller length/depth ratio than is conventional. Single'Spar Construction It was impossible to see the internal structure of the Dewoitine D511 metal-clad monoplane fighter. It is understood to follow that firm's standard practice. The wing has a single spar which runs across the fuselage. This is stiffened against torsion by the metal-covered leading edge, whilst the remaining two-thirds of the wing, also metal-covered, trail away behind, every rib being a tubularbraced cantilever. With this construction there must be a great concentration of load where the spar meets the fuselage, and the pure monocoque would be inconceivable. The stress would be too great for simple sheet to take. There is, therefore, a braced structure here which tapers out in four longerons toward the tail. Both longtitudinal and transverse members are riveted to the shell plating, the transverse ones being the deeper of the two. The Mureaux fuselages are somewhat similar, though of a more rectangular cross section. The skin acts as shear bracing for the flat rectangular panels of the sides which lie
CZECH METAL WING CONSTRUCTION The " N "-girder spar of an Avia 51. between the longitudinal stringers and the transverse frames. Even further from the monocoque but definitely within the class of the metal-clad is the big Bloch bomber. The fuselage gives an impression of straight lines and the whole structure has probably been designed by an engineer brought up in plate girder work for bridges. Although there is nothing subtle about it, an attempt was made to shroud the structional methods in secrecy. The main member of the cantilever wing is a rectangular-section box which tapers down from the centre to each tip. This has very heavy corner angles with thin sides which are stiffened by vertical Q sections. The whole is made from duralumin or Vedal, which is the French equivalent of our Alclad. It is very difficult to write of the two big Breguet machines. The low-wing cabin monoplane, designed as a commercial transport, was unfinished and the fuselage was frankly a " maquette " or " mock-up," though well made as such. The final fuselage is to be a metal-covered structure of some sort, as is the main plane. The Breguet 41M " Multiplace de Combat " is equally difficult to describe, but for another reason. It inherits
THE AMIOT BOMBER : Fuselage and wing in course of construction.
NOVEMBER 29,
1934
85
SUPPLEMENT TO
FLIGHT
12800
temperature is sufficient to braze the joint and thus lock it. On the same stand as the " Scimitar " was a small cousin from the " Avro Trainer'' family, one of the few oxy - acetylene welded fuselages in the Salon. The front and rear ends of the Letov 231 are of welded chrome molybdenum tubing. The centre portion is built of round steel tubes, squared at their ends and jointed with cover plates and tubular rivets rather after the manner of the Hawker structures. The undercarriage is very clean and all that can be seen at first A.W. double-joint tube. are two separate legs coming out from the fuselage side. Just inside the covering, however, they turn down again and run to a double hinge SPOT - WELDING : A Mureaux float being conat the centre of a floor cross-member. At the top of the structed in 18-8 stainless steel with a Sciaky machine. crook each leg is fastened to a shock absorber anchored to the bottom longeron. Two almost unnoticeable streamsomething from its smaller predecessor, the Breguet *Tcut line wires run down from the centre line to each wheel hub. Acier," that quaint machine which carried its tail on the These presumably take the horizontal components of the end of what looked like a broomstick. In the " Tout ground loads. The fuselage of the Avia 534 is " classique," which means Acier " Breguet used high-tensile steel as the only structural material. The swing back to duralumin has begun, but in tubular with wire bracing. This form of constrtction is of a kind which we know the 41M the pendulum has frozen half way. The top plane has two steel spars with vertically ribbed webs and well in England as extremely satisfactory. butterfly section flanges. Some of the ribs are duralumin Unorthodoxy and some steel. The smaller lower plane has a single box spar, of which the rear web is steel and the front of duraluHanriot has a standard fuselage adapted to three min. The vertical webs are joined by top and bottom different machinestwo trainers and an ambulance. It is surfaces of duralumin. The fuselage is entirely of duralumin, the rear end being a narrow tapering box, a compro- of square tubes but cross-braced with strips instead of the mise between the slim member of the " Tout Acier " and more usual wires. These strips are riveted into the joints at each end and have a remarkable tautness. The only the conventional form. feasible way of attaining this would appear to be by putting them in with temporary cross tubes in position, The Qerman Way these afterwards being replaced by very slightly longer and One of the most impressive exhibits was the Junkers permanent struts. Conventions are made to be broken Ju. 52-301, arranged as a seaplane. The construction of the when sufficient reasons arise, but in this case it is apparently fuselage and wing is of the corrugated duralumin cover just a question of difference. The wing spars are rect-. type which has been used by this firm for so many years. angular duralumin tubes and the ribs of three-ply. Potez has turned back to wooden construction for the Although they were not shown, Junkers have newer methods fuselage of his big 540 bomber. The wing is fabric covered, going through at the Dessau works. The Ju. 60, produced some time ago, had a flat smooth metal covering and they but with a simple two-spar structure of duralumin inside. The big Amiot bomber uses a mixed construction. The are building the Ju. 160 which is in the same class as the main plane is covered with duralumin but the basic strucHeinkel He. 70. This also has a flat covering. When higher speeds of the order of 200 m.p.h. are reached, the ture is tubular. There are three Warren girder spars, smooth covering may be worth 10 to 12 miles per hear built up of tubes, which are braced to each other, also by tubes. The joints are made up by pressings which are more. to form rigid sleeve couplings. The metal The exterior of the Heinkel He. 70 makes one suppose riveted together of the wing is clipped on to the spars and that it is covered with a particularly well finished plywood. covering A glance through the cabin dpor, however, shows a typical longitudinal members. The forward end of the fuselage and the underslung duralumin monocoque structure. When the next opporbomber's gallery is built up from duralumin channel tunity presents itself I shall solve this mystery ! The wire-braced fabric-covered biplane persists in some sections. There is a large area of window space, but quarters, and a group of very fast offensive military fighters metal covering is used in between. The shallow rear end must be mentioned in this connection. The Hawker is a long metal box built in halves and joined down the " Fury " and the Armstrong Whitworth " Scimitar " were centre line. There are four longerons with cross-members the two British representatives, and the Letov 231 and dividing the skin into almost square panels. Each panel is divided into twelve or eighteen smaller panels by light Avia 534 represented Czechoslovakia. The construction of the two British machines is so well stiff eners. It was too much to expect that the new big flying boats known that it requires no further mention. The " Fury " for the South Atlantic air mail service would appear in the was completely standard. The " Scimitar " is a development of the earlier Armstrong Whitworth fighters. Some Salon. Nevertheless, from the point of view of one brought very neat joints and specimens of the A.W. strip structures up with the sea salt in his hair, the flying boat side of the were shown. A method of making round tube from thin sheet industry was disappointing. The only flying boat was the Liore et Olivier H. 24. This eight-ton monoplane boat has strip was particularly ingenious. The two half-round strips are drawn together at their four engines in tandem pairs mounted on stilts above the beaded edges to form a tube. But in torsion, the beaded centre section. The lines and layout were not unconjoint is liable to open. A narrow copper coated strip is ventional, but the hull structure appeared on closer therefore introduced into each joint, and when the tube is examination to be somewha different. In England, if we heat-treated by the A.W. electrical resistance method the space the frames widely, we put the stringers close together,
SUPPLEMENT TO
NOVEMBER 29,
1934
FLIGHT I28of

or vice versa. It does not appear to matter which. But in this hull both are widely spaced and the shell plating seems to be uncommonly unsupported. The twin-tailed Levasseur seaplane is also odd, and can have only military significance. The snub-ended fuselage has a tubular framework covered with ribbed metal sheeting. The two attenuated floats are structurally ordinary but appear to be very narrow-waisted and may not be suitable for open sea work. The upper works of the Junkers Ju. 52.3-m. has been discussed above, but its floats may be appropriately mentioned here. In England our only experience of the verylarge twin-float seaplane has not encouraged us in its development. Both the French and Germans have gone further with this type, but their tideless seas, the Mediterranean and the Baltic, may be more encouraging than our wilder waters. The Junkers floats are a " hefty " piece of work, straightforward and solid. The only other marine flavour was given by the stainless steel-welded floats mentioned at the beginning of this article. It will be very interesting to hear how their very thin plating stands oip to the buffeting of rough water.
FRENCH BOAT CONSTRUCTION : Internal structure of LeO H.24 hull.
FIGURES OF MERIT
By R. RODGER {Continued from page 76)
12.-Typical Data
In Fig. 2 an attempt has been made to compile data relating to figures of merit for a number of the more up-todate British. American, and Continental civil aeroplanes. The writer considers the use of the word " attempt " to be advisable because the attainment of anything approaching correlative results is almost a heartbreaking task. The utter disregard for consistency in the presentation of the basic data shown by various authorities is truly remarkable. Sometimes normal power output is quoted, sometimes the maximum. Sometimes the gross weight quoted is the maximum permitted by the C. of A., whilst the performance refers to the aeroplane flying light, and exactly how light is not always mentioned. Tare weight sometimes includes cabin furniture, etc., whilst for another machine these items may be referred to as part of the disposable load. Even the main plane area seems to be a doubtful quantity as one instance arises in which the wing loading is based on main plane area including ailerons and body. Under these circumstances it is absolutely necessary to accompany Fig. 2 by a commentary to prevent, as far as possible, misapprehension of the results on the part of the reader. In the table alternative values are often quoted and in such cases there is inconsistency in the basic data, the reader being referred to the commentary for the nature of the adjustments made.
" '
13.Commentary
In compiling Fig. 2, the greatest difficulty of all has been that of assessing the power appropriate to the high speed. The powers given in column 3 are those quoted by the various authorities stated in column 21. In order that the reader may appreciate fully the nature of any differences, Fig. 3 has been prepared giving such data as one can find for the engines concerned. Airspeed Courier.There is no consistency whatever between the quoted values for weights, wing area, power, and loadings. Wing area 250 and wing loading 13.5 give .gross weight 3,370. Power 240 and power loading 15.7 give gross weight 3,770. Weight empty 2,100 and load 1,690 give" gross weight 3,790. The quoted gross weight is 3,900 which, with load 1,690, gives tare weight
2,210.
Stalling speed is quoted. Landing speed has been assessed proportionately lower taking the Envoy as a guide, q.v. The values for kLMAX seem rather high. Airspeed Envoy.Two gross weights are quoted4,900 as performance weight and 5,300 as C. of A. maximum. Wing loading appears to be quoted on maximum weight, and power loading on performance weight and normal power. I have adjusted the loadings to performance weight and maximum power. Armstrong Atalanta.Pcrwer loading 12.2 appears to be quoted on a maximum power of 470. I have been unable to discover a Jaguar type giving 470 maximum at 4,000ft.the height at which the top speed is quoted. If maximum power 470 occurs at 4,000ft., the high-speed figure is 18.7. If, however, this power refers to sea level and is not maintained to height, the corresponding highspeed figure is estimated at 21.4. Avro 642.Tare weight quoted at 7,360 is exclusive of radio equipment 90, which has been added to give tare weight 7,450. Power loading quoted at 12.8 appears to be on normal power 460 at 2,000 r.p.m. I have assumed that the engine is a geared Jaguar giving 490 maximum power at 2,200 r.p.m. The power loading has been estimated on this figure. Stalling speed is quoted. Landing speed has been assessed as for the Courier. Avro 652.Power loading quoted at 12.24 is on normal power. On the maximum power 305 the power loading is 10.8 The corresponding wing powers are 1.32 and 1.49, respectively. These adjustments reduce the high-speed figure from 27.6 to 24.4. Avro Commodore.From the quoted figures for wing area 307, wing loading 10.82, power 215, and power loading 14.45, the gross weight is returned at 3,320 instead of 3,50a as quoted, reducing the weight ratio from 1.57 to 1.49 and increasing the span loading from .40 to .42. If we accept gross weight at 3,500 and wing area at 307, then the wing loading will increase to 11.4, and the wing power to .74, reducing the high-speed figure to 20.2, and raising KLmax. to .89. The corresponding power will be 227. Blackburn Feeder &-In view of the high value of the weight ratio I suspect that trie tare weight quoted at 4,740 is exclusive of equipment, although the Ehmcanson spar
HU
2i^lLrUi
Weight lbs.
Truss.
IVltlJ Basic Data.
run
Spans.
î V l l ^
A E 8peeds. M.P.H. Weight Ratio. Figures of Merit High 8peed Fig. Speed Range Ratio. S.38 3.09 2.86 2.74 2.60 3.11 3.25 2.82 3.54 2.3fi 2.57 2.62 3.46 3.44 3.21 3.65 3.51 3.57 3.25 3.81 3.41 3.45 2.50 3.43 3.97 3.73 16 Source of Basic Data and Speeds. Span Loading. Wing Power. Remarks.
1 !
Loadings.
High.
Low.
Maker and Type.
Power Plant
Gross. Airspeed Courier Airspeed, Envoy Armstrong, Atalanta ... Avro, 642 Avro, 652 240 Lynx IVc. 2 x Wolseley A.R.9 11 203 at 2,420 r.p.m 4 x 400 Jaguar 2 x Jaguar VID 460 at 2,000 r.p.m. 2 x Cheetah 277 at 2,100 r.p.m. 215 Lynx 2 X Rapier VI 805 at 3,500 at 10,000 ft. 2 X Pegasus I. M2 2 x Jaguar VIA 450 h.p. 186/200 Gipsy Six 4 X Gipsv Six 205 at 2,350 r.p.m 2 x Gipsv Six 205 at 2,350 r.p.m. Cvclone F.2 700 at'1,900 r.p.m. 2 x Cyclone 710 at 1,950 at 7,000 ft. 2 x Menasco Buccaneer 230 h.p. Wasp S.I.D.I. 550 at 2,200 at 5,000 ft. 2 x Wasp Junior 420 at 2,200 at 5,000 ft. Wasp S.I.D.I. 550 at 2,200 at 5,000 it. Wasp S.C.I. 450 at 2,100 at 6,000 ft. Cyclone SR.1820.F.3 Approx. 700 h.p. Cvclone SR.1820.F.3 710 at 7,700 ft. Wasp S.I.D.I. 525 at 5,000 ft. 2 x Siemens Jupiter 550 h.p. 3 x Cyclone R.1820.F.3 712 h.p. B.M.W. VI. 630 h.p. 3 x Whirlwind 420 h.p.
3 O
aM Q
Tare.
Wing Ib./sq. Power, Upper,
tt
Lower. Speed Height Land- Stallb./h.p. tt in. t t in. ing. ing. tt. 47 0 52 4 90 0 71 3 56 6 37 4 58 4 54 0 54 0 39 3 64 6 48 0 48 0 85 0 42 0 42 9 55 0 42 9 41 0 48 0 48 0 48 0 91 10 90 2 48 6 54 5
9 64 37 4 54 0 54 6 0
* 1.32 0.94 1.01 0.77 0.89 0.85 0.90 0.76 0.67 1.02 0.65 0.69 0.74 0.92 1.01 1.00 0.97 0.97 0.97 0.93 1.12 0.95 1.12 0.75 1.12 1.03 1.36 20
WF Wing Flap* V.P.A. Variable Pitch Airscrews.
I
M
PI
nM n
3,900 3,790 4,900 23,000 11,790 6,598 3,500 3.320 8,600 10,500 9,400 2,400 9,200 5,000
2,210 15.6 18.25 2,100 15.15 15.8 3,077 14.55 12.18 13,600 17.7 7,450 16.2 4,192 16.1 12.2 12.04 12.24 10.8
162 170 165 160 175 130 196 195 175 170 170 165 225 210 183 230 221 225 195 221 215 200 155 202 234 224 11
S.L. S.L. 4,000 S.L. S.L. S.L. 10,000 5,000 S.L. S.L. S.L. 1,000 3,000 7,000 S.L. 5,000 5,000 5,000 6,000 7,700 7,700 5,000 S.L. 5900 S.L. 7,200 S.L.
12
48 55 56 64 50 63 60 62 48 72 66 65 61 57 63 63 63 60 58 63 58 62 59 59 60 13
55 63 64 63 14
1.77 1.59 1.69 1.58 1.58 1.57 1.49 1.82 1.72 1.57 1.59 1.67 1.55 1.75 1.61 1.70 1.52 1.49 1.77 1.76 1.65 1.59 1.74 1.71 1.75 1.71 1.68 1.50 1.42 1.36 1.82
15
0.96 1.20 1.45 1.34 1.32 1.49 0.74 0.70 1.49 1.66 1.36 1.03 1.28 1.37 1.22 1.96 1.51 1.64 1.87 1.83 1.87 1.62 1.95 1.95 1.42 0.92 2.08 1.56 1.58 2.60 17
.07 .5 .56 .35 .43 .48 .40 .42 .39 .28 .31 .64 .45 .46 .32 .40 .37 .32 .34 .32 .35 .33 .33 .33 AH .41 .32 24 18
30.1 27.8 18.7 21.4 20.8 27.6 24.4 20.2 21.3 25.4 26.2 26.8 32.9 26.3 24.4 25.0 36.1 34.3 33.8 32.1 25.4 24.1 38.1 36.2 34.6 32.9 35.7 33.9 26.0 24.7 29.8 27.4 26.0 33.0 31.4 27.5 25.0 25.3 24.0 55.8 44.3 29.1 27.9 19
Jane's 1933 Aeroplane Jane's 1933 Flight 5.4.34 Flight 7.6.34 Flight 31.5.34 Flight 7.6.34 Flight 6.4.33 Flight 7.6.34 Aeroplane 25.7.34 Flight 22.2.34 Flight 26.4.34 Flight 22.3.34 Flight 1.3.34 Flight 6.7.34 Flight 16.8.34 Aeroplane 11.4.34 Flight 16.8.34 Flight 16.8.34 Flight 8.2.34 Jane's 1933 Flight 8.2.S4 Flight 22.2.34 Jane's 1933 Jane's 1933 Jane's 1933
21
ts
w

> I-H
r
BRIT
Avro, Commodore Blackburn, Feeder Boulton* Paul, P.64 ... Boulton & Paul, Feeder British Klemm, Eagle ... De Havilland, 86 De Havilland, 89 Airplane Development Corp., Vultee V.I. Douglas, D.C. I Lockheed, Alcor Lockheed, Altair 8E ...
LOW WING CANTILEVER MONO
IXI-CXI f\
2,225 11.4 15.45 10.82 4,740 18.3 6,125 13.0 6,000 13.1 1,510 12.0 12.3
8.4 9.6
l>dkff><i
f 1
DITTO
11.7
5,520 14.35 11.2 5,950 15.30 2,851 14.9 3,102 4,275 20.0 11,875 19.2 12,080 2,665 16.7 12.2 10.2 12.7 10.2
Corrected by Capt. Broad.
48 0 10
m
o
m
7,250 18,000 4,710
R.P. V.P.A. R.P.W.F. 4 V.P.A. R.P. V.P.A. R.P.W.F. & V.P.A. R.P.- -W.F. & V.P.A. R.P.W.F. & V.P.A. R.P. V.P.A. R.P.W.F. & V.P.A. R.P.W.F. & V.P.A. R.P W.F. & V.P.A. N.P. R.P. W.F
W.F
w o
R
M
Q
5,800 9,000 5,800 4,750 7,000 7,000 7,000 17.600 19,841)
3,297 19.72 10.55 5,455 19.64 10.71 3,640 19.72 10.55 2,725 17.02 10.55 4,100 19.3 4,000 19.3 4,100 19.3 9.86 9.86 13.6
a Lockheed, Electra
o
y
at
Lockheed, Orion 9D ... Lockheed, Vega Northrop, Delta Northrop. Gamma Northrop, Victoria Dornter, Do. F Fokker, F. XX Heinkel, H.E.70A Pander, Postjager Col. 1 and 2
ft
A
1 1
DITTO DITTO MI0-WINO SEMI-CANTILEVER
10,500 14.75 16.1 13,220 19.25 5.190 18.2 5,410 6.655 25.0
6 7
rE
9.25 11.70 11.53

9.6 8
m
_ Fl
4
*
7.380 ! 12,100 5
R.P. W.F.
22
SUPPLEMEKT TO
88
NOVEMBER 29,
1934
FLIGHT
12 80 A
may be responsible. There FIG. 3ENGINE PERFORMANCE is no evidence as to what the tare weight of this machine Normal Maximum. Type. Maker. Authoritycomprises. tajft Hgt No loadings are quoted. H.P. R.P.M. H.P. R.P.M. The wing loading 18.3 is 1,900 S.L. 240 Jane's 1933. Armstrong Lynx (ungeared) 215 2,090 S.L. based on gross weight 8,600 Lynx (geared) Armstrong 2,000 S.L. 238 215 2,200 S.L. Jane's 1933. and wing area 470. The Rapier VI is quoted as giving 275 2,100 2,400 S.L. Jane's 1933. S.L. 305 Armstrong Cheetah V ... 305 at 3,500 r.p.m. at 10,000ft. Jaguar (ungeared) . . S.L. Jane's 1933. Armstrong ... 400 1,700 440 S.L. 1,870 I have no data for the Rapier S.L. 490 Jane's 1933. Jaguar (geared) 460 2,000 2,200 S.L. Armstrong ... VI but the performance quoted is the same as the Maker. Bristol 555 2,000 625 5,500 Pegasus I M2 4,500 2,800 normal performance of the VIu Swan. 600 1,520 Up to Up to B.M.W. . ... 640 1,650 Rapier II given by Swan 7,200 7,200 (see Fig. 3). Taking a risk, I 184 2,100 S.L. S.L. Flight 22.2.34. 205 2,350 DeHavilland Gipsy Six have assumed the maximum performance of the two Napier Rapier II 3,500 Swan. 305 10,000 350 3,900 10,000 engines to be the same also, 2,200 Jane's 1933. Pratt & Whitney ... Wasp Junior S.B. ... 425 5,000 445* 2,310 5,000 i.e., 350 at 3,900 r.p.m. at Jane's 1933. Pratt & Whitney .. Wasp S.C.I 450 2,100 470* 2,200 6,000 6,000 10,oooft., giving a power loading of 12.3. Wasp SI-DI 550 2,200 5,000 575* 2,310 Jane's 1933. Pratt & Whitney 5,000 The top speed is assumed S.L. Aeroplane 11.7.34. Wolseley 185 2,200 203 2,420 S.L. A.R.9 Mk. II to occur at 10,oooft., the Whirlwind 420 2,150 S.L. S.L. Jane's 1933. 440* 2,260 Wright corresponding altitude factor Wright Cyclone SR.1820.F2 . 1,900 745* 2,000 3,000 710 3,000 Jane's 1933. being introduced into the high-speed figure. Wright Cyclone SR1820.F3 . 1,900 710 7,000 745* 2,000 7,000 Jane's 1933. Boulton &- Paul P.64. * Estimated at approx. 105% normal. The power loading is on the maximum power 625. Boulton & Paul Feeder.On gross weight 9,400 and wing is based on gross weight 18,000 and engine performance 710 area 717 the wing loading is 13.1 against 12.55 quoted. at 1,950 at 7,000. which is assumed to be the operational No power loading is quoted so has been based on the maxi- height in the abrsnce of definite data. mum power of the geared Jaguar, i.e., 490 at sea level, Lockheed Alcor.I have no data for the Buccaneer so which is assumed to be the operational height in the absence have assumed high speed 183 to occur at sea level. of the data to the contrary. Lockheed A Hair.No power loading quoted. Power British Klemm Eagle.Data straightforward. No com- loading 10.55 is based on gross weight 5,800 and engine ment. performance 550 at 2,200 at 5,000, which is assumed to be D.H.86.The tare weight quoted at 5,520 is exclusive of the operational height in the absence of definite data. wireless equipment 130, lavatory 30, and cabin furniture Lockheed Electra.No comment. 270a total of 430. Adding this item to the quoted tare Lockheed Orion and Vega.Comment similar to the weight reduced the weight ratio from 1.67 to 1.55. Altair. The wing loading 14.35 is based on main plane area Northrop Delta.No power loading is quoted and the 641 inclusive of body. From the scale drawings in power at 700 approx. High speed is quoted at 8,000ft. " Flight " the wing area exclusive of body and fillets Working from the Northrop Gamma with the same engine, is estimated at 600, increasing the wing loading to 15.3 and power is taken as 710 at 7,700, on which the power loading the wing power to 137, thus reducing the high-speed is based. The operational height is also assumed to be figure from 26.3 to 24.4. 7,700. It is interesting to note here that in the formula for Northrop Victoria.Disposable load is quoted at 3,400 estimating the handicap speeds in the MacRobertson which is the same as that quoted for the Delta, plus 500 for race the wing area permitted is inclusive of ailerons, but radio, cabin, and miscellaneous equipment. Hence the exclusive of fairings and wing surfaces which form part weights have been taken as for the Delta. Power loading of the top or bottom of the fuselage. is quoted at 9.8 which gives power at 714 and appears to D.H.8Q.Main plane area 336, apparently exclusive of be an error. Power loading has been based on power 525. body, and wing loading 13.7 give gross weight 4,600, but Operational height has been taken at 5,000 in the absence of power 410 and power loading 12.2 give gross weight 5,000, definite data. as quoted. On gross weight 5,000 the wing loading is Northrop Gamma.Tare weight is quoted at 3,500 14.9, which is used in the table. Tare weight quoted at but in view of the similarity of this machine and the Delta 2,851 leaves a balance of 1,311 for cabin furniture, lavatory, I have added 500 for radio, cabin, and miscellaneous equipwireless equipment, extra fuel, and pay load. The D.H.86 ment as quoted for the Delta. Operational height has been allows T,939 as pay load for 10 passengers. The D.H.89 assumed at 7,700 in the absence of definite data. with 6 passengers has been assessed proportionately at General Note on American Machines.The powers quoted 1929 x .6 = 1,160, the difference 1,311-1,160 = 251 being added to the tare weight to give 3,102. This reduces the in Col. 3, Fig. 2 for the American engines appear to be those corresponding to the official rated engine speed, and weight ratio from 1.75 to 1.41. have been listed in Fig. 3 as normal performance. ReferVultee V.I.No loadings are quoted. Wing loading ence to Section 74 of " Airworthiness Requirements of Air 20 is based on gross weight 7,250 and wing area 362. Commerce Regulations for Aircraft " issued by the United Power loading 10.2 is based on gross weight 7,250 and engine States Department of Commerce (Aeronautics Branch) performance 710 at 1,900 at 3,000. reveals the requirement that airscrews shall be so designed No operational height is stated but the high speed 225 and adjusted that they will limit the engine speed to T05 is assumed to occur at the rated height 3,000. per cent, of the official rated engine speed at full throttle. Douglas D.C.I.Two tare weights are quoted. With 14 British engines are rated in a rather different manner, passengers 11,875 and with 18 passengers 12,080. No both normal and maximum performance being quoted. loadings are quoted. Wing loading 19.2 is based on gross Taking an average of the data for British engines given in weight 18,000 and wing area 940. Power loading 12.7 {Concluded on page 1287.)
December 27, 1934.
Supplement
1382a

Edited bv C. M. POULSEN
No 107
9th Year
Dêmbar 27, 193t
WING FLAPS
By R. P. ALSTON
Below are given the more technical aspects of the paper on " Wing Flaps and Other Devices as Aids to Landing " read by Mr. Alston before the R.Ae.S. on December 6
Wind Tunnel Tests
HE validity of wind tunnel tests at low Reynolds numbers has always been open to doubt in the matter of maximum lift coefficients, and even compressed air tunnels are not entirely above suspicion on account of turbulence. There is, however a saving grace where split flaps are concerned, which is that various different tunnels agree in giving unaltered stalling incidence when the flap is in operation. The change in AL max. due to the flap is therefore probably not much in error. Fig. 2 shows the effect of flap angle on the change in maximum lift coefficient for a split trailing edge flap running across the whole span. The curves are taken from Gruschwitz and Schrenk's experiments for a 20 and 5 per cent, chord flap; the other points are from American experiments on a 20 per cent, flap; the increase in drag coefficient at kL = 0.5 is plotted in Fig. 3. It will be noticed that no advantage in lift is gained by having a larger flap angle than 60 deg., though the drag increase continues to rise with flap angle. The effective drag increase obtainable with a flap is, of course, more than that shown in Fig. 3 since there will be extra induced drag due to operating at a higher kL. The effect of flap location is shown in Fig. 4, whence it appears that the further back the flap is, the greater the maximum lift coefficient. The effect on drag is negligible. Unless some mechanism such as the Zap is utilised to slide back the leading edge of the flap, the furthest aft location is limited by the flap chord in the closed position. The effect of varying chord for total span split flaps deflected 60 deg. is shown in Fig. 5, where the loss of lift due to forward location is seen to exceed the gain due to size when the flap chord exceeds about 30 per cent, of the wing chord. Drag continues to rise, however, as would be expected. . ^ The effect of variation of the flap span has been investigated in detail for a 20 per cent, chord flap at 60 deg.
deflection, cut away at the tips and centre in turn. The results are as one might expect ; removal of the flap from the centre causes a rapid drop in lift and so disturbs the span distribution that the induced drag is increased ; removal of flap from the tips does not decrease the maximum lift seriously. It is a pity that these experiments were not associated with any measurements of tail flow since there is a certain amount of full-scale evidence that in order to maintain sufficient downwash at the tail on low-wing monoplanes it is essential to run the flap right across the centre of the wing span. Comparison of the lift curves at equal values of maximum lift shows a less sudden drop at the stall when the centre is cut away than when the tips are cut away. One of the first machines to be fitted with a split flap was the Northrop " Sky Chief," on which the flaps ran from the end of the wing-body fillet to the wing tip. " Park bench " ailerons (which will be discussed later) were fitted. It was stated by Mr. Northrop that the stall on this aeroplane was not violent, a fact which may be attributable to the non-flapped centre portion of the wing. Before going on to discuss the various forms of special ailerons that are required if flaps are run right out to the tips it may be as well to determine what can be done with partial span flaps. It has already been shown that a flap chord of 20 per cent, and a deflection of 60 deg. need not be exceeded from a lift point of view. On a typical tapered wing monoplane, the ailerons need not exceed 40 per cent, of the span so that the central 60 per cent, is available for a flap which, on a rectangular wing, would give an increase of 0.325 in & L max. Owing to the concentration of load at the centre of a tapered wing this figure may be increased to 0.35. For a wing loading of 16 lb. per square foot and a & L max. of 0.7 (no flap) the stalling speed is 67 m.p.h., which would be reduced to 54.5 m.p.h. by the flap. The increase in gliding angle at 20 per cent, above stalling speed would be 6 deg.. which would be adequate to bring even a very clean aeroplane above the war-time standard of
90
SUPPLEMENT TO DECEMBER 27, 1934
FLIGHT
1382b

SPLIT FLAPS
VARIATION OF MAXIMUM LIFT COFFlCINr WITH FLAP ANBIE Ob
STALLING SPEEDS
ore
SPLIT FLAPS 010
VARIATION OF DRAB COEFF. AT CONSTANTKi-O-5
FI6.2.
IOJ
f 4. X
N
20*
-
(FLAPCHORD-OZc) 0-08
/ r
1
006
a
01
O SCHRCNK X TN42} (CkrkY) + TN47ifaar/<Y) A CAT./&AI4S)
0-04
0-02
7
FLAP ANGLE
FIG.3.
0,
04 06 OS 10 IZ 1-4 MAXIMUM UFT COEFFICIENT
1-6
20'
40'
60' FlAP ANGLE
80'
100'
BO'
20'
60"
8 I T 1 0 0 '
IM'
LOCATION of SPLIT FLAP
0-5
OSCHKNK
HINGE MOMENTS of SPLIT FLAPS w<a + W47Z

1
020
020
0-4
f
60'SPLIT FLAP
VARIATION OF CHORD
45'015 M 015
>
0-2SC
,0-3
Pcsv1 y r 010
010
005
'.05
/
~ 0IS<
0-1
FIG.5.
05c 0-4c 0-3c 0-2c 0-lc DISTANCE FORWARD OF TRAILING ED6E 0 0-1 c 02c 0-3c FLAP CHORD 0-4e 0-lc 0-2e 0-Sc
FLAP CHORD
FIG
20' 40" FUP ANQLE
60"
TO deg. Estimates of the effect of the flap on the Douglas D.C.2 suggest a reduction of stalling speed from 68 to TYPES of LANDING FLAP 59 m.p.h. and increase in gliding angle of 3 deg. The minimum distance to come to rest after clearing a Tooft. obstacle is calculated to be reduced from 2,000 to 1,350ft. This calculation is based on the assumption that the approach is made at the minimum speed from which complete flattening out is possible in each case and that the coefficient of friction on the ground is 0.3 with brakes. It should be noticed that any excess of speed on the approach will result in a greater increase of run for the unflapped aeroplane. It appears, therefore, that in order to make a clean, heavily loaded aeroplane reasonably easy to land it should not be necessary to have full-span flaps. This conclusion would appear to have been justified in christened " Park bench " in the U.S.A. According to the practice since no full-span flap is believed to be in use Zap Corporation, who use this type in conjunction with a full-span Zap flap, the exact location of the aileron and its commercially now except on the Pander Postjager. zero setting is very critical. A feature of this type of Special Ailerons for Use With Full'Span Flaps aileron is that its rolling moment coefficient is increased when the lift flap is down so that adequate control should If a split trailing edge flap is run right out to the wing be preserved at low speeds. Possible objections to this tip some unconventional form of aileron must be employed. type are its extra weight, drag at high speeds, and likelihood One method is to use ailerons in the normal position but of causing torsional distortion of the wing owing to its so arranged that they move upwards only, through a large far back location, but full-scale tests in this country will angle. American model tests show that the angle has to shortly settle these points. be about 70 deg. to give rolling moments comparable with those of conventional ailerons. Ailerons of this pattern Pitching Moments Due to Split Flaps have been tried in flight on the Fairchild F.22 monoplane with a full-span split flap and are believed to have been The effect on longitudinal trim when a flap is pulled abandoned on account of excessive hinge moments. An down is rather complex since it involves three factors, the alternative developed by the N.A.C.A. is a retractable backward shift of the wing centre of pressure, the lateral curved plate aileron which need involve no aerodynamic shift of the &L ~ a curve, and a change in downwash over hinge moment; these ailerons have been used successfully the tail. There is also the possibility of the turbulent on the F.22, but they must involve considerable extra wake causing a drop in tail efficiency. The simplest conweight. dition to consider is that of unaltered speed when the flap Leading edge spoilers or interceptors appear very pro- is down, i.e., kL is kept constant, which means that the mising from model tests, but when tested in flight by the incidence is reduced. Consequently there is a nose up' N.A.C.A. they gave an initial reversed control equivalent pitching moment produced by the reduced incidence of the tail. At constant kL the downwash at the tail should to a half second lag. The only remaining solution is an auxiliary aileron be practically unaltered for a full-span flap. From model mounted above the trailing edge of the wing, a type tests on split flaps of chord varying from 15 to 30 per cent.,
91
DECEMBER 27, 1934 SUPPLEMENT TO

the wing pitching moment has been plotted against incidence for various flap settings at constant kL (0.5). dkm/da from these points has a value of 0.009 per degree. An average tail plane designed to give adequate longitudinal stability may have a value of 0.007 f r dkm/da, so that there is a net nose down pitching moment of 0.002 per degree change of incidence at constant kL. For full flap angles the change in incidence is about 12 deg., giving a net pitching moment coefficient of 0.024 only. Some tests made by the Zap Corporation on models of the XOJ-i biplane and Parnall Parasol monoplane even show resulting nose up moments at a given kL and practically no change in moment at maximum, lift. In these tests it is hard to see where the additional nose up moment comes from. Failure to carry the lift right across the centre of the wing, either due to interruption of the flap or to bad body-wing interference will reduce the down-wash at the tail for a given lift coefficient and almost certainly make it impossible to attain stalling incidence if that incidence is only just attainable with no flap. In order to ensure that stalling incidence is just attainable with and without flap it may be necessary to gear together the flap and tail plane adjustment or alternatively to arrange that the gearing between elevator and stick is increased as the flap is pulled down. The former system was adopted in the Fieseler 97 in the Rundflug competition. This machine has a Fowler flap involving a very big C.P. movement and its tail plane is set down 16 deg. by full flap movement. In spite of this its minimum speed figure (engine on) only gives a kL of 1.57, so it is quite likely that there was insufficient longitudinal control to trim the aeroplane to the incidence for maximum lift, though engine power and lateral control may also have been inadequate. The sharp drop in lift at the stall that has been noticed in model and full-scale wind tunnel tests of split flaps has given rise to some anxiety on the score of a possible large loss of height in recovery from a stall, combined with acute lateral instability. Mr. Relf has suggested that on this account the longitudinal control should be limited to prevent the attainment of stalling incidence, but full-scale evidence on this point is needed.
FLIGHT
1382c
opening at high angles. For a simple hinged trailing edge flap there is a theoretical relation between hinge moment, flap angle and lift coefficient due to Ff. Glauert; this relation has been checked very closely by pressure plotting on a R.A.F. 30 flapped wing and by direct measurement on a R.A.F. 31 model. The results show a straight line law for kH against flap angle at a given kL and the value of kH at 45 deg. flap angle is identical with that for the split flap given in Fig. 6.
The Slotted Flap

The principle of operation of this flap has already been referred to, and it would be expected that the drag would be comparatively low. Wind tunnel tests on a Handley Page slotted flap of 20 per cent, chord ratio show an increase in kL max. of 0.445 at 40 deg. deflection and an increase of only 0.018 in kD at kL = 0.5. With the slot blocked up AhL max. = 0.32, Aku = 0.017 a t *L = -5- Some other tests on slotted flaps show them to be the type having the least increase in drag for a given increase in maximum lift coefficient. With regard to hinge moments, no figures are available, but it appears possible to locate the hinge in such a position that the flap is well balanced and has a suitable slot opening. There is no advantage in having the flap angle greater than 40 deg., so that the range of movement is less than that required from a split flap to give the same increase in maximum lift coefficient.
Air Brakes
All the devices discussed above have been regarded principally from their effect on maximum lift, and the drag has been left to look after itself. If KL max. is assumed to be increased by a flap from 0.6 to 1.0 then KL on the glide will be increased from 0.45 (at 15 per cent, above the old stalling speed) to 0.7 (at 20 per cent, above the new stalling speed). For an effective aspect ratio of 6 this means that the induced drag coefficient is increased from 0.021 to 0.051. The parasitic drag may be increased any amount from 0.016 for a slotted flap to 0.089 f r a 9 deg. split flap, so that the total increase in gliding angle, given by tan- 1 Akn/KL ranges from 2.8 deg. to 11.5 deg. This can be said to cover any normal requirement, ranging from the case of low drag increase where improvement in take-off is required to quite a high drag increase where short landings are required. There are other means of increasing drag alone without altering the wing lift characteristics, but they are not very effective. A retractable undercarriage which is badly faired when down, rotatable strut fairings, flaps projecting from the fuselage, and such devices are only capable of producing quite small drag increases. A variable pitch airscrew windmilling at very low pitch has been calculated to give quite a good braking effect, but the tail controls might be impaired by a " negative " slipstream. A device which acts as a pure air brake and has been tried successfully in flight is the D.V.L. gliding angle control. This consists, essentially, of two small surfaces, one on the upper surface of each wing, which are parallel to the plane of symmetry when out of action and can be rotated about a vertical axis. When turned normal to the direction of flight they break up the span distribution of loading and so increase the induced drag of the wing appreciably though their own parasitic drag is small. Flight tests on a German low-wing monoplane showed an increase of 5 deg. in minimum gliding angle with this device without appreciable increase in stalling speed.
Operating Forces for Split Flaps

Data on the hinge moment and centre of pressure of a split flap are a little scarce. Curves of hinge moment for flaps of varying chord are shown in Fig. 6. For such flaps the hinge moment is very nearly proportional to the angle of deflection. The rapid increase of hinge moment with chord shows that flaps should be kept as long and narrow as possible for a given area, a requirement in conflict with the avoidance of aileron troubles. A calculation based on these hinge moment figures shows that the moment required to hold down the flaps of a machine like the Douglas D.C.2 is about 700 lb. ft. at 75 m.p.h. This demands the use of auxiliary operating gear which may be electric or hydraulic but is bound to be fairly slow acting. A criticism of the flap installation on the Northrop " Sky Chief" was that 45 turns of a crank handle were required to get the flaps down ; this manual system has been replaced on later Northrop machines (with smaller flaps) by a hydraulic gear, and it appears that manual operation of a plain split flap of useful size is impracticable on any but light aeroplanes. It is in connection with operating force that the Zap arrangement is of principal interest ; by hingeing the flap to a link about one-third of its chord back and allowing the nose of the flap to slide back, the flap may be nearly in balance at high angles. The position of the flap C.P. is open to some doubt ; N.A.C.A. tests on plain split flaps show that, at the operating incidences, the C.P. moves back from 30 per cent, of the flap chord at 15 deg. deflection to about 40 per cent, at full deflection, whereas some tests by the Zap Corporation show the C.P. to move forward from 0.5 to 0.3 of the flap chord as the angle increases from 20 deg. to 40 deg. On the latter basis the flap is self-
Weight of Flap Installation

There is very little evidence in this respect. Experimental flap installations on the Parnall Parasol monoplane give figures of 250 lb. for the complete outfit, including controls, of a 28 per cent, chord Zap flap of full span, except for a small centre section cut away, and 651b. (estimated) for a 10 per cent, chord full span curved
SUPPLEMENT TO
92
DECEMBER 27, 1934
FLIGHT
retractable flap and associated controls. The former has there is hardly a passenger transport aeroplane not fitted already been fitted to a R. A.F. 28 section wing and involves with some form of flap, and some of their military machines " Park bench " ailerons, the latter is shortly to be fitted to are now fitted with slots and flaps. a R.A.F. 48 section wing ; both wing areas are 294 sq. ft. Where high top speed is essential, landing devices will The Zap installation weighs 8 per cent, of the all-up weight be used, not to reduce the landing speed, but to keep it the of the aeroplane and, from Fig. 4, should give a 75 per cent, same, increase the wing loading and so obtain a higher increase in Kh max. The retractable flap installation top speed. Calculations show that even with a split flap weighs 2 per cent, of the all-up weight and, from Fig. 4, used for landing only (on account of its high drag) the should give a 40 per cent, increase in A*L max. The weight performance of a typical two-seater fighter could be infigures in both these cases are for experimental installations creased by at least 10 m.p.h. on top speed and a variable on existing wings and are probably unduly heavy on that ' amount on climb according to how the aspect ratio is account, but even so a device which itself weighs only a changed. A maximum lift coefficient of 1.1 which is really tenth of the increase in lift that it gives may be regarded usable at a normal stalling angle now appears to be possible ; as good. this permits a wing loading of 20 lb. per sq. ft. for a stalling Speeds of operation of commercial and military aircraft speed of 60 m.p.h., and 30 lb. per sq. ft. at 74 m.p.h., are steadily rising owing to competition and in the U.S.A. which is not an impossible speed for a large flying boat.
SOME NOTES on the PARIS AERO SHOW

By H. J. POLLARD, Wh.Ex., A.F.R.Ae.Soc.
Mr. Pollard, as many readers will know, is on the Technical Staff of the Bristol Aeroplane Co., Ltd. Below he comments on Metal Construction, Three-ply Construction, and several other Aspects of the Machines exhibited at the Paris Aero Show
F the seventy or so aircraft exhibited at the Paris Show, the majority were, of course, monoplanes, of which less than half were " metal covered " in the usually accepted sense of the term; that is, while the main surfaces are metal-clad, the rudder, ailerons, etc., may be fabric covered, a convention which permits the use of the term " all metal," just as a biplane of which the general structure and the internal framework is of metal construction is referred to as " all metal " in spite of complete fabric covering. , A question at once arises that requires an answer. Why were most of the monoplanes shown three-ply covered? The answer most frequently given is that in small-quantity production the use of three-ply has an appreciable influence on cost. If, however, long service in all parts of the world is taken into account, then few will argue in favour of three-ply, particularly where the climate is moist and hot. At the same time large numbers of aircraft will be operating in more temperate and favourable climates; consequently the use of three-ply as a material for " stressed-skin structures " is worthy of further consideration. The fundamental expression for the critical stress at which plane panels buckle under shear or compressive forces acting in the plane of the panels is p = kE (t/b). where p = Critical stress E = Young's Modulus t = thickness of sheet b= width of sheet k =-- constant By working on a weight-for-weight basis, and using appropriate values of Young's Modulus and density, it is easy to show that three-ply in panel form is a more economical proposition than any sheet metal. Here, then, is one good argument for its use. Against it there is the very important consideration of water soakage. It is well known that the weight of a ply-covered boat hull can be very adversely affected through this cause. The same thing, to a much smaller degree, of course, happens with landplanes made from plywith ultimate disintegration of the material. In the case of large landplanes
such as cannot readily be moved into hangars, or for the largest aircraft that would have to remain out in any weather, the use of three-ply is unthinkable. The comparatively large percentage of civil aircraft exhibited in Paris, made from three-ply, forces one to review all the arguments that can be urged in favour of the material. Cost for small production (which is another way of making allowance for factories ill-equipped for metal work) has already been mentioned. The fundamentals in regard to the strength of sheeting under planar forces is a second favourable argument. A third argument often used relates to external appearance, i.e., freedom from buckles. This argument n immediately loses its force when really thin ply, say rsi . thick, is used. Thick ply, as used on the Heinkel 70 body, has always a good appearanceso long as it is not exposed to moisture for a long time. Judging, however, from the appearance of the three-ply exhibits they were no whit better, on the average, than those made from sheet metal, and the best looking body on showthe Czechoslovakian Avia 534was made of metal; the third argument, therefore, cannot be admitted.
Strength and Cost

A fourth and last point put up in favour of three-ply is that it is readily repairable with tools available anywhere in the world. A better way of stating the case appears to be that structures made from ply are less susceptible to damage, since if, for example, a bad landing is made, sufficient to cause damage, then the damage is not likely to be less than a complete "write-off," for if the forces are sufficient to buckle panels beyond repair, then the splintering of the framework used for the attachment of the panels would be a certainty. All the available evidence appears to indicate that ordinary metal monocoque constructions are adequately robust. Summing up on this point, it would appear that the main factors in favour of three-ply are low initial cost and high strength-weight ratios. As to the latter, conclusive data are difficult to obtain. From the material laws it follows that three-ply constructions should be
DECEMBER 27,
1934
9$
SUPPLEMENT TO

lighter thaii similar sheet-metal ones. The question is, are they sufficiently strong after a fairly long period of time? If, on the other hand, they are rendered quite weatherproof, does not the weight necessary for that purpose nullify the advantages? On the question of cost, it must be remembered that the '' shapes '' obtainable when three-ply is used are strictly limited : there can be no double curvature. A beautifully shaped body such as the Avia fuselage could only be obtained through the use of large numbers of narrow strips glued and screwed together. Three-ply would be good for slab-sided bodies. Low drag aircraft do not have slab sides and sharp corners. The summing up of the whole matter appears to be that the constructor who pins his faith on three-ply is, in the long run, placing himself at a disadvantage compared with the constructor who has resources sufficient to enable him to weather the expense that the development of the technique for the production of streamlined metal bodies must entail. One feels that an apology is needed for this extended consideration of an old topic, but the extensive use of three-ply at the Show certainly calls for more than passing comment. It is clear that persisting with, or even reverting to, the material is not progressive. This was one constructional aspect that was disappointing, the more so since certain military aircraft employed this material in structural form.
FLIGHT 1382c
Bucklesa Cause of Unsightly Surfaces

Another source of disappointment at the Show was the poor appearance of most of the surfaces in both metal and wood. Paradoxically, this was also a source of gratification, inasmuch as one was concerned in some small measure with the manufacture of one exhibit which was by no means free from buckles. While " good enough " for " first-off," it did not reach one's ideas of Show standard. In actual fact, however, the Bristol main exhibit was, as regards finish of surfaces in metal, second onlyand a good secondto the Avia 534. A main reason for disappointment was because Continental exhibits have had " rigid skinned" aircraft on their stands in several previous Shows and one expected them to have reached the acme of perfection by this time ; such, however, was far from being the case. Unquestionably, the surface appearance of aircraft generally will have to be improved. The time is approaching when civil aircraft will be exhibited with price tickets attached, and it will then be a poor look-out for the salesman who has machines to offer in which the surfaces have the appearance of being well trampled on after assembly. The argument that the surfaces buckle anyhow under load a long while before they rupture will not avail against such appearances, and it must be emphasised that the production of aircraft having fair surfaces is not an insuperable difficulty.
Northrop and Douglas D.C.2 aircraft, all external rivets have snap heads. One hastens to add, however, that flush-riveting he<s long been practical (our own Schneider Trophy winner, for example), and in the Show the single-spar Dewoitine D.510, together with the Polish P.2.C. aircraft, had flushriveted surfaces. Hitherto the difficulty has been the time factor, and consequent expense involved, but with the ever-insistent demand for lower drag and higher speeds flush-riveting becomes an essential, apart from the question of appearance ; for lower drag alone flat riveting is necessary over the leading portion only of wings, but once a speedy technique is developed the use of these rivets will be extended to all parts of every surface. A propos of the question of flush-riveting, it was curious that on the Bloch 211 the wings up to the front spar had ordinary snap heads, while aft of the spar all rivets, on the wing underside at any rate, were countersunk or the sheets deformed with the rivet to give a countersunk effect. Clearly, these latter had been formed on a machine as a separate bench job. The beneficial effect (if any) on this part of the wing of these buried heads was, however, nullified through the use of outstanding angle flanges provided at the transverse edge of each sheet. By riveting these angles together in pairs the assembly of the wing covering is greatly facilitated, but a much cleaner job would be obtained if the separate reinforced sheets were screwed to the internal structure of the wing without resort to the external downwardly projecting flanges on the upper surface. This matter of screwing on under surfaces is likely to become the vogue from the point of view of ease of assembly and maintenance. Resistance-welding and Fatigue Stresses Most of the troubles relating to local projections from surfaces can be overcome through the adoption of spotwelding ; this subject was dealt with by Mr. Langley last month. As, however, there are one or two additional things to say about that process, this is the appropriate place in which to interpolate such observations. The crux of the whole matter of resistance-welding lies in the fatigue strength of the welds. Let it be borne in mind that the sheets and strips are most carefully rolled and heat-treated to ensure that the proper microstructure is obtained, without which the whole of the specified physical properties of the material would be unobtainable. Sheet or strip of this material is then '' boiled-up '' and fused together, the microstructure at and near the weld then being vastly different from the carefully prepared state that exists away from the welds. The almost inevitable result is that under vibratory conditions of sufficient intensity fatigue cracks must be generated from the welds. By heat-treatment subsequent to welding the chance uf these cracks developing can be largely eliminated, but, even then, not quite wholly. In the case of a monocoque body, however, heat treatment after welding is outside the bounds of possibility. Fortunately, in the case of such a body the induced static and superimposed fluctuating stresses are very low, and perhaps no trouble need ever arise, certainly not with certain stainless and Mn-C steels, but with light alloys it is extremely doubtful whether freedom from fatigue cracking could ever be guaranteed. The proof of the pudding is, of course, in the eating, and these machines may do all that is claimed for them, bur an extended series of vibration tests is heartily recommended, the basis of comparison being the corresponding riveted joint. In the case of a highly stressed member, \ such as a wing spar, long and exhaustive tests are an absolute necessity if safety is to be ensured. The manu- r; facturers of the necessary welding apparatus exhibiting in the Show claim great advances of late. The writer has described some experiments on this work (see Aeronautical Journal for July, 1934) '< a repetition of those ex- ' periments on samples welded in one of the later type
Problems of External Projections

Almost (but not quite) as important as buckles are projections, such, for example, as external corrugations (Junkers), external stiffeners (Bloch), and protruding rivet heads (most exhibitors). Here, certainly, is a problem for early solution. Everyone wanted to know why sheeting was'not flush-riveted. The answer is that that could be done, but since very thin sheet metal is used, and since much of the riveting cannot be done by a machine, such, for example, as the De Bergue, riveted smooth-surfaced structures cannot be produced until the correct technique has been determined. This matter is being examined now, but something had to be left until the last, and flush-riveting was regarded as of secondary importance, which, of course, it is, compared with joint strength. Hitherto, smooth-riveted surfaces in very thin metal have been looked on as a refinement, and in this connection it is as well to remember that in the highly successful
SUPPLEMENT TO
FLIGHT 1382/
DECEMBER 27,
1934

body of considerable transverse curvature. It is not easy to say, however, why narrow strip was used in preference to sheet on the body of the Amiot 143, which was a large slab-sided structure. If the stresses are relatively high in the skin, then the edges must be close riveted, with consequent considerable increase in cost in the case of strip construction. Moreover, there must in such construction be increase of weight due to overlapped sheets.
machines mayand one hopes that it willcause modification of views and recasting of ideas on this subject. The reader is warned, however, not to be misled by mere static results. There is no difficulty in getting good statically strong welds. The trouble comes with high intensity alternating stresses in cases where no heat-treatment rs possible alter welding. Granting that this point need not arise in the case of sheli constructions, it does not follow that welding is necessarily the more economical manufacturing process. Consider, for example, the welded stainless steel SavoiaMarchetti amphibian flying-boat. The pitch of the hull stringers or longitudinals was about 4m. and the thickness of the sheeting was 0.007m. If the formula given at the beginning of these notes is applied for the purpose of comparing the stability of the sheeting between stringers with that of an aluminium alloy panel of the same weight per unit area, but having the longitudinals spaced Sin. apart, it will be found that the panel made from the thicker material will sustain the larger load before buckling, and although spot welding is a much quicker jointing operation than riveting, yet doubling the amount of reinforcement required must be debited against the welding process in the matter of total time taken. A more serious factor than the actual time of welding is the time required for extra jigging and fitting up prior to welding.
" Developable " Surfaces

Where large area sheets are employed the use of flat or '' developable surfaces'' is, of course, essential if " stretching " of material is to be avoided, but if a technique exists for the cheap production of non-planar, nondevelopable metal sheets, then the use of doubly curved surfaces is obviously recommendable. The beautiful appearance of the Avia 534 was due in large measure to the worked surface of the comparatively large sheets used; the curvature could not have been obtained by the use of "laid o n " sheeting. Apart from costs, the use of sheeting so curved has everything in its favour. For example, stiffening of the sheets and the elimination of the buckles that are nearly always associated with flat surfaces are two of the advantages that accrue, apart from the ease of obtaining finished structures of curved and handsome appearance. The difficulty lies in the forming process. There is unquestionably a real need in the industry for machinery for the production of such sheets. Panel beating is, of course, out of the question. " Wheeling " is too slow and indecisive, as is the " Swinging Pendulum " method. The use of large pressings, as in the automobile industry, is clearly the answer for quantity production, but one has no intention of venturing an opinion on the question of what amount of production would warrant the manufacture of the tools necessary for the adoption of the process.
Welding Duralumin
This is not intended as adverse comment on the Savoia Amphibian, which appeared to be an excellent job both in detail design and workmanship ; the criticism is directed simply against the use of steel in stressed skin structures. As to spot or shot welding monocoque fuselages, hulls, etc., made from aluminium alloy, it is as yet too early to give an opinion on the subject. There is no doubt that duralumin can be effectively welded on the Sciaky machine. Perhaps at some future Exhibition a light alloy hull or float, so assembled, may be on view ; until that time, however, one refrains from comment. It is, of course, possible that spot welding and riveting will be used jointly; certainly this was done on the Mureaux float. The two processes here had been inextricably mixed up. A first examination showed two rows of large-head rivets closely pitched along some seams, presumably for water tightness, but a close survey showed other seams secured by spot welding. For that and other reasons the motif of the combination was not clear. The technician viewing other people's work naturally examines such constructional or operational features as he is most concerned with in his everyday job. His object is either to learn something new or to satisfy himself that his methods are not out of date ; by such means progress is made. As to sheet-metal structures, the engineer making a start on metal monocoque and metal wing work meets at the outset a few problems the answers to which are not always obvious. One such question is : " How large should be the sheets that are actually secured to the framework? " (Only smooth-skinned structures being considered in these notes, it is fairly obvious that high-speed requirements will effectively debar any constructions employing external corrugations.) That question is, however, not difficult to answer ; clearly the sheets should be as large as possible consistent with the avoidance of shaping into " threedimensional" plating. If working of the sheet material, i.e., thinning by stretching or .thickening by compression, can be done in an easy way, there is no reason at all for using narrow stock. In the case of most of the sheet-metal covered exhibits, material of fair width (i8in. or more) had, in fact, been used. There were, however, one or two cases in which narrow ribbon had been employed. One such was the fuselage of the Dewoitine D.511, and this construction was evidently adopted as being the most suitable on a small
Spacing of Stiffeners
Another difficulty that arises when setting out on this class of work lies in the determination of the best spacing for the skin stiffeners. This can be assessed from the strength viewpoint, but there are many places both on wing and body metal covering where spacing must be based on experience and judgment; and no rules can be formulated as yet. The constructors who place the members too close together put themselves at a disadvantage compared with those who have an eye to riveting costs. The whole matter is bound up with sheet thickness, curvature and the form of stiffening member used. Each constructor must settle the point for himself. As a matter of fact, there was fair uniformity throughout the Show (excepting the case of Mureaux 15R2, where the free areas in the wing panelling were very small), which is only to be expected when too much riveting entails excessive costs and too little reduces strength. The inevitable result of the use of thin flat (or nearly flat) sheet and too wide spacing of stiffeners is indifferent appearance and possibly fatigue cracks due to "panting" surfaces. If saving in the number of internal framing members is desired, then advantage should be taken of any curvature given to the skin ; reinforcement of a surface having a good deal of curvature is often wasteful in material and cost; the amount of reinforcement should be varied where possible with such curvature. That there is no object in having a redundancy of internal members was shown in an exhibit on the Bristol Aeroplane Co.'s stand, which demonstrated how the stiffening effect of natural body curvature on the sheet could be utilised so as to render unnecessary the use of a large part of the internal framework normally included in these structures. This exhibit was shown in part in Flight, November 22nd, page 1250. It is hard to understand why exterior stiffeners are used at this stage, such, for example, as were employed on the fuselages of the Bloch 211 BN.4 and the Mureaux
95
DECEMBER 27, 1934
SUPPLEMENT TO

particularly the latter, since the Mureaux i8oC: had a good clean exterior. The use of exterior reinforcements, and ridges may be justified on some parts of the wings for ease of assembly, as in the case of the Bloch 211 and the Polish PZL, P24 and the P11C. The confined space of these structures renders assembly something of a problem, but no difficulty necessitating the use of exterior members should arise in the case of a roomy body. Details of monoplane wing constructions were eagerly sought, and again the Bristol Co. obliged by showing in detail sample specimens of their constructions; as these were described in the Journal " o f the Royal Aeronautical Society as recently as July, 1934 (page 660 et seq.), and by Mr. Langley on page 83 of The Aircraft Engineer, November 29, 1934, there is no need to deal with them here except to emphasise one of the chief points in the construction. This relates to the use of high-tensile steel in the spar booms. High stresses may be induced in the flanges of a monoplane spar; thus steel can be used to advantage in these cases. Also, the use of steel strip allows of extensive lamination, and variations of sectional area of spar are thus easily possible with attendant constancy of stress intensity. This form of wing construction is not put forward as the ultimate solution of the problem of monoplane wing design, but judged by present standards it is a satisfactory answer in respect of wing weight, strength and stiffness. The results already obtained will possibly be improved on
FLIGHT
as progress is made with manufacturing technique. More of the skin must be made to be effective to resist thrust without impairing its efficiency against shear forces. The problem has also other aspects. As to interior details of other monoplane wing surfaces, there is little to be said, since the exhibits shown were negligible. Of the biplanes there is nothing to say at all except to hand the palm, for appearance, to the Letov Co. for their type S231. In this, as in several other biplanes, single truss bracing wires only were used. While dealing with individual exhibits, mention must be made of the U.S.S.R. stripped aircraft, the Stal 2 The principle used in the design of the fuselage of this aircraft was "large diameter thin-walled tubes in steel." The fabrication of these members by shot welding is a step forward since the writer described structures of this type in this journal in 1928. If a fabric-faired girder is required, then the Russian designer responsible for this aircraft has got the right idea ; experience has shown, however, that light alloy monocoque bodies are superior to faired girders on most counts. The Russians are to be congratulated, as also are the Savoia-Marchetti Co., in exposing the interiors of their exhibits to view ; in this they are following the established practice of ths Bristol and Hawker Companies, and it would be to the advantage of the industry the world over if these examples were more freely followed. (To be concluded next month.)
SUMMARIES OF AERONAUTICAL RESEARCH COMMITTEE REPORTS These Reports are published by His Majesty's Stationery Office, London, and may be purchased directly from H.M. Stationery Office at the following addresses : Adastral House, Kingsway, W.C.2 ; 120, George Street, Edinburgh : York Street, Manchester; 1, St. Andrew's Crescent, Cardiff; 15, Donegall Square West, Belfast; or through any bookseller.
FLUID FLOW IN ROUGH PIPES. By A. Fage, A.R.C.Sc. K. & M.
A MODIFIED CHATTOCK GAUGE OF H I G H SENSITIVITY. By V. M.
Falkner, B.Sc, A.M.I.Mech., E., of the Aerodynamics Department, X.P.L. R. & M. No. 1589. (7 pages and 2 diagrams.) January 11, 1934. Price 6d. net.
In recent years the necessity for measurement of low speeds and for aerodynamic research at low Reynolds numbers has given rise to a demand for a gauge capable of the accurate measurement of low pressures. The standard Chattock tilting gauge has a limit of accuracy of about 0.000065 inches of water, added to a general error of about 1 part in 1,000. Later experiments have reduced the former figure to about 0.00004 inches of water. The gauge may be converted into a highly sensitive instrument by a simple modification which consists in arranging the liquids so that there is a bubble of the lighter liquid in the heavier liquid (xylene and water). This arrangement differs from all previous ones in that the forces of surface tension and gravity on the bubble are in opposition and the control exerted by the bubble can be reduced to zero. The standard glass-work used with modification to the frame has then a limit of accuracy of 0.0000065 inches of water.
A s EXPERIMENTAL INVESTIGATION OF THE WAKE BEHIND AN ELLIPTIC
net.
No. 1585. (11 pages and 13 diagrams.) October 17, 1933. Price is.
A body is considered to be hydrodynamically rough when irregularities on its surface affect to a measurable extent the resistance to motion. The effect of roughness on resistance depends not only on the geometrical form of the irregularities but also on the nature, of the flow over the surface. In practice, surface roughness does not commonly assume a simple geometrical form, but arises from an indefinite number of irregularities whose size and form cannot be specified with exactness. The resistance of such a surface can only be obtained by measurement. Even when the irregularities have a simple geometrical form, the effect on resistance cannot be predicted theoretically and must be measured Experiments are described which show that the resistance, of a rough pipe obeying the square law arises from the normal components of the pressures on the irregularities constituting the roughness, and not as in a smooth pipe from surface friction. Observation with an ultramicroscope shows that the flow in a rough pipe is much more disturbed than that in a smooth pipe. Evidence is obtained of considerable agitation in the flow near the irregularities, and of the creation of eddies comparable in size with the irregularities.
THE STRESSING OF AN AEROPLANE FUSELAGE UNDER COMBINED BEND-
CYLINDER.
By G. J. Richards, Ph.D., A.R.C.Sc., D.I.C.
Communi-
cated by Professor L. Bairstow, C.B.E., F.R.S. R. & M. No. isoo. ( < > puses and 8 diagrams.) June 28, 1933. Price 9d. net.
The form of the wake behind an elliptic cylinder was examined at a number ol value; of Reynolds number, and the longitudinal and lateral spacings together with the velocity of the vortices formed were found. The first was found to be constant in any one experiment, whilst the lateral spacing increased with time. The longitudinal spacing and the initial value of the lateral spacing were found to vary with R. The spacing ratio was found to have an initial value of 0.32, whilst it increased exponentially with time, that is down the wake. There also appeared to be a functional relation between the increase of spacing ratio and the velocity of the. vortices, the one increasing with the other, and both differing in seemingly girnilar experiments.
MEASUREMENT OF THE FULL SCALE WATER RESISTANCE OF A I I I F SEAPLANE IN STEADY AND ACCELERATED MOTION. B y E . T. Jones, M.Eng.
ING AND TORSION.
By A. G. Pugsley, M.Sc. Communicated by the
Director of Scientific Research, Air Ministry. R. & M. No. 1586. (11 pages and 3 diagrams.) June 24, 1933. Price 9(1. net.
In designing a fuselage structure it is customary to consider its strength as a freely jointed frame for a number of loading cases, in each of which is contemplatod either (i) an up or down load at the rear end of the fuselage, or (ii) a side load on the fin and rudder. For cases of the type (i), owing to symmetry of structure and loading, the forces in the members of a fuselage, are usually statically determinate, but for a case of the type (ii) in which a fuselage is subjected to combined bending and torsion, most fuselage structures are highly redundant and some approximate method of stressing becomes desirable for practical use. The present report is concerned with the problem of providing such a method. . The essential structure of a normal fuselage is analysed and the statically mdeterminatenature of the external 'oadino: system demonstrated. Thisindeterminancy is then expressed in terms of the distribution of the reactions at the forward end bulkhead of the fuselage, and the factors governing this distribution are discussed The proposed revised method has some advantages over the existing method
m
(i)1t provides a mechanism for meeting the main theoretical objection to the ing the arbitrary " rigid wail " assumption by distribution
Communicated b y the Director of Scientific Research, Air Ministry. R. & M. No. J591. (30 pages a n d 18 diagrams.) December 6, 1933. Price rs. gd. net. The full scale water resistance of a seaplane has been measured to determine how closely the law of corresponding speeds applied to the results of models tested in a tank predicts the full scale resistance. It has been measured directly by means of a special design of undercarriage in steady and in accelerated motion. The apparatus has proved to be satisfactory and the measurements are reasonably consistent. The float resistance at speeds between 0.3 and 0 8 of the takeoff speed is almost constant and equal to about one-fifth of the displacement at rest. Compared at the same attitude, the full scale resistance is about 10 per cent, higher than that of a 1/lOth scale model. The difference of resistance between model and full scale is not necessarily all due to scale and further tank tests are required to show how t he resistance of a smooth float increases as rivet heads and other small protuberances, such as lap joints and seams, are added. The curves of resistance in accelerated motion determined by the direct and indirect methods are in fair agreement. Compared at the same attitude, the resistance in accelerated motion is about 10 per cent, less than the steady motion resistance at all speeds greater than 35ft./sec.
T H E FORM OF A HEAVY FLEXIBLE CABLE U S E D FOR T O W I N G A HEAVY
BODY BFI.OW AN AEROPLANE. above, is more convenient for routine use.
By H. Glauert, F.R.S.
Communi-
cated bv the Director of Scientific Research, Air Ministry. R. & M.
96
SUPPLEMENT TO
FLIGHT
DECEMBER 27,
1934

Potentials were measured between various metals and a calomel electrode taken as standard. Actual potential differences between any pair of metals were then obtainable by difference. Most determinations were made at 25CC, but a short series was carried out at 40C. Of the metals tested, stainless steel of the 18:8 type and monel metal were the most positive followed in order by brasses and bronzes, " Twoscore " type stainless steels, 13 per cent, chromium stainless steels, duralumin and copper-containing aluminium alloys, ordinary steels, aluminium alloys free from copper, cadmium zinc and finally the magnesium alloy D.T.D.88 which was the most negative of the metals tested. The use of metals of widely different potentials in contaci is liable to result in serious corrosion of the more negative metal especially where saline water is likely to be encountered.
W I N D TUNNEL TESTS ON A MODEL GLOSTER TROOP CARRIER. W I T H AND W I T H O U T SLIPSTREAM. By W . G. A. Perring, R.N.C., a n d C.
No. 1592. (8 pages and 10 diagrams.) February 9, 1934. Price is. net.
The mathematical expressions for the form of a heavy cable in a wind have been known for many years, but not systematic numerical results are available. Calculations have been made to derive a family of curves, depending on the weight-drag ratio of the cable, which should suffice to cover all practical problems, involving the towing of a heavy body. The use of the curves is illustrated by a typical numerical example.
T H E EFFECT OF W I N D ON THE TAKE-OFF OF SEAPLANES. By E. T.
Jones, M.Eng. Communicated by the Director of Scientific Kesearch, Air Ministry. R. & M. No. 1593. (14 pages and 16 diagrams. January 12, 1934. Price is. net.
Take-off times have been observed on four seaplanes in winds varying from zero to 25 m.p.h. and correction formulae have been deduced from the results. Some results were also obtained from take-offs made down wind. The validity of application of the formula to all seaplanes has been examined and theoretical support to the formula is given. The effect of wind on the maximum weight at which a seaplane can take-off is also examined. The take-off time and distance in zero wind for all seaplanes can be fairly accurately given by formulae quoted in the paper. The maximum weight at which a modern seaplane fitted with a fixed pitch airscrew can rise from the water is almost independent of the direct effect of wind speed up to the highest wind in which the fully loaded seaplane is capable of being handled on the water. Gusts and the effect of wind in rippling the surfare of the water have not been included in the calculations. The results of practical tests show that there are occasions when a flying boat is unable to rise from a glassy sea but can take-off in a light wind from a lightJy disturbed sea.
LANDING AND TAKE-OFF SPEEDS OF AEROPLANES. By R. S. Capon.
Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1594. (7 pages a n d 3 diagrams.) January 19, 1934. Price 6d. net. In landing, the pilot usually stalls the aeroplane when the wheels are at a smal' distance from the ground. Since the aeroplane is decelerating it follows that in a stalled landing the speed at the moment of contact will usually be less than the stalling speed. The effect on the landing speed is considered in relation to the height at which the aeroplane is stalled : it is shown that in a typical aeroplane the landing speed will be 6.5 per cent, less than the stalling speed if the wheels are 6in. from the ground at the stall. The possibility of taking off at speeds below the stalling speed is considered. In the type of take-oft where a sudden increase of incidence is made at theendof the run (tail up take-off), a proportion of tbr weight of the aeroplane is borne by the undercarriage, and in certain types of undercarriage there may be available stored energy to project the aeroplane upwards when the incidence is increased. It is shown that if the whole weight of the aeroplane is borne by the undercarriage just prior to the pull off, the take-off speed may be 13.5 per cent, below the stalling speed (eugine-on), or in the more practical case when half the weight of the aeroplane is airlx>rne prior to the pull-off, 8.5 per cent. 3n many undercarriages, however, there is little available stored energy. ACCURACY OF PERFORMANCE MEASUREMENT. By J. L. Hutchinson,
Callen. Communicated b y t h e Director of Scientific Research, Air Ministry. R. & M. No. 1618. (15 pages and 11 diagrams.) October, 1930. Price is. net. The tests have been carried out to supply data for design purposes. Without slipstream the maximum lift measured was 0.520 without tailplane and would be 0.524 with tailplane set to trim. The slipstream increased the maximum lift by 0.10 with the tailplane in position. The maximum lift occurred at a model incidence of about 18, and over the range of speed, 40 to 80 feet per sec. it showed practically no scale effect. The minimum drag occurred at about 0 incidence, when the drag coefficient was 0.01i)l without tailplane and 0.0223 with tailplanc at a wind speed of 00 f.p.s. Tested at a wind incidence of 0.6 the drag with tailplane decreased from 0.02419 measured at 40ft. per sec. to 0.02175 measured at 80ft. per sec. The slipstream increased the effectiveness of the elevator control surfaces, and at a wing incidence of 8.8, corresponding to a kj, of 0.3 the moment produced by an angular movement of the elevator was, under full thrust conditions, 35 percent, greater than the moment produced for the same angular movement of the elevator without slipstream. The tests without slipstream show that an angular movement of about 12 was sufficient to produce a rudder power of 10 at 10 above the stalling incidence. The slipstream increased by nearly 100 per cent, the moment produced by the rudder, but had practically no effect on the yawing moment due to yaw for zero rudder angle.
COLLECTED REPORTS ON B R I T I S H H I G H SPEED AIRCRAFT FOR THE 1931
SCHNEIDER TROPHY CONTEST.
With an introduction by H. M.
B.A., and E. Finn, B.Sc. Communicated by the Director of 'Scientific Kesearch, Air Ministry. R. & M. No. 1601. (5 pages and 4 diagrams.) February 7, 1934. Price 6d. net.
A number of flights were made on two aircraft to determine the consistency of repetition tests. The consistency of measurement of rate of climb was determined from the range of possible mean curves cf rate of climb against height which could be deduced according to present practice in routine performance reduction from five climbs of Fox aircraft. This variation amounted to 40 ft./mill, at sea level and 90 ft./min. at 15,000 ft. The consistency of measurement of level speed was determined from five tests of Hart aircraft. The variation between the mean curves was less than 1J m.p.h. at all heights. The accuracy of speed measurement depends, however, on the stability and controllability of the aircraft, which characteristics in the cas of the Hart are very favourable for accurate flying. The above-mentioned accuracy of level speed measurement could not, therefore, be expected with all types. From a general discussion of the separate sources of error in performance measurement the conclusion is drawn that the actual performance in the test flight is measured accurately and that the largest errors arise in allowing for variations from ideal condit.ins. (>f these variations, vertical currents introduce the most serious errors and it ts 00 this account that a number of tests are necessary.
W I N D TUNNEL TESTS ON A BRISTOL FIGHTER MODEL WITH SLOTTED
R.A.F.34 SECTION WINGS.
By K. W. Clark, B.Sc., D.I.C.
Com-
municated by the Director of Scientific Research, Air Ministry. K. & M. No. 1609. (5 pages and 4 diagrams.) April 14, 1934. Price 6d. net.
Lift coefficient and normal slat forces have been measured in the wind tunnel on a modl Bristol Fighter F.2B biplane with slotted R.A.F.34 section wings, for comparison with full scale experiments recorded in Kefs. 1 and 2. The model was 1/10 scale, and was represented in full detail except for the airscrew and bracing wires. The wings were square tipped and the slats extended over the whole span except for the centre section of the lower plane. The slot chord was 14.5 per cent. of the wing chord. The maximum lift coefficient of the aeroplane was 0.77 at 27J degrees incidence compared to 0.84 at 30 degrees full scale. The normal force coefficients on the model reached maxima of 1.81 at 25 degrees incidence for the mid semi-span position, and 1.91 at 30 degrees for the tip position. The coefficients measured full scale were still increasing with incidence at the great incidence tested, with values c' 2.35 at 27 degrees incidence for the mid semi-span position and 1.8 at 30 Jegrees incidence for the tip position. THE E.M.F. BETWEEN METALS IN SEAWATER. By J. W. Willstrop,
contact with each other or by a third conducting material, there will be a flow of current from one metal to the other and this will be accompanied by solution or corrosion of the negative metal. This corrosion is not to be confused with any corrosion the metals might suffer independently as a result of sea-water attack. The tendency for this electrolytic corrosion to occur is dependent on the E.M.F. or potential difference between the metals when immersed in sea-water.
B.Sc., A.I.C. Communicated by the Director of Scientific Research, Air Ministry. R. & M. No. 1611. (10 pages and 1 diagram.) June, 1934. Price 9d. net. If two dissimilar metals immersed in sea-water are connected either by actual
Trophy Contest of 1931, the preparations for the Contest, the actual Contest, and the successful attempts on the speed record after the Contest. The monograph is mainly concerned with the technical aspects, although a tribute is paid to the great skill of the, pilots, without which the successes could not have been achieved. The monograph is divided into sections, each written by the person or persons mainly responsible for the work described, and although the monograph forms a connected whole, the individual reports may be read without reference to the rest of the monograph. Section 1 is an introduction by H. M. Garner, giving a brief description of the history of the 1931 Contest and a summary of the contents of the monograph. In Section II the development of the design and construction of the S.6A and S.6B are described by R. J. Mitchell, the chief designer of Supermarine Aviation Co., (Vickers) Limited. Although the design was based on the S.fi, the Schneider Trophy winner of I!t29, there was a large number of problems which required prompt solution, and these were very ably dealt with by Mr. Mitchell in co-operation with the Air Ministry and National Physical Laboratory staffs. One particularly difficult problem was the provision of adequate water and oil cooling for the engine. Section III describes the development of the engine by Messrs. Rolls-Royce Ltd. The problem was to extract more power from the engine used in 1929, and although the external shape of the engine was hardly altered, almost the whole of the working parts of the engine had to be redesigned. The airscrews were all of the Fairey-Keed type, designed and constructed by the Fairey Aviation Co., and a brief description of the development of the airscrews with an account of the methods of construction is given in Section IV. Section V describes the wind tunnel tests, is written by W. L. Cowley, A. McMillan, W. S. Walker and Sylvia W. Skan. It was desirable that models as large as possible should be used, so as to make the scale effect as small as possible. The tests were therefore made on as large models as possible in the Duplex Wind Tunnel at the N.P.L. (size 14 ft. X 7 ft.), the largest wind tunnel constructed at that time in this country. For the first time in the history of these Contests a large amount of full scale data was collected. This is described by R. K. Cushing, who was the Technical Officer for the High Speed Flight, in Sections 6 and 7 and also in a separate Report (R. & M. 1472). Section I S discusses the best method of turning, and shows that the best turns are made with a relatively small acceleration. Spectacular turns are of no value. Section 7 is a collection of airscrew performance data. Considering the difficulties of the tests, the standard of accuracy reached was very high, airscrew efficiencies being obtained with a probable error of less than 3 per cent. The highest efficiency obtained was 88 per cent., a high value when the high tip speeds are borne in mind. R. & M. 1472 describes the determination of the position errors of the high speed aircraft, a very important piece of work based on the automatic timing apparatus developed by the R.A.E. in 1929. The absolute accuracy of the speed measurements depends finally on this apparatus. Wing Commander Orlebar summarises the flying experiences of the High Speed Flight in Section 8. The greatest difficulties were encountered in taking off and landing. The take-offs of the S.6A and S.6B fitted with certain airscrews were particularly difficult, because of the large turning tendency. The medical aspect of high speed flying is discussed in Section 9, by Wing Commander G. S. Marshall. The opinion is advanced that the ordinary Royal Air Force medical examination needs little alteration when applied to high speed pilots. The monograph concludes with Section 10, giving a short descriptive account of the Schneider Trophy Contest and the two speed-record flights. It is evident that the speed in the Contest could have been improved had not Flight Lieutenant Boothman been instructed to take no risks on the turns and to keep the water temperature of the engine at a safe level by throttling. The speed record of 4071 m.p.h., made on the second attempt by Flight Lieutenant Stainforth was a very good achievement, only made possible by superb piloting. The monograph illustrates the many aspects of aeronautical research which have to be considered in the development of racing aircraft. Close co-operation between the different specialists was necessary in order to achieve success, and the monograph shows how close this co-operation was in many of the problems. The large expenditure of time and money, made possible by the generosity of Lady Houston, resulted in the retention of the Schneider Trophy by Great Britain and in a new speed record. The effort was therefore well worth while in its hmnediate results. The effect of such work in a broader sense, as a stimulant to aeronautical development in general, is almost incalculable.
Garner, M.A. R. & M. No. 1575. (96 pages and 80 diagrams.) January, 1934. Price 10s. net. This monograph describes the development of the British aircraft for the Schneider-

The Aircraft Engineer 1934

Uploaded by

Copyright:

Available Formats

The Aircraft Engineer 1934

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Aircraft Engineer 1934

Uploaded by

Copyright:

Available Formats

January 25, 1934

JANUARY 25, 1934

THE AIRCRAFT ENGINEER

rputto xm/o/r j t i w aauewa

Fig 8.Typical examples of sodium-filled valves

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

Of HOT 6AS. So*ie Gerosir mu. ftHTM He#E

aezc.I BUILO IM Of O**osrr\

TMEVE M*Y SM //y This Oue Tb /rsS*LL

- O>t/s//iG Ore* He*Tt*e

JANUARY 25, 1934

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

PLANCHETTE: FOR MARKING LINES ON

THE AIRCRAFT ENGINEER

LAP JOINT -SINGLE RlVfS

H/KLF BLOCK MOCtEL

OOUBlt MVETS TREBLE RIVETS

JANUARY 25, 1934

THE AIRCRAFT ENGINEER

Tons/ sq. Tons/sq " in. in. Tons Tons 8 12 12 17

February 22, 1934

EZ WTW CCHNJ3T fOSt

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

FEBRUARY 22, 1934

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

HEAT TRANSMISSION THROUGH AND RECTANGULAR PIPES. By

THE FLOW PAST CIRCULAR CYLINDERS AT LOW SPEEDS.

THE RADIALLY-BRACED AIRSHIP RING.

Bairfltow, C.B.E., F.R.S.

By Prof. L. Communicated by D.S.R.,

March 29, 1934

THE AIRCRAFT ENGINEER

k " >" - T V I t C K H I AT!O

MARCH 29, 1934

THE AIRCRAFT ENGINEER

Method of Test, in shear.

Fig. 2, Appendix II : Sequence of sheets tested. 306 c

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

The stress at the point A is given as (Ref. 10, page 5)

hL,(KV) //( cotatan 8)s where/,,, = (Sw/kxt) 1/sin a cos a

and the virtual length I1 is given by =C3Z

MARC H 29, 1934

THE AIRCRAFT ENGINEER

THE AIRCRAFT ENGINEER

& It" \ 17a \

Fig. 2 : Diagram of the Antoinette cooling system.

THE AIRCRAFT ENGINEER

April 26, 1934

FLIGHT" ENGINEERING SECTION

April 36, 1934

(Oonchuied from page 24.)

THE AIRCRAFT ENGINEER

TMEVE MY SM //y This Oue Tb /rsSLL

- O>t/s//iG Ore* HeTte