REPORT 1255 AN ANALYSIS OF THE STABILITY AND ULTIMATE COMPRESSIVE STRENGTH OF SHORT SHEET-STRINGER PANELS WITH SPECIAL REFERENCE TO THE INFLUENCE OF THE RIVETED CONNECTION BETWEEN SHEET AND STRINGER * By Josarx W, Sexontax and Jans P. Perensox SUMMARY A method of strength analysie of short sheet-stringer panels subjected to compression is presented which takes into account the effect that the riveted attachments between the plate and the stiffeners have on the strength of panels. An analysis of experimental data shows that panel sirength is highly influenced by rivet pitch, diameter, and location and that the degree of ‘influence for a given riveting depends on the panel configuration and panel materia INTRODUCTION Rivets have been used extensively for attaching the cover skin to the stringora and webs of aireraft wings. These rivets have beon designed, to a largo extent, by rule-of- thumb methods; yet, extensive experimental work of which reference 1 is representative has shown that the compressive strength of stiffened panels is greatly influenced by varia tions in diameter and pitch of tho rivets. References 2 to 4, in which the mode of instability of plates in compression known as wrinkling or foreed crippling hes been analyzed, show that the panel strength is influenced also by the location (rivet offset) as well as the pitch and diameter of the rivets. This mode of instability results from the existence of a flexible attachment between the plato and its supporting members and has occurred more frequently as the compres sion skins have become heavier and the supporting members lighter. ‘Tho purpose of the present report is to evaluate the strength of short compression panels and in particular to determine the influence of the riveting used to fasten the stringers to the plate on the strength of the panel. ~ Figure 1 shows the variation in panel strength with rivet piteh and names the various modes of failure involved. Only rivet itch is considered to bo varied in figure 1 but variations in strength could be obtained also by varying the rivet diameter or the rivet offset. When the rivet pitch is sul, the panel ‘of figure 1 fails in the local mode; for larger pitches, it may fail in cither the wrinkling or the interrivet mode, Failures in the interrivet mode are not usually permitted in contem- porary design; whereas, failures in the wrinkling mode are common. The problem of evaluating the effects of riveting on tho strength of panels becomes, therefore, primarily a study of the wrinkling mode of failure. The localmode section of the curve of figure 1 is shown as e horizontal line. It is recognized that there may be somo gain in strength with favorable change in riveting after the riveting (pitch in fig. 1) is such that the local mode is obtained. ‘The available ‘test data indicate that the gain in strength is small and it is noglected in the analysis presented herein. ‘A study of the wrinkling mode is made with the use of the procedures established in references 3 and 4 in connection with the caleulation of the strength of multiweb beams in ending. These procedures make use of » new structural parameter termed the “effective rivet offset” which plays an important role in determining the strength of riveted struo- tures such as compression panels and multiveb beams and makes possible relatively simple structural anelysis. The effective rivet offset is evaluated by using a relatively rigorous analysis of the initial instability of compression panels supplemented by experimental data and is applicable to the anlaysis of multivveb beams as well as panels. A semi- empirical maximum-strength analysis of panels which utilizes the effective-rivet-offact concept is made and com- pared with a large number of test fesults to show the accuracy. and generality of the analysis. ‘The analysis is exemplified in the appendix. SYMBOLS, be ‘width of attachment Mingo of stitoner (seo fg 2), in. b width of outstanding flange of stiffener (ovo fig. 2), in ba width of top of ht for hat-seotioy stiffeners, be geometric rivet offset (seo fig. 2), in. by ‘stiffener spacing (see fig. 2), in. by depth of wob of stiffener (Gee fige2), im. + a rivet diameter, in f effective rivet offset (ove Bi. 5), io Bey Duckling-stress coefficient ke failing-stress coefficient > rivet pitch, in. De allowable rivet pitch, in, radius of bend between attachment flange and web of stiffener (see fig. 2), in. ts plate thickness (seo fig. 2), in te stiffener thickness (seo fig. 2), in. uperedea NAGA Teche Note 381 by esp W Smmonbn and ames P Peta 108, 1314 REPORT 1255—NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS crippling) _--Interrivet Rivet pitch Fraune 1.—The influence of rivet piteh on the strength of a short sheet-atringer panel showing the three predominant modes of failure, As cross-sectional area of Z-section stiffener, in.* Ds plate flexural stiffness per unit width, Exes /120.—y), in-kips Dy flexural stiffness per unit width of web, “Baty?/12(1—1, ius ips E ‘Young’s modulus, ksi Ewe secant modulus, si Ew tangent modulus, ksi Bs ‘Young’s modulus of plate material, ksi By ‘Young's modulus of stiffener material, ksi R rivet tensile strength, kips Be required rivet tensile strength, kips @ rotational stiffness per unit length (see fig. 5), kips 3 lateral deflection of plate, in. 7 plasticity factor Ver +1 » duekle length, in. » Poisson’s ratio oer buckling stress, ksi x average stress in panel at failure, ksi average stress in panel at failure in local mode, ksi ow failing stress of plate, ksi rer crippling-strength of Z-section stiffener, ksi + ’ defleotional stiffness per unit length, ksi ‘The designation for the various aluminum alloys has re- cently been changed. ‘The old designation and the corre- sponding new designation for the aluminum alloys mentioned in this report aro as follows: ogg | ng fet, | ite STRUCTURAL ANALYSIS A pandl typical of those analyzed is shown in figure 2. ‘The pane! is considered to bo short enough so that the column bending mode can be neglected yet long enough so thet various local modes can form freely without end effects. ‘TheCOMPRESSIVE STRENGTH OF SHORT SHEBT-STRINGER PANELS Fraune 2.—A shoot-stringor panel. panel is considered to be wide with many equally stringers but the results of the analysis can be applied to panels with as fow as four stringers without appreciable ‘The analysis is presented in four sections. ‘The first section develops an initial-instability analysis which together with available experimental date is used in the second section to establish the effective rivet offeet as a function of approprinte panel parameters. ‘Tho values of effective rivet offset. thus established are used in the third section to formulate a semi- empirical maximum-strength analysis. Finally, the fourth section is devoted to developing criteria which limit the piteh and diameter of rivets required to achieve the predicted strength of panels. INITIAL INSTABILITY OF PANIES ‘The panel shown in figure 2 usually will buckle into either the local mode which has been analyzed in reference 5 or the wrinkling mode which will be analyzed herein, Another mode termed the “torsional cum local” mode was analyzed in refereneo 6. ‘This mode may become the predominant mode when the width of the outstanding lange of the stif- fener becomes small (cay bp<0.45n) s0 the flange does not have enough stiffness to prevent the line of intersection be- tween the flange and the web of the stiffener from translating when the panel buckles. ‘The wrinkling mode of instability can be anslyzed by eonsidering tho plato to be supported by elastic springs with a doflectional stiffness per unit length of panel yas indicated in figure 3. A cross section of the plate through sn up- Fraune 3.—Idealized structure used in analysis of sheet-tringer panel. 15 buckle is shown in figure 4." ‘The stability criterion for the plate is given in roferonce 3 as Q Teo i= cosho ‘This expression has been solved and values of k,, are plotted against values of Nb, for various values of the parameter Vis /x4D, in figure 7 of reference 3. ‘The deflectional stiffaess provided by a stringer of the ‘same material as the plate is given by reference 4 as Pee ‘where the rotational stiffness a is a function of the web stress and the buckle length and can be taken from reference 7 which uses the symbol 4S" to defino this stiffness. ‘The assumptions implied in the use of the above formulas have been given in reference 4 but aro reviewed here for com- ploteness of the present report. Besides the restrictions on length and width of panel as discussed earlier, the implied assumptions are: (1) Deflec- tions aro small, (2) the structure is elastic, and (8) the stringer stiffness can be obtained from the idealization shown in figuro 5. ‘This idealization is based on the assumptions: (2) The effective rivet offset can be defined as the distance from the web of the stringer to longitudinal line along which the rivets effectively clamp the attachment flange to the plate, (b) the longitudinal bending stiffness of the attach- ‘ment flange can be neglected, and (c) the web ean be assumed to be simply supported at the bottom. ‘This last assumption will be good for webs of normal proportion as long as the width of the outstanding leg of tho Z is about O4by. At much larger values, it ean become the unstable clement and thereby initiate buckling; at much smaller values, it will not hhave enough depthwiso stiffness to provide simple support to the web. For webs with small width thickness ratios a flange width greater than 0.45, is required to support the web ageinst translation. For such cases the criterion of reference 8 should be used to design the outstanding flange.116. REPORT 1255—NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS - () Loads on stringer. () Ldewized strings, (© Distortion of idealized stcingor. ‘Frauen 5.—Loads and deformaticns used in the caloulation of the deflectionsl stiffness of short Z-section stringers, "Bauatons G) and (2) have bean solved and the results | ‘This value of» i the value given by Stowell (af. 9) for ong are given in figure 6. The buckling coefficient ke, is plotted | simply supported flat plates in compression. against the parameter for various values of the parameter | — Local-buckling curves from reference 6 for by/by=0.4 and Siby. The buckling cosfiicient is related to the buckling | ty/ty=0.63 and 1.00 have been plotted in figure 6 for com- stress by the relation __ parison with the wrinkling curves. 7 Tt will be noted that the buckling coofficient ky, for the (8) | wrinkling mode is determined by the two parameters f/byy and f oven though theso parameters aro not sufficiont to: A value of the plasticity factor 7 that has been found to give | determine the panel configuration. The local-buckling good correlation between test and caloulation is curves, for instance, require the additional parameter ty/ts ve to fix their location on the plot of ke against 8. ‘This pho- wake GA re (@, | 2omenon mas pointed out in reference 4 in connection with 2 x oat? ) ): | the calculation of wrinkding coefficients for multiweb beams. 2 a - Ey 5 aa i 0.90) and the chart should not be used for much smaller values of rivet diameter without confirmation. ‘Figuro 8 is applicable to multiweb beams as wall as panels ‘and can be used in the epplication of the formulas and design charts of reference 4 to the analysis of the bending strength. of multiweb beams. ‘Tho failure of short compression pancls usually results from e growth of either local or wrinkling typo of buckles, Less frequently, failure may result from rivet failure or growth of an interrivet type of buckle. ‘The first two types of failures will be discussed in this section and the last two types ov Panes118 5 REPORT 1250—NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS ‘TaBLR 1 [TEST DATA AND PROPORTIONS OF 7075-78 ALUMINUN-ALLOY PANELS - tena] ; ae | ae wniin [toe | me | tte | tote | ote | aie | AS |e | ett | ots owen x» | ae sa | ae mo | we | age i BR] ie ho) a ee | he | 3 B | HE a | a8 Ho) Be | ae Bl a wm | 33 Ho] ae | ae Bl is | BB Ho] BP | dG 8 | iG | a8 Poe nt ee 8) ie eo) 8 nm | ae | Bo B | 3 eo) ae Bo) ge | at B| Hi ho] ES Bo] gt | ah BR] ie fo] us eo] i | ae B | HS | gs feo) Ba | ae B | Hi | gs ge as B| BS ee) Rs BS we B | att fo) 8 as B | Ht eho] ES Fe ao B | Ht fo) a tt ae B | it | 88 Fe ai B | as he | is i ae B | Hi ee] 8 8 ae S| He eee | aes u aS Be ee] BS BB a Bl RG Ho] gs BB wi Bl 3 fo] te Bs as B | ate ee | 8 BB Ee B | HS ge | gS $8 aa Bae ee | 48 ue a 3 | He He || a ai BR] Bs | a8 BE a3 8] Bs a | 8 Bf a3 | 83 Ho) a8 a Bt 8 | 83 | 8 Be ay B | Re a | 88 E 3 “Stringers with rajlw=1.0 wore extruded. All others were forned. SAll vote woro 2117-79 fatshoad rivets with NAGA countarsink on the plate side, "The depth of countersink for tho Kr Key Hey Hr, and Yerinchediameter rivets was 0.040, 0.050, 0.000, 0.070, and 0.080, respectively. will be considered in the next section where rivet criteria are developed that ean be used to prevent such failures. Failure in the wrinkling mode—Panels which buckle initially in the wrinkling mode usually fail in a similar mode. ‘The plate configuration at failure, however, is simpler then tho initial buckling configuration because, as the initial buckles grow with an inerease in applied load, the plate buckle shape becomes more and more cylindrical until at failure it may be assumed to be cylindrical and the plate may be treated as ‘2 column on an elastic foundation. ‘The plate in the column mode sppears much like the well-known interrivet mode excopt the length of buckle is greater than the rivet pitch. ‘The stringer, however, has a very different configuration. In the interrivet mode the stringer cross section may remain essentially undistorted while the plate and stringer separate. In the wrinkling mode of failure the attachment flange of the stringer follows the plate contour and causes the other plate clements of the stringer to distort also. ‘The similarity etween the appearances of the wrinkling mode and the intersivet mode hes caused investigators to make strength caloulations with interrivet-type formulas on panels which failed in the wrinkling mode. (Gee, for instance, ref. 12.) ‘The panels of this reference evidently failed in the wrinkling mode and the strength of the panels can be calculated by the methods developed herein. ‘The stability criterion for tho plato in the wrinkling mode of failuro is given as (see ref. 13) bt ay Ry © ‘The support stiffness was determined by trial to givo tho best correlation between panel strength and calculated strength. It was found that the support stiffness could bo taken as 12/4 f vee fs) © (he) CH) ‘This equation is identical to equation (2) exeapt the rotational stiffness « ™ has been replaced by a constant value of 3. In the trial calculations used to determine the support sti oder ale were ed ine the apart value as given by equation (2), ; considered to give the best agroement between calculated strength and panel strength over a wide range of panel pro- portions. Tt gave particularly superior correlation compared ‘with the apparent value when the webs of the stiffeners were relatively unstable because the epparent value (eq, (2)) gave the restraint at the onset of buckling of the webs and not theCOMPRESSIVE STRENGTH OF SHORT SHBET-STRINGER PANELS 19 u Ww 1 [| | y | +4 [| tt ° |_-+~| |_| —|—| 8 E S Coe a ae 7 ai —_— Ts |_| [ts] 7 |_| | + rt a [+t + tw 5 5 4 3 2 7 a eg £ a ‘Frouns 8 —Experimentally determined values of effective rivet offset for % and channel stringers and full-depth channel webs. restraint offered the skin at pang failure. ‘The value a PE —3 vas also used in reference 4 to calculate the strength of multi- web beams in bending. ‘With the simplification implied by equation (6), that the support stiffness is independent of the buckle length, equa- tion (6) ean be simplified to read after ky is minimized with respect to buckle length. Equations (8) and (7) have been solved and the results are presented in figure 9 which gives the maximum stress that the plate can carry in the wrinkling mode. At this stress, the Interal deflections of the plate, and therefore the lateral forces on the stringers, become large and destroy the capa- city of the stringers to carry additional load except for un- usual panel proportions. Experience in testing panels and multiweb beams indi- cates that a plate in the wrinkling mode suffers 2 relatively moderate redistribution of stress after initial buckling. The lond-shortening curve for a plato in the wrinkling mode, therefore, nearly coincides with the stress-strain curve of the plate material until just prior to plate failure. ‘The stringer on a panel which has buckled in the wrinkling mode eppears 96876—6t—0 very much like a stringer on a panel which has buckled in the local mode and evidently suffers much the same redis- tribution of stress and loss of axial stiffness. In order to calculate the strength of a panel, it is necessary to know the load carried by the stringers at panel failure. (The plate load is given by fig. 9.) ‘The load carried by the stringers depends on the proportions of the panel. If the stringers are relatively sturdy (B<1), they will be stressed the samo as tho plato, If the stringers aro unstable (8>1), the stringers will not bo loaded as heavily as the plate. An approxima- tion which gives predictions which are slightly high when the stringers are unstable but which gives satisfactory results over the entire practical range of panel proportions is that tho stringers take the same stress as tho plate as long as that stress is not greater than the stringer crippling stress, jin which case the stringers take their crippling stress. In addition, the calculated load carried by the panel must always be greater than the crippling load of the stringers tested without being fastened to the plate. This criterion takes care of the case when the area of the stringers is large compared with the aren of the plate and the attachment between the plate and the stringer is so flexible that wrinkling occurs at a lond less than the crippling load of tho stringers. For this case, the lateral forces on the stringers are com- paratively small and do not affect the strength of the stringers. Furthermore, at the shortening necessary for the120 ° ‘2 # S ‘8 0 2 Ficus 0,—Maximum stress cooMoients for sheet-stringer panels that fail by wrinkling. stringers to achieve their crippling stress, the load being carried by the plate has fallen to a negligible quantity and it may be assumed that the entire load is being carried by the stringers. ‘The value of the plasticity factor » to be used with figure 9 is given by equation (4). ‘The uso of a plasticity factor which is a function only of the stress-strain curve of the plate material and is applied to the average stress in the plate at failure may seem to be rather arbitrary for panels on which the proportions are such that the panels buckle at loads that are considerably less than the loads that the panels ultimately carry. Panels which buckle in the local mode, for instance, experience @ severe redistribution of stress as the panel is loaded beyond the buckling lod. The factor may not be too arbitrary for panels which fail in the wrinkling mode, however, because a plate in the wrinkling mode of failure is under relatively uniform stress across the width of the plate; that is, the stress is not peaked at the stringers as for a plate which has buckled in the local mode. ‘Tho correlation between test and calculation obtained by using the plasticity factor given by equation (4) will be given REPORT 1255—NATIONAL ADVISORY COMMITTEE FOR ABRONAUTICS” 8 6 Be eRe a sx het, (By 3 aI— a bs) later and indicates that the factor is satisfactory even for panels with a large post-buclding strength. ‘When figure 9 is used to calculate the strength of # panel, the strength in the local mode as well as the strength in tho wrinkling mode should be calculated and the lond the panel can be expected to carry will be the lowor of the two loads. ‘The strength of panels in the local mode will be discussed in the next seetion Failure in the local mode.—Panels which buckle initially in the local mode may fail as a result of tho growth of the local bucldes. (Gee fig. 1.) A few pancls have been ob- served to buckle in the locel mode and to switch from local Duckling to wrinkling at a higher stress level and eventually fail in the wrinkling mode. ‘The data from such panels evidently would plot near the value of rivet piteh in figure 1 where the local mode ends and the wrinkling mode starts. ‘A study of the available date on compression pauels on which the pitch and diameter of the attachment rivets were varied indicates that the gain in strength corresponding to a decrease in pitch or an increase in diameter of the rivets after the local buckling range has been reached is small.COMPRESSIVE STRENGTH OF SHORT SHEET-STRINGER PANELS Consequently, for riveted panels there is a panel strength which is relatively independent of changes in riveting thet corresponds to failure of the pancl in the local mode. "This characteristic has been recognized for a long time (see, for instance, ref. 11) and is responsible for the numerous inves gations in the past on “strongly riveted panels” (the inves gation of ref. 5, for instance). When these investigations were applied to the design of panels, however, the riveting required to make the panel behave as a strongly riveted panel was not known and rather severe rivet criteria had to be used. (Geo criterion of ref. 14.) ‘The present analysis alleviates this difficulty by relating the strength of panels to the piteh, diameter, and offset of the attachment rivets. Reference 5 shows that the ultimate strength of panels which buckle locally at high stresses is closely related to the buckling load and can be calculated by the buckling charts ‘The particular curves for a value of 7 of 0.4 and values of 7 of 0.63 and 1.00 are reproduced in figure 6, Referenco 15 gives # mothod of predicting the strength of a panel in the local mode provided the strength of a nominally identical pancl of another material is known. ‘With the help of these references and the test data of refer ences 1, 10, and 16 to 20, the strength of some panels which fail in the local mode was estimated and is given in figuro 10. In the construction of figure 10, the method of reference 5 dotermined the indicated strongth of the panels when the failing stress is high (usually panels with values of 8 of about unity and with small values of by/ts and bite). ‘These par- ticular panels require the most severe riveting eriteria in order to foree the panel to fail in the local mode and con sequently their strengths aro the most difficult to obtain experimentally. ‘The available experimental data, supple- mented by the procedure of reference 15, sufficed to deter mine the strengths of the other panels considered. of that reference, RIVET CRITERIA ‘The maximum-strength analysis of compression panels given in the proceding section requires certain limitations fon the pitch and strength of rivets in order that the panel will carry the predicted load. ‘The rivets must be spaced closely enough and have adequate strength to make the stringer flange follow the plate contour. If the spacing is too large, the panel may fail by interrivet buckling. If the strength is insufficient, the panel may fail prematurely because of rivet failure. Rivot pitch—An expression for buckle length which is consistent with the maximum-strength formulas (6) and @ is »_ [fz be Vie ® ‘Tho allowable rivet pitch which must not be exceeded in order that the stringer flange follow the plato contour can logically be related to the buekle length as given by equation (8). It was found by trial that, if the rivet pitch was less than 90 percent of the calculated buckle length, wrinkling would occur rather than interrivet buckling. Hence, the rivet pitch must satisfy the criterion 121 F<0.009)% © Rivot strength—The lnteral force required to hold the compressed plate in its deflected position is proportional to the support stiffness and the lateral deflection of the plate. ‘The force on a rivet near the crest of a buckle may be expressed approximately as Reyop (19) where 8 is the Iaterel deflection of the plate at the crest of a buckle. ‘The value of y may be taken from equation (A19) of reference 4. An appropriate value for the rotational stiff- ness $B in this equation ic $L=8 GL In order to express formula (10) as a rivet-strength criterion, the value of lateral deflection must be known or assumed. Figure 8 of referenco 21 indicates that, for an idealized H-section column, maxi- ‘mum load is reached before the lateral deflection is one-fifth of the column (or plate) thickness provided the buckling stress is at least half of the compressive yield stress of the column material. (For panels which bucklo carly a value Inger than one-fifth should be used.) If this value is used in formula (10), the required tensile strongth for a rivet becomes ace o sue ‘The tensile strength of a rivet is defined as the load required to cause any failure; it may be the load required to break the shank but more often it is the load required to pull the countersunk head through the plate or, when the stiffener gage is small, to pull the rivet head through the stiffener. Reference 22 gives the strength of protruding-head rivets. Reference 23 gives strength data on NACA countersunk and conventional countersunk rivets. Additional rivet-strength date ean be found in references 24 and 14. Bxpression (11) gives the tensile strength of the attach- ment rivets thet is required in order that the predicted strength of the panel in the wrinkling mode can be achieved. Obviously, when the panel fails in the local mode, expression (11) does not apply. ‘Tho available date indicate that for this case the rivet strength need not be any greater than that required when feilure is in the wrinkling mode and the stress levels at failure in the two modes are equal. Ey ts, Ba>: P qu) EXPERIMENTAL VERIFICATION ‘The date presented in this section have already been used to establish the empirical factors in the analysis presented carlier and will now be compared with the analysis to assess its validity. ‘The data were taken, in large part, from pub- lished NACA panel data obtained from panels which were five bays wide (6 stringers) end hed a slenderness ratio L/p of 20. The ends of the panels were ground flat and parallel in a special grinder prior to testing and the panels were tested flat-onded in a hydraulic testing machine. A lange amount of wrinkling data is available on panels made from 2024~T3122 REPORT 125§—NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS 50 fe 20 ao _ =| Ss 40 | LT —— 3 7 - : -—{— 2 20 | wo mil mtd titi 50] o E 20 30 > | ——_| . ee j-———_| 40 t 20 : ae ae 20 oo - + + oa, : : co * % () feftg=008 © enfa=.00. Fiovae 10.—The strength of 2024-78 suminumally Zatfenod panel inthe lea! mode. byfby=O4D (previously designated 248-2) aluminum alloy. ‘The mum- ber of tests on panels which failed in the wrinkling mode and which were made from 7075-T6 aluminum alloy is much smaller for two reasons: (1) The investigation on the effect of riveting on panel strength was made on 2024~T3 slumi- num-alloy panels first and later on 7075~T6 aluminum-alloy panels. ‘The knowledge gained from the early experiments could be applied to the later tests and thereby reduce the number of tests required. (2) The tests on 7075-16 alu- minum-elloy panels were made on panels with extruded stringers with smell fillets s0 tho rivet line could be moved in close to the web of the stiffener and thereby provent the wrinkling type of failure. In order to roliovo the shortage of data on 707516 aluminum-alloy panels, a series of panel tests wore made in the present investigation and are reported herein.COMPRESSIVE STRENGTH OF SHORT SHEWT-STRINGER PANELS ‘Data on panels with riveting which do not satisty the cri- teria of expressions (9) and (11) and the additional eriterion that the ratio p/d must be less than 15 will not be given in the presentation which follows. The latter criterion is in- cluded because available data on panels for which the failure was definitely wrinkling were considered to bo inadequate to establish design curves for these high values of p/d. ‘The restriction on panel design imposed by this criterion, however, is not considered to be severe because contemporary design rarely allows such large rivet pitches. {ROL-TY ALOMINUMCATLOY PANELS. ‘The data of references 10, 16, and 17 are shown in figures 11, 12, and 13, respectively, whero the average stress in the panel at failure 2, is plotted against the rivet parameter p/d. ‘The data of rofereneo.6 for panels with a bs/ls greater than 50 are not given because itis relatively easy to rivet such panels so that the panel will fil in the local mode. NACA counter- sunk rivets were used to assemble the panels. Other perti- nent dimensions are given in the figures. The data plot against the parameter p/d with e small amount of scatter. ‘This characteristic is responsible for the uso of the 7-param- eter on the f-chart of figure 8. 59) 40] 2. 2 0) tote ) tells0.03. Frovnn 11,—Compariton of ealeulated and experimental falling stresses of 2024-73 Zatiffoned panels of reference 10 for five values of felts. Byll=20; Dalle=25; boltr=6.6;tr=0.004 in. 123 0 ‘ol ina oe ain o hit 8 ye ie 2 We £38 Bin 10 le we ° % 40| 2 ry ksh 20) io Reo aps (© tnite=0.79. @ feltg= 1.00. (©) tlle 1.28. Frauas 11.—Coneluded. ‘The curves in figures 11 to 13 represent predicted panel strengths. ‘The wrinkling section of the curves was obtained with the use of figures 8 and 9. For the panels represented by the data in figure 13, where the stringers are relatively ‘unstable, the crippling strength of the stringers was required to obtain the panel strength in the wrinkling mode. ‘Tho stringer crippling strength was taken from reference 25; the data were extrapolated when it was necessary. ‘The curves predict the trend as well as the magnitude of the data within the accuracy of the panel tests: experionee in testing panels124, indicates that strength tests on two nominally identical panels usually give strengths which differ by less than 5 ereent from the average strength although differences as great as 10 percent have been obtained. ‘The wrinkling ‘curves miss the middle of the scatter band of the date in some instances by about 5 percent. It is believed that such discrepancies are largely a result of neglecting the difference jin material properties and panel parameters (particularly Jolly) between one group of panels and another. ‘The panels represented by the data of figures 12 and 13 wero built in groups similar to the grouping used in tho presentation of the date and are therefore particularly susceptible to errors common to a group of data, These differences were neg- lected in the presentation of the data because of the result- ing simplicity and because only nominal values of the rivet offset Bp were known. ‘The local-mode section of the curves in figures 11 to 13, ‘was obtained from figure 10 for the panel proportions cov- on (@) Balter 0. Fiavar 12—Comparizon of calculated and experimental faling stresaes of 2024-T3 Zetiffened panels of reference 16 for four ‘values of Balls at two values of tyjls. bwitw=20; bollr=6.0; t= 0.004 in, REPORT 1255—NATIONAL, ADVISORY COMMITTEE FOR AERONAUTICS ered by the figure. The local strength of the panels with a Byity of 25 and 50 which are not covered by figure 10 wore obtained by interpolation and extrapolation of the data from figure 10 by using tho present data as a guide. A study of figures 11 to 13 indicates that the strength of a panel in tho local mode becomes increasingly difficult to attain ns tw/ts is inereased or as bs/ts is decreased. Accordingly, the closest riveting used in the investigation (p/d=3.0) was just ado- quate to attain the strength in the local mode of the panels of figure 11 with a thickness ratio tr/tg=1.00 and was inadequate to attain tho local strength of the panels with énits=1.25. For panels with much smaller values of bs/ts than were used in figure 11, it would be impossible to rivet the panels s0 that the local strength is obtained without the ‘uso of smaller values of the rivet offset Bo. Some test data from reference 26 on panels with hat~ section stiffeners are given in figure 14. ‘Tho averngo stross in the panel at failure @, is plotted against bs/ts whero 2b, is the distance between similar locations on two adjacent hhat stiffeners. Only data for the thickness ratio ¢y/ts=0.30 59) so} 324 a = £ an. _ og S 9 Ye vot 8 8 2 8/32 © 3yi6 8 ial to} C #083 ° on 40] 5° 39 Gl ©) balle=36, ‘Frouan 12—Continued.COMPRESSIVE STRENGTH OF SHORT SHUET-STRINGBR PANULS 125 “ ~ 40} 40} - | : = fed 2G Sot 2 Sok 2 So 3 HE ioe see iue ioe 1088 ’ ‘ & 5 “ ee “ 20] oo ane 3” ee Seal Foo ls B a 3 © beat steno, owe Gt Toon 8 Sone, fare shown because they are considered to be sufficient to | the data. The shaded test points represent panels which establish the concept that for panels with unequal stiffener spacings an average spacing can be used for predicting the ‘maximum load of the panel in the wrinkling mode of failure. ‘Tho particular thickness ratio tyr/ts=0.38 was chosen rather than some other because tho panels with other thickness ratios had stiffer attachments between the het-seetion stiff- eners and the plate so that most of these panels failed in the local mode rather than in the wrinkling mode, ‘The data for panels with clements having a width-thickness ratio Yt greater than 50 have not been shown. ‘Tho calculated curves in figure 14 aro based on an average measured value of bo/ty rather than the nominal value. {E:T ALUMINUMALLOY PANES "The data of table are shown in figure 15 where the average stress in the panel at failure is plotted against the parameter & ‘Tho predicted panel strengths are indicated by the curves and agree with the teat date within the accuracy of had stringers with « value of 4 of 6.0. Since the stringers had o valu of 2 of 20, these stringers had a value of F# of 0.30. These points all appear high on the figure and indi- cate that the attachment between the stringer and the plate was actually stiffer than figure 8 indicates. It is not known whether the test loads were high because ralfr was large or whether it was because r4/by Was large or both. Inesmuch 1 the chart gives conservative predictions in this range, the “uncertainty is not serious. DISCUSSION OF RESULTS ‘A method has been developed whereby the strength of panels is related to the design of the attachment between the plate and the stiffener. ‘The method makes use of an experimentally determined effective rivet offset f which is an important dimension in the determination of te strength of panels, ‘The importance of this dimension as well as126 other panel dimensions on the strength of pancls can be readily seen from the equation Ly bw 1 igtiy SV gL bw B(Q) He ‘This equation gives the strength of a plate in the wrinkling mode and is equivalent to the chart of figure 9. It is seen that the failing stress of the plate is approximately inversely proportional to f°. Equation (12) has been used to esti- mate the strength of compression panels covering a wide range of the structural parameters ty/s, bs/te, and by/ter and was found to give satisfactory correlation with test results. ‘The f-chart of figure 8 was constructed from data of tests on 2024-73 and 7075-76 aluminum-elloy panels and multi- web beams which were assembled with 2117-T3 (previously designated A17S-T3) aluminum-alloy rivets. Since the tivel stiffness is a contributing factor in the determination of a a 50) 0] ea 4,10 othe 8 ise 86 2 8/32 2 37i6 Siva ° 50 ao} = 8 s '9} by for100 te) SO eS 2 z (8) bwite=28. Fiours 13—Comparison of ealeulated and experimental failing stresses of 2024-73 Zstiffoned panels of reference 17 for four ‘values of byflyr and two values of trite, Dall t=0.004 in REPORT 1255—NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS the effective rivet offset f, changes in rivet material can bo expected to make corresponding changes in f which would show up in a panel test as a change in panel strongth. Reference 14, however, indicates that very littlo inerease in panel strength can be expected from the use of rivet materials with » higher modulus of elasticity and strength than those of 2117~T3 aluminum alloy but reports on panels with one rivet material (RS-1 magnesium) which had o smaller modulus of elasticity and strength and which failed at loads that were consistently less than those of the pancls with 2117-8 aluminum-alloy rivets. Similarly, reference 14 reports on panels with blind-type Cherry rivets (AN 463) which failed et loeds less than those of the panels with 2117-78 aluminum-alloy rivets. Figure 8 should be used ‘with caution, therefore, for rivet materials whose modulus of clasticity nd strength aro less than those of 2117-T3 aluminum alloy when used with aluminum-alloy sheet. ‘The panels of reference 14 that were assembled with 1100-F (previously designated 25-F) aluminum-alloy rivets do not satisfy the strength criterion of expression (11) and their low strengths aro attributed to the low tensile strength of tho rivets. +0 “7 Evo te coe 2 4, in s 8 Sie rol Si 2 oie ean ol fw r Wy fooss ° sor of Neeser x € 20 of Ea f-100 lo ° ‘os = ? (©) brtr=30. Froure 18—Continued.COMPRESSIVE STRENGTH OF SHORT SHERI-STRINGER PANELS 127 50, 50 ad 49 aq 3 a 4, in Beet Se S 2 oe ° ie e vo 8 i 2 238 2 ese BT £38 ‘0 i wh bia ; isos E083 ° ° : Py 5 40 40} oS xk 20 20] vof be ss00 10 i ie : ae an oF ——— oA é z (© bolt. (2) brfty=60. ‘Fiouns 18—Continued. Frequently panels aro assombled by using extruded stringers which havo a right-angle exterior corner between th web and the attachment fiango. ‘The uso of such stringers usually climinates the wrinkling mode excopt for very unusual proportions for two main reasons: (1) ‘The smell fillet between the web and the flange of the stringer allows the rivet to be moved in close to the web so that the rivet offset bo is reduced and as # consequence fis also reduced and (2) the deflectional stiffness of such a stringer is greater then that of a bent-up stringer of similar proportions with the same rivet offset because of the large stiffness when the plate buckles toward the stringer. For this case, the stiffness may be more nearly that of the web rather than thet of the cantilevered flange because tho plato ean bear directly on the web. ‘The effective stiffness which determines the rivet offact / is some combination of this stiffness and the stiffness for the case when the plate buckles away from the stringer 1s shown in reference 4. ‘The number of available tests are insufficient to establish a chart such as figure 8 for extruded stringers, ‘These tests (from refs. 14, 19, and 20) indicate that figure 8 can be used to obtein a conservative estimate of the effective rivet offset. Expressions (9) and (11) for ‘490876—o1_—10 Frawar 13.—Concluded, the required pitch and strength of rivets can also be used. Previous investigations of the effect of riveting on the strength of panels of which reference 14 is the most recent have developed a rivet criterion whereby the strength of a panel with e given riveting (given pitch and diameter) is related to the strength of a similar but strongly riveted panel (panel which reaches its potential strength) by a master curve. ‘The master curve is based on the lower limit of test daia from panes of various configurations that were con- structed of 2024~T3 and 7075T6 aluminum alloy and were assembled with rivets of various materials. ‘The present investigation has mede a more detailed study of the data for panels with the smaller rivet pitches—the data on panels with p/d greater than 15 as well as the data on panels which developed interrivet buckling have not been analyzed. With this restrictive scope and the help of recently developed procedures of stress analysis, it was possible to make more accurate correlation of the strength of these panels with the riveting used to assemble the panels. For instance, the present investigation utilizes the concept that, after a certain critical value of ftw hes been reached by decreasing the rivet pitch and/or offset and/or increasing rivet diameter, little128 Aosaek or 0 40 / ‘0 19} © ° Ce : ao 3580 (2) Ditw=19. ©) barltn=29. (© bultr=30. Fiavee 14.—Comparison of caloulated and experimental failing stresses of 2024-T3 hat-stiffened panels of reference 28 for three ‘values of byfty and four values of by/by. tpt 020; pld= 16/3; Boft=11.05 b=0.040 in, or no additional gain in panel strength can be expected by further changes in rivet pitch, diameter, and offset. ‘This critical value of fftw is different for different panel configure tions. Tt is more difficult, to achieve when the thickness ratio tw/ts is large or when the parameter bts is small and in: REPORT 1255—NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS tw 15 Ficunn 15—Compariton of ealeulated and experimental falling stresses of 7075-10 Zetiffened panels of tablo I. ‘Tho shaded poluts represent panels whioh hed stringors with ra/lw==6.0. extreme cases mey be impossible to achieve in riveted panels. ‘The present investigation also makes uso of tho fact that variations in panel strength for a given change in rivoting can be much greater for 7075~T6 aluminum-alloy panels than for 2024~T3 aluminum-alloy panels because plasticity may play @ much smaller part in determining the strength of the 7075-16 aluminum-alloy panels. CONCLUSIONS A method of strength analysis of short compression panels hes been presented which relates the panel strength to the pitch, diameter, and location of the rivets used to assemble the panel. A large number of panels heve beon analyzed with this method. ‘These panels covered a wide range of panel configurations. They had clements with aspect ratios bjt which ranged from 20 to 50 and wore assembled with rivets which had pitch-diameter ratios p/d of from 3 to 16. Both 2024-T3 and 7075~T6 aluminum-alloy panels were considered. ‘The following conclusions can be made from these studies: 1. Panel strength is highly influenced by variations in rivet piteh, diameter, and location. 2. Favorable variations in the pitch, diameter, and loca- tion of rivets for a given panel results in increased panel strength until the riveting is adequate to force failure in the Jocal mode; further variations in riveting will produce negligible increases in panel strength. 3. Tho minimum riveting specifications that will force the andl to fail in the local mode depend on the panel configura~ tion and on the panel material. ‘Lanauey Azzonavricar Lanorarony, Namonaz Apvisory Cosamrrre ror Apronavrics, Lananey Fiexp, Va., January 17, 1965.APPENDIX NUMERICAL EXAMPLE ‘The use of the design charts and design procedures set forth in the body of the report are exemplified by analyzing short, 2024-T aluminum-alloy, Z-stiffened, compression panel which is similar to tho ono shown in figure 2 and has tho following dimensions and structural parameters: ty=0.064 in. rft=40.0 f=0.102 in, Bry =0.40 tylts=0.68 —Balty=8.0 Jsltg=30.0 —Dofty=5.0 ‘The panel is assembled with 8/82-inch, brazier-head (AN 456), 2117-13 aluminum-alloy rivets spaced at 1 inch, The rivets have the manufacturer’s head on the plate side and a shop-driven head on the stiffener side. Young's modulus of clasticity E is assumed to be 10,600 ksi and Poisson’s ratio «is taken as 0.32. ‘Additional parameters and information that can be ob- tained after the panel proportions are given and which will bo useful in tho analysis which follows are the parameters 6 and 2/d, the area of a stringer z, and corresponding area of plato bss, the local crippling stress of a stringer gs.j., and tho strongth of the panel in the local mode #.,- These par rameters are as follows: ,__bulty_ 40.0 bate 30018 ple 10.7 ‘3/32 }= (0.064)*[40.0-+-8.0-+-0.40 (40.0)] = 0.262 in gto?! tg=80.0(0.102)*=0.912 in? gp 2.5 Kei (tof. 25) Fry 31.8 ksi (fig. 10) When p/d and boty are given, fitw can be read from figure 8 as Sit =6.98 ‘The value of f[by is computed as follows: 6.98. 0.175 From figure 6, rr=2.88 (wrinkling) =2.45 (local buckling, extrapolated) ‘Phe panel should wrinkle at (see formula (3)) erl9=25.5 ksi and since the plasticity factor for 2024~T aluminum alloy is unity at this stress p= 25.5 ksi ‘From figure 9 ky and oxy is computed as 64 ouly=39.2 With the use of a curve for ¢ against o/y for 2024~T3 alumi- num alloy with a compressive yield stress (0.2-percent offset stress) of 43.6 ksi, the plate failing stress is found to bo oy=34.1 kei Since oy is greater than the local crippling stress of the stringer found earlier, the load that the panel will carry in the wrinkling mode is determined by adding the loads car- ried by the stringers and the plate. ‘The average stress in the panel is the panel load divided by the panel area; that is, = _MubststorengAs _34.1(0.312)+27.5(0.262)_. . "tetas OBI20.262 ST st ‘The stress % is less than %.,4, found earlier so the panel should fail by wrinkling provided the criteria on rivet pitch, and strength are met. By expression (9), the maximum allowable rivet pitch py is given as z 90-9) 3-43 (80-0)(0.102)=2.04 in. ‘The actual rivet pitch of 1.00 inch is therefore small enough, to prevent interrivet buckling, ‘The allowable rivet strength ia (expression (11)) P 3os00_ 1 car O04 O10, 4 a a9 ve 139] Gow L 3(6.08)+ 160 | 5 1-00=0.280 kips ‘The load required to break tho shank of a %,-inch rivet based upon an allowable stress of 57.0 ksi is 0.394 kips Reference 22 shows that the rivet in question will shear its 129130 ‘REFORT 1255—NATIONAL ADVISORY COMMITTEE FOR ABRONAUTICS head at 68 percent of the lond required to break the shank; therefore, BR: 68(0.304)=0.268 kips which is adequate rivet strength. ‘The predicted buckling ‘and failing stresses are those given previously. REFERENCES 1, Dow, Norris F, and Hickman, Wiliam A.: Effect of Variation in Rivet Diameter and Pitch on the Average Stress at Maximum Load for 248-13 and 758-6 Alunioum-ANoy, Flat, Z-Stiffened Panels That Fail by Locel Instability. NACA TN 2139, 1950, 2. Bijlaard, P. P., and Johnston, G. S.: Compressive Buckling of Plates Due to Forced Crippling of Stifeners, Preprint No. 408, SALE. Fund Paper, Inst. Aero. Se, Jan. 1958. 3, Andorcon, Roger A., and Semonian, Joseph W.: Charts Relating the Compressive Buckling Strex of Longitudinells: Supported Plates to the Efective Deffectional and Rotational Stiffness of ‘the Supports. NACA Rep. 1202, 1954. (Supersedes NACA TN 2987.) 4, Semonian, Joeeph W., and Anderson, Roger A.: An Analysis of the Stability and Ultimate Bonding Strength of Multiveb Beams With Formed-Channel Webs. NACA TN 3282, 1954 5. Gallaher, George L., and Boughan, Rolla B.: A Method of Caleu- lating ‘the Compressive Strength of Z-Stiffened Panels ‘That Develop Local Instability. NACA TN 1482, 1947. 6. Argyris, J. H., and Dunne, P. C.: Part 2. Structural Analysis, ‘Structural Principles and Date, Handbook of Aeronautics, No. 1, Pitman Pub. Corp. (New York), 1952. 7. Kroll, W. D.: Tables of Btiltneas and Carry-Over Factor for Flat Rectangular Plates Under Compresion. NACA WH 1-208, 1943. Formerly NACA ARR SK27_) 8 Anderson, Melvin 8.: Compressive Crippling of Structural Seo- tions, NACA TN 3553, 1056. 9, Stowell, Blbridge %: A United Theory of Plastic Buckling of Columns and Plates. NACA Rep. $98, 1948, (Supersedes NACA TN 1556.) 20, Dow, Nomis F,, and Hickman, William A.: Effect of Variation in Diameter and Pitch of Rivets on Compressive Strength of Panels With ZSection Stiffoners. T—Panels With Close Stiffener Spacing That Fall by. Local Buckling. NACA WR I-44, 1945. (Formerly NACA RB L5G03,) U1. Dow, Norris F,, and Hickman, William A.; Preliminary Tnvestiga- tion of the Relation of the Compressive Strength of Sheot- Stiftener Panels to the Diameter of Rivet Used for Attaching Stifleners to Sheet. NACA WR L-61, 194. (Formerly NACA RB Latia) 39, Holt, Marshall: Results of Edge-Compresion ests on Stffened FlatGheet Panels of Alclad and Nonclad 18-76, 249-73, and 758-76 Aluminum Alloys. NACA TN 303, 1954. 13, Timoshenko, 8.: Theory of Hlastio Stability Con, Ine, 1988. 14. Dow, Nomis F,, Hickman, William A., and Roson, B. Waltor: Effect of Variation in Rivet Strength on the Averago Stross at Maximum Load for Aluminum-Alloy, Flat, ZStffoned Com- pression Panels That Fail by Loeal Buckling. NACA TN 2003, 1958. 18. Dow, Norris F., and Anderson, Roger A.: Prediction of Ultimate Strength of Skin-Stringer Panels From Load-Shortoning Curve Proprint No. 431, S.M.P. Fund Proprint, Inst. Aoro. Sel, Jan. 1054. 16. Dow, Norris F, and Hickman, William A.: Effect of Variation in ‘Diameter and Pitch of Rivets on Compressive Strongth of Panole With 2-Section Stilfeners—Pancls of Various Sttfonor Spnelngs ‘That Fail by Local Buckling. NACA TN 1407, 1947, 17, Dow, Norris F, and Hickman, Wiliam A.: Bifoot of Variation in ‘Diameter and Pitch of Rivets on Comprosslve Strongth of Panel With ZSeotion Stiffenere—Pancls That Fail by Local Buckling and Have Various Values of Width-to-Thiokness Ratio for tho Webs of the Stiffoners. NACA TIN 1737, 1948, 48, Schuette, Evan H.: Chatta for the Minimum-Weight Design of 248-T Aluminum-Alloy Flat Compression Panels With Longl- ‘tudinal ZSection Stifeners. NACA Hop. 827, 1045. (Supor- sedes NACA WR 1-107.) 19, Hickman, Wiliam A., and Dow, Nortis F.: Data on the Compros- sive Strength of 755-16 Aluminum-Alloy Flat. Panols Having Small, Thin, Widely Spaced, Longitudinal Extruded 2-Section Stiffeners. NACA 'TN 1078, 1949, 20, Hickman, William A., and Dow, Norris F.: Data on the Compres- five Strength of 765-76 Aluminum-Alloy Flat Panole With Longitudinal Bxtruded %Section Stifoners. NACA TN 1820, 1049, 21, Wilder, Thomas W., 111, Brooks, Wiliam A. Jr. and Mathauser, ‘Eldon E.: ‘The Effect of Initial Curvature on the Strength of an Inclastic Column. NAGA TN 2872, 1953. 22, Schuette, Evan H., Bartone, Leonard M., and Mandel, Mervon W.: ‘Tensile Tests of Round-Head, Flat“Head, and Braslor-Head Rivets. NACA TN 030, 1944. 28, Mandel, Merven W., and Bartono, Leonard M.: Tensile Teste of NACA and Conventional Machine-Countersuak Flush Rivets, NACA WR I-176, 1944. (Formerly NACA ARR L4P06,) 24, Schuette, Evan H.,'and Nuss, Donald B.: Data on Optimum ‘Length, Shear Strength, and Tensile Strength of Ago-Hardenod 178-T Machine-Countersunk Rivata in 758-T Shoot. NACA ‘TN 1205, 1947. 25. Lundquist, Eugene E., Schuotto, Evan H., Holmer), GoorgoJ., and Roy, J. Abort: Column and Piato Compressive Strengths of Alr- craft Structural Materials—248-T Aluminum-Alloy | Sheet, NACA WR 1-100, 1045. (Formerly NACA ARR LSF01.) 26. Hickman, William A., and Dow, Norris ¥.: Compressive Strongth, of 4ST Alumioum-Alloy Flat Panels With Longitudinal Formed Hat-Section Stifoners Having Four Ratios of Btifencr ‘Thickness to Skin Thickness. NACA TN 1653, 1048. McGraw-Hill Book