Fig. 81Rib shear flow In four-tange tube and Ry, The corresponding shear lows inthe walls, observing that they are atcally determinate in single cell tbe, may be Cxpressed in the 2 a5 4 : wher gat te en th ng nd pene With the condition of zero torque about the chosen reference point 4 and Ty is the torque about this ans (se p. 56). There appears lite virtue in ileal uniform tubes or tubes with similar longttadinal variation of Stress carrying material to select atthe shear centre Ee 10 Write gu 88 an aeeee (os) so p. $6). Similarly its oF no. advantage in tubes with arbitrary tidial variation off and B to split 9, in (bd), ina quastengineers theory shear flow gg for 2870 rate of twist and a Bredt-Batho shear ow for a torque Ty fo be found by a, (360), Naturally, n multecel tubes itis 'on the other hand preferable t use the presentation (ba) for ‘calculation of gx Note that since we assume that flange loads Vary linea fetween rib satis, systems (Ol) and (b2) only represent ETB, At any cross-section between rib station if the Mange areas are constant ‘r vary siilary thin each bay. "To compete the sttially equivalent stress system we have to calculate the rib shear flows. Consider to ths effect rg. 62 where rib 1 i shown Isolated under the action of the equilibrating forces: in tis ti assumed that the moments May My are applied as ange leads EMulh &Malhy For a trapezodsl rib the shear flow is no, of course, constant but se- solving farces horizontally for the upper or lower half ofthe rib we find the fverage sheat flow M. ana 9) where 4a, nd gare the shear flows in the covers of the /—1 and bays respectively. The selfequllbvating stress systema Y, We show in gs. 63 and 64 the Range loads and shear Rows due 10 i and’ note that they extend only over the bays (Fl) and f. these figures reprocuee in'a more convenient form forthe present purpose the information given in fc. 43, The system is connected tothe boom load function P of Argyris and Dunne by tt (hte) r=? (v6 “The Y-aystems are determined from (nt) equations of the typ (2828) which take ere the form 3 8ari 48-0 ceceeees OD Where 5. isthe relative warping at the tb £ due to the chosen statically ‘quivalesc stress system. Having the F redundancies we find the exact” ‘tess distribution by superposition of the statically equivalent stress System and that due tothe Yr cd em Fig. She ‘Arrows Indicate actual direetion of shear flows for positive ¥ systems The coeficients 5, and By We write al 8 coeficients inthe split form 325,45,43,.. - ew where he ster 37 refer othe anges (hear), walls and ib espe Te coins Fo tube with sma gaa ation Ys age reas icc toy Busi sevo unc the diet sane snot wa ange ade) are at any cross-section those of E.T.B. and hence generate only planar Hiplcaras oe fans Om he ter bases Seay Cala Ate aes note oi ees rr ett pcr Poa se TOE Sa tab ae at Be ea a 9) winere Pray Pan ate the (nearly) varying fange loads due 10 ¥, ‘external TSads respectively. "The integration sm (G9) it restricted to bays (=i) and since the iatuence of Fis restricted to this region. In general, the contribution of (68) to Bs nepligble and could only become sgn cantina bad design, “The general formula for 5. follows immediately from the unit lead methods, f ft#iae ae (10) here ¢, ate the shear lows duc to Y==1 and the subscript» denotes inte- 3in ration over the walls Again we need only conser bays (/—1) and (2) {End We obusin immediately wih the notation of ho. Bag E OmnulldenneBnmsh lon Baonsliad wn were (12) and the subseripts (and (/—1) indicate thatthe expresions in the square ‘rackets have to be evaluated inthe and (1) bays respectively Note ‘Using Ea, (b) in each of the expressions inthe sua brackets we find Si eneritonnaBass ‘This formula may be simplified considerably if we refer the torque to a point $ whose co-ordinates, 95 are aq teen ] cia) 5, 3 [of yenya=0 . where 5, she shear foot Eq (B13) rede thes to : 3 2 lorena Bie (130) where Ts the torque about $ and = £ eB leans - (15) “The coeficent Bn of Ea (I1) can now be writen conisly a 1 Buu Gg Ta —Toa ous Sis, of course, the er0 warping cenre* for the bay ¢ ofthe tube. The expressions given above for its co-ordinates allow forthe eet of diferent {Spor of the Manges (within the present assumptions). For the uniform bay Eas. (b14) reduce to the well Known result (ora) where D. of Eq (a4 is obtained by statis, seep. 53. “The att contribution to 2 is that ofthe ribs. Fic. 64 shows that CE] anol fee) 1 4(4#)( qt) (42), Sas (Be) “Boul. 19 ne jig ans anlezty gondii a ods an ites of heh Seca cenete enous Bem Birt Birt Bie ++ (wy) The conficents By Fics. 63 and 64 indicate that 8, and 8, can only arse for nimble “The contribution By of the direct strains in the flanges may be expressed vnfrr.$ Spaefe . wn ser on Fo anon Fah eat vaing ne ns dt vist calipers oi 1 we assume that the Nange areas and hence @ are constant i each bay swe find immecitely, BucaymBene Tenens utdeeinersae (raeBann GS foams (018) “he Hatin of constant flange areas mech bey hat been dias on pa. "Broceding now to the conteibution of the Bi, we note that its generat expression Be feist 20) nd derive easy rom hat , | oan where ‘ . v © loten Bmomrs (21), orth st erm Br. in om no that tie bt or Di Weel az a4 HE, FS Ga 0 wy maa tee ley 1h) -eein eH) amsrig Tl earcna th) aia") tis) ealies*t) terra) | scr (22) Finally we find the total By coefficient from Ba-Bas t Ban D ir “This concludes the derivation ofall formule (or the B-coeicientsin the Eqs, {BN forthe unknown Y, Selution of the latter system of equations ‘elds the total stress dstrbution nthe tbe. We lustate our analysis on Xrumencal example forther alow: Exfet of Cucouts Fig. 65.—Syrtem Yo. 1 when rb ks removed I eb kis removed bays k-1 and & merge into a single bay and the calulation proceeds by the previous analysis observing thatthe number of nknowns bof souret, aso reduced by one, Fg. 69 shows schematially the farge loads doe to" Y)-» the lniluence of which extends now over the three bays A -2, 4 Land’ there no system Ys Tin the (k “1) bay a wall is removed it is agin possible to use our analysis wath Tow alterations. Thus, {e) the statically equivalent stress system Paw gen the open tube bay i ‘Sicalated with the engineer” theory and tSrsionbending theory." onFig. 66.—Flange loads and shear flows in extended Y, system when web ‘or cover is remaved Ini I) bay (©) the unknown system Ye, and Ys merge into a single system which we ‘denote by Ys and ilustrate in rc. 66 In ths system the fange loads ‘are constant in bay Ld to the kinematic freedom of Our ‘Open, tube, ‘Observe that the system of (n—2) unknown Ys consists now of Yu Yo Vass Va Yast soseceee Fort (6) ther sno term By su In genera, the coefficients 5, are again cither nero or negligible. Should they be important they may be calculated with Eq, (G9) noting that for Byer the integral extends from sits k-2 to A+ 1 and that in the open bay ‘Prfncludes the toraion-bending flange loads. "The terms Bio, may be obtained from formula (bI1) or (bla) but for im Which it Zero and Bq which is calulated from (see also mo. 66) slldonmesBrvassh—[donmesBmcalad cou) Similarly, the Bee are found from Eq (B16) except f0F Byu, Which is, bunt Ui). G) J-i1&)..- (BI ) there are no terms Bese Unde the assompton that lange ares are constant within each bay 1 Te Bund Som Bunge Pret iOal+ be Ory (oie) All other Bu and By a, waectd excep tht Bum Formulae (0204) or be ete become in the case of, eo) a Fig 81—Namercal example of mes fows In spar Flange loads and shear All other By, and Bi are given by formulae (6203). Formulae (b21) for 8» ete. take for mk the form, 14 5 ssa SFR Grdica Iealce 1 awe —a Ses tsleeane testis) Means tl Beran buns leat Gaal i) + then eal} ay ra BERS RUG REGEN) een 9 ‘The temaining 8, coefficients ate derived as before from (62). I willbe seen later under ibe matrix analysis in @) how much mors convenient the tse ofthe H matrix device i forthe treatment of cut-outs, Ths concludes the long-hand analysis bythe 8, method. In what follows we Uusrate it bya numerical applet. 2, Mmerical Example of the Bu Method Consider the singly symmetrical cylindrical tube with six bays shown in ‘0. 67 under the action of tb loads Ray and Ry AI-bays have the Sime Tength 1=20%n ‘The overall cross-sectional dimensions are given in Fo. 67 from which we derive, using Mes. 63 and 68, the functions 3s, y= +1/6,04= ~116, y= +1/10, eo4= 1/10 not 249 45/88, c= —1/16, eye 3/80, y= 1/16 int. "(028)“The ratio G/E, of shear modulus of walls to Young's modulus of an Biaken Similarly we fine for tne contributions 8, of Eg, (B16) using the valves (Gt) in ante G)E=0-385 in al bays ee “Tan | gives the Fange areas Band moment of inertia J atthe ib 0} faeieen} stations, the wal thicknesed oye nen bay. and the tikes and | Sheard ofthe i 2 cation of tunes acces 0|-| austin | Ee, we asume constant flange areas in each bay ancompute these a he train values ofthe ange areagat ihe rib stains. 0! | eaitiins ‘Having cteuated the tatios ay ofthe walls an the functions ® and yf Eas (bi9) and (elt) we dicrmine easly the cowficents Buy and 0} | aici! Su of Eas (B18) an (ba). These are also gen ane. saul presen ncaa Sqn ses tom rom which we 0} | @aicinn derive the er Bn of Ea (01) ‘Wit the formation piven 1 ZABLS 1 and IF we can obtain the ce 1] | GalGrde dtundartes ¥,t0 for the ese when the ribs are taken as finitely rig 4 ‘The relevant eguatons ae in mati notation [seltn r 1x} 14-108 -3-560 0 o 0 oo! *! Foust) yw re 0 0 0 | ly =a 1 o 3-116 31-635 = 2-380 0 o | ny — 2821 0 9 23m ss 262 0 | | % ner 0 00-2682 sis 1998] | vs |— se Fo 0 0 0 ts ae} |, | a wn | aL I a and Ye may now compute etl By and 8 concise tn [os Si iSong | sas] . 5 sie fume sw oe ee) Tx -10. | co Jas [sr wie 20s oo | | x - mal pon <1 atin <7 200 0 | | x | som aan — aval | 0 rom 1072 «74m -109 208] | x ca 0 0 208-1054 2081-10-20] | the ation of which in in : 0 0 9 208-020 sr20/ | x Yom 1953 Ym, Ye BS. Ym ABST Yom 39 (029) FL" [Next we consider the effect of rib deformability. Eqs. (2) for Bu, are easily ‘aleulated direcly but it may be convenient ais couse thelr matex form lation for their evaluation (sce also Ea. (258) on p. 41). We have (continaed on nest page) 1 7 | L ; | L J ‘CoBREEEN AE pte ii ro and rn ge “TABLE I of Example 9b ‘Cras-sectonal dimensions and By Say coeticients | (o19) | ontoss | ontts6 2) [Sate | Sage Fra" RirciO® | beqoisay | 7-031 Beran 108, 5 ons 5-490 390 wags | raa7 | 20-247 4753 = Ti rors | ta-ist | 14-208 oriei0 | o-1887 | 0°2350 sous | ris | 68 F755 Sam | Lerore | F076 | 78 ‘TABLE TL of Example 9 Stately equivalent stress system and 5. coefficients iw a laos By a 7 ¥ a Come 7 459 Frit) abourspar‘a’ | 75600_| 70800 | 2800 | 19200 Bio To) 26 ww Bum Palle) 10390 | 9920 10 | 920 Piss | Sesto | otaso_| aie | $50 | 20 SovilbTn Heo | 9 | Te | TS | aT Sabin) maaeg | 377-2 | “3163 | =133-2 | <8 Saree lin) Higeo | 157-3 | "Ss | “Ore | ibs Soin T1069] SOF] BW | wT] aT cree aw | | | S| SE lo] GariGuty 10" | "sor | 564 sai | 303_| 100 - 4 leading to the three or ive joint eavations provides a practical and power: sana] {ul mtd forthe detenation a the pater nace ene ne "ure However, in more complicated cases, with mote tha One ngundeey 6522 arsing i ach by, the more atm achnigue of te mtr orm ton and solution fs the only natural tool Even forthe uc borer ne asita tte ana under a general foncing andthe enelotne ot oe eet sae (628) requires the use of mares fore practical raliaton -1 ‘Therefore, for comparison with the long-hand solution and t itlutrate {he way in which the method applied in practice we snalye then tae no 1 pct une an ae ors api ate orton lity matrix forthe points at spars inter = 09:2 "we apply the Homatrs dev we apy the Hratrs devise of p39 to hnd theses ce ibuiion when one web Bld f removed," ‘Which yield (i bin.) Y= 21898, ¥,=297, Y= 11065, Y= = 1243, Yo —721, Y= — 844 (629) Having the Y; we can find the total stress distribution in the given struc- ture, Thus, forthe Mange loads at tne rib station f Poi Pamit Vtg (030) and forthe shear ows in bay £ ant (Ye Yas) 5 fom 6 “Typical results for th Mange load P and shear flow gare plotted in 3. Application ofthe Matrix Force Method o the Same Example ‘The long-hand method of formulating the equations for the redun- éancies in the analysis of the four-boom single-cell cube Jost described, “4 “Akhough the matrix method is naturally associated with the use of sutomaieuigtl computers and large nurtbet of aunty Sek broveextemely uel nsmalr problems mihi morecememne ess Sf computation, Thus, the eakultions destted nese at ee agora! Sk Ope cng machine nd ay te et {hat the compete cacultion—up to and including the fen ae ‘occupied @ total time of about 40 man hours. nine es Sext ‘The applied loads are numbered consecutively from | to 2 a8 shown in no, 6 which ees lo the numbering ofthe ements. A rettoncd inthe intoducton tod) we consider as base systems he npr oe independent as beams and tis obvious thatthe tp sed ote Sas sod ii ose ltd i sch yam, The ip Couns ened neither i the base system nor due tothe redundanees and need as ‘considered at all ve lila Du tour choice of base system tt apparent that inthe mats @ considerable number of zeros wil aris and cla sain ee we ‘sdvance the pattioning of Bsn Fa: O53) sage fares nae es Se ee ee‘where the numbers above the columns refer to the loads R,. Sufies 2 and ‘bare used to denote front and rear spars respectively In addltion, we take advantage wherever possible ofthe symmetry of the structure by writing {erms for the top surface only. Thus Bug isthe matrix of lorgitudinal fange lads in the top ange of spar a’ due to loads Ry--. Ry adh ‘only 12 rows, Le, one foreach end of each eoment of ange (sce zcusion (om p. 40) Numbering of elements is from root to tip m each case cee Fig. 6) “The six redundant force systems of the Yype are applied asin the previous analysis and the mattix by i partitioned ina similar way t0 by » co32) un “The corresponding form forthe Rexbility matrix € ofthe unassembled clement (633) u 4 where fia fy are obtained from Eqs. (257) and tof, fom Eas. (258) Carrying Gut the various matrix multiplications we Can put the products fb, and fb, of Ea. (253) inthe forms, bte.-[1b:f0a5. Abu] [lth 1M {bya (B/)+[b",)4 (byt), Mylan + [6/',), (35) Io the same way, the product Fy by'fby which isthe xibilty mate of 1th: basic system under the extefnalfoads (see Eq. (230) 18 [ota © | [ieee ° eo ee u a ‘The above partitioning ofthe various matrices and the computation of, the Qexbiities asthe sums of the component fexbilites isnot ofcourse riety necessary. However, it usualy happens thatthe number of stucs tural elements considerably greater than either the number of external forees or the number of redundaneies and hence controls the order of matrices to be handled. For example, in the prevent problem the Tall by matrix has 49 rows whereas in the scheme used above the highest order matix is 1212 determined by f,. Although the reduction i ecessary hete for manual computation i is Gbvious tat such a method ls requzed even with automatic computation Tor actual aiverat structures where the ‘degree ofreundarey may be ofthe order of 200 oF 0. ‘With the extremely simple basic system chosen the b, submatrices ean ‘now be written down By inspection. Asa typical case we find forthe 1p flange of spar a the matrix By. of tensile flange loads due to R Pda] eons 9 bth bet (636) rrsess r 7 123456 o1r23a4s o123s4s | 2 Joor2sae foor236 3 ooo 2s on oo ora: ‘ oo oor? ooo 012 oo oo008 oo 0004 6 | 0 00000 L 4 “The numbers in tales over the columns of the matrix refer to the exter= nal loads fy... -y and the numbers opposite the pits of rows are the fhumber of the sments ofthe lange a in each pair of rows; the fist one IS for the end of the clement nearer the root and the second row for the end away from the root (See HG. 68 and p. 40) 6Fig f(righo)—Four-Qange cube. Flange loads and web shear flows due to Snes Yond f= 00's application SY Mara Toren ome For the web of spar’ the matrix Brew of shear flows due tothe loads R Inthe basic systems dese bere brads | orn is eo oor |e os coords 00001 Ie ing matrices for spr "are identical with those of Eas (c3ihand 38) except that they ae mailed Sythe es SE nd “IMO instead at 20/6 and = 11 essetey Using. 68 which shows the prt ofthe sructre affected by cach ele TE Bis Fs 6-82 —8:341 1028 L189 1-599 1-989 2-389 0.006 —3:419 6-838 —10-294 Row SiN 4178 5-244 0 00s =$'5i4 4-088 “lo-ese 0 0-100 1-886 3-253 4.919 6.586 8 0 0-083 4-253 3-307 “12-381 900-181 2-000 4:000 6:000 0 0 O08 —HBIs “S-81s 9 0 VO Osta F118 4-118 0 0 0 “0 O08i 5173 0 0 BO 0063 2000 L . a 66388 1-3916 0.0347 —0-0888 —0-0132 —0-0003 13916 2-9840 0-7156 0-024) —0-0244 ~0-0083, stoyrmiorx | O-OMT 09-7136 2538 0-5454 —O-0088 0-018 y*=10% | posse 0-028 93494 23167 O-466i | —0-0007 =p:0132 —0:0244 00088 O-a6s1 2088S 0° 3578 =0-0003 —0:0083 —0-0184 —0-0007 03578 1-812 L 4 D2 3 8s 6 r 11-320 27-415 43-613 59-689 75-572 91-509 —2-595 6-598 —10-780 —14-873 —18-934 —20-995 ] 2.389 14-956 30-050 45-395 60-395 75-870 —0-575 9-087 8-896 —13-867 —18-802 —25-721, ctevtoy-apctmm | 9-063 2581 14176 28-202 42-159 36-034 “0-022 0-546 5-700 —11-782 —17-980 —26-142 a 8-095 0-102 1938 13-$00 27196 40-097 0-021 0-036 278 64a 12-990 — 19 Ho:023 0128 0-179 1-477 f3-608 28-050 0-008 0-033 0-033 1-249 6-227 —12-012 0001 =0-019 =0:103 0.227 1-593 14328 0-009 0-007 0-08 0-013. 0-755 4-942 u UL -(bt3a) 12 3 6 5 6 7 8 9 wo nu oR . ~ 1-447 2098 2-725 3-385 4-071 4-749 0-409 —Ls00 1-797 2479 —3.186 —3-4r2]1 0-398 —O-mst Ci-gss 2-434 “3-208 —4-088 —0-096 —o-681 —I-482 —3-3L 3-134 —3-954 2 2-011 0-430 —0:971 —1-965 2-978 —3-994 —0-004 0-164 —0-980 —1-964 —2:997 —-024 | 0-016 —0-017 0-323 —1-080 —2-134 3-182 0-008 —0-006 —0-213 —1-053 —2-188 —3.239 | 4 =0:004 —0:021 —0:030 0-246 —1-065 —1-992 0-001 0-006 —0-006 0-207 1-038 —2:002 | 5 0°00 0003 —0-017 —0:038 0°266 —0-945 0-000 0:00! 0-007 0-002 —0-126 0-825 | 6 . So eee 1 7 34 5 6 7 8 9 © HW oR r - 100-15 —101-78 ~95-87 9222 —87-62 83-13 —10-s2 —13-08 9-82 —5-24 —0-69 3-81 | wi Th 102-21 88-99 771i 7-6) 95-44 2-8 “N61 —16-63 10-856 4-38 3-18) w2 0-83 13-97 102-93 90-20 48-74 87-88 0.22 4-98 —28-08 27-58 26-20 —21-53 | ws 0:38 ‘014 "11-03 —103:98 “95:87 89-35 “0-08 0:36 6-47 7-35 —38-02 33-65 12 0:57 0-4 "887 “jou TE 99:7 0-0) O14 O38 6:34 38-49 “Sod 0 0-10 “033 1518 8-30 95-20 0-00 0-08 0:21-0:07 3:93 —25.74 | ws 4 oun Note “The numbers above columns refer tothe loads R, 10 Ri: In’, the numbers opposite rows refer to valees at the rib stations, fn the By and by matrices two values were used at each rb station to denote the (Cqual) loadin the adjoining fange element sections. Here in By, these two values are mere. Ing, the numbers opposite rows refer to values in the web feds ‘The additonal factor 2 is introduced to take cae ofthe lower flange since be‘and by contain only the loads for the top flange. This applies slo for the Mangos of spar a" and forthe covers ‘Canning out the multiplications and assembly indicated in Eq. (035) we bain the complete matrices, fb, and b, fb, and ence, on inverting By, the columa of redundancies ( Roe (43) These results are printed in fll above except for by'f, which is Ideatical with the matrix D of the 8a coeticents in Eq (28)andneednot be repeated here The calculation of the reciprocal mati (b 1b,) wa using the Jordan technique described on p. 28 and require imately 8 hours for completion -by an unpractsed and compara+ ‘ively nse computor. ‘The complete set of fange loads and shear flows due to the forces Ris obtained now with the b, and b, matrices ateady given and the solution (yas S=bR=[b,—bi(by)-'by1b,)R wn, Limitations of space prevent the presentation of the complete matrix, but a8 typical results we show above the flange Toads in the uppet flange of spar ‘a’: baie Bya(by fo -*by th] = (44a) and the shear ows inthe web of spar a: bee (Bove —Broslbs 15, "bs fb.) (oi) Calculation of the ange foads and shear lows forthe particular set of oforces R, used inthe preyious analysis gives results in agreement withthe ‘aloes edleulsted thete. Such differences as aise are’ due to the neglect Df the terms By in the By analysis. (See footnote on p63). [Asa further fustraion, the hange loads and shear flows given in Tame 1 due to a single load fy 1,000 To, are easly obtained from the thie column ofthe final b mauris of Ea. (D44,) "These values are also depicted graphically in rc. 69. fect of a Cut-Out “To show now the application of method of treating cut-outs developed lonp, sl we determine the effet of removing the web of spat ain the thins Bay Ge. element v3), This, of course, converts this bay into a fout-boom ‘pen tube which, if taken inco account initally woud have required the ‘troduction of the pesal extended Y system over the cut-out diseussed fon p. 62 and destroyed the simplicity and regular pattern of both B, and ‘rimaries By our artic of imposing inital stains in the cut-out element ‘Ub the stress thre s aero we eliminate all such special considerations. Moreover, when the cut-out i introduced subsequent to the intial stress analysis, Gur method does aot require any adcitional special stress systems but works solely with th existing information, e web earied only Shear stress, the matrt of inital strains has here only ane row and the Fesult is obiained with remarkable speed and simplicity, the adational ‘matrix fo be inverted being of order 1 “The shear flow in the element to be eliminated, Le. By, is given by the third row Bra (Eq. (440), (printed in full on p. 67) and the matrix Bay Of the shea ow inthis element due fo unit redundancies issmulary the {hed row of Bins (Eq (O40) 12, 1 basppO1 100)... os) Now the equation for the set of Yy redundancies due to inital (shear) strain H in this element is (Ea. (236) Or (202), ‘° DY yb yn =D, (646) and therefore the additional sess (shear flow) isthe elemeat due to H is BisY y= —j,D-tb,/H oan ‘Adding this to the stress inthe clement already calculated, BR, and equa {ng the recutant to zero, we find, using the above by and the B already calculated, the required initial strain Ha(b,0-'by "BR, 192 sayy . (oss) Hence from Eq. (046) the additional Yy system required to nullify the sees in the cutout element is q o-0238 o-1817 0° 5380 Yy=-192 DAR cc ccccceeeseseee (O49) 0.4650 —0-1203 0-0087 Increments of redundancies Yy due 0 cutout, Pee ee i 5 6 7 0-006 -0-064 0471 0-412 0-405 0-402 0-001, fee 3591 3-147 3-096 3056 0-008 Ye 435 10°57) 9-266 9-115 9.037 Oars 247 3-189 —§-083 —7-922 0-020, 0-019 0-323 —2:378 —-084 —2-050 0-001 0-013 ~0:092 0-081 0. Exact eiblty of structure in in... gape e [aise 27-62 35-67 4142 48-35 85-32 4-43 Be TE6E 103-4 152-9" 16h" 194-8 10-82 356) 103-8 2d 259 MBI AIL 1730 482 1329 265-9-435-3 S60 73-2 23-71 $36 Iho 403 3160 Rotor [55-32 1948 8191 722-2 1090-3 1810-6 36-42 151-6 359-3 649-9 995-9 443 10-82 17-30 25-71 30-05 "36-42 1O-8t 1-2) 36:50 65-22 94:30 122-9 1St-6 17-21 18-42 67-28 136-1 210-9 285-1 2543 9895 212-8 355-1 S19 38 129-0) 80-4 SOK3 147-7 339:3. 25-53 932 159-7 3659 BSSL 99TH 1375-0 F261 Teg 3804 6775 10369 ‘The mumerical result for the Yi. mates is printed in full below. 1 should Ge emphasized that this matrix gives the complete solution for a fenetal st of forces R, It only remains to premultply with by to obtain the flange and shear flows, Thus, taking only a single force 1,000 Tb. ‘we obtain Irom the third columa the additional fange loads and shear ows eiven in Taste Il cue to the removal ofthe weo w3. Adding these {o the values already given for the strocture without the cut-out, we obtain the total values fr te structure with cutcou. Fic. 69 lustrates the results traphically, “This paricular example of the application of the H matrix has been worked out rather more ful than i srctly necessary in practice in order {Dilestate the way e works, In fact, the final result forthe streses inthe fracture with cutout is given directly by Eq. 268) on pi. However, the mount of work involved is astonishing small, consisting ofa few simple _matix mltplieations and a single inversion. IN more general cases where ‘eral elements are removed atthe same time the addtional matrix to be liverted sof order mm where mrs the number ofsteses loads or stress flows) which have to be nulited, The Fexbity Marri “To conclude the present example we calculate now the complet flexi- bility matrix of the structure Tor the points and directions ofthe forces RO Be a Eq (3128) for the fesbility matrix, F (Ao, bib "b,)"B/, ees ssvvnee (50) we have already available the products fb, bf and (b,b,)-%b/fb, and therefore we only have to perform two further matix multiplications and one addition, The complete matrix obtained is gen In fil below, Eq. (5800), Tike numerical values given there (a in all other numerical matries in this example) have been founded off to fewer significant gues than were {sed [nthe actual computation “Rithough of eouree the F matrix is symmetrical, this characteristic was not used fo shorten the computations and all values vere calculated ine ‘pendently, the symmetry Being then used to givea check on the numeri- tal work, The same applies, naturally, to the computations of any other ‘Syrametrical mates like D. The whole question of numerical accuracy and ‘Nowance of computational errors s obviously one of extreme importance Jnthe pracieal application of these methods and it isessential (9 havecor= brchetaive methods of texting or checking calculated results. We hope to ‘igcuss these aspects more fully in Part Ill ofthe series with special refer Chee to the use of auvomatc dighal computers. ‘For comparison with the present analysis, the Nexiblty matrix was also ‘computed on the assumption that direct sireses in the Ranges are given Gy ELTIB. je the statialy equivalent system as used in the 5 long hand shaljsi. Ths . est) were b, is, the corresponding statically equivalent matrix, Particular {Mention is drain tothe fact toa Eq, (DS1) includes al shear strains asso~ ited with Be The usual artiferal separation of deexions into those due fo bending shear nd twist doesnot even arise in these calculations. In the following table we compare some ofthe /, calculated by Ens. (080) and (B51). a 9 mM oR 5 0-023 0108 0-125 0-120, O-112 O73 0.804 1-961 0914 0-856 0-509 2367 2.830 269 2-519 0-42 —2.088 2-439 -2-339 -2-189 | R ‘O-114 ~0:332 ~0:636 0-008 0-567 =o.021 —0.025 —0:023 —0-072 - esses (490) so mo Way Be 2843 3238 3650 67-23 95-28 129-0 159-7 65:22 1361 212-8 289-4 365-9 $430 2109 385-1 343 651 843-6 1000-3 30-05 122-9" 245-1 SOLD 747-7 997-3 110 17-2 "25:53 “29-88 36.25 42°61 BD Fe Or 6 16e9 | FBT TT BEL 36-2 30-4 649-3 D088 1074 22-1 HIB 530-6 677-5 936-9 36:35 136-1 306-2 530:5 783-1 1036-9 1287 | J (6500)TABLE IV of Example 9 Comparison of ‘exaet” and approximate flexibilities fy in.b.) eee Tis fa Toa af Eaaae SS aa 75-6529 Hons DT Approx x10" 10°50 29°32 49°61 10611843 N415-1 Error “21% 33% —20% 20% -0-7% Ene Ke oo ‘As would be expected, the errors ae much greater for points neat tothe root, Towards the tip they become negligible butt should be remembered {het the structure analysed Bas quite a high aspect ratio. The example {Rows ail too cleily that for modern tow aspeet ratio wings it wall be Seeessry not only to ake account of the shear strains associated with He Barto base the ealevlations Of flexibility on the exact sess dis- {ribution, Moreover, as shown by cur simple example the computations Wading appear inte ratcethod atrial ep of ese analysts ‘Thermal Stresses ‘We consider now the stresses which arse inthe fourflange tube du t0 noncaniform tnperatur. For this purpose we assume that the upper Range BP tare hes the temperature distribution shown in ria, 70, The. varia~ Goris taken to be linear Between nodal points and hence is, defined Bo"ene valves at the nodal points, Since also the flange loads of the 1 systems vary lineay between these points the argument on p43 leading {o the sesond of Eqs (27) may be used here 00. Thus in place of the tRatrice (eq. 281) ofthe direct strains at the nodal points duet the loads Rin the badie syatem we introduce here the thermal strains (20) at the ame points. As transverse anges donot enter into the present considera~ Alans fre matrix I simplifies considerably and observing that fi the same Imnexch bay, becomes for the affected flange 2100000) 1410000 s 1000 14 1ao ose) o 140 aoa ooo 1? 4 ene, the matrix Bi fom Ea. 28. [ | oe, oe Drona) 20, | ene enncene 8 2. | 2. | 22, | Ls Ou On... --- On 9H he tempeatres atthe nodal points se 0.7) In Bia the two equal values in the elements on either side of the nodal Point are merged into one so that b, «gives bere single lange loads atthe dal points. "The 1000000 o 100000 ,{ oo 10000 Gus | 9 9 0 1 0 0 0 oooo100 oooo010 (60) L 4 aao) e.08/0) 2 wean) ant87e and taking 223% 10" per degree centigrade we find = as “ ° 0 ” 2 soceee 59) Using the inverted matrix Dt of Eq. (b#3) ve obtain the column Yo of eadundancie dus 10 the given temnperatre distribution as (in Tb in.) Yor(—41340 —36860 —28740 —18210 ~13980 ~11220) .. (B61) fom which premultipiying by the appropriate b, matrix, we derive the ‘ange toads and shear lows, ‘These are given graphically in M0. 70. 4, Examples of More Complex Structures Having investigated at some length the relatively simple example of a four-hange tube both by the hy, and matric methods we tir our attention {o some more complex and representative aieraft structures where the {hate method alone is indiated. In What follows we discuss mainly the Becton of the appropriate. seif-equilrating stess systems for the Selingup of the by matin; (ee also the more general a and 39) ‘We seart with a two-spar sitflange tube, the two additional anges being inqroduced to represent more accurately the diect stress carrying ability Fike covers. Bos 71 ilusrates astrucre of this ype fully supported Sk the soot with six bays and no cutouts, The total number of redun- ‘Guages cary seen to be n=-6%3=18 which may be confirmed also by ‘Bar 47a) on p. 387 thus, nalB-+N—4)=66-+1—4) 18 eoosees (062) “The calculation of the By matrix is straightforward and the very simple ‘Tks of to independent spars made in GQ) Is here too the most ad> ‘vantageous. Fos areus-sction with no axis of symmetry we may form the by matrix for the following redundant systems? ooe Fig. 1_—StesNange, single-cell tube, Assembly of X and Y systems 6 Y-systems applied to the comer anges, ) 6 X systems in the top cover and 6 X sytems in the (63) bottom cover 4s indicated om no. 71; see also nos, 41,43 (or (63) and (6) fora full tsenption including shear flows of these systems, TF the cros ection ofthe tobe in mo. 7 has an axis of symmetry the number of redundancies for it oads and torque is me=6x2=12 ‘As redundant systems we choose in this case 6 Y-systems applied tothe comer anges oo) 6 Faystems ‘An ¥ stem consists ofan antsymmettical combination of two equal and ‘opposite X systems in the top and bottom covers as ilustrated on 4G. 72 ‘These coupied X systems were not mentioned on p. 38 ‘Next we discus the sx-Aange tube with intermediate web shown in rng. 3. There are again six bays and for the moment we assume that thee fre no cutouts When the roti fully supported the degree of redundancy is from Eq, (47a) eaB LNA 6§642—4) 24 eorceereeeseers (6D) “The b, matrix may be oblained simpy from the basic system consisting Of three independent spars. As redundant systems we select, 6x2=12 ¥eytems applied to the aoining four flange tubes) 6 Z systems } 6) {6 X systems in the top (or bottom) cover J “The structural elements affected by these systems are shown schematically Inno. 73 Fics. 42,43 (or (63) and (64) and (44) describe fully the Ay ¥, ‘systems. Naturally half ofthe Ysystems may alternatively be based on the tbe formed by th four corner fanges. "Assume now that in one bay the intermediate web is missing and that no stiff jointed closed frame le provided to restore partally the lost shear ‘filfaes ofthe set. The numberof redundancies ls obviously reduced by ‘one, The b, matix may be determined fr the basic system consisting of ‘the spars and the two ribs shown in Fic. 74 OF the redundancies given fn ‘system (66) we lose 0 (se an epee Hspteme Tears a ane tan Fig. 7.—Singly symmetrealshange sin ‘Rand Y systems Fig. 7.—Siellange, two-el ube. Aster of, Yand Z systemssyste patie Fig. 75.—Extended X system applied a top cover the one Z and one X system in the affected bay sn, and the four ¥ systems acting atthe adjoining ib stations ‘We have a net lose of six standard redundant systems and need hence five new syotems. At such we choose (see Mos. 4 and 73) fone stretched ¥ system (as shown in 0, (66) two standard ¥ systems applied tothe four comer anges at the adjoining tx “= two stretched X systems a the top and bottom covers thown in no. 75) ° . Alternatively we could select two stretched Y systems (1wo a ‘ange tubes) bat since these are linearly related to the two above Y systems Applied atthe comer ange itis then necessary to drop one ofthe later Y (oss) F Fig, Th-sheftange, oo-cll ube, Modified Vand Z nystems when rib of systems. On the other hand it may be best to sick tothe choice (B68) for the bay withthe missing web and to work also In the other bays with T ‘tems applied at the four comer flanges. Ifthe cosssection as an axis of symmedry and we consider only torgue toads we lose through the missing web in one bay five of the redundant systems (066), be ‘one Z system in the afected bay co) ‘two standard ¥ systems applied tothe four comer anges atthe adjoining ribs (6a) ‘one stretched ¥ system (equal and opposite stretched X ‘ystems ae the top and bottom covers) ‘Again itis better to adopt the Y systems applied at the four comes flanges aval rib stati ‘Another internal cut-out may oocur through a missing rib, for example in the right hand side cell ee 16,76), Here we lose in the affected (wo ‘ays five ofthe redundant systems (065), i. ‘the two adjoining Z systems } four Y systems at the adjoining rib stations “The required four new systems may be taken one stretched Y system three ¥ systems applied atthe affected and adjoining two f (>) ibe in the same cell “The required four new systems may be chosen as (See no. 76) two ¥ systems extending over three bays and applied atthe aajoining nibs inthe it hand sie el Ge mo. 6) | wm) ‘one Z system extending over the affected two bays cone X system 4 TEE giemetamersnes “Sa bivmranteesaeemrs symmetry the X system of (b70) is replaced by an ¥ system, ai aterm cem incom we mehasieceations om mie two Ppsems at aoiing ibs theta of 06. ay be sued appli tothe opposite cove and BRASS EU So ne edundancla ne ttched system in th feed ell om n ‘one stretched X system at te afected coverase Fig) (see Fa 78), Tr—Sitange, wo-cell cube: Modified ¥ and Zaystems when ‘The above discussion of the effet of missing shear fields and the choice cof appropriate redundancies may be appied with lite alteration to the ‘ore general multtcst structures considered jn Re. However, we require Invadgition for tubes wich three or more eel the stretched Z systems use {tated in ic. 78 when an intermediate wed or cover # mising. “The above discussion shows that every type of cut-out requires special onsideraton and inevitably disturbs the pattern of the b, and by matrices, ‘We emphasize also that particular cave is necesary inthe selection of the rneseseary number of redundancies at cut-out regions to avo the danger Of linear dependency. Its evident from the previous example aa cutout that all thee complications are avoided by the use of the new technique ‘on p. Al, With the help ofthis powerful method we may cary out te stress Analysis in the eontinugus structure (always simpler and smoother pro- rss) and derive Hence (by the indirect use of titi srans in the ements {o be removed) the eres distribution in the actual structure With Cut out. ‘Naturally we introduce inthis approach additional redundancies but this should in seea of ite ar no importance when a Sigil computer In concluding Part I we point out again that the matrix force method ‘and ilustrated here om some very simple problems is directly applicable to delta wings and in Tact to any wing configuration a long as the grid of spars and ribs is ortho ‘onal or nearly 20. Emphasis in che examples has been on the (ores method of analysis since it, in general, more suited to aiceaft applica Tipas. Although the a, ay matrices ofthe displacement metiod (Section 8D) may be slightly simpler to ores thar the b,, by matrices ofthe free rrethod i is.on the other hand easier to) write‘down the f than the Ie ‘ate ofthe unassembled elements. Possibly the greatest advantages of the force method in wing and fuselage problems lie in the better cor Aitioning ofthe Eqs. the much greater accuracy in the. derived stress and the smaller number of unknowns. n 2oaystem Fig, 1.—Exeended Z system in mulecell tube when one ovPart Il. Applications to Thermal Stress Problems and St. Venant Torsion By J. H. Argyris and S. Kelsey 1, INTRODUCTION follow up and amplify the general theocems and principles teveloped in Part by presenting here a number of applications. Its hoped thatthe problems considered may prove of more than academic interest: Three of them are concerned wih siresses doe to dermal sraiing-—a current fashion inthe ateraft structural feld—nd were originally provemted in an A.RC. paper “Te Hit deals with the stresses arising ia citeula ring due to-a pheraly varying temperatare rite on the inside of the ring, which i= Bhcary caste Solution in terms of the bending moment and normal force i the ring i Obtained easly by a straightforward appleation of the Espetho! deverbed in Section 8 of Pat “The second example fs concerned with the same problem but the material of the ring fs taken to have a nonlinear elastic stressatain ‘elation. The anaes is based on the unit load formulation ofthe virtual {ores principle (Section 6 of Part 1). [ti of some intrest that this not {oo trivial problem is amenable to a complete solution without excessive aigebeic and namercal work Fhe use of the general principle of virtual fores for thermal stress problems is illustrated in the thied example. An approximate solution is Ebtained for the stresses in a rectangular plate due to a, temperdture Aistribation symmetrical about the Tongitudinal axis of the plate and Uniform along the feng. in the fourth example, the twin principles of vitual displacements and virtual forces afe both used to obtain approximate solutions t0 the tuations [or the approximate tiping and stress functions are de hd applied to rectangulay and double wedge cross-sections, Use of the io mute ps upper nd lowe Bounds to he oro conan a newimate of he maximum shear tres” The closeness ofthese bounds “ind comparison with some exact solutions indicate thatthe eslts achieve athigh degree of accuracy. 2, FIRST EXAMPLE “Thermal Stree ina Circular Rlog ‘A uniform eirclar cing of symmetrical cross-section and mean radius + is guested on the inside circumference toa temperatre dstbution Oy Foyt = 2c +-conmmpy or ~ZpcZ \ © @=0 over the remainder ofthe circumference the temperature varying linearly across the depth f ofthe section and faling to ser0 on the outside circumference (FG, 18). "The bencing moment and normal force distribution in the ving are to be found withthe assumption that (a) the material follows Hooke's law: (6) the bending deformations of the ring may be calculated by the tngineers' theory of bending for straight beams; and (6 the deformations avising fom the normal force are to be considered but the shearing deformations are to be neglected. Aaational Notation ‘A: aea of cross-section of rng 1; moment of inertia of cros-section of ring -M bending moment (-+ye if producing tension ousise) 1: normal force (+ tension) ‘The method of calelaton follows closely Section 8 of Part From the symmetry of structure and loading, the shear fore in the ring on the axis of symmetry ie zero. Hence as the basic or staialy {sterminate system we select the structure shown in ic. le. As Tedune daneies we choose the bending moment X, and normal fore X, atthe ‘laste eantre,® which inthis cae isthe cere ofthe rng Forces and moments basic system dus 1 spectively ind Xp re System Xy ® 9 MynroonbiN a ene total moments and normal frees: MEMAX HNN, 1 wo Noun, } “ Strsining of basic sytem due to temperature @ (see #10. 1c). Relative Fotation ude of two cross-sections at distance ds apart duo temperature fradiont Oh across the section Dene Sm! ar a el HenBek of net Fig. Fire Examplesthermal ‘renee in crcolar ing (6) thermal straining of ring B0 “The negative sign arses since positive bending strain ¢ corresponds to Increased length on the outside of the ring. Increase of length de at ceonteold (at mid-height of Section) 20 en? és 6 ‘The two redundancies Xy and Xy are calculated from the compatibility ‘equations (182) given in Part I which take here the form Buk +80 } } o BuaNatbeg=0 and ence bu waht ° From the definition of the -coafiients or equations (203) and (204) of Part BM [Ma Bom [Mydade+ [Nona U0 for p=1 and 2. The sign {(.-.}4s denotes integration around the cteum- erence ofthe ring. Using (2), (3), (5) and (6 ia (9) and (10) we fin er suc Bane tw J and eft mp sen tn Burn —292 (14 [rca cot Yl a 20, ,#) mgt SED an } tus : bu_so ft wy $808, ay and yn fea 5 TE omewsacwass 9 woe b onaenc aah andemshy ay Subsiuing (13) and (14) into (4) we obtain the Hal Bending momen and normal force N in the actual structure as " M in) 1 cle) 18 cos ¥] rc) in Zheng an THe eos ¥] sa) 0. | SOF TEE cosy . 7) 1 is often possible to nesleet the deformations due to normal force, ete pau gree rman soa ts ‘in Eqs. (16) to (1 ponding " 7 (On the other hand no difeulty arises ifthe effect of sear deformations ‘on By has to be considered.» This term is, howeve, usually nealigibe. 'N&it we exleulate the bending moment setup in the ring if the thermal straining i completly prevented. Thus: "4 ..—Bending moment and normal force distribution in cirular ring “ive to thermal strain, m= 8+ @,ttec083))/2 » ae iy Fig.2—Bending moment and normal force distribution in elrcuar rin . ‘ue to thermal tering w= * M~ $= 29Et BEN + cos mp) and a $0 ate «is Hence MME con mp) for ZR Eo cecsccee 0) Comparison of (20) wit (16) shows tat inthe case =I the thermal straining is completely prevented if we ignore the deformations due to normal Fores By a1) and (ed ally othe bending moment an Sd sofa ALO TB oe] assseeces QD vi snlm) 1B a hon a mera Apations “Te uration of bed mement and noma fe around the Ft a Seep aS) Een Be Me tt Ltn of or mea Wales of Mate and Te oe ea a oa 3. SECOND EXAMPLE ‘Thermal Stresses fn a Citolar Ring with Non-Linear Elasticity Consider again the ring of the frst example subjected to the sa temperature datibution, The crostsection of the rng is the idealized [Erection shown in rig. 4 and the materi assumed (9 Obey non-linear resestrain eationhip of the form eB + $0) oDa7 tnd nil A—Second Example: idealized tection where Sle) an odd funtion of (ame stressstrain dlayam in com Pression and enson). Moreover, fedBiae=0 for e™=0 whch ensures yuna! gear relation forvery Sal stresses Ge rc. 5). “rk bending moment and normal force istibutons in ie i ae 1 be found wth he assumptions that: ier Only the Ranges of he ection cary direct stresses, which are taken ri te tees ant be constant in each Range (oy the diet strana inthe Manges derive both from bending moment and normal free (c) Shear deformation ofthe web are neil ‘Aduonal Notation UA As area of inner aod outer fange respectively 1A total irc tes earring area 1 Ps loads in ier and outer Range respectively “The structure is twice redundant as nthe frst example and as base atte ake once are hat of i I Alto a fedundancis we elect Seem is crmorment and themormal force Xa a the cere a nee unknowns and X, tf convenient 10 work in fms of ange loads an sain Flange loads due to, Nadi Mey Made, Me Py oN Mes Pym ie a) [Note that M, i taken about the centroid ofthe crosesetion. Using the sete tet Ge Ns and Mof Ege 2) and 3) (24) we ob forsyatem f=: pan ih as ran (Hef) co Pax (4 -f)eonv en secs nl hang nds and can be epee 28 \ an Suistiuting these expressions for the stresses in Eq, (23) the elastic strains? in the flanges are found to be baat Path; Pat) 7 PakytPaake , g(Pakvt Pose) Padysataag( Rabie Patt) | yeaee yne ‘Noting that in the actoal system no relative movements take place at the cut Ende and applying the unt load method as given by Eq. (71) in Part | we obtain the allowing (wo equations We? ra )ds=0 1 ‘ an NaPatesPtaOPakle=0 f eB bePadesdy 0 | \ I. (30) eRe +P Ms 45, 10 08 06 g 6 04 o2 0s cE 15 20 c Fig. -Nominear sressstea where By faO Prats Big SOO Pads =. op are the relative displacements of the point of application of X and Xy ate to ee thermal expansion ofthe Inner ange, ting Ea 8), an 2) or he Rage ons and wri in 0) Byki +BiakatlPihy + Povbele Bye" 0) Bay X yt BeaNe tl Pah + Prabdds + 5, f = Ban f(ME ME) acand bandas (BAM) as are the sual displacement coecent 8 avsig from the linear part. 0/6, Sfincserain c andarethe same anin the irstexample 1a fct,By4"On1=6 ue to the selection of the elastic centre a8 ongin of X'and y and Ege. (82) simply to By Ay HPs Praga Bye 0 oo BesXa bl Pugit Paptds+By90 j “These reduce 10 Eqs. (7) of the first example when $0, 5 ~ coefcents bu 00. i mo . on ban 0144) oP an which are essentially the same as in (12, but fora small dilerence in Bay Ming rom the geometry ofthe erossscction chosen. Furthermore der 5 78 BE Bae BUM cece 6) where 1 Aad ve paand 1-44 asia SBOE 2 oD qs, (36) are identical with (11) inthe ist example. Nontineae terms of Eg. 4) Ook Bas ay and Gh for PP. and Eqs. (2) and (3) for MN the angoneets ofthe p-tunetion in Ui are convententy writen in te form paral pateeerycer)) ee ano brea en o(1-44)]} ‘Substitution of G8) nto (24) shows that fora given defunetion we obtain Seer ation, with respect to 9. two non-linear equations inthe un Fae rend The solution of these equations determines the com- plete sess dation inthe ring 105 } 04 X 03 el [4 02 —] o4 | 10 20 30 inearly elastic ving - bSecond Example: thermal stresses In ‘VEldes of redundsncless, 3nd" "? a =e rationship eo-EE) cy o where a, is an arbitrary reference stress (, may be defined asthe stress at Srnich the total strain i tice the linea? stain component) and ns a bd integer: By varying oy and rts possible to obtain versuse diagrams ‘with a wide varty of shapes (ee Fo. 3), ‘We assume also that The defunctions of Eqs. (38) take now the orm pile txy+B) cos OPE | foo” te gt gin oy rad hal pen, Saye Bek pepe. see “ satire oe tn ate) fd [otesmerf Ete trdt +P) cos bina fet) cos pS Jragademor§ Sf try txdl—B) cos Prey : oe wy Jeusbtepnfa +009 tx, +clt +) cos cos fs Jenade=prft 20% xr +xdt =P) cos $2 cos which can be integrated without dificult since mi an integer "Faking nod Eas) redace othe suancous cule equations Defoe deytt34ie,41—-L =o 21+ By + phegLM2+ Boxy? +3 + 682+ 8 x28) — 5) 6 20, Fig. 1—Bending moment and normal force detibutions in nonlinearly laste ing ue anton tey~ Dep tpteyl6xyt43x,3}-1-0 The linearly elastic ease is represented by p=0. Then singel which valves may also be obtained from Eqs. (13), (14) and (19). 0) \ (46) ‘Numerical Results Ege, (46), which have only one pair of real roots, have been solved for vsiues of'p beween O-and 3 and fic. 6 shows values of xy and xp Bleed apna Tne rots ofthe simuancos cb equations ware DSbtained by an eration method, successive values of xy tnd xy being falculated, using the previous approxiations forthe ters nthe brackets For p03, successive values fr and x, caleulated in this way ae: O'S) 0-433 O52) O-449 0-449 Xr 0-5 0-390 0-416 0-410 0-412 ‘The intial approximation inthis ease was the nearly elastic solution: for further valuce af . extrapolation of previous results gives satisfactory starting points, Since successive approximations are alternately greatet find less than the inl values of xy and ry its quite easy to improve the fhtural convergence of the process by ‘a litle judicious anticipation, Direct solution of Eqs. 46) was also obtained. Eliminating xy an equation cubic in px," and ty is obtained, which can be solved for ether para- ‘eter and te corresponding values ofp. x, and x, calculated, “The bending moment and-notmal force dlstTbuton around the rng is plotted in rcs 7 Tor p0 (linear elastciy), p™=I"O, and p=2°0, it the on-dimensional form M/My and NF/My As would be expected, the ect ofthe nonlinear sresesran curve (which azide decreases with) is generally to reduce the maximum bending moment produced by a given temperatute rte and also to smooth out the variation around the cing. 4. THIRD EXAMPLE “Thermal Streses ina Rectan Consider & thin ectangular plate of length 2a and width B with a temperature distribution 0) an) symmetrical about the longitudinal acs ofthe plate and constant along thelength see ric. 8). The stresees in the plate are to be found with the Sssumptions that they are constant across the thickness (plane stress and that the material i otropic and follows Hooke’ Iw. Additional Notation 2a, bs length and width respectively of plate ‘xy: citesian co-ordinates with rexpect to the ‘entre ofthe plate Gesla, nmyib,b ‘non-dimensional co-ordinates i): thermal strainff: fonations of » Ay B,C: constants ($2 function of oF £ +4: roots of auxiliary equation Smabla “The thickness ofthe plate i taken as unity. “The problem can be solved exactly but we shall derive here an approxi= mate abloion using the principle of virtual fores in the form discussed in Part I, Section 6E. order frat a pate of infinite length for which the stress distribution ig Sel Known Thus. forthe symmetrical temperature istrbution con- Sldored here, the selequllbrating stress system ) 20 Ba0 [sto)dn—st0)] ' . “a enon 0 J fs ct up, To obtain th stress distribution ina plate of nite length, we can Apply a stegeystem one atthe (Wo ends sa and superimpose the A2et i (llanion) stregeystem on (48). We sall, however, adopt here i siferent apprsich in order to illustrate the application of Our ‘Seinepl of vital forces im the presence of thermal strains. “Gaded by the solution (48) for an infrte plate, we assume that i is possible to express the tongitudinal stress inthe form ca OSE vm we Se ee sn and. fs a fonction only of. We know that Ea. (49) cannot, in general Fete he re ston Sze he cost rb il ae ‘ary with 2 However, fort amoath function 9) ican Be expected that Uy yells sulicenty accurate answer * "Pie itibution (49) ean be represented bythe tess function a) : "sn wish pe he ress er ry a pe can 3 lt o YF not pf ou BOE IS sce nC) “The Guy te distributions are symmetrical and the ay distribution anti= symmer al ith respect tothe axis. Pom Bq. (52) and the conditions : 63) 2 canieymoesicaldstibtion) 6 the fnaton fis [sdnrendy foes on We apply the principle of vetual forces in the form of Eq. (804) of pag spb the sai c Ea, (806) of [Jane +easopracdy—0 . . (38) we ttl oe ste BF sare tte semen he BF=EAObY S$ asrieasensy - 3) Substituting for F and © fom Eq, (SI) and (47) respectively, and ) we obtain from Ea (58) Sth, at ah dy at, a [er Bese] raeearo 0 ign my be writen, on iterating with epect 10 Sed agadhfiew-n pe @ here the cous, B,C We ra Note that 4, Band ae al postive. be (6 ea ont be sted for an arity 36 if ABB FhrGicg—0-0 wy which isthe required differential equation in "The particular solution of (64) i get. . cousins: 6) whichis, of course, the solution for an infinite plate and the complementary Finstion ean be writen in the form P=K, cosh By + Ks cosh Bob + Ky sinh ByE+K, sinh Bye .... (66) were Bus= 4 « are the oot ofthe airy equation asap fic=0 . vo 8) Seas Te Slot oncopce tac a arted Gate 2% Sse Meteo ere smoot 0) 1 Ee om Taw cal” EAT fenone VE) palcosh yt +g) sin «ol —€)-+oosh 1 =€) sin ool + £9) 4-1 Hee win u(t +€)c08 w(t ~€)-rsinh wll =) cos w(t +8) rome a) which, atthe centre of the plate (E=0) reduces t0 n