APFDL-TR-72-59 [FINITE ELEMENT MODELING AND OPTIMIZATION OF AEROSPACE STRUCTURES. 4 WALTER J. DWYER GRUMMAN AEROSPACE CORPORATION — F33b(s. 7/-C- lob TECHNICAL REPORT AFFDL-TR-72-59 AUGUST 1972 ‘This document has been approved for public release and sale; its distribution is unlinited, = USAF. d rw.) AIR FORCE FLIGI DYNAMICS LABORATORY AIR FORCE SYSTEMS COMMAND WRIGHT-PATTERSON AIR FORCE BASE, OHIOFOREWORD This report was prepared by the Structural Mechanics Section of the Gruman Aerospace Corporation, Bethpage, New York. It documents work done under USAP Contract F33615~T1-C-1N66, which was initiated under Project No. 167, "Structural Optimization Methods Por Aerospace Vehicles,” Task No. 146701. ‘The effort was administered by the Air Force Flight Dynanies Laboratory, Air Force Systems Comand, Wright Patterson Air Force Base, Ohio, Dr. Vipperla B. Venkayya (AFFDL/FBR) was the Project Monitor. The Project: Gngineer and Principal Investigator of the present effort wea Dr. Walter J. Dwyer, Structural Methods Engineer, Structural Mechanics Section. This report covers work from 15 April 1972 to 15 April 1972 and was submitted to the Alr Force in April 1972. The technical report has been reviewed and is approved. Francis J. Tahiky ot. Chief Theoretical Mechanics Branch Air Force Flight Dynamics Leboratory atABSTRACT This report documents a study made of the optimization of typical aircraft structural components using the Automated Structural Optimization Progrem (ASOP) described in AFFDL-TR-70-118. ‘The structures examined were a fuselage structure for a proposed space shuttle orbiter and a wing structure for the same vehicle. Exemples are given of practical finite element modeling of these structures. atTABLE OF CONTENTS ‘INTRODUCTION DISCUSSION A Wing Design Stuay h 3. 4 Structural Idealization Loading Conditions Design Criteria Optimization Results B Musolage Design Stuay 1. Structural Téealigation 2, Loading Conditions 3. Design Criteria 4. Optimization Results REFERENCES APPENDIX A APPENDIX B iv PAGE eorrer 11 B 1 45 30 37 38 45 46 ay6. MW. 2B. 19. al. 23. ak List of Figures Gexeral Structural Arrangement of the H-37 Orbiter Idealization of Orbiter Wing Portion of Typical Wing Cross Section Finite Element Idealization of Typical Cover Elenent Ultimate Loads Per Side Acting on the Orbiter Wing Distribution of Applied Loads on the Orbiter Wing Conventional Cover Allowable Compressive Stress Thicknesses of Optimized Orbiter Wing Top Cover Elements Left Half of Orbiter Fuselage Finite Element Model Fuselage Bulkhead Idealizations Fuselage Bulkhead Structural Arrangements Mission Profile H-32 Orbiter Applied Loading Conditions Summary of Applied Loading Conditions Ha30 Orbiter Fuselage ULtimate Bending Monent Envelope He3L Orbiter Fuselage ULtinate Vertical sear Envelope H-3T Orbiter Fuselage Ultimate Axial Load Envelope Modulus Ratio for Buckled Elastic Web Allowable Compressive Stress for Bar Elements Main Longerons of H~3! Orbiter Fuselage, Left Halt Aveas and Critical Loading Conditions for the H-3T Main Longerons Starting from Hand Calculated Sizes Areas for the H-3! Nain Iongerons After Optimization Starting with Uniform Sizes Integrally Stiffened Panel Appendix 8 a. 32 33 3h 35 36 39 aa he a7 4BCantracks Approved for Public ReleaseINTRODUCTION Automated structural optimization methods using finite-elenent analysis are now availble with sufficient capability to be used in practical design studies. Yo examine their use in an actual design, the optimization of two components of a shuttle orbiter vehicle is described. ‘he computer progran used is the Automted Structural Optimization Program (ASOP) described in Reference 1, Several extensions of this program which were developed to increase ite generality are also described. Many different orbiter and booster configurations were examined during Grumman's initial studies of the space siuttle system. In these studies the ASOP program was used extensively on the wing ani teil eurfaces. However, because of the lack of experience with the optimization of fuselage structures, no automated optimization of the fuselage was initially attempted. For the study reported herein, use of the program for optimization of both the wing aud the fuselage was undertaken to denonstrate the versatility of ASOP. Results obtained from the automated design procedure are compared with the structural design obtained by more traditional methods. The particular design chosen for this study 1s known as the H-3T configuration; its general structural arrangement is shown in Figure 1. ‘he orbiter vehicle for this design has a delta wing and two main propulsion engines. Four air breathing engines are also provided for flight within the atmosphere, ‘The liquid hydrogen for the main propulsion engines is carried in two external disposable tanks attached to the side of the fuselage. The crew compartment is also jettisonable in an emergency and is not structurallypart of the fuselage. Access to the payload is through clam shell doors along the top of the fuselage. ‘These doors are also not part of the primary structure. Tais report shows how both the wing and fuselage structures are idealized ‘nto finite element mathematical models. This will serve to guide the program user in establishing the idealizations needed for the optimization process. To denonstrate the program's usefulness, results obtained from the automated calculations for the elenent gages are compared to material distributions obtained by the more traditional non-automated methods. To demonstrate the program's convergence characteristics, the fuselage structure was optimized twice starting with different material distributions, vith essentially the same final results.4eyFqH0 LE-H Om JO qemeouEay TEGONTAS TUONO T amity ARES oe meansDIscUssION A. Wing Design 3! 1. Structural Idealization The general planform of the wing is shown in Figure 1. The seni~ span from the centerline of the fuselage to the tip is about 650 inches; ‘the root chord is approximately 600 inches. The development of the finite- element model of the prinary structural box evolves naturally from the defined positions of the spars and ribs as shown in Figure 2. Each of the covers, outboard of the root, is idealized es anistropic membrane quadrilateral and triangular elements which carry all shear and axial stress - i.e., no bar elenents are used (see Appendix B). The webs of the spars and ribs are ‘treated as shear panels separated ty posts which enable transfer of external loads to the structure. Inboard of the root, the type of construction dictates the need for introducing bar elenents to represent the carry-through beams. Safficient attention to the details of the carry-through structure 4s required to properly account for the boundary effects on the outer portion of the wing. The configuration employs covers of stringer-reinforced titanium sheet. The internal construction uses standard titanium spars and ribs. By using the spar and rib inter section as the nodal points for the finite element model, the exposed wing is broken into ten spanwise segnents and an average of five segments in the chordwise direction. Since the distribution of pressure in the chordwise direction is expected to be smooth and since there are no internal cutouts in the covers, the subdivisions in the chordwise direction should be adequate to represent the true state of stress. ALL the elements in the cover of the wing are well shaped, that is, the triangularam 4 a te SEES LEP SQLS GEE” L UPPER COVER Figure 2 Idealization of Orbiter Wing Eselenents are approximately 45° isosceles triangles and the quadrilateral elements are all of aspect ratio of two or less with the exception of the last element in the trailing edge root region. This element is not connected ‘to a fixed point on the boundary and the strain energy in the element can be expected to be lov. For this reason the discrepancy between the predicted and true stress distributions can be expected to be unimportant to the overall behavior of the structure. The spars and ribs are represented by shear panels. The axial load-carrying capacity of the web miterial near the covers ie assumed to be accounted for in the adjacent cover elements, a factor which mst be considered in the final interpretation of the results obtained in the computer program. In constructing the geometry data for structures where the primary stresses arise trom benling moments, for example wings and tails, a pre- Lminary estimate mst be made of the radius of gyration so that the behavior of the structure will be accurately reflected in the finite element model. Consider the portion of a wing cross section shown in Figure 3. x (f Comssotded bets Figure 3. Portion of a Typical Wing Cross Section 6Here 4 is the spacing between stringers, b 1s the depth of the stringer from the center of the cover sheet, and x is the height of the cover sheet above the centroidal axis of the wing. The moment of inertia of the finite element model about this axis should be the same as the actual wing, that is: roa ae ng eg age eats age ® where y is the height cf the element ebove the neutrel exis of the finite element model, A is the totel crose section area of the cover sheet of width 4 and thickness t. I,, is the mment of inertia of the stringer, A, is the shest area between stringers ond A,, is the area of the atringer (it is assumed that the centroid of the stringer is at b/2 from the center of tue cover skin). Making ‘the crces eection area of the model and the actual structure the same and defining Ras the ratio of stringer area to sheet area, one obtains 2 2 ree 2, Teyariyt BaP eae sR Ba 1, or neglecting b” compared to x° and therety neglecting 1,,, and 7g A,t° as compared to the remaining terms in the expression for inertia, ¥ (a+R) = (14 R) ROX 202. me oe (oR) Here, y is the desired height at which the finite element nodal point should be above the centerline. The thickness of the elements is now + t+ A /t, the average thickness of the cover end the elenent idealization would be as shown in Figure |.Figure 4 Finite Element Idealization of Typical Cover Element We have now replaced the actual structure by a finite clement idealization ‘that will have the same bending stiffness, ne should note that there is a Slight decrease in torsional stiffness because the two covers are now closer to- gether, acl there is a slight increase in torsional stifmmess because the covers contain material that is normally in the stringers, In general, if these effects are regarded as significant, the covers should be represented by a combination of bars and membrane elenenta. 2, Toading Sontitions Because of the preliminary design nature of these studies, exact pressure Gistributions oa the wing were not available. Consequently, simplified but realletic distributions had to be assumed. Five loading conditions were selected as being critical. These include: two maximum dynamic pressure conditions at positive and negative angles of attack; a 2.5¢ pullout condition on descent fron orbitsand two landing conditions, The magnitudes of the loads are shown in Figure 5, The distribution of the loads on the exposed surface of the wing was assumed to be as shown in Figure 6, ‘These distributions consist of a subsonic configuration avi a transonic configuration. The subsonic distribution peaks rather sharply at the Leading edge end tapers to zero at the trailing edge, with the 8ee D _ © P Location| Condition L v D vig ete 6 1. Max a (+) 515,00 - - B 2. Max qa (-) ~615,000 - B 3. 2.5 g Pullout 148,000 - - A 4, 2-Pt Landing with 198,000} 236,000 170,000 A Spin-Up 5. @-Pt Tending with 198,000 | 265,000 | -290,000 A Spring ~ Back Figure 5 Ultimate Loeds Per Side Acting on the Orbiter WingC.P. Locations Wing Box Wing-Fuselage Interface A 2Ap - - 44 1365 A= Subsonic C.P. B= Transonic C.P. tse “450 Spanwise Distribution: ass iE 5S P “ge Transonic bows f¥ oy Ger bk—ec cue ee Transonic re Figure 6 Distribution of Applied Loads on the Orbiter Wing 10spanwise distribution decreasing uniformly from root to tip. ‘The transonic distribution, in contrast, has a linear distribution from leading to trailing edge and 2 constant magnitude from root to tip. These distributions are adjusted to give the correct total load and center of pressure locations for each of the loading conditions. The determination of the loads applied to a particular nodal point is based on the pressure acting at that nodal, point and its associated cover area. Design Criteria The material assumed for the optimization is titanium Ti~-6A1-6V-28n annealed having the room temperature properties given below: Fy 7 155 kat Ruy 7 145 kot Pay = 148 kod Fyy = 2100 ksi B= 17.0 x 10st In using the ASOP program, curves of allowable stress versus the appropriate structural index must first be developed, tabulated, and then applied ‘to account for the instability of the cover panels (see Reference 2). The allowable stress curve used for this particular example is the curve for Y-stiffened panels shown in Figure 7. A length of thirty inches is assumed between supports. aweoryg aapeserdiog 2TqUACTTY OI TeUOF@UOATOD J, OTE4, Optimization Results The wing idealization was optimized using the fully stressed design option of the ASOP program. Three iterations were performed. After three iterations the elenent sizes change only in the third significant figure, Therefore,for this idealization no more iterations are necessary. The top cover thicknesses are shown in Figure 8. The thicknesses incresse, as expected from tip to root. The only exceptions are at a few locations near the tip where the local applied loads are high and the wing is still fairly thin, resulting in high bending stresses. The thicknesses of the spars and ribs are almost ell at the minimm of .02 inches and for this reason are not shown, 2Bfoes 025 hoes [eas 02S {025 8 josz |03 oso 225 fors 028 {035 |.o% | ong | 25% (ors 996 |o% |ove | 050 |.ose | O55 039 (ax o31\037 |.043 [og 057 |os# |.062 |,0@0 268, ooo IF oat Figure 8 Thicknesses of Optimized Orbiter Wing Top Cover Elements cg3B. Fuselage Design Study 1, Structural Idealization The idealization of the orbiter begins at fuselage reference station 408, (See Figure 1 ). Forward of this point the structure is essentially an acrodynanic fairing and is not considered as primary structure, The bulk- head station 408 is treated as a simple combination of bars, beams, and shear panels as shown in Figure 9 . The shear panels and beams allow the shear, axial losd and moment from the forward tip of the fuselage to be introduced into the structure as concentrated loads at this station. The bulkhead and frame idealizations at éach of the important fuselage stations is shown in Figures 10 and 11. It should be understood that the idealization includes lumped properties of the afjacent menbers in the real structure and not only the properties of the particular bulkhead or frame alone. Some upecific comments on these structural components are now given. The next bulkhead in the idealization is at station 548. The corresponding structure is shown in the structural arrangement drawing in Figure 11, The important point to note when comparing the idealization and the actual structural arrangement (Figure 11) at this station is the 6level of detail preserved in the idealization. The nunber of shear panels used in the bulkhead 1s the minimm necessary to preserve the size and Location of the cutout for the forward oxygen tank support, and to provide the backup structure for the external hydrogen tank support, There are ‘two beam elenents to maintain the approximate external contour across the top of the bulkbeed. Another point of interest is station 608. This is the point of attach- nent of the cabin to the fuselage. The ber element representing this attachment can be seen in the idealization. ‘The frame is composed only of ‘beams around the outside and bar elements on the interior, The bar elements are needed to support node 52 and to represent the lumped areas of the shear panels between the adjacent bulkheads, ‘The next bulkhead at station 686 serves mainly as a point of application for a portion of the distributed loads on the fuselage and also serves to break up the shear panels along the side of the fuselage into elements of reasonable aspect ratio. ‘Station 825 4s at the bulkhead between the cabin and the cargo bay. Because the cargo bay covers are non-structural, the bulltead is idealized only up to the hinge line of the cover. The aft cabin support is at node 63. This dictates the form of the subdivision of the bulkhead into shear panels. Again the outside of the bulkhead is defined by beam elements, ‘The bulkheads at stations 955 and 1083 are similar. Each is a series of beam elements around the fuselage countour end bar elements defining the central keel of the fuselage and the floor of the cargo bay. In this region the fuselage is essentially a two cell box beam, The structural idealization at at‘these two stations can be compared to the typical construction in this region shown in the structural arrangement drawing at station 1136 (Figure 11). The next section of interest is at station 1213. This bulkhead supports the aft end of the main oxygen tank ani receives the most forward wing spar. This is also the beginning of the air breathing engine bay. The bar from node 159 to 173 is added to stabilize one of the fuselage longerons that ends at this bulkhead ami it is not structural. At this station the bulkhead has again been divided into as few shear panels as is necessary to preserve the basic behavior of the bulkhead. The next idealized bulkhead 14 at 1326 where the air breathing engines are connectéd to the structure, ‘he bulkhead at station 1400 is the aft booster ettackment bulkhead. At this point the booster thrust loads are transferred to the orbiter making the longeron just forward of the interstage attachment point one of the heaviest members in the structure. ‘The bulkhead at station 1502 provides aft support for the external hydrogen tank ettachment, The beam elements that make up the perimeter of ‘the frame portion do not represent actual meubers in the structure, but are the result of lumping together the properties of the frames on both sides of station 1502. ‘The thrust structure for the main orbiter propulsion engines 1s provided vy buliheads at 1572 and 1622, The structural arrangement drawing (Figure 11) is presented only at 1572 which 1s also the ond of the cargo bay and the forward attachment point for the vertical fin, There are several interesting details in the structural idealization of ‘this bulkhead thet should be acted. One of the shear panels in the original Adealization had to be replaced by a bar element because the adjacent sides of 18‘the element were almost parallel. The main thrust structure is represented by @ horizontal beam between 1572 and 1622. The forward side of this beam is represented by the bars aud shear panels between nodes 273 and 257, and 27h and 259. The shear webs for the beam are represented by shear panels between the two bulkheads. Because this bulkhead represents the end of the cargo bay, Support mst be provided for the shear panels comprising the inner shell of the bay, These supporting nodes and the shape of the inside of the cargo bay can be seen by tracing out the nodes 27h, 263, 261 and 279. The bulkheed idealization at 1622 shows the rear end of the thrust struc- ture beam, Also attached at this bulthead is the support for the on orbit 10, ‘tank which occupies most of the space above the center of the bulkhead. ‘The last two bulkheads are almost rings because of the large cutouts for the rocket exhaust, The tops of these two are attachment points for the fin. 19va ; 0 Mae ie] "38 Spear Le ree wo Booster Saeee at esee aa 63 Lop Fs 686 F3008 Figure 10 Fuselage Bulkhesd Idealizations 2023 Lo F5o0s FS 955 A Soa FS 6s, Figure 10 continued. Fuselage Bulkhead Idealizations alWine ATTACHMENT Ponrs ast, saosren Arracioonr: FS 1508 suse yo wine ZDOMENT pons: ATICHMENT PowT? Figure 10 contirued. Fuselage Bulkhead Idealizations 22want Evie Pinus oneTs, srtaciaeeuT vant 08 = ROLE FoR, ON ORB ENGINE. re me FS 1788 Figure 10 contimed. Fuselage Bulkhead Tdealizations 23STAx825 Figure 11 Fuselage Bulkhead Structural Arrangements ah5TAx 113. AG Hee) AET 102 TANK SUPPORT WING BEAM CARRY THRU $7 x1213 Figure 11 contimed Fuselage Bulkhead Structural Arrangements 25—WING BEAM CARRY THRU 574x/326 Figure 11 continued Fuselage Bulkhead Structural Arrangenents 26MAIL GEAR WHEEL WELE MUG BEMA CARRY THRY 8TA x/400 AFT INTERSTAGE PT Pigure 12 contimed Fuselage Bulkhead Structural Arrangements eTENGINE SUPPORT BEAM STRUT BETWEEN 41509 #1572 THRUS | nea PAYA Nene arr rave sueporr STAKIS72 Pigure Li continued Fuselage Bulkhead Structural Arrangements 28squemeSaerry TerMonNS PuSTATNE OFeTOMY peTUTAHOS TT OMIT bBL/~ LS ld! ¥OLS pe ora sax0dMNS Ws—" 292, Loading Conditions Sixteen loading conlitions were selected as possibly being critical for different portions of the orbiter fuselage. A schematic of the mission Profile is shown in Figure 12, The points where the different loading conditions occur are indicated. A summary of these loading conditions, with the magnitude of the safety factor for each condition, is given in Figure 13. The magnitude of the loads in each of the conditions and other relevant informtion is given in Figarell. To make this information more meaningful, ‘the envelopes of ultimate bending moment, shear, and axial loads are plotted in Figwes15, 16, ani17. In these figures the critical losding condition for each point along the fuselage is also indicated. As can be seen, the critical bending monent conditions over mich of the fuselage is load condition 5 during boosted flight when the booster thrust line is several feet below the orbiter centerline. The interface loads between the wing and fuselage and between the vertical fin and fuselage were obtained by calculating the reaction loads of the wing and tail when optimized independently as cantilevered structures. For a more rigorous analysis of these interactions it would be necessary to do a coupling analysis of the wingstail ani fuselage or alternatively analyze the entire vehicle as one structure. ‘he interaction loads for the orbiter and booster were obtained froma previous anslysis of the two vehicles together.eeu ‘or's eTrro4d UoTssIN ZT emTBTE opngTaty ayaao woraeag eoeds 7 (at (Et somBry 998) uoy;Tpuoo SuypuoT oqwuaysap sxoqUTN, 31suOFSFBIOD SUTPEOT PETTsdy Te4TAI0 LEH ET OMIT onmay TOAST senog “BoM ATU OUTST sUPTE ‘FeNAGL TEAS] Texog “Bzeuy ATuQ oupsuy 4z0T (6) mH ‘BuTpUeT yowg-Bupadg ‘gouduy awep esoy Surpuer dq-updg ‘onda ze9p sson Suppuer dq-wrds “Toast ‘aupog 2 Suppuey yoeg-Fupadg ‘unog Trek ‘aurog z Buppuey dq-upds ‘usog Trey “auror Z HOOTINE O5°S woraTuaT so5Tqz9 800g eqs00g Jo pur DE THHFFUL (a) & om pum THEL fo xe PUM PUSH ‘xe xeW spUuTH pmoxD (-) ITA spoT oumwy noriaTeossa eT at wt et et ot or st st wt ue vt vt vt at rT wiv dame noraagadgag soIErGu0osvoraytaep urbeor peTTédy 30 Airmnmg ¥T ome 33sdoToats sureaoy FoTDUeR SqmmTSTN OBUTEsNE TATA LEH ST amin "NI “"W1S 39WaSnd a eal fee | (ft ‘8 *8.855 208) slopas}uoo|gopreoT sao ee jastarae leawrpaE = peT=sFo| i 00p- O5e- o0¢- 0S2~ 002 Osi- O0f = OS ~ 0S oof SETHI gOT X AW “WOW *qN3aadoTeAtE sways THOTIIA PgvaTTT PRETOSNE LOgTGLO BE-H OT omy Aa ir Taiie 8 NI "WLS a Asn o a A aq & & & S88 Elah— GOR qT *8,34a) feaworTar| SdI¥ “YV3HS 3‘edoToaus PROT Terxy SquUTaTN eBeTeME zeATQI0 TE-H LT emis “NI ‘vs 39vTasnd OOp- Oseé- (at ‘Srypeot FeoraTs. oor SdI¥ ‘GVO WIX¥ 36|. Design Criterta The orbiter fuselage is assumed to be made of the cane aaterisl as the ving as given previously, But vecause in the fuselage large portions of the skin are in diagonal tension the elastic properties of most of the shear Panels mst be nodified. the only exceptions are several panels near the min Propulsion engines which are heavily loaded ani therefore thick enough to resist shear buckling, {fo modity the shear stiffness of the panels in diagonal tension, te shear modulus was reduced to 75% of its normal value. The justification for tais is based on the graph in Figure 18 which 1s copied from Reference 3. 1 234 6 1 20 4060 100 200 600 @ = Figure 18 Modulus Ratio for Buckled Elastic Web aTIn this figure 1/1, is the ratio of shear stress to shear buckling stress. Gipp/G 4s the ratio of effective shear modulss for the buckled panel to shear modulus for an unbuckled panel. A,,/dt is the effective area of the upright (4m this particular case @ ring) to the cross section area of the aheet. For a wide range of 1/1,,, the shear stiffness ratio is about .75 when the upright aren ie large compared to the sheet area between uprights. In the ASOP program the input is in terms of & (Young's modulus) and y Poisson's ratio and it is therefore not possible to specify G directly. Instesd orthotropic material properties are specified for all shoar panels in diagonal tension. ‘he failure criteria used for the fuselage are the standard criteria in ASOP, thet ic,maximm shear stress in the shear panels and maximum allowable axial stress in the bar elements that are puckle-resistant, The backLing stress criteria used for optimization for those bar elements that are not well supported, for example the hydrogen tank attactment atruts,is shown in Figure 19. 1, Optimization Results ‘Two separate optimizations of the fuselage were performed starting with different initial element sizes, The initial optimization started with a size distribution that was generated by hand. First, the structure was sized using simple strength of materials formlas ant the ultimate epplied moment and shear envelopes. Using these rough sizes a finite element analysis was performed to obtain accurate internal loads. With these internal loads the structure was again resized by hand, The weight of the resulting finite elenent model was 17,200 pounds. Except for a few minor exceptions, all the shear panels in the structure vere at the minimm allowable thickness. The stracture was then optimized using ASOP, Three iterations were performed using ‘the fully stressed design option in the program, The weights of the structure for the iterations were 12,150 Ibs, ,12,200 Ibs, ani 12,080 lbs. The major 38eo The optimmn stress in a tubular column 1s given by the expression, ino 2B. — ef? tet: shantey, C Wh,E7 7% —"“Wedght-Strength Analysis of Air Geo een craft Structures" Substituting 72 and rearranging terms ne oc | fea ra wet (a? 2 - 63 ost woo ec ow & 2 “ao Mais is the form in which the stability table 60 for sper caps ani diagonals is entered into ‘the FSD optimization program. 6 secon Fas, v0 ss \t . Sn Mea =e 726 750 Bee 2s 300 Figure 19 Allowable Compressive Stress for Bar Elenents 39, 350difference between the final structure velghing 12,080 1bs. and the initial structure weighing 17,200 1bs. was the areas of the frame elenents which were not Anttially sized for the applied loads, insteed only approximate values were used. The most interesting reauite of the optimization are the sizes of the main longerons. for convenience these longerons have been identified in Figure 20. the areas of these longerons at the start of the optimization and after the final iterations are given in Figure 21, Also the critical Joeding conlition 1s indicated. As can be expected from an exanination of ultimate bending moment envelope, load condition 5 is the critical condition over most of the length for those longerons near the top and bottom of the fuselage. In general, the longerons near the forward ani aft ends of the fuselage ere at minim area. From a comparison of the starting size ami the final size 48 can be seen thet the initial celeulations using strength of materials theory and a single finite elenent analysis gave a material distribution that is fairly close to the final distribution. This is important because it indicates that the ASOP program 4s generating material distributions similer to those that would be generated by more traditional methois. After this first optimization of the structure wes completed, another one was performed starting with uniform elenent gages. ‘The results of this exercise are shown in Figre 22 where the longeron steas are given in a form similar to Pigure 21, This time the areas for the first resizing and the Yast resizing are given. Seven iterations were performed in all, ‘The eriticel loading conditions are the same as before and are not repeated in the figure. ‘The weight of the structure after three iterations is essentially the sane a5 the final weight in the previous case. The weight efter seven Ateretions is 11,750 pounds. An examination of these final elenent sizes and wosozTg peyeTROTO PUY Moxy PupyTEys sTONSBIOT UTE LEH 043 Jor SUOTATPUOD FuTpeOT TWOTHTIN pue swery Tz SIMI Zz] 2z{ 2 “s Fy) si] 7] *] 3] 7] 2] & ae om! o--| o4-| on) i28| zor] £87| cov|4e-|05-7| 2-71 29] ow low | a on| ov-| o4-| 04 | 907] 92-1| woz | ove |za't | 00’ | a57| o2- los. low. wzmo7 F] es par] ep $ o> eps] ey om | bt | 0%! om] ws: | 28:0 woe] se) b9°2| £07] 68° | oH | o%-|] on | g | Tr¥OR+IO ow | o9- | ow | ow | 2-7] #-£r| 8:5 | 2s-4| 947) SH7|65- | 09- | oF: | ow: wae? a s | s = on | oz-| bs-| om) on] om: a: | — 58°| ert] 89°| om | om) om *) 2 Z ortos |e] 8 | aT aT “| @-| 0%.| o%-| sovl@0-| oF-| on-| on | on-| on-| on] om] +s | ap: | was) 957! 59° | bs | 19-| 95- | on-| ow-| om | ow- eet tee coral ane Raeg a —— wee) 5E'9| p<] 2 bo Sel £69) +1 ey Zalee eel ees are Sr Ts) 63:| da| g2-| 98- a ae 2 ML on | 4e'| 29°] £9°| Bs Lakd 7] 37] St om:| of | ow a ‘a 19:| acl ss- Pccttet| it Zz) #7 = 1 | Sa} oo on ° an: | sm | sh on| on| zx'| zw | em] zm] on- Zz Zz 5a ed be ee a Ske w21| 0€'2| 59°2| 09°2| o87| éo'\s0s-| om] o”°| ov'| ow: 739" = 997| #0°t| 0G'E| 26"E| cf Z2]007| 05:| on-| om | ow wren wil a) 9) 2] 9 wea ‘| 7 on| #e-| £b°| o87| 947] 98-7) £3-7| tar] 44-7] ee) go7| ees sex|\ | 4 | axvoesre ow | es-| 88:| en se| 25+| s97| 99%| 09-7] ev 977) 09 seaer a] >] 2; #7 3) 5 Saas por e7 om] eae] cert x4 27) e¢7] 095 sargloeew avvogns w | io | 92: | eeu rore| Sere] ase] seadoee dee uzeen aw\|ea«ele|awloleltlele|ols|ele NerLw 207 XIGAON Aesezts wroxyan U3TM Bupyzeag worqerTMy|do seqzy suozeHuO] Lf-H oy} FOF sway 2g omBty 7994 wamo7 axvee ino wamo7 ye xe7| Bear : Taw 16-1) 38° | ban leer | Tee | OFF wadan IECHLYHNEEAW Tae 4 Z| Fea) Sb 425) Bes Se ers 4b | B51) 187) £87) Le7| 4E4 water ST oe Sea ET EET =| 43] tor] S62] E27] SET = laren ; oe B46: | ory | 264 sesyie ee: wlelele)lzi|o gt | wouw207) wegnan 43‘the final sizes for the previous example shows that the two designs are ‘essentially the same even though the starting points were different.REFERENCES 1. Dwyer, W., Emerton, R., and Ojalvo, Z., "An Automated Procedure for ytimization of Practical Aerospace Structures", Volume I Theoretical Development and Users Information, AFFDL TR-70-118, April 1971. 2. Shanley, F. R., "Weight Strength Analysis of Aircraft Structures" Dover Publications, Inc., New York, N.Y. (1960). 3. Bahn, P., and Peterson, J., "A Sumary of Diagonal Tension Part 1 Methods of Analysis", NASA TN 2661, May 1952.APPENDIX A Modificutions for Anisotropic Distribution Equation 4-2 pages 48 and 49 of reference 1 has been modified to give the correct load distribution for aniactropic construction. This has been accomplished by using the values of E in the direction of the element edges in formmla 4.2.6, Elastic Property Axis from Input Data This is done by pre end post mltiplying the elastic constant Aij by the transformation matrix [cos y | sin? y | siny+ cosy where y ssin’ y | cosy | siny + cosy sin 2y j-sin 2 | cos 2y 4s the angle between the elastic property axis and the side of the element. Mads elves @ nev clastic property matrix XK ,,. Fron %,;, B can be easily extracted,‘APPENDIX B ORTHOTROPIC BEHAVIOR OF INTEGRALLY STIFFENED PANELS A noneisotropic structure frequently encountered in present day aircraft structural analysis 1s the integrally stiffened panel, Fig. 23. The following derivation is made with the purpose of applying the anistropic finite elenents herein described to problems of analysis of such structures, where the stiffeners form a regularly distributed orthotropic gridwork. This work does not consider the flexural behavior of such structures but does describe the in-plane mechanics of deforsation. Fig. 23. Inte Stiffened Panel b = Stiffener spacing in x direction agit 6060) a = Stiffener spacing in y direction 2 AL shyt c airecti Alt Bary) 4 Ryt, eres of stiffener in x direction 2 2 Ag = hgt, area of stiffener in y direction «te, Suttoner arn = x dtretton Pea Plate area - x direction p,=:2-= Stiffener area - y direction 2 Bt” Blate area - y direction MtConsider the strain energy of the module pictured on P. 47 a & uaB [ i (Geox +Oper thay Be, Jae dy te (a) Waere Os OF + Tye Cx Ey> Uy are stress and strain components in the plate. gj , 6, are stress snd strain of the stiffener in the x direction while g- , €, are the stress and strain of the stiffener in the direction, sumptions: 1. Stress, and therefore strain, does not vary with coordinates (4.e. independent of x, y). 2, For compatibility of the structural components we mst assume €,=€, and €,= €y Stress-Strain Relations: f G os g, es (1) Ter Tey aye may é; nae =: (b2) is Le éSubstituting the stress-strain relations into the strain exergy equation and performing the indicated volune integrations, we obtain the following expression for strain energy. vate Bee eee 6] © In onder to obtain generalized stress components we must define a strain energy density function with respect to the plate ‘volume in order that integration over the plate volume yields the total strain energy in the stiffened panel. @ ve f U, dv 2 ato vee. FF It must be noted that the modified Hooke's Zaw that follows is based on this step, When the modified Hooke's Law is used to derive a stiffness or flexibility matrix, the integration mst be done over the plate volume only, The strain energy density is therefore: Ee ep BAe, EK cre, +6 YE ©) The generalized stress-strain relations are obtained by differentisting the strain energy density with respect to the components of the strain tensor, for examples G&= 3% a) 4gThe resulting modified stress-strain relations are: @) oy Fron assumption (2) and equations (b) and (n) ve can obtain the actusl stress in the plate end ribs, % 4 & Tuy z i ’ x, @ i Parvo 4 : CAP?) | ORD | 0 t % 50DOCUMENT CONTROL DATA-R&D {Sect lavtictin oft of bast on nding anstton stb steed hn the orca rapertcfaa Unclassified Grumman Aerospace Corporation Bethpage, New York Finite Element Voteling ant Optimization of Aerospace Structures Ta ear aSy Walter J. Dwyer August 1972 F33615~72-C-1N66 Distribution of thts document 1s unlimited. ‘Air Force Flight Dynamtos Laboratory Wright-Patteraon AFB, Ghio eae Mats report doounents 2 study nate of the optinization of typical alzeraft structured, componente using the Astouated Structural Optinization Progran (A502) described in AFFDI-TR-70-118, ‘The structures examined were a fuselage atructure Zor e proposed apace aiuttle orbiter ald a wing suructure for the tene vehicle. Examples are given of practical finite element modeling of these structures, (U) Unclassified canter Caress ————Finite Elenent Method Structural Optimization ‘Automated Design sus.ar ‘ring Otten: 1972 — 750-489/277 vee atthe overnmant Printing Oe Y Tey Cease