ENERGY THEOREMS AND STRUCTURAL ANALYSIS A Generalised Discourse with Applications on Energy Principles” of Structural Analysis Including the Effects of Temperature and Non-Linear Stress-Sirain Relations. by J. H. ARGYRIS, D.Sc. (Eng) Profesor of Aeronautical Structures, ‘University of London, Imperial College of Scence and Technology Co-author of Part II S, KELSEY, BSc. (Eng) Lecturer in Aeronautleal Structures, Imperial College of Seience and Technology NEW YORK PLENUM PRESS LONDON BUTTERWORTHSPublished inthe US.A, by PLENUM PRESS ‘dion of| PLENUM PUBLISHING CORPORATION 227 West 1th Suet, New York,NY.) Fis published by Buwerworth & Co, (Publahec Lid, Reprinted 1968 Originally published in a series of articles in ‘AIRCRAFT ENGINEERING Ost. Nov., 1954; Feb, March, April, May, 1955, JO, Arayee 1960 ‘Reprinted in rest Brian photo io by Page Bros (Norwich Lid.En Lg arto 29/323 239 PREFACE T Present work was originally published as a series of articles in Aircraft Engineering between October 1954 and May 1955. ‘The ‘purpose of these papers was two-fold. Ftsly to generalize and extend but atthe same time alio to unify the fundamental enerey principles of | analysis of elastic structures. Although much of the corresponding theory has been available fora numberof years, to the best of the author's knowledge it has not besa given before in such generality. Asan example, ‘whilst keeping within the small daection theory the arguments have been developed ab initio to include nonlinear elasticity and arbitrary initial strains eg. thermal strains. The first assumption introduces naturally the twin concepts of work and complementary work frst put forward bby Engeser. The author has attempted in this connexion to refer fall relevant and historically important papers. Since the appearance of the present articles, afew papers have been publithed which touch upon the same subject but suffer, unfortunately from a rather incomplete list of references. Seconaly, the writer developed in considerable detail practical methods ‘of analysis of complex siructures—in particular for aeronautical engineer- Jing applications. The most important contributions are the matex methods ‘of analysis. Since they are only edrsorily refered to in the Introdsetion, it may be appropriate here 1o describe their use and origin in greater detail. The matrix formulation besides providing an elegant and concise expression of the theory of such structure, is ideally suited for modern automatic computation because of the systematic ordering of numerical operation which the matrix ealeuls aflords. The necessary programming or the digital computer is simplified since it can be preprogrammed to carry out matrix operations with only simple orders as to location and size of the matrix concerned and the operation ¢0 be performed. The specific programming fora particular problem may therefore be written comparatively quickly and easily and, moreover, follows closely the algebraic analysis [As developed here, the matrix methods of analysis follow from particular forms ofthe two fundamental energy principles applicable to structures ‘made up as an assembly of discrete elements. The one principle leads to ‘an analysis in terms if displacements as unknowns (displacement method), while the second leads to an analysis in terms of force (force method) Besides revealing more ciearly the duality of the two methods, this erivation shows also the close connexion between the aproximate ‘methods (like the Rayleigh-Ritz method) for continuous systems and the matrix methods for finite assemblies This ie partculaly valuable in providing suitable techniques for establishing the basic properties — stiffess and fexibilty—of the individual elements ofa complex structure Where these elements have to be assigned simplified stress or strain patterns. But in stressing the advantages of a unified approach to these diverse problems, a word of caution is necessary against cacrying over into the modem methods too many ideas associated with practical calculations by the established or classical methods, The ability to tackle succesfully problems in which the number of unknowns is measured in hundreds carries with it the necessity of rethinking one's practical approach if maximum advantage is to be gained from modern computational teche niques. Inthe force method of analysis the choice of basi sytem and ofthe redundant forces must be governed primarily bythe requirements of simplicity and standardization, in orger to reduce the manual preparation of data toa minimum, and reduce the probasility of errors At the time of publication of the original articles it was intended to reprint them a a single volume and to follow up the Parts | and if, contained here, with further parts dealing specifically with the practical pplication of the matrix methods. Unfortunately it was not posible, for ‘a number of reasons, to complete this plan and the acticles have for some time teen unavailable. ice there appears to be @ persistent interest in them the present reprint has been produced to mest the deficiency Grateful thanks and acknowledgment are due to the Fitor of direraft Engineering for permission to reprint the articles in this form. The method fof reproduction has not permitted complete rearrangemnt of the text {nto book form, so thatthe divisions into monthly instalments are stil marked by blank spaces. However, errors in the text ave been corrected as far a5 possible, and the pages have been renumbered consecutively ‘to make for easier reference. Grateful thanks are du to Miss J. A. Bereg for her care and skilinffecting these changes. The author would also like to thank here those correspondents who have writen 10 point out textual errors and misprints A list of references to further work is also appended. These are all concemed with the matrix methods of analysis whose basie theory is developed here. In particular, Ref. 6 is an expanded and developed orm of part of the work which was initially planned for the original FURTHER REFERENCES TO RECENT WORK 1,3. H. Argyrie and S. Kelsey. "Structural Analysis by the Matrix Fores Method, with applications to Aircraft Wine”. Wssenschft= Miche Gesellchaf fur Lififaher, Jahrbuch 1936, pT. 2. J.H. Argyris and S. Kelsey. "The Matrix Foree Method of Structural ‘Analysie and some new applications". Bri. eon. Research Counet, RU&'M. 3034, February, 1956 3. 4.H. Argyris, “Die Matrizen-Theorie der Statik”, Ingeniur Archin, Vol. 25, No.3, p. 174, 1957 4, JM. Army. “On the Analysis of Complex Elastic Structures” “Applied Mechanics Reviews, Vol 11, NO. 7, 1988. 5, J. Hi Atgyels and S, Kelsey: “Note on the Theory of Aircraft Structures". Zettschit fur Fligwissenschafien, Vol. 7, No. 3, 1938, J. H, Arayrs and 8, Kesey. “The Analysis of Fureages of Arbitrary GrossSeetion and Taper. aircraft Engineering, Vol. XXX, No. ‘6h p62; No. 362, pr IO; No. 363, p.135: NO, 364, p. 169: No ‘368, p 192; No. 366 ps 244; No. 367, . 272; 1999.” (To be pubished Jn Book form by Butterworths Seientine Publications)be Part I. General Theory By J. H. Argyris 1. INTRODUCTION HE increasing compleiy of ster structures ad the many exact. oF approximate methods svaable for ther anus demged at intersted view ofthe whole subject, ot only In der 0 samp their appatons but also 10 vscover some move general tthe el Inethods Threat ls oter reson demanding more comprehens discussion of the base theory. We mention onl the Insessingsietion pad to temperature stesses and the realiation othe importance of poke linear eet: When viewed from al thee aspect the den of presrtng a ne aan er ha mesg ‘Wit hs presen paper we sot OU Yo Svcop a comprehensive system for the determination of stresses and deformations in ease, ‘structures. Based on two fundamonal energy pmeiies: Although mush fe tis) ven has naturally been known fr mary Years we believe tht sone of te theorems and he eneralty ofthe resulc are ew, Te loading ters considered are of ar arbcry nature and include ab tuto the eft of temperature or other inital rans, Nether do we restrict ourselves 10 ‘asl bodies obeying Hooke's law bu ake account of purely clastic nom linear steswsein favs, This possibly not of very grea importance a presat but may have wider appieations in the future No problems of gas bhiywil be Touched pon inthe presen sete of ares and any oer considerations of lepe-delon tery ar, In neal ome, Those Durpase sto investigate, thin the ari-detesion theory, the sss nd deformations in ease bodies not necessarily obeying inca ste Sain law and under any load and temperate bution Byaainc effects ar intially not considered and ence tt asurned for he presen thatthe Toads and temperature are of the quissate typ When re ‘esigating thermal stam ef we ought sty fo base Ye analy thermodynamic consicetions: Thee ate, however ony slighty thes Sponere sin all theoretical work, we start by discussing the exact implications and equations derived rom the inal touring, bt we donot fest ‘urselves hereto this apect On the conta, we pay close atention te {pproninate methods of arava bsed on te py coneeps of wok andstrain energy. In parucular we tempt to give upper and lowes bounds eral proper ofthe srdture sch ats snes No steps Imada oefimate the eto af sess and formations at any Peicalat pai, “This series of papers orginally aro? from lectures given by the author singe 194530 at the ImprilCollegsy Unversity Lovdon Natural, the scope ofthe pretem work has groom beyond the narrower oncet of undergraduate teaching, butte bass of he analy dacs back ISR dine prone te pint ot hat crn tthe base ides ovate with Enger’ who unfortunatly docs nt stem to have followed thm up. We tele, ofcourse 0 Te oo somplementay cor cepts of work and compiemcntary work. I we consider an ordinary ford Siepicement dagram, then, even Hf he fenict ourseives to smal di Dlacements, tis may be curvilinear, the material follows 9 eat fersestran law. Works the are Getoen the dapaceet ans tn he Gurve le complementary work i hat includ betwen te fone ats tnd the cure, Thu the vo areas complement each She inthe fe tEngulr aren (res) placement) whieh woud be the work he ulate bree were ating with te ful intensity rom the beginning of the {isplacement. Naturally inthe ase of «ody flowing Hooke’ les, Le {wo complementary areis are equal, bust sul etl for ie purpose ot nayss to Keep them apart Since wing a previous paper's On he ‘bet the author has had the opportunity of consulting the most Ie ‘eresing test book of Stephen Pinsheno, Tere a reterene rade fo the work of Wesergards who indeed hs eveopes further te bask das of Engtser, But not on gute such a gencral basis se here: Sines approximate methods igre prominently i is pape elerene ight fo iemade to the work of Proper and Synge. They f00 st ou 1 Gaelop Selma the determination ‘of upper and lower lit to sain fSreray,restncing themarves, however, 0 Hooke? lw and cxclvang femperatire ects, Moreover, it sppeirs that although sanyo thee 2 {2 Sree Feat as eH 38 tS ine Rh ow of Sd de i son Se od 1) Toni, Sand Goode. Thur 9 hy, nde Mace Ne Yr, (13) Are Thema Sire dnt and wre Perens A.C. Va, Dee 98%, Fede ta tS aml eee de cat iaoaam seem Rrcaietdecmigcrth g o lik Setoartat ras ke tan a ae fit hes Powis arc eae sraiog Ta tee See bance ee ones (roi eetemaim meen dead sate meee ane ee es ier ha eet aya cr ee five ged atta cheer er iS ante etre atta ten taet fotos otter ce ns i Pecan akc Misra ce ees ath te at ected vata cacctrawe arte Rien Ai gansare nanan iden: ar asminaente tn eres Mie incu tte cal emanate ees sameeren irae ane gr Setgen eet Seaecinmaraiantast steer et lager recency tag eg name sain enastormias see setae Enc VECME Na 1 Dace 3Firstly the principle of virtual displacements may always be used to derive, for any particular Structural problem, the governing aiferental ‘auations and te appropriate static Boundary conditions in terms of the ‘isplacements, This method, however, i not commended in general asa {Substitute for the derivation from consideration of equlirium and elastic eommpatity, ‘Next the principe of virtual work is used to derive Castgliano’s theorem Patt I, generalized for thermal eects. Asis well known, this priniple plies not ony for non-finear strese stain laws But also for ge dis placements. Outline of argument leads us then naturally to the principle ‘Gf minimum strain energy for a fixed set of displacements and a given temperature distribution. This theorem applies also for nonlinear stress: fran las and is of great interest for approximate calculations In terms ‘Of assumed forms of sisplacements. It shows us tot, while the strain nergy is fora given se of displacements a minimum when the compaible State alto one of equilibrium, it on the other hand s maximum for 8 {ven set of forees under the same conditions, These theorems were Rrst ‘eveloped for linearly elatie bodies by Lord Rayleigh more than seventy five years ag0. They are shown to apply also in the presence of thermal ‘Main andlor nonlinear elastiiy. Inthe remainder ofthe chapter we i Yestgate in more detal approximate methods of analysts Using the Rayleigh-Ritz procedure and'it isin such applications thatthe principle fof irtual displacements shows its greatest power. The pariculae form of the Rayleigh-Ritz procedure known as the Caleckin method is also dit fussed. ft it of importance when the assumed deformations sais) all Soundary conditions. The methods indicated apply again In the presence Of theqndl srsine and non-linear stresesteain aus. The next, Section 5, {ive imple illustrations to the method of virtual dsplacemen’, "The sesond fundamental principe is developed in Section 6. We cal it the principle of virtual forees or complementary virtual work. Here We considera sate of equilibrium, apply 2 statically consistent aad infinitely Smal virtual Torce and stres system and find, by using the Idea Of com plementary work, the second principle This i @ necesary and sufficient Eonaition that the position of equllorium is also one of elastic com- patibility. Again this theorem may be used to derive the differential ‘Squatons of any particular problem, this mein tems of stresses or stress Tesultants, However, our commen on the parallel method inthe ease of the virtual displacements are equally applicable here. It should never be uted asa substitute for more physical and geometric reasoning, ‘Next, we derive whet is esentially a generalization of Castilino’s Il theorem. Contrary to| what is geeally Beived this theorem Joes ‘apply for nonlinear stress-strain laws as long as we replace strain energy ‘ycomplementary stain energ), whieh is defined in the same way a6 ‘complementary work. Its extended to include temperature eects. We proceed then withthe gencralization of Castigliano's principle of imu nn energy (or lease wrk) for non-linear stress-stra favs and thermal Strains. Some interesting Gevelopments derive from this and are given it the form of maximum and maimnum theorems complementary fo those developed under the virtual displacement method. They do not seem (O fave been given previously inthis form and provide a useful background fo approximate methods. They show us that any assumed statically equ alent stres dstibution must always under-esimate the sifines. This is tet valuable for practical purposes and is exacly opposite tothe effect ‘of astumed dipbeement distibstions which always overestimate the ‘Sips. The to in conjunction give us hence lover and upper bounds fw overall charagerstics of the Structure such a8 Ks stiffpes. Tp tg seodon we Giseus also the Unit Load Method whieh, as mentioned Bos, vig th as or oe of he more amvenient meds or {he ealeultion of daplacements and of tedundant forces Its shown {0 be applicable to stractures with nonlinear stese-strain laws. Section 7 resents some simple ilustraions of the principle of virtual Tore. Tn the last section we develop a slightly more generalized version of the 8a method of Mcler-resiau, These equations lead themselves readily Toresentaton in matrix form, Next we obta thecortespondingequations ‘wien displacements and not forces ae introduced asthe unknowns. “A Note on the Mathematics ‘The mathematics used in this paper is in general elementary and shouldbe familia to any university graduate. We have avoided the more formal application ofthe clculut of variations which ean be singularty, Unatiraetive to those more physialy inspired. Chapter 3 and parts of Chapters and 6 may prove ats rather dificult fora student. How. ‘ever it always posible to gain an understanding of the basic ideas by Substituting simple examples e frameworks) forthe necessarily more ‘eral proofs gen here “Te later parts ofthis series of papers will present a number of appli tions of the basie methods developed ere 2. BASIC EQUATIONS AND NOTATION body forees (e8. gravity fore) per unit ncn bof Paxalel toa bn dn ds surface forces per unit surface Co-ordinate system Tin dtetion cosines of external normal t© [9x Oy, O: sie Js nen a2 fon displacements J Normal (umany te Lo yon St} total shear strains | }e initial direct stains (eg. thermal strains) Ten Ton Te Stone Noe‘ shear strains Grape ete, elastic direct strains clastic shear strains even dVededydz clement of volume as element of surface a linear coefficient of thermal expansion (may vary wth 8) 2 rise of temperature E Youne’s modulus G Shear modulus May vary with @ ’ Poisson's ratio J 97 ~B1iYart Ont t BYP + Ouse cee “The corresponding explicit expressions for ay and oe ae obtained by substituting the stains ype te and fee. espectively TOF Pr w work of external forces U.=—Weeonst, potential (energy) of external forces uw steain energy (or potential energy of elastic de- Formation) complementary work, complementary potential of ex: ternal foress and compiementary potential enetey Of caste deformation us complementary potential energy of total deformation From & consideration of equilibrium on an elenent dV =dxdye, las. trated Tor he nirestion mF. Ys we Ue Ut b wiimrscortereneenenonnerss[Bs Fig. 3.—nternal equilibrium conditions From a consideration of equilibrium onthe surface (en. 2) 4) It Rn baby I tat bt mn, Over pact ofthe surfice the boundary conditions may be expres in eins asa 6s Hote ate boundary Sonos aod See ee Ia: ines of dpcenet oe ano or Beet Soundary condos) Natura the boundary conitins ny ba Both yaa ovr the shine prt of ihe sacs Censor esas he {be shown naa essed fay but in at tes (eB ead es Sttbeupte=. Rien ne vn pane but heey Restos toucnsees Sutot th pla seeatecoa ‘ a wale. both state and kinematic conto, “To denote infitesinal ements of eo properties of the structure (eg, co-ordinates, area, volume) we ue the standard symbol "Fo. denote infinitesinal increments of Toros, streste, placement, sai eed eae sete pena L "Hits UP iy aseming engineers theory of bending to hola (eR) -r-0 : wD and te Guo frm 0 vu workeqution [CS e) a} ae-0 (ie tis easy to se that for displacement funetons (80) stsfving al boundary conditions the Galerkin and Rayleigh-Ritz methods must yield the same ‘quations for oy be cand hence also the same ceformations. We need ‘nly realize that inthis ease Eqs. (57) are indeed equivalent tothe principe Of virtual work. Heace substitution of %, » in Eqs (57) must ge he Same result as substation into su-0 “The advantage of Galerkin's method les in a more direct derivation of the equations in a,b,c However, contrary to svhat usually assumed, this advantage (ssl if we ealelate BU directly. We note also thal Galerki's method allows only such approximate functions as satisfy al boundary conditions, while the Rayleigh Ritz procedire requires only the “stsacton ofthe Kinematic boundary conditions,Consider the cantilever under transverse load shown in fia, 10. To ‘obtain an approximate expression for the defieions » by the Raleigh Ritz procedure we need only select a function or sequence of functions Bing However, when applying the Galerkin method Ea, (61a) must satisfy aso the static conditions at #=F, fe. Shear force Bending moment =0 °° (8). (3) an 5. ILLUSTRATIONS OF THE METHOD OF VIRTUAL DISPLACE- MENTS In this section we present a number of applications ofthe principle of virtual work, These are not meant to give the shortest possible solutions 1 the problems considered but merely to illsrate the way in which the method can be applied in some simple cates, More complicated problems are investigated in Part 1, (@) Continuous Beam with Non-Linear Spring Support ‘The uniform beam of bending stffest EY shown in rig. 11 caries 8 uniformly distributed load p and is simply supported atthe ends, At the czntre an additional support consists ofa spring wth the load displase. ‘ent lw pearfierstyg) seccnnanstees ta “ru te ini springs an te pet x whi sng ae Wie eat of he Enso Tory of Beng eas y pans ee faire aig ah by Iie dec Se ba ane Wc en gaan Meee Pas 2 Norinesr = “pring! r—te, Fig. 1.—Virtual displacements: Exam; * “Wien nonlinear spel {G1 Simoly support beam ‘The fist pat of the problem is, ofcourse, trivial and the result known to.any undergraduate, but we wan to show here how the Vital Dspace ‘ments method can be applied in suc a case, We consider virtual placements consisting of small arbitrary ad Sonal tins (ofthe beam fom i egitim po Principle of Virtual Displacements then express the equilorium con: ition inthe form 3u,-8W=0 o @® ‘The virtual displacements must satsy certain kinematic boundary con- tions namely Byn0 for z=0and 2=2! eee « and since both structure and loading are symmetricel we nocd only con Sider symmetrical virtual displacements, Hoses AB de=(69Y=0 for 21 Strains and stresses in beam due to «es as) and therefore the vitual strain de to Bris Bem FBV") = F150)" coon . (26) ‘The increment of strain energy inthe beam due to bending is thus 2f fo-Berdd-de= 24 | 5 which becomes, om integrating twice by pacts, hv") de @ remy bry | v6) [he Br de) esc 8) ‘The increment of strain eneray in the spring is Py a) ‘and the increment of work done by the distributed lod is BWA Bocce svisivessss 1D) 10 Since otherwise By is arbitrary we conclude that to saity Eq all) (E1740, NED") ,p~P=0 . fn snd» must satisty the differential equation Eby. cesceeeeseeee (al) Eqs (a2) and (a13) together with the kinematic conditions ra=O and (¥).0)=0 aa ‘ve all the necessary infurmation forthe determination of» lng rating Ea. (a3) and using the boundary conction (a) andl) wwe find finaly Yor ¥ the quadratic equation PAH UHH 0) KIVAVIV + IYO fas) bere WekOYGEL, Vim Splt24 : (©) Plane Redundant Framenork ‘A lane framework consisting of single joint connected by a number ‘of hinged bars toa rgié foundation is loaded by forces X ane along the axes Ox, Oy respectively (Ho. 12}. In addition, the bars are heated te arbteary temperatures and have also inital trains due to ertors a ‘manufacture, ind by application ofthe Principle of Virwal Displacements the forces Inthe bars. et u » be the dlspiacements of the loaded hinge, measured from the position for which all ars have the correct length and ate-ac 2606 tenes perature. Then the cota direct stain inthe rth bar fr these displacements secon sia 8, po tehet co wy The total stan ig made up ofthe ease strane ogether with the thermal stain ena) and the tal oa pr nh ie a ee el snes a) bing th atonal enth of tar (neato he corrc lent) due to manufacturing errs or oer eases, ems he ante sale tas oe andi dnt ent aa Menta ne[eem behind.) a I we now impose on the joint the virtual displacements Bu, By there arises an increment of strain energy 8U,, and an increment of work 317 of the applied foros, where Area 4,, Length & Fig, .—Virtal displacements: Example (b) Redundant yscem of bare