CEMENT and CONCRETE RESEARCH. Vol . 6, pp. 773-782, 1976. Pergamon Press, Inc.
Pri nt ed i n the United St at es.
ANALYSIS OF CRACK FORMATION AND CRACK GROWTH IN CONCRETE BY MEANS OF FRACTURE MECHANICS AND FINITE ELEMENTS A Hi l l er bor g, M Mod~er and P-E Petersson Di vi si on of Bui l di ng Mat er i al s Lund I n s t i t u t e of Technology, Lund, Sweden (Communicated by Z. P. Bazant) (Received August 24, 1976) ABSTRACT A method i s presented i n which f r act ur e mechanics i s i ntroduced i nt o f i n i t e element anal ysi s by means of a model where st resses are assumed to act across a crack as long as i t i s narrowl y opened. This assump- t i on may be regarded as a way of expressi ng the energy absorpt i on G c i n the energy balance approach, but i t i s al so i n agreement wi t h re- sul t s of t ensi on t est s. As a demonstrati on the method has been appl i ed t o the bending of an unrei nf orced beam, which has led to an expl ana- t i on of the di f f er ence between bending st r engt h and t ensi l e st r engt h, and of the var i at i on i n bending st r engt h wi t h beam depth. Une m6thode est pr esent , e, par l aquel l e l a m6chanique des rupt ures est i nt r odui t e dans l ' anal yse des ~l~ments f i n i s ~ l ' ai de d' un module, oO l es cont r ai nt es sont suppos6es d' op6r er sur l es c6t6s d' une f i ssur e t ant que cet t e f i ssur e est ~t r oi t e. Cette hypoth~se peut et r e consid~r~e comme un moyen d' expri mer l ' ab- sor pt i on G c d' 6ner gi e en usant l ' appr oche de l ' ~q u i l i b r e d' 6ner gi e. Cette hypoth6se est aussi j u s t i f i 6 e par l es r 6sul t at s des essai s de t ensi on. Pour en prouver l a v al i di t Y, cet t e m~thode a ~t~ appl i qu~e au f l 6chi sse- ment d' une poutre non arm~e et f our ni t une expl i cat i on de l a di f f 6r ence ent r e l a r ~si st ance au moment de f l exi on et l a r ~si st ance ~ l ' e f f o r t de t ensi on, ai nsi que de l a var i at i on de l a r ~si st ance au moment de f l exi on en f onct i on de l a profondeur de l a pout re. 773 774 Vol . 6, No. 6 A. Hi l l er bor g, M. Mod4er, P-E. Petersson Importance of cracks and crack growth Crack f or mat i on and crack growth pl ay an i mpor t ant par t i n the performance of unr ei nf or ced and r ei nf or ced concr et e. Examples of t h i s are crack spaci ng and crack wi dt h i n bending shear chracks and t h e i r ef f ec t on shear capaci t y cr acki ng moment of r ei nf or ced and unr ei nf or ced beams mi crocracks i n compression and compression f a i l u r e . A r at i onal desi gn i n these cases ought t o be based on r e a l i s - t i c t heor et i c al models, which take crack f or mat i on and crack propa- gat i on i nt o account. So f ar no such models have been av ai l abl e. Con- sequent l y t he desi gn methods have had t o be based on empi r i cal re- search, supported by s i mp l i f i e d models. Recent advances wi t h i n f r act ur e mechanics and f i n i t e element methods (FEM) have now gi ven us a p o s s i b i l i t y of anal ysi ng crack growt h. Fr act ur e mechanics gi ves the fundamental r ul es f or crack propagat i on and FEM makes i t possi bl e t o appl y these r ul es t o comp- l i c at ed cases. The cases we wi sh t o anal yse are r at her compl i cat ed, as t hey i nvol ve di ver se phenomena, such as f or mat i on and propagat i on of cracks two or more par al l el cracks bent shear cracks shri nkage s t r ai ns i nt er ac t i on between concret e and r ei nf or cement i n t e r a c t i o n between cement mat r i x and aggregat e. I t i s t her ef or e necessary t o use FEM and al so t o t r y and f i nd a method which s i mp l i f i e s the anal ysi s as much as possi bl e. Proposed approach There are many methods t o choose from f r act ur e mechanics, e. g. the st r ess i n t e n s i t y f ac t or approach the energy balance approach the " s t r i p - y i e l d " model accordi ng t o Dugdale the cohesi ve f or ce model accordi ng t o Bar enbl at t . The d i f f e r e n t methcds are known t o gi ve coherent r es ul t s . In the st r ess i n t e n s i t y f ac t or approach t he st r esses near the crack t i p are st udi ed. These st r esses t h e o r e t i c a l l y a ag_j~roach i n f i n i t y at the crack t i p accordi ng t o t he expr essi on o = K/V2~r, where r i s the di st ance from the crack t i p and K i s a c o e f f i c i e n t , the st r ess i n- t e n s i t y f ac t or , depending on the l oad, t he crack di mensi ons, et c. When K reaches a c r i t i c a l val ue K c, t he crack propagates. The st r ess i n t e n s i t y f ac t or approach has been used a gr eat deal i n FEM anal ysi s. The d i r e c t method r equi r es a FEM mesh wi t h ver y small elements cl ose t o the crack t i p , whi ch l i mi t s i t s a p p l i c a b i l i t y t o Vol . 6, No. 6 775 FRACTURE MECHANICS, ANALYSIS, CONCRETE compl i cat ed problems. I n d i r e c t and speci al methods permi t t he use of gr eat er el ements. The methods cannot expl ai n t he f or mat i on of cracks, onl y the pr opagat i on. In t he energy balance approach i t i s assumed t hat a cer t ai n amount of energy G c i s absorbed by the f or mat i on of a uni t area of crack sur f ace. When a crack propagates a cer t ai n amount of st ored energy i s rel eased. The crack propagates when the rel eased energy i s equal t o or gr eat er than the absorbed energy. FEM has been used t o determi ne the energy rel ease r at e i n the energy balance approach, see e. g. / 5 / . Thi s enables the use of a FEM mesh wi t h r at her l ar ge el ements. The f or mat i on of cracks cannot be expl ai ned. In the Dugdale model i t i s assumed t hat t here i s a p l a s t i c zone near the crack t i p accordi ng t o Fi g. I . Wi t hi n the p l a s t i c zone a st r ess equal t o t he y i e l d st r engt h ~y act s across t he crack. The Bar enbl at t model i s s i mi l ar t o the Dugdale model, but t he st r ess i s assumed t o vary wi t h the def or mat i on. I t does not seem t o have been used i n f i n i t e element anal ysi s. The basi c i dea of the model we propose i s demonstrated i n Fi g. 2. I t i s i n some respect s s i mi l a r t o t he Bar enbl at t model. The model i s descri bed onl y f or mode I (t he opening mode), but i t may al so be appl i ed t o modes I I and I I I . e l a s t i c c r o c k | r e a l c r a c k ~ p i a s t i c z o n e ] m crct ck l e n g t h i L , I Fi g. 1 The Dugdale model f or crack t i p p l a s t i c i t y Fi g. 2 Proposed model The crack i s assumed t o propagate when the st r ess at the crack t i p reaches t he t e n s i l e st r engt h f t " When t he crack opens the st r ess i s not assumed t o f a l l t o zero at once, but t o decrease wi t h i ncr eas- i ng crack wi dt h w, f or example accordi ng t o Fi g. 3. At t he crack wi dt h w I t he st r ess has f al l en t o zero. For t hat par t of t he crack where W<Wl, t he "cr ack" i n r e a l i t y corresponds t o a mi crocraced zone wi t h some remai ni ng l i gament s f or st r ess t r ans f er . As t her e i s a st r ess t o be overcome i n opening t he cr ack, energy i s absorbed. The amount of energy absorbed per uni t crack area i n wi deni ng t he crack from zero t o or beyond w I i s 0 and cor r esponds t o t he area between t he curve and the coor di nat e a xi s ' in Fig. 3. 776 Vol , 6, No. 6 A. Hi l l er bor g, M. Modeer, P-E. Petersson Cr f t I 0 w! W Fi g. 3 Assumed v ar i at i on of st r ess o wi t h crack wi dt h w, general case We now choose the curve i n Fi g. 3 so t hat 7 1 d w = G ( I ) C 0 which means t hat the energy absorbed per newformed uni t crack area i s t he same as i n the energy balance approach. The model of Fi g. 2 may thus be looked upon as a way of expr essi ng t he energy balance approach. At the same ti me the assumption of Fi g. 2 may be looked upon as a r e a l i t y . St resses may be present i n a mi crocracked zone as long as t he correspondi ng di spl acement i s smal l . Thi s has been c l ear l y demon- st r at ed i n t ensi on t es t s , usi ng a very r i g i d t es t i ng equi pment, e . g . , by Evans and Marathe / 4 / ; cf . Fi g. 5. By the appl i cat i on of the proposed model the curve ~(w) may be chosen i n d i f f e r e n t ways, e. g. accordi ng t o Fi gs. 4a, b or c, which al l show si mpl e mathematical r el at i ons . For t y pi c al y i e l d i n g mat er i - al s, l i k e mi l d s t eel , Fi g. 4a seems t o be the best choi ce. I t cor r e- sponds ex ac t l y t o the Dugdale model wi t h f t = ~y. and ~I = COD at i n i - t i a t i o n of crack growt h. The d i s c o n t i n u i t y may gl ve r l se t o some prob- lems by the appl i c at i on i n FEM, but t hey are not ser i ous. w I w f l L W W I W b) C) Fi g. 4 Examples of pos- s i bl e assumptions of v a r i a t i o n of st r ess ~ wi t h crack wi dt h w i n pr act i cal appl i cat i ons For concret e i t seems t hat Fi g. 4c i s the best choi ce as i t corresponds reasonabl y wel l wi t h t ensi on t es t r esul t s / 4 / , cf . Fi g. 5. I t i s al so si mpl e, cont i nuous and sui t abl e f or FEM anal ysi s. For our purpose we have t her ef or e chosen Fi g. 4c. We then obt ai n ~dw = f t wl / 2 0 or from ( I ) , w I = 2Gc/f t For or di nar y concret e Gc/ f t seems t o be of the order 0.005 - ( 2 ) Vol . 6, No. 6 777 FP~ACTURE MECHANICS, ANALYSIS, CONCRETE O.Ol mm, cf . / I / , and thus w I of the or der 0.01 - 0.02 mm. In t he appl i c at i on we f u r t h e r assume t hat the concret e i s l i n e a r - e l a s t i c unt i l f t i s reached. Fi g. 5 shows a comparison between our assumptions wi t h Gc/ f t = 0.01 mm, ~ / f t = I 0 000 and a t ensi on t es t from / 4/ wi t h a gage l engt h of 1 (25 mm). Thi s corresponds t o a t heor et i c al average el ongat i on over thergage l engt h when ~ reaches O, i . e . w = Wl, of 2. 0. 01/ 25 = 800-I 0 - . The assumptions seem t o agree reasonabl y wi t h t he t es t r es ul t . A l ower val ue of E/ f t would have improved t he agree- ment, but from t he poi nt of vi ew of t he energy balance approach t he E-val ue correspondi ng t o unl oadi ng i s most i mpor t ant and t h i s j u s t i - f i es the choi ce E/ f t = I 0 000. Fi g. 5 A t es t r es ul t from a t e n s i l e t es t accordi ng t o Evans & Mara- the / 4 / , compared t o a correspondi ng assumed r e l a t i o n by t he anal ysi s ft o r , , , ~ . 0 200 ZOO 600 BOO Strai n x 10 6 (1" gage l engt h) A speci al f eat ur e of t he proposed method i s t hat i t expl ai ns not onl y t he growth of e x i s t i n g cr acks, but al so the f or mat i on of new cracks, as i t i s assumed t hat cracks s t a r t f ormi ng when the t e n s i l e st r ess reaches f t , i . e . , t h e same c r i t e r i o n i s used f or f or mat i on and propagat i on of cracks. The anal ysi s can be performed wi t h a r at her coarse mesh, as i n the example bel ow, because t her e are no st r ess s i n g u l a r i t i e s and t he amount of absorbed energy i s not very s ens i t i v e t o t he mesh si ze. The p o s s i b i l i t y of usi ng a coarse mesh means t hat r at her comp- l i c at ed problems can al so be t r eat ed wi t hout usi ng t oo many el ements. Appl i cat i on t o an unr ei nf or ced beam i n bending In order t o st udy the a p p l i c a b i l i t y of the method t he f ol l ow- i ng case has been anal ysed / 2 / . An unr ei nf or ced concret e beam wi t h a const ant r ect angul ar cr oss- sect i on i s loaded by a pure bending moment M accordi ng t o Fi g. 6. When t he bending moment reaches a val ue M o t he t e n s i l e st r ess i n t he bottom f i b r e reaches f t - As we assume t hat t he concret e cannot t ake hi gher t e n s i l e st r esses than f t , cracks wi l l form and s t a r t opening when M i s i ncreased above M o. We wi l l now st udy how these cracks grow when t he bending moment i ncr eases. In order t o s i mp l i f y t he cal cul at i ons we assume t hat onl y one crack opens, and t hat t hi s happens at t he sect i on of symmetry. The f i n i t e element mesh used f or the c al c ul at i on i s shown i n Fi g. 6. The bending moment M i s appl i ed as a coupl e of f or ces at the 778 Vol . 6, No. 6 A. Hi l l e r b o r g , M. Mod#er, P-E. Pet ersson C M ~ 4 J 4 2 41 4 0 J 9 3'8 J7 .16 3 5 3 3 3 2 3 t Fi g. 6 Bent r ect angu- l a r beam and a cor r espondi ng FEM r epr esent a- t i o n l e f t end of t he beam. The cr ack i s assumed t o open at t he sect i on t o t he r i g h t , whi ch i s t he sect i on of symmetry. Mois t he moment whi ch gi ves ~31 = f t , where ~31 i s t he st r ess at poi nt 31. Mo woul d be t he f a i l u r e moment i f t he mat er i al were el as - t i c and p e r f e c t l y b r i t t l e . When M i s r ai sed above M o t he cr ack s t a r t s openi ng at poi nt 31. At t hat poi nt we i nt r oduc e a f or ce cor r espondi ng t o t he r e l a t i o n between st r ess ~ and cr ack wi dt h w accor di ng t o Fi g. 4c. Wi th t h i s new f i n i t e el ement system we can c a l c u l a t e t he st r ess at poi nt 32 and we can det er mi ne t hat val ue M = M I , whi ch gi ves a st r ess 032 : f t " We can now i nt r oduc e anot her f or ce at poi nt 32 and c a l c u l a t e a moment M = M 2, g i v i n g 033 = f t et c. By pr oceedi ng i n t he same way we get a r e l a t i o n between cr ack dept h and appl i ed moment accor di ng t o Fi g. 7. 1,4 N I t Mo 0 i i c r a c k d e p t h n o d e p o i n t Fi g. 7 Cal cul at ed bendi ng moment M ver sus crack dept h When t he cr ack grows t he cor r espondi ng bendi ng moment reaches a maximum val ue Mma x whereupon i t s t a r t s decr easi ng. As t he maximum val ue i s reached t he s t r u c t u r e becomes unst abl e i f M i s kept c ons t ant , and i t f a i l s suddenl y as t he cr ack pr opagat es. The r e l a t i o n Mmax/M o i s t he same as t he r e l a t i o n between ben- di ng st r engt h and tensTTe s t r e n g t h , as M o i s t he moment whi ch makes t he maximum bendi ng st r ess i n t he uncracked sect i on equal t o t he t en- s i l e s t r engt h. Vol . 6, No. 6 779 FRACTURE MECHANICS, ANALYSIS, CONCRETE I t can be shown t hat t he behavi our of the beam depends on t he parameter H/ I , where H i s t he beam depth and 1 i s a c r i t i c a l l engt h, def i ned by c c : E G / f ~ ( 3 ) I c As the r e a l t i o n EG c = K 2 C holds f or pl ai n st r ess and appr oxi mat el y f or pl ai n s t r a i n , we may al so wr i t e 2 I c = ( Kc / f t ) (4) Fi g. 8 shows the r esul t s of the above anal ysi s as wel l as of an anal ysi s where shri nkage s t r ai ns ~s accordi ng t o Fi g. 9 have been taken i nt o account. Fi g. 8 Theor et i cal v ar i a- t i on of r a t i o be- tween bending and t e n s i l e st r engt h wi t h beam depth H and I c = ( Kc / f t ) ~ = : EGc/f ~ Fi g. 9 2 , 0 - 1 . 5 1.0 0 . 5 Assumed d i s t r i - but i on of shr i nk- o age s t r ai ns o B e n d i n ~ st r engt h Tensi l e st r engt h ! J . ~ o O t ~,E ~ T .a~ ~ . . . . ~ c ~ j ,, . . . . . < r L 2 3 4 H/ I c I t must be remembered t hat the r es ul t s i n Fi g. 8 correspond t o a si mpl e FEM model where onl y one crack i s assumed t o open, i n- dependent of t he st r esses i n the ot her par t s of t he beam. A more r e a l i s t i c model wi t h cracks opening i n al l pl aces where f t i s ex- ceeded wi l l gi ve somewhat d i f f e r e n t r es ul t s wi t h hi gher val ues of Mmax/M o, es pec i al l y where shri nkage s t r ai ns are pr esent . 780 Vol . 6, No. 6 A. Hi l l e r b o r g , M. Mod~er, P-E. Pet er sson Rel ati ve bending s t r e n q t h 1,0 - e - : = ~ , 0 . 5 - - 0 0 100 200 300 R e a g 4 t , Wi H I s . 1931 7 6 8 Vat ~@s I K t L l e r m o m n , 1932 2/, V a { u e s = Wh r i g h t . C~ r wo o d IO~Z 111 V~ L u e $ O N i e { s e n 195,~ 71 Vat t JeS L i n d n e r . S p r a ~ u e l g $ 5 120 V a L u e J WaLk er ~ [ O l m 1957 ~ VQI ~es Me y e r 1963 ~8 V a L u e s 500 H mm Fi g. I 0 Test r e s u l t s of bendi ng s t r engt h ver sus beam dept h, summarized by Mey- er / 3 / , compared t o t h e o r e t i c a l cur - ve f or 1 = I 00 nTn C Rel at i ve bending stren( th 1,5 , . o o , , 1 ! o i loo o wOt o d r i e d O.O0"J'~ * H 2 P, ~urs ~' dr ded O, Ot S. H 2 h ~ r s [ g s % c o n f i d e n c e k t t 4 r ~ l I I i i b L i " ! 2OO I I - 300 H mm Fi g. I I Test r e s u l t s of bendi ng s t r engt h ver sus beam depth f or one q u a l i t y of concr et e, t est ed wet and dr i ed i n 45% RH, compared t o t h e o r e t i c a l curves f or 1 = I 00 mm C Vol. 6, No. 6 781 FRACTURE MECHANICS, ANALYSIS, CONCRETE Fig. I0 shows a comparison between t heoret i cal values according to Fig. 8 and t est resul ts summarized by Mayer / 3/ . The t heoret i cal curve is shown f or I c = EGc/f ~ : I0 000"0.01 = I00 mm, corresponding to the values used in Fig. 5. I t has been assumed t hat there i s no shrinkage. Fig. 11 shows a comparison between t heoret i cal values according to Fig. 8 and our own t est resul t s. Regarding the i nfl uence of shri nk- age i t must be noticed that the t est specimen had a square cross-sec- t i on, dryi ng in al l di r ect i ons, whereas the t heoret i cal curve i s val i d for a specimen drying only upwards and downwards, and t hat creep was not taken i nt o account in the cal cul at i ons. In spi t e of i t s si mpl i f i cat i on, the model seems to be able to expl ai n the t est resul t s. Conclusion The proposed method of combining f ract ure mechanics and f i ni t e element anal ysi s seems to yi el d r eal i st i c resul ts regarding crack f or - mation and propagation as well as regarding f ai l ur e even i f a coarse element mesh i s used. This opens up the possi bi l i t y of studying comp- l i cat ed problems wi th a l i mi t ed amount of computer work. References / I / Welch, G B and Haisman, B "Fracture toughness measurements of concrete," Uni versi t y of New South Wales, Sydney, Aust r al i a, Uni ci v. Report No R 42, January 1969. / 2/ Petersson, P-E and Mod~er, M "Model based on f ract ure mechanics f or the cal cul at i on of crack propagation in concrete," (i n Swe- dish wi th English summary), Di vi si on of Bui l di ng Technology, Lund I nst i t ut e of Technology, Sweden, Report No 70, 1976. / 3/ Mayer, H "Die Berechnung von Durchbiegung von Stahlbetongbau- t ei l en, " Deutscher Ausschuss fur Stahlbetong, H 194, W Ernst & Sohn, Berl i n, 1967. / 4/ Evans, P H and Marathe, M S "Microcracking and st r ess- st r ai n curves f or concrete i n t ensi on, " Materi al s and Structures (RILEM), No I , pp 61 - 64, 1968. 151 Salah EI-Di n, A S and EI-Adawy Nassef, M M "A modified approach f or esti mati ng the cracking moment of rei nforced concrete beams," ACl Journal No 7, Jul y 1975, pp 356 - 360.