ACI 224R-90

Control of Crackingin Concrete Structures

Reported by ACI Committee 224

The principal causes of cracking in concrete and recom-mended crack control procedures are presented. The cur-rent state of knowledge in microcracking and fracture me-chanics is discussed. The control of cracking due to dryingshrinkage and crack control for flexural members, layeredsystems and mass concrete are covered in detail. Long-term effects on cracking are considered, and crack controlprocedures used in construction are presented. Informa-tion is provided to assist the engineer and the constructorin developing practical and effective crack control pro-grams for concrete structures.

Keywords: adiabatic conditions; aggregates: air entrainment; an-chorage (structural); beams (supports); bridge decks; cement-ag-gregate reactions; cement content; cement types; compressivestrength: computers; concrete construction; concrete pavements;concrete slabs; concretes; conductivity: consolidation; cooling;crack propagation; cracking (fracturing); crack width and spacing:creep properties; diffusivity; drying shrinkage; end blocks; expan-sive cement concretes; extensibility; failure; fibers; heat of hydra-tion; insulation; joints (junctions); machine bases; mass concrete;microcracking; mix proportioning; modulus of elasticity; moisturecontent; Poisson ratio; polymer-portland cement concrete; pozzo-lans; prestressed concrete; reinforced concrete; reinforcing steels;restraints; shrinkage: specifications; specific heat; strain gages;strains; stresses; structural design; temperature; temperature rise(in concrete); tensile stress; tension; thermal expansion; volumechange.

ACI Committee Reports, Guides, Standard Practices , and Com-mentaries are Intended for guidance in designing, planning,executing, or inspecting construction, and in preparing speci-fications Reference to these documents shall not be made inthe Project Documents. If items found in these documents aredesired to be part of the Project Documents, they should bephrased in mandatory language and incorporated into the Proj-ect Documents.

Copyright 0 1990, American Concrete Institute. All rights reserved includingrights of reproduction and use in any form or by any means, including the making ofcopies by any photo process, or by any electronic or mechanical device, printed or

224R-1

ContentsChapter 1 - Introduction, page 224R-2

Chapter 2 - Crack mechanisms in concrete,page 224R-22.1 - Introduction2.2 - Microcracking2.3 - Fracture

Chapter 3 - Control of cracking due to dryingshrinkage, page 224R-93.1 - Introduction3.2 - Crack formation3.3 - Drying shrinkage3.4 - Factors influencing drying shrinkage3.5 - Control of shrinkage cracking3.6 - Shrinkage-compensating concretes

Chapter 4 - Control of cracking in flexuralmembers, page 224R-164.1 - Introduction4.2 - Crack control equations for reinforced concrete beams4.3 - Crack control in two-way slabs and plates4.4 - Tolerable crack widths versus exposure conditions in re-

inforced concrete4.5 - Flexural cracking in prestressed concrete4.6 - Anchorage zone cracking in prestressed concrete4.7 - Tension cracking

Cbapter 5 - Long-term effects on cracking,page 224R-215.1 -5.2 -5.3 -5.4 -5.5 -

IntroductionEffects of long-term loadingEnvironmental effectsAggregate and other effectsUse of polymers in improving cracking characteristics

written or oral, or recording for sound or visual reproduction or for use in any knowledge or retrieval system or device, unless permission in writing is obtainedfrom the copyright proprietors.

ACI COMMITTEE REPORT

Chapter 6 - Control of cracking in concretelayered systems, page 224R-236.1 - Introduction6.2 - Fiber reinforced concrete (FRC) overlays6.3 - Latex modified concrete (LMC) overlays6.4 - Polymer impregnated concrete (PIC) systems

Chapter 7 - Control of cracking in mass con-crete, page 224R-267.1 - Introduction7.2 - Crack resistance7.3 - Determination of temperatures and tensile strains7.4 - Control of cracking7.5 - Testing methods and typical data7.6 - Artificial cooling by embedded pipe systems7.7 - Summary - Basic considerations for construction controls

and specifications

Chapter 8 - Control of cracking by correctconstruction practices, page 224R-368.1 - Introduction8.2 - Restraint8.3 - Shrinkage8.4 - Settlement8.5 - Construction8.6 - Specifications to minimize drying shrinkage8.7 - Conclusion

Chapter 9 - References, page 224R-429.1- Specified and/or recommended references9.2 - Cited references

Chapter 1 - IntroductionCracks in concrete structures can indicate major

structural problems and can mar the appearance ofmonolithic construction. They can expose reinforcingsteel to oxygen and moisture and make the steelmore susceptible to corrosion. While the specificcauses of cracking are manifold, cracks are normallycaused by stresses that develop in concrete due tothe restraint of volumetric change or to loads whichare applied to the structure. Within each of thesecategories there are a number of factors at work. Asuccessful crack control program must recognizethese factors and deal with each of them, in turn.

This report presents the principal causes of crack-ing and a detailed discussion of crack control pro-cedures. The body of the report consists of sevenchapters designed to help the engineer and the con-tractor in the development of effective crack controlmeasures.

This report is an update of a previous committeereport, issued in 1972.1.1 The original report wassupplemented by an ACI Bibliography on cracking,1 . 2

also issued by this committee. In the updating pro-cess, many portions of the report have undergonesizeable revision, and the entire document has beensubjected to a detailed editorial review. Chapter 2,

Chapter 2 - Crack mechanisms in concrete*

on crack mechanisms, has been completely rewrittento take into account the experimental and analyticalwork that has been done since the completion of thefirst committee report. Chapter 6, on crack control

in concrete layered systems, is new to the reportand deals with a form of concrete construction thatwas in its infancy at the time the first report wasdrafted. Individual chapters on crack control in re-

inforced and prestressed concrete members havebeen condensed into a single chapter, Chapter 4, on

crack control in flexural members. The resulting pre-sentation is more concise and, hopefully, more usefulto the structural designer. Chapter 5, on long-term effects, details some interesting findings on thechange of crack width with time. Chapters 3, 7, and 8, which consider drying shrinkage, mass concrete, and construction practices, respectively, have beenexpanded and updated to take into account the mostrecently developed procedures in these areas. In ad-dition, new sections have been added to Chapters 7and 8 which provide specific guidance for the devel-opment of crack control programs and specifications.

The committee hopes that this report will serve asa useful reference to the causes of cracking and as akey tool in the development of practical crack con-trol procedures in both the design and the construc-tion of concrete structures.

References1.1. ACI Committee 224, “Control of Cracking in Con-

crete Structures,” ACI JOURNAL, Proceedings V. 69, NO.12, Dec. 1972, pp. 717-753.

1.2. ACI Committee 224, “Causes, Mechanism, and Con-trol of Cracking in Concrete,” ACI Bibliography No. 9,American Concrete Institute, Detroit, 1971, 92 pp.

2.1 - IntroductionBeginning with the work at Cornell University in

the early 1960s,2 .1 a great deal has been learnedabout the crack mechanisms in concrete, both at themicroscopic and the macroscopic level. Of special in-terest during the early work was the realization thatthe behavior of concrete, under compressive as wellas tensile loads, was closely related to the formationof cracks. Under increasing compressive stress, mi-croscopic cracks (or microcracks) form at the mortar-coarse aggregate boundary and propagate throughthe surrounding mortar, as shown in Fig. 2.1.

During the first decade of research, a picture de-veloped that closely linked formation and propaga-tion of these microcracks to the load-deformation be-havior of concrete. Prior to load, volume changes incement paste cause interfacial cracks to form at themortar-coarse aggregate boundary.2.2,2.3 Under short-term compressive load, no additional cracks form un-til the load reaches approximately 30 percent of thecompressive strength of the concrete.2.1 Above thisvalue, additional bond cracks initiate throughout thematrix. Bond cracking increases until the loadreaches approximately 70 percent of the compressivestrength, at which time microcracks begin to propa-gate through the mortar. Mortar cracking continuesat an accelerated rate until the material ultimatelyfails. For concrete in uniaxial tension, experimentalwork indicates that major microcracking begins atabout 60 percent of the ultimate tensile strength.2.4

‘Principal author: David Darwin.

CONTROL OF CRACKING 224R-3

$EGlrgyjm 0.0012 m 0. CKI

STRAIN STRAIN

Fig. 2.1 - Cracking maps and stress-strain curvesfor concrete loaded in uniaxial compression. *

*From S. P. Shah, and F. O. Slate, “Internal Microcracking,Mortar-Aggregate Bond and the Stress-Strain Curve of Con-crete,” Proceedings, International Conference on the Structure ofConcrete (London, Sept. 1965), Cement and Concrete Association,London, 1968, pp. 82-92.

Studies of the stress-strain behavior and volumechange of concrete 2.5 indicate that the initiation ofmajor mortar cracking corresponds with an observedincrease in the Poisson’s ratio of concrete. The term“discontinuity stress” is used for the stress at whichthis change in material behavior occurs.

In general, it has been agreed that the micro-cracking that occurs prior to loading has very littleeffect on the strength of concrete. However, workby Brooks and Neville 2.6 indicates that the effect ofearly volume change on microcracking of concretemay result in a reduction of both tensile and com-pressive strength as concrete dries out. Their studyshows that upon drying, the strength of test speci-mens first increases and then decreases. They postu-late that the initial increase is due to the increasedstrength of the drier cement paste and that the ulti-mate decrease in strength is due to the formation ofshrinkage induced microcracks.

Work by Meyers, Slate, and Winter 2.7 and Shahand Chandra2.8 demonstrates that microcracks in-crease under the effect of sustained and cyclic load-ing. Their work indicates that the total amount ofmicrocracking is a function of the total compressivestrain in the concrete and is independent of themethod in which the strain is applied. Sturman,Shah, and Winter2.9 found that the total degree ofmicrocracking is decreased and the total strain ca-pacity in compression is increased when concrete issubjected to a strain gradient.

At about the same time that the microcrackingstudies began, investigators began applying fracturemechanics to the studies of concrete under load. Thefield of fracture mechanics, originated by Griffith2.10

in 1920, serves as the primary tool for the study ofbrittle fracture and fatigue in metal structures.Since concrete has for many years been considered abrittle material in tension, fracture mechanics is con-sidered to be a potentially useful analysis tool forconcrete by many investigators. 2. .12

The field of fracture mechanics was first appliedto concrete by Kaplan2.11

in 1961. The classical the-ory serves to predict, the rapid propagation of amacrocrack through a homogeneous, isotropic, elas-tic material. The theory makes use of the stress in-tensity factor, KI , which is a function of crack geom-etry and stress. Failure occurs when KI reaches acritical value, KIc , known as the critical stress-in-tensity factor under conditions of plane strain. KIc isthus a measure of the fracture toughness of the ma-terial. To properly measure KIc for a material, thetest specimen must be of sufficient size to insuremaximum constraint (plane strain) at the tip of thecrack. For linear elastic fracture mechanics (LEFM)to be applicable, the value of KIc must be a materialconstant, independent of the specimen geometry (asare other material constants such as yield strength).

The earliest experimental work utilized notchedtension and beam specimens of mortar and con-

crete.2.11-2.14 The crack resistance was expressed interms of the strain energy release rate at the onsetof rapid crack growth, G, which is directly related tothe fracture toughness of the material. Later in-vestigations evaluated the crack resistance of paste,mortar and concrete in terms of the fracture tough-ness, itself.2.15 Work by Naus and Lott2.16 indicatedthat the fracture toughness of paste and mortar in-creased with decreasing water-cement ratio, but thatthe water-cement ratio had little effect on the frac-ture toughness of concrete. They found that KIc in-creased with age, and decreased with increasing aircontent for paste, mortar, and concrete. The effec-tive fracture toughness of mortar increased with in-creasing sand content, and the fracture toughness ofconcrete increased with an increase in the maximumsize of coarse aggregate.

Additional work by Naus,2.17 presented just priorto the previous committee report,1.1 indicated thatfracture toughness was not independent of speci-men geometry for tensile specimens of paste, mortarand concrete and that fracture toughness was a func-tion of the crack length. These observations lead tothe possibly erroneous conclusion that fracture me-chanics may not be applicable to concrete. Becausecertain size requirements must be met, before frac-ture mechanics is applicable, these results may onlyindicate that the test specimen did not satisfy all ofthe minimum size requirements of linear elastic frac-ture mechanics.

The balance of this chapter describes some of themore recent studies of crack mechanisms in concreteand gives a somewhat different picture from thatpresented in the previous committee report.

-5

224R

l/30woz

z- 10

OUNCOATEDAGG.l COATED AGG.

1600 2000 2400 2800 3200MICROSTRAIN

Fig. 2.4 - Stress-strain curves as influenced by coating aggregates (Reference2.36).

seemed to indicate a very large effect, thus empha- sizing the importance of interfacial strength on the behavior of concrete. These studies utilized rela- tively thick, soft coatings on the coarse aggregate to reduce the bond strength. Since these soft coatings isolated the aggregate from the surrounding mortar, the effect was more like inducing a large number of voids in the concrete matrix.

Two other s tudies 2.36,2.37 which did not isolate thecoarse aggregate from the mortar indicate that theinterfacial strength plays only a minor role in con-troll ing the stress-strain behavior and ult imatestrength of concrete. Darwin and Slate2.36 used athin coating of polystyrene on natural coarse aggre-gate. They found that a large reduction in interfacialbond strength causes no change in the initial stiff-ness of concrete under short-term compressive loadsand results in approximately a 10 percent reductionin the compressive strength as compared to similarconcrete made with aggregate with normal inter-facial strength (see Fig. 2.4). They also found thatthe lower interfacial strength had no appreciable ef-fect on the total amount of microcracking. However,in every case, the average amount of mortar crack-ing was slightly greater for the specimens madewith coated aggregate. This small yet consistent dif-ference may explain the differences in the stress-strain curves.

Perry and Gillott 2.37used glass spheres with dif-ferent degrees of surface roughness as coarse aggre-gate. Their results indicate that reducing the inter-facial strength of the aggregate decreases theinitiation stress by about 20 percent, but has verylittle effect on the discontinuity stress. They also ob-served a 10 percent reduction in the compressivestrength for specimens with low mortar-aggregatebond strength.

Work by Carino,2.38 using polymer impregnatedconcrete, seems to corroborate these two studies.Carino found that polymer impregnation did not in-crease the interfacial bond strength, but did increasethe compressive strength of concrete. He attributedthe increase in strength to the effect of the polymeron the strength of mortar, thus downgrading the im-portance of the interfacial bond.

The importance of mortar, and ultimately cementpaste, in controlling the stress-strain behavior ofconcrete is illustrated by the finite element work ofBuyukozturk2.37 and Maher and Darwin. 2.31,2.32 Usinga linear finite element representation of a physicalmodel of concrete, Buyukozturk was able to simulatethe overall crack patterns under uniaxial loading.

Mortar

Fig. 2.5 - Stress-strain curves for concrete model. **From A. Maher. and D. Darwin, “Microscopic Finite Element

Model of Concrete,” presented at the First International Confer-ence on Mathematical Modeling (St. Louis. Aug.-Sept. 1977) .

224R-6 ACI COMMl=lTEE REPORT

w-

\. _

-- Normal fnterfaclal Str.- ._ I nfuxte I nterfaciaf Str

._ Zwo Tenstle and CoheweI nterfaclal Str.

- - - Z e r o InterfacIal Str.

S t r e s s . PSI4 lMPar

L_._ 1

*l. 0 *2. 0 0. 0 1.0

Strdln. 0.001 In/in

,’

A g g r e g a t eMortar

.?. 0

Fig. 2.6 - Stress-strain curve for finite elementmodel of concrete with varying values of mortar-ag-gregate bond strength (Reference 2.32).

However, his finite element model could not dupli-cate the nonlinear experimental behavior of thephysical model using the formation of interfacialbond cracks and mortar cracks as the only nonlineareffect. Maher and Darwin 2 .31,2.32 have shown that byusing a nonlinear representation for the mortar con-stituent of the physical model, a very close represen-tation of the actual behavior can be obtained. Theresults for Buyukozturk’s model are shown in Fig.25. .

The inability of linear elastic models2.25,2.26,2.39 toduplicate the nonlinear behavior of concrete utilizingmicrocracking alone has been explained as being dueto the fact that concrete is really a “statistical mate-rial.” When the proper statistical variation is se-lected, the nonlinear behavior of concrete can be

v MORTAR@21 v0 CONCRETE 0

ti* OS I I I I I *

lJ4 l/2 314 1(6.4) (12.7)(19. 1)(2X 4)

NOTCH DEPTH, INCHES (mm)

Fig. 2.7 - Effect of notch depth on flexure strength(Reference 2.42).

duplicated2.25 While the statistical variations un-doubtedly play a part, the major nonlinear behaviorcan also be matched by considering the non-linearities of the mortar constituent.2.31,2.32 Fig. 2.6 il-lustrates the results obtained for a highly simplifiedmodel of concrete under uniaxial compression usinga nonlinear representation for mortar. The stress-strain curve for the model without cracking differsvery little from that of models that have a normal,or above normal, amount of microcracking. Micro-cracks have a relatively minor effect on the primarystress-strain behavior of the models. The dominanteffect of microcracking is to increase the lateralstrain. In every case the failure of the model is gov-erned by “crushing” of the mortar which occurs atan average strength below that of the mortar alone.

Newman2.5s and Tasuji, Slate, and Nilson2.40 lhave

observed that the principal tensile strain in concreteat the “discontinuity stress” appears to be a functionof the mean normal stress, 0, = (0,+0,+0,)/3. Intheir study of the biaxial strength of concrete, Ta-suji, et al., observe that the final failure of theirspecimens consists of the formation of macroscopictensile cracks. They also observe that the stress atdiscontinuity occurs at approximately 75 percent ofthe ultimate strength in compression and at about 60percent of the ultimate strength for those cases in-volving tension, matching the levels at which mortarcracking begins.2.3,2.4 l Their work seems to point verystrongly toward a “limiting tensile strain” as thegoverning factor in the strength of concrete.

Overall, the damage to cement paste seems toplay an important role in controlling the primarystress-strain behavior of concrete under short-termaxial load. In normal weight concrete, aggregateparticles act as stress-raisers, increasing the initialstiffness and decreasing the strength of the paste.For cyclic and sustained loading, a great deal of thebond cracking results from load induced volumechanges within the paste, but has no significant ef-fect on strength. A number of investigators feel thatthe onset of mortar cracking marks the “true” ulti-mate strength of concrete.2.6-2.8,2.33,2.34,2.41 l Whethermortar cracking itself controls the strength of con-crete or whether it only signals intimate damage ofthe cement paste remains to be seen. Additionalstudies in this area are clearly warranted.

2.3 - FractureSince the publication of the previous report, a

number of investigations have shed additional lighton the applicability of fracture mechanics to con-crete and its constituent materials.

Shah and McGarry utilized flexure specimens sub-jected to three-point loading.2.42 Their work indicatesthat while paste is notch sensitive, neither mortarnor concrete are affected by a notch (Fig. 2.7). Shahand McGarry also ran a series of tests using notchedtensile specimens and determined that paste speci-

CONTROL OF CRACKING 224R-7

mens, and mortar specimens made with fine aggre-gate that passed the #30 sieve, are notch sensitive,but that mortar specimens containing larger sizes ofaggregate are not notch sensitive.

Brown utilized notched flexure specimens anddouble cantilever beam specimens of paste and mor-tar2.18 8 His tests show that the fracture toughness ofcement paste is independent of crack length and istherefore a material constant. The fracture tough-ness of mortar, however, increases as the crackpropagates, indicating that the addition of fine ag-gregate improves the toughness of paste. This be-havior is similar to the behavior found in structuralsteels that exhibit a plane strain-plane stress transi-tion. Because the plane strain-plane stress transitionoccurs beyond the limits of LEFM, the analysis ismore complex. To re-establish the applicability ofLEFM, larger test specimens must be used withtougher materials such as mortar.

Mindess and Nadeau investigated the effect ofnotch width on KI for both mortar and concrete.2.20

Utilizing notched beam specimens of constant lengthand depth, with varying widths, they found thatwithin the range studied, there was no dependenceof fracture toughness upon the length of crack front.Since their work utilized small specimens with adepth of only about 50 mm (2 in.), there is some in-dication that rather than measuring the fracturetoughness of the material, they were simply measur-ing the modulus of rupture.

The applicability of these results, and much of theother fracture mechanics work, has been broughtinto perspective based on the experimental work byWalsh. In separate investigations of notched beamspecimens2.21 ’ and beams with right angle re-entrantnotches2.22 Walsh has demonstrated that specimensize has a marked influence on the applicability oflinear elastic fracture mechanics to the failure ofplain concrete specimens. As illustrated in Fig. 2.8,

0.10 L I I

1 4a/a0 (log scale)

Fig. 2.8 - Relationship bet ween test results andtheory for notched concrete beams (Reference 2.22).

for specimens of similar geometry but below a cer-tain critical size, the specimen capacity is governedby the modulus of rupture of concrete, calculatedfrom the linear stress distribution. For specimensabove this size, the strength is governed by the frac-ture toughness, which he approximated as a functionof the square root of the compressive strength of theconcrete. Walsh concluded that, for valid toughnesstesting of concrete, the depth of notched beamsmust be at least 230 mm (9 in.). This type of behav-ior is also observed in metals, i.e., for valid fracturemechanics test results, the test specimens mustmeet minimum size requirements (ASTM E 399).These size requirements are dependent upon thesquare of the toughness levels being measured. Thusa material whose toughness level is twice that ofanother material (all other properties being equal),must have specimen dimensions four times that ofthe first material for the test results to be equallyvalid.

Gjorv, Sorensen and, Arnesen2.23 investigated the

L-__ -- ~_. - -~ 1

Fig. 2.9 - Effect of notch depth on flexural strength(Reference 2.23).

notch sensitivity of paste, mortar and concrete usingthree-point bend specimens similar to those used byShah and McGarry2.42 As shown in Fig. 2.9, they de-termined that both mortar and concrete are notchsensitive, but less sensitive than cement paste. Theyconclude that the disagreement with the earlier re-sults is due in part to their improvement in the load-ing procedure. They feel that linear elastic fracturemechanics is applicable to the small specimens of

ACI COMMITTEE REPORT

paste, but not to the small size specimens of mortarand concrete. Even the small specimens of mortarand concrete, however, have some degree of notchsensitivity since the failure is not consistent with themodulus of rupture based on the net cross section.Citing Walsh’s earl ier work,2.21 they agree thatLEFM is applicable to large concrete specimens, butthat it is not applicable to small specimens.

Hillemeier and Hilsdorf2.43 utilized wedge loaded,compact tension specimens to measure the fracturetoughness of paste, aggregate and the paste-aggre-gate interface. They feel that, while the failure ofconcrete in tension and compression is controlled bymany interacting cracks rather than by the propaga-tion of a single crack, fracture mechanics does offeran important tool for evaluating the constituent ma-terials of concrete. They found that paste is a notchsensitive material and that the addition of entrainedair or soft particles has only a small affect on K I c .Their work indicates that the KIc values for inter-facial strength between paste and aggregate is onlyabout one-third of the KIc value for paste alone, andthat the characteristic value of KIC for aggregate isapproximately ten times that of paste.

Swartz, Hu, and Jones2.24 used compliance mea-surement to monitor crack growth in notched con-crete beams subjected to sinusodial loading. Theyconclude that this procedure is useful for monitoringcrack growth in concrete due to fatigue. Based onthe appearance of the fracture surface, which showsa combination of both aggregate fracture and bondfailure, they feel that fracture toughness is not apertinent material property. However, they statethat an “effective” fracture toughness might be asignificant material property if related to specificmaterial and specimen variables such as aggregatesize and gradation, and proportions of the mix, andif the calculation considers the nonlinear material re-sponse of concrete.

A number of investigators do not feel that theGriffith theory of linear fracture mechanics is di-rectly applicable to all concrete2.23,

2.24* 2.42 (ASTM E399). Some like Swartz, et a1.2.24 feel that the theoryhas application when the limitations and specificnonhomogenous effects are taken into account.Clearly, specimen size requirements must be givenmore attention. Of key interest in future work arethe observations by Walsh2.21’ 2.22 that show that ifthe specimens are large enough, the effects ofheterogeneity are greatly reduced and that concretemay approximate a homogenous material to whichthe principles of fracture mechanics can be applied.

References2.1. Hsu, Thomas T. C.; Slate, Floyd O.; Sturman, Ger-

ald M.; and Winter, George, “Microcracking of Plain Con-crete and the Shape of the Stress-Strain Curve,” ACIJOURNAL Proceedings V. 60, No. 2, Feb. 1963, pp. 209-224.

2.2. Hsu, Thomas, T. C., “Mathematical Analysis ofShrinkage Stresses in a Model of Hardened Concrete,”ACI JOURNAL, Proceedings V. 60, No. 3, Mar. 1963, pp.371-390.

2.3. Slate, Floyd O., and Matheus, Ramon E., “VolumeChanges on Setting and Curing of Cement Paste and Con-crete from Zero to Seven Days,” ACI JO U R N A L, Pro-ceedings V. 64, No. 1, Jan. 1967, pp. 34-39.

2.4. Evans, R. H., and Marathe, M. S., “Microcrackingand Stress-Strain Curves for Concrete in Tension,” Mate-rials and Structures, Research and Testing (Paris), V. 1,No. 1, Jan. 1968, pp. 61-64.

2.5. Newman, Kenneth, “Criteria for the Behavior ofPlain Concrete Under Complex States of Stress,” Pro-ceedings, International Conference on the Structure ofConcrete (London, Sept. 1965), Cement and Concrete Asso-ciation, London, 1968, pp. 255-274.

2.6. Brooks, J. J., and Neville, A. M., “A Comparison ofCreep, Elasticity and Strength of Concrete in Tension andin Compression,” Magazine of Concrete Research (London),V. 29, No. 100, Sept. 1977, pp. 131-141.

2.7. Meyers, Bernard L.; Slate, Floyd O.; and Winter,George, “Relationship Between Time-Dependent Deforma-tion and Microcracking of Plain Concrete,” ACI JOURNAL,Proceedings V. 66, No. 1, Jan. 1969, pp. 60-68.

2.8. Shah, Surendra P., and Chandra, Sushil, “Fractureof Concrete Subjected to Cyclic and Sustained Loading,”ACI JOURNAL, Proceedings V. 67, No. 10, Oct. 1970, pp.816-824.

2.9. Sturman, Gerald M.; Shah, Surendra P.; and Winter,George, “Effects of Flexural Strain Gradients on Micro-cracking and Stress-Strain Behavior of Concrete,” ACIJOURNAL, Proceedings V. 62, No. 7, July 1965, pp. 805-822.

2.10. Griffith, A. A., “The Phenomena of Rupture andFlow in Solids,” Transactions, Royal Society of London,No. 221A, 1920, pp. 163-198.

2.11. Kaplan, M. F., “Crack Propagation and the Frac-ture of Concrete,” ACI JOURNAL, Proceedings V. 58, No. 5,Nov. 1961, pp. 591-610.

2.12. Glucklich, Joseph, “Static and Fatigue Fractures ofPortland Cement Mortars in Flexure,” Proceedings, FirstInternational Conference on Fracture, Sendai, Japan, V. 2,1965, pp. 1343-1382.

2.13. Romualdi, James P., and Batson, Gordon B., “Me-chanics of Crack Arrest in Concrete,” Proceedings, ASCE,V. 89, EM3, June 1963, pp. 147-168.

2.14. Huang, T. S., “Crack Propagation Studies in Micro-concrete,” MSc Thesis, Department of Civil Engineering,University of Colorado, Boulder, 1966.

2.15. Lott, James L., and Kesler, Clyde E., “Crack Prop-agation in Plain Concrete,” Symposium on Structure ofPortland Cement Paste and Concrete, Special Report No.90, Highway Research Board, Washington, D.C., 1966, pp.204-218.

2.16. Naus, Dan J., and Lott, James L., “FractureToughness of Portland Cement Concretes,” ACI JOURNAL,Proceedings V. 66, No. 6, June 1969, pp. 481-489.

2.17. Naus, Dan J., “Applicability of Linear-Elastic Frac-ture Mechanics to Portland Cement Concretes,” PhDThesis, University of Illinois, Urbana, Aug. 1971.

2.18. Brown, J. H., “Measuring the Fracture Toughnessof Cement Paste and Mortar,” Magazine of Concrete Re-search (London), V. 24, No. 81, Dec. 1972, pp.185-196.

CONTROL OF CRACKING 224R-9

Chapter 3 - Control of cracking due to dryingshrinkage*

2.36. Darwin, David, and’ Slate, F. O., “Effect of Paste-Aggregate Bond Strength on Behavior Concrete,” Jour-nal of Materials, V. 5, No. 1, Mar. 1970, pp. 86-98.

2.42. Shah, Surendra P., and McGarry, Fred J., “GriffithFracture Criterion and Concrete,” Proceedings, ASCE, V.97, EM6, Dec. 1971, pp. 1663-1676.

2.19. Evans, A. G.; Clifton, J. R.; and Anderson, E.,“The Fracture Mechanics of Mortars,” Cement and Con-crete Research, V. 6, No. 4. July 1976, pp. 535-547.

2.20. Mindess, Sidney, and Nadeau, John S., “Effect ofNotch Width of KIC for Mortar and Concrete,” Cementand Concrete Research, V. 6, No. 4, July 1976, pp. 529-534.

2.21. Walsh, P. F., “Fracture of Plain Concrete,” IndianConcrete Journal (Bombay), V. 46, No. 11, Nov. 1972, pp.469-470, 476.

2.22. Walsh, P. F., “Crack Initiation in Plain Concrete,”Magazine of Concrete Research (London), V. 28, No. 94,Mar. 1976, pp. 37-41.

2.23. Gjorv, O. E.; Sorensen, S. I.; and Arnesen, A.,“Notch Sensitivity and Fracture Toughness of Concrete,”Cement and Concrete Research, V. 7, No. 3, May 1977, pp.333-344.

2.24. Swartz, Stuart E.; Hu, Kuo-Kuang; and Jones,Gary L., “Compliance Monitoring of Crack Growth in Con-crete,” Proceedings, ASCE, V. 104, EM4, Aug. 1978, pp.789-800.

2.25. Shah, Surendra P., and Winter, George, “InelasticBehavior and Fracture of Concrete,” ACI JOURNAL, P ro -ceedings V. 63, No. 9, Sept. 1966, pp. 925-930.

2.26. Testa, Rene B., and Stubbs, Norris, “Bond Failureand Inelastic Response of Concrete,” Proceedings, ASCE,V. 103, EM2, Apr. 1977, pp. 296-310.

2.27. Darwin, David, Discussion of “Bond Failure and In-elastic Response of Concrete,” by Rene B. Testa and Nor-ris Stubbs, Proceedings, ASCE, V. 104, EM2, Apr. 1978,pp. 507-509.

2.28. Spooner, D. C., “The Stress-Strain Relationship forHardened Cement Pastes in Compression,” Magazine ofConcrete Research (London), V. 24, No. 79, June 1972, pp.85-92.

2.29. Spooner, D. C., and Dougill, J. W., “A QuantitativeAssessment of Damage Sustained in Concrete DuringCompressive Loading,” Magazine of Concrete Research(London), V. 27, No. 92, Sept. 1975, pp. 151-160.

2.30. Spooner, D. C.; Pomeroy, C. D.; and Dougill, J. W.,“Damage and Energy Dissipation in Cement Pastes inCompression,” Magazine of Concrete Research (London),V. 28, No. 94, Mar. 1976, pp. 21-29.

2.31. Maher, Ataullah, and Darwin, David, “A FiniteElement Model to Study the Microscopic Behavior of PlainConcrete,” CRINC Report-SL-76-02, The University ofKansas Center for Research, Lawrence, Nov. 1976, 83 pp.

2.32. Maher, Ataullah, and Darwin, David, “MicroscopicFinite Element Model of Concrete,” Proceedings, First In-ternational Conference on Mathematical Modeling (St.Louis, Aug.-Sept. 1977), University of Missouri-Rolla, 1977,v. III, pp. 1705-1714.

2.33. Karsan, I. Demir, and Jirsa, James 0.. “Behaviorof Concrete under Compressive Loadings,” Proceedings,ASCE, V. 95, ST12, Dec. 1969, pp. 2543-2563.

2.34. Neville, A. M., and Hirst, G. A., “Mechanism ofCyclic Creep of Concrete,” Douglas McHenry Symposiumon Concrete and Concrete Structures, SP-55, AmericanConcrete Institute, Detroit, 1978, pp. 83-101.

2.35. Nepper-Christensen, Palle, and Nielsen, TommyP. H., “Modal Determination of the Effect of Bond BetweenCoarse Aggregate and Mortar on the CompressiveStrength of Concrete,” ACI JO U R N A L, Proceedings V. 66,No. 1, Jan. 1969, pp. 69-72.

2.37. Perry, C., and Gillott, J. E., “The Influence of Mor-tar-Aggregate Bond Strength on the Behavior of Concretein Uniaxial Compression,” Cement and Concrete Research,V. 7, No. 5, Sept. 1977, pp. 553-564.

2.38. Carino, Nicholas J., “Effects of Polymer Impregna-tion on Mortar-Aggregate Bond Strength,” Cement andConcrete Research, V. 7, No. 4, July 1977, pp. 439-447.

2.39. Buyukozturk, Oral, “Stress-Strain Response andFracture of a Model of Concrete in Biaxial Loading,” PhDThesis, Cornell University, Ithaca, June 1970.

2.40. Tasuju, M. Ebrahim; Slate, Floyd 0.; and Nilson,Arthur H., “Stress-Strain Response and Fracture of Con-crete in Biaxial Loading,” ACI JO U R N A L, Proceedings V .75, No. 7, July 1978, pp. 306-312.

2.41. Shah, Surendra P., and Chandra, Sushil, “CriticalStress, Volume Change, and Microcracking of Concrete,”ACI JO U R N A L, Proceedings V. 65, No. 9, Sept. 1968, pp.770-781.

2.43. Hillemeier, B., and Hilsdorf, H. K., “Fracture Me-chanics Studies of Concrete Compounds,” Cement and Con-crete Research, V. 7, No. 5, Sept. 1977, pp. 523-535.

3.1 - IntroductionCracking of concrete due to drying shrinkage is a

subject which has received more attention by archi-tects, engineers, and contractors than any othercharacteristic or property of concrete. It is one ofthe most serious problems encountered in concreteconstruction. Good design and construction practicecan minimize the amount of cracking and eliminatethe visible large cracks by the use of adequate re-inforcement and contraction joints.

Although drying shrinkage is one of the principalcauses of cracking, temperature stresses, chemicalreactions, frost action, as well as excessive tensilestresses due to loads on the structure, are fre-quently responsible for cracking of hardened con-crete. Cracking may also develop in the concreteprior to hardening due to plastic shrinkage.

Information presented in this chapter concernsonly the subjects of cracking of hardened concretedue to drying shrinkage; factors influencing shrink-age; control of cracking; and the use of expansive ce-ments to minimize cracking.

The subject of construction practices and specifica-tions to minimize drying shrinkage is covered inChapter 8 (Sections 8.3 and 8.6) of this report.

*Principal author: Miloss Polivka.

224R-10 ACI COMMITTEE REPORT

3.2 - Crack formationWhy does concrete crack due to shrinkage? If the

shrinkage of concrete caused by drying could takeplace without any restraint, the concrete would notcrack. However, in a structure the concrete is al-ways subject to some degree of restraint by eitherthe foundation or another part of the structure or bythe reinforcing steel embedded in the concrete. Thiscombination of shrinkage and restraint develops ten-sile stresses. When this tensile stress reaches thetensile strength, the concrete will crack. This is illus-trated in Fig. 3.1.

ORIGINAL LENGTH

I I

UNRESTRAINEDSHRINKAGE t-

RESTRAINED SHRINKAGEDEVELOPS TENSILE STRESS

IF TENSILE STRESS ISGREATER THAN TENSILESTRENGTH, CONCRETE CRACKS

Fig. 3.1 - Cracking of concrete due to dryingshrinkage.

Another type of restraint is developed by the dif-ference in shrinkage at the surface and in the inte-rior of a concrete member, especially at early ages.Since the drying shrinkage is always larger at theexposed surface, the interior portion of the memberrestrains the shrinkage of the surface concrete, thusdeveloping tensile stresses. This may cause surfacecracking, which are cracks that do not penetratedeep into the concrete. These surface cracks maywith time penetrate deeper into the concrete mem-ber as the interior portion of the concrete is subjectto additional drying.

The magnitude of tensile stress developed duringdrying of the concrete is influenced by a combinationof factors, such as (a) the amount of shrinkage, (b)the degree of restraint, (c) the modulus of elasticityof the concrete, and (d) the creep or relaxation of theconcrete. Thus, the amount of shrinkage is only onefactor governing the cracking. As far as cracking isconcerned, a low modulus of elasticity and highcreep characteristics of the concrete are desirablesince they reduce the magnitude of tensile stresses.Thus, to minimize cracking, the concrete should havelow drying shrinkage characteristics and a high de-gree of extensibility (low modulus and high creep) aswell as a high tensile strength. However, a large ex-tensibility of a concrete member subjected to bend-ing will cause larger deflections.

3.3 - Drying shrinkageWhen concrete dries, it contracts or shrinks, and

when it is wetted again, it expands. These volumechanges, with changes in moisture content, are aninherent characteristic of hydraulic cement con-cretes. It is the change in moisture content of the ce-ment paste that causes the shrinkage or swelling ofconcrete, while the aggregate provides an internalrestraint which significantly reduces the magnitudeof these volume changes.

When cement is mixed with water, several chem-ical reactions take place. These reactions, commonlycalled “hydration,” produce a hydration product con-sisting essentially of some crystalline materials (prin-cipally calcium hydroxide) and a large amount ofhardened calcium silicate gel called “tobermoritegel.” This rigid gel consists of colloidal size particlesand has an extremely high surface area. In a hard-ened cement paste, some of the water is in the capil-lary pores of the paste, but a significant amount is inthe tobermorite gel. Shrinkage is due to the loss ofadsorbed water from the gel. On drying the first wa-ter lost is that which occupies the relatively largesize capillaries in the cement paste. This loss of wa-ter causes very little, if any, shrinkage. It is the lossof the adsorbed and inter-layer water from the hy-drated gel that causes the shrinkage of the paste.When a concrete is exposed to drying conditions,moisture slowly diffuses from the interior mass ofthe concrete to the surface where it is lost by evapo-ration. On wetting this process is reversed, causingan expansion of the concrete.

In addition to drying shrinkage, the cement pasteis also subject to carbonation shrinkage. The actionof carbon dioxide, CO2, present in the atmosphere onthe hydration products of the cement, principally cal-cium hydroxide, Ca(OH)2, results in the formation ofcalcium carbonate, CaCO,, which is accompanied bya decrease in volume. Since carbon dioxide does notpenetrate deep into the mass of concrete, shrinkagedue to carbonation is of minor importance in theoverall shrinkage of a concrete structure. However,

CONTROL OF CRACKING 224R-11

carbonation does play an important role in theshrinkage of small laboratory test specimens, partic-ularly when subjected to long-term exposure todrying. Thus, the amount of shrinkage observed on asmall laboratory specimen will be greater than theshrinkage of the concrete in the structure. The sub-ject of shrinkage due to carbonation is discussed indetail by Verbeck.3.1

3.4 - Factors influencing drying shrinkageThe major factors influencing shrinkage include

the composition of cement, type of aggregate, watercontent, and mix proportions. The rate of moistureloss or the shrinkage of a given concrete is greatlyinfluenced by the size and shape of the concretemember, the environment, and the time of dryingexposure. These and other factors influencing magni-tude and rate of shrinkage are herein discussed.

3.4.1 Effect of cement - Results of an extensivestudy made by Blaine, Arni, and Evans,3.2 of the Na-tional Bureau of Standards on a large number ofportland cements indicate that it is not possible tosay that a cement, because it conforms to the re-quirements of one of the standard types of cements,will have greater or less shrinkage than a cementmeeting requirements for some other type of ce-ment. Their results on neat cement pastes showed awide distribution of shrinkage values especially forthe Type I cements. The 6 month drying shrinkagestrain of the neat pastes ranged from about 0.0015to more than 0.0060 with an average for the 182 ce-ments tested of about 0.0030. They found that lowershrinkage of pastes was associated with: 1. lowerC

3A/SO

3 ratios, 2. lower Na

2O and K

2O contents,

and 3. higher C4AF contents of the cement. Tests by

Brunauer. Skalny, and Yudenfreund3.3 show that forshort curing periods Type II cement pastes exhib-ited considerably less shrinkage than Type I pastes.However, the shrinkage of pastes cured for 28 dayswas about the same for the two types of cements.

Tests made by the California Division of High-ways 3.4 on mortar or paste as a measure of behaviorin concrete indicate that Type II cements generallyproduce lower shrinkage than Type I cements, andmuch lower than Type III cements. Tests by Lerch1.5

show that the proportion of gypsum in the cementhas a major effect on shrinkage. Cement producersmoderate the differences in shrinkage due to cementcomposition by optimizing its gypsum content.

The fineness of a cement can have some influenceon drying shrinkage. Tests by Carlson3.6 showed thatfiner cements generally result in greater concreteshrinkage, but the increase in shrinkage with in-creasing fineness is not large. His results show thatthe composition of the cement is a factor and thusfor some cements an increase in fineness may showlittle change and in some cases even a lower con-crete shrinkage.

TABLE 3.1 - Effect of type of aggregate onshrinkage of concrete3.6

Specif icl-year

Absorption, shrinkage,Aggregate gravity percent percenttSandstone 2.47 5.0 0.116Slate 2.75 1.3 0.068Granite 2.67 0.8 0.047Limestone 2.74 0.2 0.041Quartz 2.66 0.3 0.032

3.4.2 Influence of type of aggregate - Coarse andfine aggregates, which occupy between 65 and 75percent of the total concrete volume, have a majorinfluence on shrinkage. Concrete may be consideredto consist of a framework of cement paste whoselarge potential shrinkage is being restrained by theaggregate. The drying shrinkage of a concrete willbe only a fraction (about l/4 to l/6) of that of the ce-ment paste. The factors which influence the abilityof the aggregate particles to restrain shrinkage in-clude (a) the compressibility of aggregate and the ex-tensibility of paste, (b) the bond between paste andaggregate, (c) the degree of cracking of cementpaste, and (d) the contraction of the aggregate par-ticles due to drying. Of these several factors, com-pressibility of the aggregate has the greatest in-fluence on the magnitude of drying shrinkage ofconcrete.

The higher the stiffness or modulus of elasticity ofan aggregate, the more effective it is in reducing theshrinkage of concrete. The absorption of an aggre-gate, which is a measure of porosity, influences itsmodulus or compressibility. A low modulus is usuallyassociated with high absorption.

The large influence of type of aggregate on dryingshrinkage of concrete was shown by Carlson.3.6 As anexample some of his shrinkage data for concreteswith identical cements and identical water-cementratios are given in Table 3.1.

Quartz, limestone, dolomite, granite, feldspar, andsome basalts can be generally classified as low-shrinkage producing types of aggregates. High-shrinkage concretes often contain sandstone, slate,hornblende and some types of basalts. Since the ri-gidity of certain aggregates, such as granite, lime-stone or dolomite, can vary over a wide range, theireffectiveness in restraining drying shrinkage willvary accordingly.

Although the compressibility is the most impor-tant single property of aggregate governing concreteshrinkage, the aggregate itself may contract an ap-preciable amount upon drying. This is true for sand-stone and other aggregates of high absorption capac-ity. Thus, in general, aggregate of high modulus ofelasticity and low absorption will produce a low-shrinkage concrete. However, some structural gradelightweight aggregates, such as expanded shales,

224R-12 ACI COMMITTEE REPORT

+ 119 142 166 1905 0.060u

% 0.050

I

," 0.020zz 0.010is 200 240 280 320

WATER CONTENT OF CONCRETE

kg/m3

Ib/yd3

Fig. 3.2 - Typical effect of water content of con-crete on drying shrinkage (Reference 3.8).

clays, and slates which have high absorptions, pro-duced concretes exhibiting low shrinkage character-istics.3.7

Maximum size of aggregate has a significant effecton drying shrinkage. Not only does a large aggre-gate size permit a lower water content of the con-crete, but it is more effective in resisting the shrink-age of the cement paste. Aggregate gradation alsohas some effect on shrinkage. The use of a poorlygraded fine or coarse aggregate may result in anoversanded mix, in order to obtain desired work-ability, and thus prevent the use of the maximumamount of coarse aggregate resulting in increasedshrinkage.3.4.3 Effect of water content and mix proportions -The water content of a concrete mix is another veryimportant factor influencing drying shrinkage. Thelarge increase in shrinkage with increase in watercontent was demonstrated in tests made by the U.S.Bureau of Reclamation.3.8 A typical relationship be-tween water content and’drying shrinkage is shownin Fig. 3.2. An increase in water content also re-duces the volume of restraining aggregate and thusresults in higher shrinkage. The shrinkage of a con-

400’(237)

3 5 0(208)

3 0 0(178)

2 5 0(148)

200

2.5 19.0 37.5 75 150 m m

(119) 3/8 3/4 1 1/2 3 6 in.

MAXIMUM SIZE OF AGGREGATE

Fig. 3.3 - Effect of aggregate size on water require-ment of non-air-entrained concrete (ACI 211.1).

crete can be minimized by keeping the water con-tent of the paste as low as possible and the total ag-gregate content of the concrete as high as possible.This will result in a lower water content per unitvolume of concrete and thus lower shrinkage.

The total volume of coarse aggregate is a signifi-cant factor in drying shrinkage. Concrete propor-tioned for pump placement with excessively highsand contents will exhibit significantly greatershrinkage than will similar mixes with normal sandcontents.

Tests reported by Tremper and Spellman3.4 showthat the cement factor has little effect on shrinkageof concrete. Their data show that as the cement fac-tor was increased from 470 to 752 lb/yd3 (279 to 446kg/m3) the water content remained nearly constant,while percentage of fine aggregate was reduced.

The amount of mixing water required for concreteof a given slump is greatly dependent on the max-imum size of aggregate. The surface area of aggre-gate, which must be coated by cement paste, de-creases with increase in size of aggregate. The largeeffect that the maximum size of aggregate has onthe water requirement of concrete is shown in Fig.3.3. The data plotted in this figure, taken from ACI211.1 shows, for example, that for a 3 to 4 in. (75 to100 mm) slump concrete, increasing the aggregatesize from 3/4 in. (19 mm) to 11/2 in. (38 mm) decreasesthe water requirement from 340 to 300 lb/yd3 (202 to178 kg/m3). This 40 lb (24 kg) reduction in watercontent would reduce the 1 year drying shrinkage byabout 15 percent.

Also shown in Fig. 3.3 is the effect of slump onwater requirement. For example, the water require-ment of a concrete made with 3/4 in. (19 mm) size ag-gregate is 340 lb/yd3 (202 kg/m3) for a 3 to 4 in.slump, but only 310 lb/yd3 (184 kg/m31 for a 1 to 2in. slump (25 to 50 mm). This substantial reductionin water content would result in a lower dryingshrinkage.

Another important factor which influences the wa-ter requirement of a concrete, and thus its shrink-age, is the temperature of the fresh concrete. Thiseffect of temperature on water requirement as givenby the U.S. Bureau of Reclamation3. is shown inFig. 3.4. For example, if the temperature of fresh

concrete were reduced from 100 to 50 F (38 to 10 C),it would permit a reduction of the water content by33 Ib/yd3 (20 kg/m3) and still maintain the sameslump. This substantial reduction in water contentwould significantly reduce the drying shrinkage.

From the above discussion it must be concludedthat, to minimize the drying shrinkage of concrete,the water content of a mix should be kept to a min-imum. Any practice that increases the water re-quirement, such as the use of high slumps, high tem-peratures of the fresh concrete or the use of smallersize coarse aggregate, will substantially increaseshrinkage and thus cracking of the concrete.

CONTROL OF CRACilNG 224R-13

00

4.4 10.0 15.6 21.1 26.7 32.2 378 OC310084)

300(I 78)

290(I 72)

280(166)

270(160)

260(154140 50 60 70 80 90 100 OF

TEMPERATURE OF FRESH CONCRETE

Fig. 3.4 - Effect of temperature of fresh concreteon its water requirement (Reference 3.8).

3.4.4 Effect of chemical admixtures - Chemical ad-mixtures are used to impart certain desirable prop-erties to the concrete. Those most commonly usedinclude air-entraining admixtures, water-reducingadmixtures, set-retarding admixtures, and accelera-tors.

It would be expected that when using an air-en-training admixture, the increase in the amount of airvoids would increase drying shrinkage. However, be-cause entrainment of air permits a reduction in wa-ter content with no reduction in slump, the shrink-age is not appreciably affected by air contents up toabout 5 percent.3.8 Some air-entraining agents arestrong retarders and contain accelerators which mayincrease drying shrinkage by 5 to 10 percent.

Although the use of water-reducing and set-re-tarding admixtures will permit a reduction in thewater content of a concrete mix, it will usually notresult in a decrease in drying shrinkage. Actuallysome of these admixtures may even increase theshrinkage at early ages of drying, although the laterage shrinkage of these concretes will be about thesame as that of corresponding mixes with no admix-tures.

The use of calcium chloride, a common accelerator,will result in a substantial increase in drying shrink-age, especially at the early ages of drying. Testsmade by the California Department of Transporta-tion3.44showed that the 7 day shrinkage of a concretecontaining 1.0 percent of calcium chloride was aboutdouble that obtained for the control mix without ad-mixture. However, after 28 days of drying, theshrinkage of the concrete containing calcium chloridewas only about 40 percent greater than that of thecontrol mix.3.4.5 Effect of pozzolans - Fly ash and a number ofnatural materials such as opaline cherts and shales,diatomaceous earth, tuffs and pumicites are pozzo-lans used in portland cement concrete. The use ofsome natural pozzolans can increase the water de-

Fig. 3.5 - Rates of drying of concrete exposed to 50percent relative humidity (Reference 3.9).

wiiiaa

-0 4 8 I2 16 20 24 28 in.w DEPTH BELOW CONCRETE SURFACE

mand as well as the drying shrinkage of the con-crete. Also, it was observed that the use of some ofthese pozzolans increased drying shrinkage althoughthey had little effect on the water content of theconcrete. Some fly ashes have little effect on dryingshrinkage, while others may increase the shrinkageof the concrete. All of these observations are basedon results of tests made on laboratory size speci-mens. However, as noted in Section 3.4.7 and Fig.3.6, the larger the concrete member, the lower the

3.4.7 Influence of size of member - The size of aconcrete member will influence the rate at whichmoisture moves from the concrete and thus in-fluence the rate of shrinkage. Carlson3*’ has shown

shrinkage. This may explain the negligible differencein shrinkage cracking of field structures, with andwithout pozzolan, despite clearly greater shrinkageof the concretes with pozzolans in laboratory testson small size specimens.3.4.6 Effect of duration of moist curing - Car1son3.6reported that the duration of moist curing of con-crete does not have much effect on drying shrink-age. This is substantiated by the test results of theCalifornia Department of Transportation3.’ whichshow substantially the same shrinkage in concretethat was moist cured for 7, 14, and 28 days beforedrying was started. As far as the cracking tendencyof the concrete is concerned, prolonged moist curingmay not necessarily be beneficial. Although thestrength increases with age, the modulus of elastic-ity also increases by almost as large a percentage,and the net result is only a slight increase in thetensile strain which the concrete can withstand.

Steam curing at atmospheric pressure, which iscommonly used in the manufacture of precast struc-tural elements, will reduce drying shrinkage (AC1517). Also, because stream curing will produce ahigh early-age strength of the concrete, it will re-duce its tendency to crack, since the pre’cast mem-bers are unrestrained.

224R-14 ACI COMMITTEE REPORT

7.5x7.5 10 x 10 10x12 5 12.5x 15 cm

I I

I I3x3 4 x 4 4 x 5 5x6 in

AVERAGE END AREA DIMENSION OF CONCRETE PRISM( LOG SCALE )

Fig. 3.6 - Effect of specimen size on drying shrink-age of concrete (Principal author’s data).

that for a concrete exposed to a relative humidity of50 percent, drying will penetrate only about 3 in.(75 mm) in 1 month and about 2 ft (0.6 m) in 10years. Fig. 3.5 shows his theoretical curves for thedrying of slabs. Hansen and Mattock3.10 made anextensive investigation of the influence of size andshape of member on the shrinkage and creep of con-crete. They found that both the rate and the finalvalues of shrinkage and creep decrease as the mem-ber becomes larger.

This significant effect of size of member on dryingshrinkage of concrete must be considered when eval-uating the potential shrinkage of concrete in struc-tures based on the shrinkage of concrete specimensin the laboratory. The rate and magnitude of shrink-age of a small laboratory specimen will be muchgreater than that of the concrete in the structures.Test results of several studies carried out to com-pare the shrinkage of concrete in walls and slabs inthe field with the shrinkage of small laboratoryspecimens have shown, as expected, that the shrink-age of the concrete in a field structure is only a frac-tion of that obtained on the laboratory specimens.Even in laboratory tests the size of the specimenused has a significant influence on shrinkage. As anexample of the effect of specimen size on shrinkageis the data presented in Fig. 3.6, giving the resultsof shrinkage tests obtained on four different sizeconcrete prisms. It will be noted that the shrinkageof the prisms having a cross section of 3 x 3 in. (7.5x 7.5 cm) was more than 50 percent greater thanthat of the concrete prism having a cross section of 5x 6 in. (12.5 x 15 cm).

3.5 - Control of shrinkage crackingConcrete tends to shrink due to drying whenever

its surfaces are exposed to air of low relative humid-ity. Since various kinds of restraint prevent the con-

crete from contracting freely, the possibility ofcracking must be expected unless the ambient rela-tive humidity is kept at 100 percent or the concretesurfaces are sealed to prevent loss of moisture. Thecontrol of cracking consists of reducing the crackingtendency to a minimum, using adequate and prop-erly positioned reinforcement, and using controljoints. The CEB-FIP Code give quantitative recom-mendations on the control of cracking due to shrink-age, listing various coefficients to determine theshrinkage levels that can be expected. Control ofcracking by correct construction practices is coveredin Chapter 8 of this report, which includes specifica-tions to minimize drying shrinkage (Section 8.6).

Cracking can also be minimized by the use of ex-pansive cements to produce shrinkage-compensatingconcretes. Shrinkage-compensating concretes are dis-cussed in Section 3.6.

3.5.1 Reduction of cracking tendency - As men-tioned previously, the cracking tendency is due notonly to the amount of shrinkage, but also to the de-gree of restraint, the modulus of elasticity, and thecreep or relaxation of the concrete. Some factorswhich reduce the shrinkage at the same time de-crease the creep or relaxation and increase the mod-ulus of elasticity, thus offering little or no help tothe cracking tendency. Emphasis should be placed,therefore, on modifying those factors which producea net reduction in the cracking tendency.

Any measure that can be taken to reduce theshrinkage of the concrete will also reduce the crack-ing tendency. Drying shrinkage can be reduced byusing less water in the mix and larger aggregatesize. A lower water content can be achieved by us-ing a well-graded aggregate, stiffer consistency, andlower initial temperature of the concrete. As dis-cussed in Section 3.4.4, however, the reduction ofwater content by the use of water-reducing admix-tures will not usually reduce shrinkage.

Another way to reduce the cracking tendency is touse a larger aggregate size. A larger aggregatesize allows an increase in aggregate volume and areduction in the total water required to obtain agiven slump. The larger aggregate also tends to re-strain the concrete more, and although this may re-sult in internal microcracking, such internal crackingis not necessarily harmful.

A third way to reduce the cracking tendency is toapply a surface coating to the concrete, which willprevent the rapid loss of moisture from within. Thismeans of controlling cracking has not been used toits full potential and should be given better consider-ation. However, many surface coatings such as all-purpose paints are ineffective, because they permitthe moisture to escape almost as fast as it reachesthe surface. Chlorinated rubber and waxy or resin-ous materials are effective coatings, but there areprobably many other materials which will slow theevaporation enough to be beneficial. Any slowing of

CONTROL OF CRACKING 224R-15

3.6 - Shrinkage-compensating concretesShrinkage-compensating concretes made with ex-

pansive cements can be used to minimize or elimi-nate shrinkage cracking. The properties and use ofexpansive cement concretes is published in numer-ous papers and reports.3-11* 3*12 Of the several typesof expansive cements produced, the Type Kshrinkage-compensating expansive cement is mostcommonly used in the United States.

In a reinforced concrete, the expansion of the ce-ment paste during the first few days of curing willdevelop a low level of prestress inducing com-pressive stresses in the concrete and tensile stressesin the steel. The level of compressive stresses devel-oped in the shrinkage-compensating concretesranges from 25 to 100 psi (0.2 to 0.7 MPal. Whensubjected to drying shrinkage, the contraction of theconcrete will result in a reduction or elimination ofits precompression. The initial precompression of the

the rate of shrinkage will be beneficial, because con-crete has a remarkable quality of relaxing under sus-tained stress. Thus, concrete may be able to with-stand two or three times as much slowly appliedshrinkage as it can rapid shrinkage.3.5.2 R e i n f o r c e m e n t - Proper ly placed re-inforcement, used in adequate amounts, will not onlyreduce the amount of cracking but prevent unsightlycracking. By distributing the shrinkage strains alongthe reinforcement through bond stresses, the cracksare distributed in such a way that a larger numberof very fine cracks will occur instead of a few widecracks. Although the use of such reinforcement tocontrol cracking in a relatively thin concrete sectionis practical, it is not needed in massive structuressuch as dams due to the low drying shrinkage ofthese mass concrete structures. The minimumamount and spacing of reinforcement to be used infloors, roof slabs, and walls is given in AC1 318.

3.6.3 Joints - The use of joints is the most effectivemethod of preventing formation of unsightly crack-ing. If a sizable length or expanse of concrete, suchas walls, slabs or pavements, is not provided withadequate joints to accommodate shrinkage, it willmake its own “joints” by cracking.

Contraction joints in walls are made, for example,by fastening to the forms wood or rubber stripswhich leave narrow vertical grooves in the concreteon the inside and outside of the wall. Cracking of thewall due to shrinkage should occur at the grooves,relieving the stress in the wall and thus preventingformation of unsightly cracks. These grooves shouldbe sealed on the outside of the wall to prevent pene-tration of moisture. Sawed joints are commonly usedin pavements, slabs and floors.

Joint location depends on the particulars of place-ment. Each job must be studied individually to de-termine where joints should be placed.*

STEEL\_B--- _----

ORIGINAL LENGTH

t

T A b T___++IC~*___

EXPANSION PUTS STEEL INTENSION AND CONCRETE INCOMPRESSION M

STRESS LOSS DUE TOSHRINKAGE AND CREEP

RESIDUAL EXPANSION OR, -+jSMALL CONTRACTION

.Qrl 3 7 -.concretes.

Basic concept of shrinkage-compensating

CURINGr/ .p- DRYING

SHRINKAGE- COMPENSATINGCONCRETE, p = 0.16Ym

PORTLAND CEMENT

;CONCRETE

, I I I I0 50 100 150 2oc

AGE OF CONCRETE, DAYS

Fig. 3.8 - Length change characteristics of shrink-age-compensating and portland cement concretes(Relative humidity = 50 percent).

concrete minimizes the magnitude of any tensilestress that may ultimately develop due to shrinkage,and thus reduce or eliminate the tendency to crack-ing. This basic concept of the use of expansive ce-ment to produce a shrinkage-compensating concreteis illustrated in Fig. 3.7.

A typical length change history of a shrinkage-compensating concrete is compared to that of a port-land cement concrete in Fig. 3.8. The amount of re-inforcing steel normally used in reinforced concrete

*Guidance on joint sealants and control joint location in slabs is avail-able in ACI 504 and in ACI 302, respectively.

224R-16 ACI COMMITTEE REPORT

Chapter 4 - Control of cracking in flexuralmembers*

3.8. Concrete Manual, 8th Edition, U.S. Bureau of Re-clamation, Denver, 1975, 627 pp.

3.9. Carlson, Roy W., “Drying Shrinkage of Large Con-crete Members,” ACI JOURNAL, Proceedings V. 33, No. 3,Jan.-Feb. 1937, pp. 327-336.

made with portland cements is usually more than ad-equate to provide the elastic restraint needed forshrinkage-compensating concrete. To take full advan-tage of the expansive potential of shrinkage-com-pensating concrete in minimizing or preventingshrinkage cracking of unformed concrete surfaces, itis important that positive and uninterrupted watercuring (wet covering or ponding) be started immedi-ately after final finishing. For slabs on well satu-rated subgrades, curing by sprayed-on membranesor moisture-proof covers have been successfully uti-lized. Inadequate curing of shrinkage-compensatingconcrete may result in an insufficient expansion toelongate the steel and thus subsequent cracking dur-ing drying shrinkage. Specific recommendations andinformation on the use of shrinkage-compensatingconcrete are contained in ACI 223.

References3.1. Verbeck, George J., “Carbonation of Hydrated Port-

land Cement,” Cement and Concrete, STP-205, AmericanSociety for Testing and Materials, Philadelphia, 1958, pp.17-36.

3.2. Blaine, R. L.; Arni, H. T.; and Evans, D. N., “Inter-relations Between Cement and Concrete Properties: Part 4- Shrinkage of Hardened Portland Cement Pastes andConcrete,” Building Science Series No. 15, National Bu-reau of Standards, Washington, D.C., Mar. 1969, 77 pp.

3.3. Brunauer, S.; Skalny, J.: and Yudenfreund, H.,“Hardened Cement Pastes of Low Porosity: DimensionalChanges,” Research Report No. 69-8, Engineering Re-search and Development Bureau, New York State Depart-ment of Transportation, Albany, Nov. 1969, 12 pp.

3.4. Tremper, Bailey, and Spellman, Donald L., “Shrink-age of Concrete - Comparison of Laboratory and FieldPerformance,” Highway Research Record. Highway Re-search Board, No. 3, 1963, pp. 30-61.

3.5. Lerch, William, “The Influence of Gypsum on theHydration and Properties of Portland Cement Pastes,”Proceedings, ASTM, V. 46, 1946, pp. 1252-1297.

3.6. Carlson, Roy W., “Drying Shrinkage of Concrete asAffected by Many Factors,” Proceedings, ASTM, V. 38,Part II, 1938, pp. 419-437.

3.7. Reichard, T. W., “Creep and Drying Shrinkage ofLightweight and Normal Weight Concrete,” Monograph74, National Bureau of Standards, Washington, D.C., 1964,30 pp.

3.10. Hansen, Torben C., and Mattock, Alan H., “In-fluence of Size and Shape of Member on the Shrinkage andCreep of Concrete,” ACI JOURNAL, Proceedings V. 63, No.2, Feb. 1966, pp. 267-290.

3 .11 . ACI C o m m i t t e e 2 2 3 , “ E x p a n s i v e C e m e n tConcretes-Present State of Knowledge,” ACI J OURNAL,Proceedings V. 67, No. 8, Aug. 1970, pp. 583-610.

3.12. Klein Symposium on Expansive Cement Concretes,SP-38, American Concrete Institute, Detroit, 1973,491 pp.

4.1 - IntroductionWith the regular use of high strength reinforcing

steel and the strength design approach for re-inforced concrete, and higher allowable stresses inprestressed concrete design, the control of crackingmay be as important as the control of deflection inflexural members. Internal cracking in concrete canstart at stress levels as low as 3000 psi (20.7 MPa) inthe reinforcement. Crack control is important to pro-mote the aesthetic appearance of structures, and formany structures, crack control plays an importantrole in the control of corrosion by limiting the possi-bilities for entry of moisture and salts which, to-gether with oxygen, can set the stage for corrosion.

This chapter is concerned primarily with crackscaused by flexural and tensile stresses, but temper-ature, shrinkage, shear and torsion may also lead tocracking.”4.1 Cracking in certain specialized struc-tures, such as reinforced concrete tanks, bins andsilos, is not covered in this report. For informationon cracking concrete in these structures, see Refer-ence 4.2 and ACI 313.

Extensive research studies on the cracking be-havior of beams have been conducted over the last50 years . Mos t o f them are repor ted in ACIBibliography No. 9 on crack control.4.3 Others arereferenced in this chapter. Reference 4.1 contains an

extensive review of cracking in reinforced concretestructures. Several of the most important crack pre-diction equations are reviewed in the previous com-mittee report. 1.1’Additional work presented in theCEB-FIP Model Code for Concrete Structure givesthe European approach to crack width evaluationand permissible crack widths.

Recently, fiber glass rods have been used as areinforcing material.4.4To date, experience is lim-ited, and crack control in structures reinforced withfiber glass rods is not addressed in this report. It isexpected, however, that future committee docu-ments will address crack control in structures usingthis and other new systems as they come into use.

4.2 - Crack control equations for reinforced con-crete beams

A number of equations have been proposed for theprediction of crack widths in flexural members; mostof them are reviewed in the previous committee re-port1.1Pand in key publications listed in the refer- ences. Most equations predict the probable max-imum crack width, which usually means that about90 percent of the crack widths in the member arebelow the calculated value. However, research hasshown that isolated cracks in beams in excess oftwice the width of the computed maximum can

*Principal authors: Edward G. Nawy and Peter Gergely.

CONTROL OF CRACKING 224R-17

sometimes occur,*-4 though generally the coefficientof variation of crack width is about 40 percent.4-1Evidence also exists indicating that this range incrack width randomness may increase with the sizeof the member? Besides limiting the computedmaximum crack width to a given value, the designershould estimate the percentage of cracks above thisvalue which can be tolerated.

Crack control equations recommended by ACICommittee 224 and the Comite Euro-Internationaldu Beton (CEB) are presented below.4.21 ACI Committee 224 recommendations - Re-quirements for crack control in beams and thick one-way slabs in the ACI Building Code (ACI 318) arebased on the statistical analysis4-6 of maximumcrack width data from a number of sources. Based onthe analysis, the following general conclusions werereached:

1. The steel stress is the most important variable.2. The thickness of the concrete cover is an impor-

tant variable, but not the only geometric considera-tion.

3. The area of concrete surrounding each re-inforcing bar is also an important geometric vari-able.

4. The bar diameter is not a major variable.5. The size of the bottom crack width is influenced

by the amount of strain gradient from the level ofthe steel to the tension face of the beam.

The equations that were considered to best p r ediet the most probable maximum bottom and sidecrack widths are:

W* =

w, =

whereW* =

w, =

fJ =A =

tb =t, =P =

h 1 =

0.091 v-a p (f, - 5) x 10-3 (4.la)

0.091 rt,, Al

-1-G. t,l&.. (f, - 5) x 1 0 - 3 (4.lb)

most probable maximum crack width at bot-tom of beam, in.most probable maximum crack width at levelof reinforcement, in.reinforcing steel stress, ksiarea of concrete symmetric with reinforcingsteel divided by number of bars, in.2bottom cover to center of bar, in.side cover to center of bar, in.ratio of distance between neutral axis andtension face to distance between neutral axisand centroid of reinforcing steel = 1.20 inbeamsdistance from neutral axis to the reinforcingsteel, in.

Simplification of Eq. (4.la) yielded the followingequation

w = 0.076~fs ~AX D3 (4.2)

whereW = most probable maximum crack width, in.dc = thickness of cover from tension fiber to

center of bar closest thereto, in.

When the strain, Ed, in the steel reinforcement isused instead of stress, f,, Eq. (4.2) becomes

w = 2.2 p L, V-JX (4.3)

E, = strain in the reinforcement

Eq. (4.3) is valid in any system of measurement.The cracking behavior in thick one-way slabs is

similar to that in shallow beams. For one-way slabshaving a clear concrete cover in excess of 1 in. (25.4mm), Eq. (4.2) can be adequately applied if p = 1.25to 1.35 is used.

AC1 318 Section 10.6 uses Eq. (4.2) with p = 1.2 inthe following form

2 = f,cQi- (4.2a)

Using the specified cover in AC1 318, maximiumallowable z = 175 kips per in. for interior exposurecorresponds to a limiting crack width of 0.016 in.(0.41 mm).

The Code allows a value of z = 145 kips per in.for exterior exposure based on a crack width valueof 0.013 in., (0.33 mm), which may be excessive basedon Table 4.1. While application of Eq.

(10.4) of AC1 318-771 to beams gives adequate crackcontrol values, its application to one-way slabs withstandard 3/4 in. (19 mm) cover and reinforced withsteel of 60 ksi (414 MPa) or lower yield strengthresults in large reinforcement spacings. However,the provisions of Code Section 7.6.5 indirectly limitthe spacing of such reinforcement in one-way slabs.

AC1 340.1R contains design aids for the applica-tion of Eq. (4.2a).

4.2.2 CEB recommendations - Crack control recom-mendations proposed in the European Model Codefor Concrete Structures apply to prestressed as wellas reinforced concrete and can be summarized asfollows:

The mean crack width, wm in beams is expressedin terms of the mean crack spacing, s

rm such that

Kn = L&n

where

and represents the average strain in the steel.

(4.4)

(4.5)

f s f II =

K =

steel stress at the cracksteel stress at the crack due to forces causingcracking at the tensile strength of concretebond coefficient, 1.0 for ribbed bars, reflectinginfluence of load repetitions and load duration

224~018 ACI COMMITTEE REPORT

The mean crack spacing is

S rm(4.6)

where

c =S

x2 =

x3 =

QR =

At

=

clear concrete coverbar spacing, limited to 15d,0.4 for ribbed barsdepends on the shape of the stress diagram,0.125 for bendingA, /A,effective area in tension, depending on ar-rangement of bars and type of externalforces; it is limited by a line c + 7d, from thetension face for beams; in the case of slabs,not more than halfway to the neutral axis

A simplified formula canbe derived for the meancrack width in beams with ribbed bars,

fw, = 0.7 _“-

d3c + 0.05 -!

ES QR(4.7)

A characteristic value of the crack width,presumably equivalent to the probable maximumvalue, is given as 1.7~~.

4.3 - Crack control i n two-way slabs and platesCrack control equations for beams underestimatethe crack widths developed in two-way slabs andplates4.7 and do not tell the designer how to spacethe reinforcement. The cracking mechanism in two-way slabs and plates is controlled primarily by thesteel stress level and the spacing of the re-inforcement in the two perpendicular directions. Inaddition, the clear concrete cover in two-way slabsand plates is nearly constant [3/4 in. (19 mm) for inte-rior exposure], whereas it is a major variable in thecrack control equations for beams.

Analysis of data in the only major work on crack-ing in two-way slabs and plates4s7 has provided thefollowing equation for predicting the maximumcrack width:

&,sI:w= ws (4.8)

n

where the radical rl = db,s21et, is termed the gridindex, and can be transformed into

]k = fracture coefficient, having a value k = 2.8 x

lO-5 for uniformly loaded restrained two-way

P =

f =s

db1 =

s1 =

s2 =

46 "1 =

Qrl =

=

w =

action square slabs and plates. For concen-trated loads or reactions, or when the ratioof short to long span is less than 0.75 butlarger than 0.5, a value of k = 2.1 x 1O-5 isapplicable. For span aspect ratios 0.5, k =1.6 x 1O-s(as defined in Section 4.2.1) 1.25 (chosen tosimplify calculations though varies between1.20 and 1.35)actual average service load stress level, or40 percent of the design yield strength fy,ksidiameter of the reinforcement in direction“1” closest to the concrete outer fibers, in.spacing of the reinforcement in direction “l”,in.spacing of the reinforcement in per-pendicular direction “2”, in.direction of reinforcement closest to theouter concrete fibers; this is the direction forwhich crack control check is to be madeactive steel ratioArea of steel A, per ft width_ - V - - P - - - -

12 (dbt + 2CJ

where Cl is clear concrete cover measuredfrom the tensile face of concrete to the near-est edge of the reinforcing bar in directionb& VW1crack width at face of concrete, in., causedby flexural load

Subscripts 1 and 2 pertain to the directions of re-inforcement.

For simply supported slabs, the value of k shouldbe multiplied by 1.5. Interpolated k values apply forpartial restraint at the boundaries. For zones of flatplates where transverse steel is not used or when itsspacing s2 exceeds 12 in., use s2 = 12 in. in theequation.

If strain is used instead of stress, Eq. (4.8)becomes

(4.9)

where values of the kl = 29 x 100~ times the kvalues previously listed.

References 4.8 and 340.1R contain design aids for

the application of these recommendations.

4.4 - Tolerable crack widths versus exposure condi-tions in reinforced concrete

Table 4.1 is a general guide for tolerable crackwidths at the tensile face of reinforced concretestructures for typical conditions and is presented asan aid to be used during the design process. Thetable is based primarily on Reference 4.9. It is im-

portant to note that these values of crack width are

CONTROL OF CRACKING 224R-19

TABLE 4.1 - Tolerable crack widths,reinforced concrete

Exposure conditionTolerable

crack width, in. (mm)

Dry air or protective membraneHumidity, moist air, soilDeicing chemicalsSeawater and seawater spray:

wetting and dryingWater retaining structures*

0.016 (0.41)0.012 (0.30)0.007 (0.18)

0.006 (0.15)0.004 (0.10)

*Excluding nonpressure pipes

not always a reliable indication of the corrosion anddeterioration to be expected. In particular, a largercover, even if it leads to a larger surface crackwidth, may sometimes b e preferable for corrosioncontrol in certain environments. Thus, the designermust exercise engineering judgment on the extent ofcrack control to be used. When used in conjunctionwith the recommendations presented in Sections4.2.1 and 4.2.3 to limit crack width, it should be ex-pected that a portion of the cracks in the structurewill exceed these values by a significant amount.

4.5 - Flexural cracking in prestressed concretePartially prestressed members, in which cracks

may appear under working loads, are used exten-sively. Cracks form in these members when the ten-sile stress exceeds the modulus of rupture of theconcrete (Sfl to 90 under short-term conditions).The control of these cracks is necessary mainly foresthetic reasons. The residual crack width, after re-moval of the major portion of the live load, is small[about 0.001 in. to 0.003 in. (0.03 to 0.08 mm)] andtherefore, crack control is usually not necessary ifthe live load is transitory.

The prediction of crack widths in prestressed con-crete members has received far less attention thanin reinforced concrete members. The available ex-perimental data are limited and, at the same time,the number of variables is greater in prestressedmembers.

4.5.1 Crack prediction equations - One approach tocrack prediction, w h i c h r e l a t e s i t t o t h e non-prestressed case, has two steps. First the decom-pression moment is calculated, at which the stress atthe tension face is zero. Then the member is treatedas a reinforced concrete member and the increase instress in the steel is calculated for the additionalloading. The expressions given for crack predictionin nonprestressed beams may be used to estimatethe cracks for the load increase above the decom-pression moment. A multiplication factor of about1.5 is needed when strands, rather than deformedbars, are used nearest to the beam surface in the

prestressed member to account for the differences inbond properties.

The difficulty with this approach is the complexityof calculations. The determination of the decompres-sion moment and, especially, the stress in the steelis complicated and unreliable unless elaborate meth-ods are used.4.10 For this reason, approximate meth-ods for crack width prediction are attractive. Theseare not much less accurate than the more com-plicated methods, and the lack of sufficient data, cov-ering large variations in the variables, precludesfurther refinements at this date.

The CEB Model Code has the same equation forthe prediction of the crack width in prestressedmembers as in nonprestressed members (see Section4.2.2). The increase in steel strain is calculated fromthe decompression stage. Several other equationshave been proposed.4.11-4.“0

Limited evidence seems to indicate that unbondedmembers develop larger cracks than bonded mem-bers. Nonprestressed deformed bars may be used toreduce the width of the cracks to acceptable levels.The cracks in bonded post-tensioned members arenot much different from cracks in pretensionedbeams.

4.5.2 Allowable crack widths - Some authors statethat corrosion is a greater problem in prestressedconcrete members because of the smaller area ofsteel used. However, recent research results4.“’ indi-cate that there is no general relationship betweencracking and corrosion in most circumstances. Fur-

thermore cracks close upon removal of the load, andthe use of crack width limits should depend on thefluctuation and magnitude of the live load.

4.6 - Anchorage zone cracking in prestressed con-crete

Longitudinal cracks frequently occur in the ancho-rage zones of prestressed concrete members due totransverse tensile stresses set up by the concen-trated forces.4.22T 4.23 Such cracks may lead to (or incertain cases are equivalent to) the failure of themember. Transverse reinforcement (stirrups) mustbe designed to restrict these cracks.

Two types of cracks may develop: spalling crackswhich begin at the end face (loaded surface) andpropagate parallel to the prestressing force, andbursting cracks which develop along the line of theforce or forces, but away from the end face.

For many years stirrups were designed to takethe entire calculated tensile force based on the anal-ysis of the uncracked section. Classical and finite-ele-ment analyses show similar stress distributions forwhich the stirrups are to be provided. However,since experimental evidence shows that higherstresses can result.4.23 than indicated by these an-alyses, and the consequences of under-reinforcement

224R-20 ACI COMMITTEE REPORT

4.2. Yerlici, V. A., “Minimum Wall Thickness of CircularConcrete Tanks,” Publication No. 35-11, International Asso-ciation for Bridge & Structural Engineering, Zurich, 1975,p. 237.

References4.1. Leonhardt, Fritz, “Crack Control in Concrete Struc-

tures,” IABSE Surveys No. S4/77, International Associa-tion for Bridge and Structural Engineering, Zurich, 1977,26 pp.

4.8. Nawy, Edward G., “Crack Control ThroughReinforcement Distribution in Two-Way Acting Slabs andPlates,” ACI JOURNAL, Proceedings V. 69, No. 4, Apr.1972, pp. 217-219.

4.9. Nawy, Edward G., “Crack Control in ReinforcedConcrete Structures,” ACI JOURNAL, Proceedings V. 65,No. 10, Oct. 1968, pp. 825-836.

can be serious, it is advisable to provide more steelthan required by this type of analysis.

More recently, designs have been based oncracked section analyses. A design procedure forpost-tensioned members using a cracked section an-alysis4.24 has found acceptance with many design-ers. For pretensioned members, an empirical equa-tion has proven to be quite usefu1.4.25

Spalling cracks form between anchorages andpropagate parallel to the prestressing forces andmay cause gradual failure, especially when the forceacts near and parallel to a free edge. Since analysesshow that the spall ing stresses in an uncrackedmember are confined to near the end face, it is im-portant to place the first stirrup near the end sur-face, and to distribute the stirrups over a distanceequal to at least the depth of the member to fully ac-count for both spalling and bursting stresses. Pre-cast blocks with helical reinforcement may be usedwhen the prestressing forces are large.

4.7 - Tension crackingThe cracking behavior of reinforced concrete mem-

bers in tension is similar to that of flexural mem-bers, except that the maximum crack width is largerthan that predicted by the expressions for flexuralmembers.4.26T 4.27 The lack of strain gradient, a n dresultant restraint imposed by the compression zoneof flexural members, is probably the reason for thelarger tensile crack width.

Data are limited but it appears that the maximumtensile crack width may be expressed approximatelyin a form similar to that used for flexural crackwidth.

w = O.lOf,&tA x 10-3 (4.10)

4.3. ACI Committee 224, “Causes, Mechanism, andControl of Cracking in Concrete,” ACI Bibliography No. 9,American Concrete Institute, Detroit, 1971, 9.2 pp.

4.4. Nawy, Edward G., and Neuwerth, G. E., “Behaviorof Concrete Slabs, Plates and Beams with Fiber Glass asMain Reinforcement,” Proceedings, ASCE, V. 103, ST2,Feb. 1977, pp. 421-440.

4.5. Clark, Arthur P., “Cracking in ReinforcedConcrete Flexural Members,” ACI JOURNAL, ProceedingsV. 52, No. 8, Apr. 1956, pp. 851-862.

4.6. Gergely, Peter, and Lutz, Leroy A., “MaximumCrack Width in Reinforced Concrete Flexural Members,”

Causes, Mechanism, and Control of Cracking in Concrete,SP-20, American Concrete Institute, Detroit, 1968, pp.87-117.

4.7. Nawy, Edward G., and Blair, Kenneth W.,“Further Studies on Flexural Crack Control in StructuralSlab Systems,” Cracking, Deflection, and Ultimate Load ofConcrete Slab Systems, SP-30, American ConcreteInstitute, Detroit, 1971, pp. 1-41.

4.10. Nilson, Arthur H., Design of Prestressed Concrete,John Wiley and Sons, New York, 1978, 526 pp.

4.11. Abeles, Paul W., “Cracks in Prestressed ConcreteBeams,” Proceedings, Fifth IABSE Congress (Lisbon,1956), International Association for Bridge and StructuralEngineering, Zurich, 1956, pp. 707-720.

4.12. Bennett, E. W., and Dave, N. J., “Test Perfor-mances and Design of Concrete Beams with LimitedPrestress,” The Structural Engineer (London), V. 47, No.12, Dec. 1969, pp. 487-496.

4.13. Holmberg, Ake, and Lindgren, Sten, “CrackSpacing and Crack Widths Due to Normal Force andBending Moment,” Document D2:1970, National SwedishCouncil for Building Research, Stockholm, 1970, 57 pp.

4.14. Rao, A.S.P.; Gandotra, K.; and Ramaswamy, G.S., “Flexural Tests on Beams Prestressed to DifferentDegrees of Prestress,” Journal, Institution of Engineers(Calcutta), V. 56, May 1976.

4.15. Bate, Stephen C. C., “Relative Merits of Plain andDeformed Wires in Prestressed Concrete Beams UnderStatic and Repeated Loading,” Proceedings, Institution ofCivil Engineers (London), V. 10, Aug. 1958, pp. 473-502.

4.16. Bennett, E. W., and Chandrasekhar, C. S., “Cal-culation of the Width of Cracks in Class 3 PrestressedBeams,” Proceedings, Institution of Civil Engineers(London), V. 49, July 1971, pp. 333-346.

4.17. Hutton, S. G., and Loov, R. E., “Flexural Behaviorof Prestressed, Partially Prestressed, and ReinforcedConcrete Beams,” ACI JOURNAL, Proceedings, V. 63,No. 12, Dec. 1966, pp. 1401-1410.

4.18. Krishna, Raju N.; Basavarajuiah, B. S.; andAhamed Kurty, U. C., “Flexural Behavior of PretensionedConcrete Beams with Limited Prestress,” BuildingScience, V. 8, No. 2, June 1973, pp. 179-185.

4.19. Stevens, R. F., “Tests on Prestressed ReinforcedConcrete Beams,” Concrete (London), V. 3, No. 11, Nov.1969, pp. 457-462.

4.20. Nawy, E. G., and Huang, P. T., “Crack andDeflection Control of Pretensioned Prestressed Beams,”Journal, Prestressed Concrete Institute, V. 22, No. 3,May-June 1977, pp. 30-47.

4.21. Beeby, A. W., “Corrosion of Reinforcing Steel inConcrete and Its Relation to Cracking,” The StructuralEngineer (London), V. 56A, No. 3, Mar. 1978, pp. 77-81.

4.22. Gergely, Peter, “Anchorage Systems in Pre-stressed Concrete Pressure Vessels; Anchorage ZoneProblems,” ORNL-TM-2378, Oak Ridge NationalLaboratory, U.S. Atomic Energy Commission, Oak Ridge,Tenn., 1969, pp. l-49.

4.23. Zielinski, J. L., and Rowe, R. E., “An

CONTROL OF CRACKING 224-21

Chapter 5 - Long-term effects on cracking*

Investigation of the Stress Distribution in the AnchorageZones of Post-Tensioned Concrete Members,” TechnicalReport No. 9, Cement and Concrete Association, London,Sept. 1960, 32 pp.

4.24. Gergely, P., and Sozen, M. A., “Design of Anchor-age Zone Reinforcement in Prestressed Concrete Beams,”Journal, Prestressed Concrete Institute, V. 12, No. 2,Mar.-Apr. 1967, pp. 63-75.

4.25. Marshall, W. T., and Mattock, A. H., “Control ofHorizontal Cracking in the Ends of Pretensioned ConcreteGirders,” Journal, Prestressed Concrete Institute, V. 7,No. 5, Aug.-Oct. 1962, pp. 56-74.

4.26. Broms, Bengt B., “Crack Width and CrackSpacing in Reinforced Concrete Members,” ACI JOURNAL ,Proceedings, V. 62, No. 10, Oct. 1965, pp. 1237-1256.

4.27. Broms, Bengt B., and Lutz, Leroy A., “Effects ofArrangement of Reinforcement on Crack Width andSpacing of Reinforced Concrete Members,” ACI JOURNAL,Proceedings V. 62, No. 11, Nov. 1965, pp. 1395-1410.

5.1 - IntroductionCracking in concrete is affected by the long-term

conditions to which the concrete element is sub-jected. In most cases, long-term exposure and long-term loading extend the magnitude of cracks in bothreinforced and plain concrete. The discussion in thischapter summarizes the major long-term factorswhich affect the crack control performance of con-crete.

5.2 - Effects of long-term loadingAs discussed in Chapter 2, both sustained and

cyclic loading increase the amount of microcrackingin concrete. The total amount of microcracking ap-pears to be a function of the total strain and islargely independent of the method by which thestrain is induced. Microcracking due to long-termloading may well be an effect, rather than a majorcause, of creep, and microcracks formed at serviceload levels do not seem to have a great affect on thestrength or serviceability of concrete.

The effect of sustained or repetitive loading onmacroscopic cracking, however, may be an importantconsideration in the serviceability of reinforced con-crete members, especially in terms of corrosion ofreinforcing steel and appearance.

The increase in crack width due to long-term orrepetitive loading can vary between 10 percent and1,000 percent over the span of several years. 5.1-5.8 While there is a large scatter in the data, informa-tion obtained from sustained loading tests of up to 2

.7,5.8 and fatigue tests with up to one millioncycles5.4, 5 .5,5.8,5.9 indicate that a doubling of crackwidth with time can be expected. Under most condi-tions, the spacing of cracks does not change withtime at constant levels of stress.5.4,5.7,5.8

An excep-

*Principal authors: David Darwin and Ernest K. Schrader.

tion to this occurs at low loads or in beams withhigh percentages of reinforcement, in which case thetotal number and width of cracks increase sub-stantially after the loading has begun.5.2,5.4,5.8 Thelargest percentage increase in crack width is thenexpected in flexural members subject to low levelsof load, since the cracks take more time to develop.

For both prestressed and reinforced concrete flex-ural members, long-term loading and repetitive load-ing seem to give about the same crack widths andspacing.5.9 The rate of crack development, however,is considerably faster under repetitive loading.5.5,5.8-5.10

As discussed in Chapter 4, crack width is a func-tion of cover. For short-term static and fatigue load-ing, surface crack width is approximately propor-tional to the steel s t ra in 5.7,5.8,5.10 Cracks grow inwidth under sustained loading at a decreasing rate.However, the rate of growth is faster than the aver-age observed surface strain at the level of the steel.For long term loading, crack width is proportional tothe steel strain (including the effects of creep), plusthe strain induced in the concrete due to shrink-age.5.7

Under initial loads, cracks adjacent to re-inforcement are restricted by the bond between thesteel and the concrete,5.7-5.11 and thus the width ofsurface cracks do not provide a good indication ofthe exposure of the reinforcing steel to corrosiveconditions. Over a period of time, however, the ad-hesion bond between the steel and the concrete un-dergoes breakdown. After about 2 years, the crackwidth at the reinforcement is approximately equal tothe crack width at the surface.5.7 At this stage,cracks in flexural members are triangular in shapeincreasing in width from the neutral axis to the sof-fit, and are approximately uniform across the widthof the beam. Therefore, after a few years, the widthof a surface crack provides a good estimate of thecrack width at the level of the reinforcing steel.

Many questions remain as to the importance ofcrack width on the serviceability of reinforced andprestressed concrete members . 5.12-5.14 Addedcover is generally acknowledged as a method of im-proving the corrosion protection for reinforcingsteel. Since additional cover also results in addedsurface crack width, and since this surface crackwidth appears to provide a good estimate of thecrack width at the level of the steel, the entire ques-tion of the importance of crack width on corrosionprotection remains open. It does seem clear thatcrack widths predicted on the basis of short termstatic tests do not provide a precise guide to crackwidths in structures actually in service.

5.3 - Environmental effectsThe long-term effects of an adverse environment

in both producing and in enlarging concretecracks5.15,5.16

can be damaging to both concrete and

224R-22 ACI COMMll-i=EE REPORT

reinforcement. If concrete is not resistant to freezingand thawing when critically saturated, it will de-velop cracks when frozen. The lack of such resis-tance may be due to either the use of non-frost-resis-tant coarse aggregate or the failure to produce asatisfactory air-void system or failure to protect theconcrete from freezing prior to the reduction of thefreezable water content by maturity to a tolerablerange. The achievement of critical saturation in non-frost-resistant concrete may be facilitated by thepresence of preexisting cracks which allow entry ofwater more readily than would be the case other-wise. The initiation of D-cracking near joints orother cracks in pavements is a good example. Inmore extreme cases, it is not uncommon for cracksin the roadway deck of dams and navigation locks(caused either by thermal stress or shrinkage of thericher topping mix) to spall due to water whichfreezes in the cracks themselves (independent of thefrost resistance of the concrete). On the otherhand,preexisting cracks may also function to allow con-crete to dry below critical saturation before freezing,when this might not occur in the absence of suchcracks. Hence, the role of cracks as they effect thedeficiencies in frost resistance will vary with the en-vironmental conditions (e.g., typical time of dryingafter wetting before freezing), crack width, ability ofcracks to drain, etc.

If the aggregate used in the concrete is durableunder freeze-thaw conditions and the, strength of theconcrete is high, the concrete durability will bet-ter. (AC1 201.2R). Field exposure tests of reinforcedconcrete beams5*17 (subjected to freezing and thaw-ing and an ocean side environment) indicate thatthe use of air-entrained concrete made the beamsmore resistant to weathering than the use of non-air-entrained concrete. Beams with modern de-formed bars were found to be more durable thanthose using old-style deformations. Maximum crackwidths did not increase with time when the steelstress was less than 30 ksi, (210 MPa) but did in-crease substantially (50 to 100 percent) over a 9 yearperiod when the steel was 30 ksi (210 MPa) or more.

5.4 - Aggregate and other effectsConcrete may crack as the result of expansive re-

actions between aggregate and alkalis derived fromcement hydration, admixtures or external sources(e.g., curing water, ground water, alkaline solutionsstored or used in the finished structure).

Possible solutions to these problems include limita-tions on reactive constituents in the aggregate, limi-tations on the alkali content of cement, or addition ofa satisfactory pozzolanic material. The potential forsome expansive reactions, e.g., alkali-carbonate, isnot reduced by pozzolanic admixtures. AC1 201.2Rand Reference 5.18 give details on identification and

evaluation of aggregate reactivity.

Based on reports of AC1 Committees 201 and

212 5.15.5.169 the possible hazard of using calcium chlo-

ride in a water-soluble salt environment warrants arecommendation against its use under such circum-stances. Also, the use of calcium chloride in re-inforced structures exposed to unusually moist envi-ronments is to be avoided regardless of the presenceor absence of water-soluble salts in adjacent watersand soils.

Detrimental conditions may also result from theapplication of deicing salts to the surface of hard-ened concrete. When such applications are neces-sary, calcium chloride or sodium chloride should beused and only within recommended application rates.Concrete subjected to water soluble salts should beair entrained [6.5 to 7.5 percent for normal 3L4 in. (19mm) MSA concrete and 4.5 to 5.5 percent for F/2 in.(38 mm) MSA concrete], should have adequate cover(about 2 in.), and should be made with a high-qualitymix yielding low permeability.

5.5 - Use of polymers in improving cracking c h a r -acterisitics

Extensive work is available on the use of polymersin modifying the characteristics of concrete.5*1gy 5.20p5.21 Polymer-portland cement concretes have a largedeformation capacity, high tensile and compressivestrengths and negligible permeability. The tensilesplitting strength can be as high as 1550 psi(10.7 MPa).5-22 Polymer impregnation is anothermethod of introducing beneficial polymer systemsinto concrete. This procedure creates a ‘layer’ of highquality material to the depth that has been im-pregnated. These materials are discussed in greaterdetail in Chapter 6.

Because of these desirable characteristics, it is ex-pected that structural elements made with polymermodified concrete will exhibit superior serviceabilityin cracking, deflection, creep, shrinkage, and per-meability.

Referenees5.1. Bate, Stephen C. C., “A Comparison Between Pre-

stressed Concrete and Reinforced Concrete Beams UnderRepeated Loading,” Proceedings Institution of Civil Engi-neers (London), V. 24, Mar. 1963, pp. 331-358.

5.2. Brendel, G., and Ruhle, H., “Tests on ReinforcedConcrete Beams Under Long-Term Loads (Dauerstandver-suche mit Stahlbetonbalken),” Proceedings, SeventhIABSE Congress (Rio de Janeiro, 1964), International As-sociation of Bridge and Structural Engineering, Zurich,1964, pp. 916-922.

5.3. Lutz, LeRoy A.; Sharma, Nand K.; and Gergely, Pe-ter, “Increase in Crack Width in Reinforced ConcreteBeams Under Sustained Loading,” ACI JOURNAL, Pro-ceedings, V. 64, No. 9, Sept. 1968, pp. 538-546.

5.4. Abeles, Paul W.: Brown, Earl L. II; and Morrow,Joe W., “Development and Distribution of Cracks in Rect-angular Prestressed Beams During Static and FatigueLoading,” Journal, Prestressed Concrete Institute, V. 13,No. 5, Oct. 1968, pp. 36-51.

CONTROL OF CRACKING 224R-23

.

Chapter 6 - Control of cracking in concretelayered systems*

5.18. Woods, Hubert, Durability of Concrete Construc-tion, Monograph No. 4, American Concrete Institute/IowaState University, Detroit, 1968, 187 pp.

5.5. Bennett, E. W., and Dave, N. J., “Test Perfor-mances and Design of Concrete Beams with Limited Pre-stress,” The Structural Engineer (London), V. 47 No. 12,Dec. 1969, pp. 487-496.

5.6. Holmberg, A., and Lindgren, S., “Crack Spacing andCrack Width Due to Normal Force or Bending Moment,”Document D2, National Swedish Council for Building Re-search, Stockholm, 1970, 57 pp.

5.7. Illston, J. M., and Stevens, R. F., “Long-term Crack-ing in Reinforced Concrete Beams,” Proceedings, In-stitution of Civil Engineers (London), Part 2, V. 53, Dec.1972, pp. 445-459.

5.8. Holmberg, Ake, “Crack Width Prediction and Min-imum Reinforcement for Crack Control,” Dansk Selskabfor Byaningsstatik (Copenhagen), V. 44, No. 2, June 1973,pp. 41-50.

5.9. Rehm, Gallus, and Eligehausen, Rolf, “LappedSplices of Deformed Bars Under Repeated Loadings(Ubergreifungsstosse von Rippenstahlen unter nicht ruhen-der Belastung),” Beton und Stahlbetonbau (Berlin), No. 7,1977, pp. 170-174.

5.10. Stevens, R. F., “Tests on Prestressed ReinforcedConcrete Beams,” Concrete (London), V. 3, No. 11, Nov.1969, pp. 457-462.

5.11. Broms, Bengt B., “Technique for Investigation ofInternal Cracks in Reinforced Concrete Members,” ACIJOURNAL, Proceedings, V. 62, No. 1, Jan. 1965, pp. 35-44.

5.12. Atimtay, Ergin, and Ferguson, Phil M., “EarlyChloride Corrosion of Reinforced Concrete - A Test Re-port,” ACI JOURNAL, Proceedings V. 70, No. 9, Sept. 1973,pp. 606-611.

5.13. Beeby, A. W., “Concrete in the Oceans - Crackingand Corrosion,” Technical Report No. 1, Cement andConcrete Association (London), 1978.

5.14. Beeby, A. W., “Corrosion of Reinforcing Steel inConcrete and Its Relation to Cracking,” The StructuralEngineer (London), V. 56A, No. 3, Mar. 1978, pp. 77-81.

5.15. Mather, Bryant, “Cracking Induced by Environ-mental Effects,” Causes, Mechanism, and Control of Crack-ing in Concrete, SP-20, American Concrete Institute, De-troit, 1968, pp. 67-72.

5.16. Mather, Bryant, “Factors Affecting Durability ofConcrete in Coastal Structures,” Technical MemorandumNo. 96, Beach Erosion Board, Washington, D.C., June1957.

5.17. Roshore, Edwin C., “Field Exposure Tests of Rein-forced Concrete Beams,” ACI JOURNAL , Proceedings V. 64,No. 5, May 1967, pp. 253-257.

5.19. Brookhaven National Laboratory, “Concrete Pol-ymer Materials,” BNL Report 50134 (T-5091, 1968.

5.20. Polymers in Concrete, SP-40, American ConcreteInstitute, Detroit, 1973, 362 pp.

5.21. Polymers in Concrete, SP-58, American ConcreteInstitute, Detroit, 1978, 420 pp.

5.22. Nawy, Edward G.; Ukadike, Maurice M.; andSauer, John A., “High Strength Field Polymer ModifiedConcretes,” Proceedings, ASCE, V. 103, ST12, Dec. 1977,pp. 2307-2322.

6.1 - IntroductionA “layered” concrete system can be created by a

mortar or concrete overlay (topping) placed on anexisting concrete surface. The use of “layered” con-crete systems has been increasing during the last10 years in the renovation of deteriorating bridgedecks, strengthening and/or renovation of concretepavements, warehouse floors, walkways, etc., and innew two-course construction of decks and pave-ments. The overlay can be portland cement lowslump dense concrete (LSDC), polymer-portland ce-ment concrete (PPCC), more commonly referred toas latex modified concrete (LMC), fiber reinforcedconcrete (FRC), or internally sealed concrete. A “lay-ered” system can also be created by impregnatingthe upper portion [l/z to 3 in. (10 to 80 mm1 ] of exist-ing concrete with a monomer system that requirespolymerization after soaking.

The major sources and types of cracking in theselayered concrete systems are:

1. Differential shrinkage cracking2. Reflective cracking (stress cracking)3. Differential temperature cracking4. Edge curling and delamination5. Incorrect construction practicesLong term observations 6.1-6.3 of many “layered”

concrete systems have shown that differentialshrinkage cracks are by far the most common andmost likely to increase and widen with time.

6.2 - Fiber reinforced concrete (FRC) overlaysWhen properly proportioned, mixed, and placed, a

crack resistant topping layer of FRC can be the solu-tion to certain field problems. Fibrous concrete over-lays of highways, airfields, warehouse floors, walk-ways, etc., have been used since the early 1970s.Fibers are usually steel with lengths between 10and 60 mm (l/2 to 2l/2 in.). The effects of fibrousconcrete on cracking in a “layered” system dependlargely on the field conditions of each situation.Some typical observations for similar field or labo-ratory conditions are discussed below.6*2-6*76.2.1 Bond to underlying concrete - During early fi-brous concrete overlay work, it was thought that a“partially bonded” layer was the ideal system. Theterm “partially bonded” means that no deliberate at-tempt is made to bond or to debond the toppinglayer to the underlying material through agents, fas-teners, polyethylene sheet, etc. The surface to beoverlaid is cleaned of all loose material, usually byhosing, and generally left in damp condition. Afterthe evaluation of partially bonded projects, this pro-cedure has become the least desirable technique to

*Principal authors: Alfred G. Bishara and Ernest K. Schrader.

224R-24 ACI COMMITTEE REPORT

use. Over a period of several years many partiallybonded FRC overlays have shown noticeableamounts of reflective cracking and edge curling. Thecurled edges are typical in thin overlays [less thanabout 3 in. (76 mm)] and can result in cracks if sub-jected to long-term dynamic loading.

If the base slab is relatively crack free, or if theoverlay is of sufficient thickness and strength to re-sist the extension of cracks in the original slab, abonded layer with matched joints is generally thebest approach. If the FRC layer is of sufficient thick-ness, a totally unbonded overlay is generally bestwhere severe cracking is present or may develop inthe base slab. Essentially unbonded systems havebeen constructed satisfactorily where FRC is placedover an asphalt layer. The asphalt itself will act as adebonding layer if it has a reasonably smooth sur-face without potholes. This type of constructionlends itself particularly well to deteriorated airfieldslabs which have been resurfaced with asphaltic con-crete but require additional rigid pavement to takeincreased loads imposed by heavy aircraft. Anothertechnique, which has been used when the base mate-rial to be overlaid is reasonably smooth, consists ofplacing the FRC over a layer of polyethylene sheet.On irregular, spalled, or potholed surfaces a thin lev-eling and debonding layer of sand or asphalt is desir-able.

6.2.2 Fiber size and volume - The crack arrestingmechanism on which the basic theory of FRC isfounded depends on fiber spacing.6.8 Although fibersize and volume have little effect on the formation ofthe first crack they are major factors influencingsubsequent crack development. As fiber diameter in-creases for any given volume percentage, the num-ber of fibers decreases and the spacing between fi-bers increases. Also, as the volume percentagedecreases, the spacing increases. If the fiber spacingbecomes relatively large [more than about 5 mm (0.2in.)], the crack arresting mechanism is limited. Re-gardless of the reason, as the fiber spacing in-creases, the number of small cracks decreases, butthe number and width of larger cracks increase. Forconcrete with 20 mm t3/4 in.) aggregate, about 0.9percent fibers by total volume will provide sub-stantial crack resistance. For concrete with 10 mm(3/8 in.) aggregate about 1.2 percent is normal, andfor mortar, 1.4 to 1.8 percent is adequate. If fibercontents much greater than these are used, or if ag-gregate gradations are not suitable, high cement andwater requirements result and the FRC layer is sus-ceptible to shrinkage cracks.

6.2.3 Fiber type and shape - Because of their in-creased resistance to pullout, deformed steel fibershave an advantage over smooth ones with regard toboth pre- and post-cracking behavior. However, theadvantage is not always worth the additional ex-pense.

The basic crack theory is applicable to both glassand metallic fibers, but the two types do exhibitsome difference in physical crack behavior. Test+*have shown that glass FRC has less ability to storeenergy after its failure in flexure than steel FRC.Also, microcracking in the general vicinity of a ma-jor crack is typically more prominent with steel thanglass. The failure (crack) zone for glass is more local-ized.6.2.4 Fibers in open cracks - There has been con-siderable discussion about the condition and effec-tiveness of steel fibers that bridge over or through acrack. At the time of cracking, the fibers lose theirbond to the concrete but continue to provide a “me-chanical resistance to pullout.” This post-crackingstrength is one of the most important characteristicsof FRC. The “obvious” problem is that after cracking,steel fibers will oxidize and provide no long-termbenefit. However, the majority of investigations6.3,6.5,6.6 have shown, that if the cracks are tight(0.001 - 0.003 in. (0.03-0.08 mm)], the fibers will notoxidize, even after several years of exposure. Long-term evaluations are currently underway.6.3 6.2.5 Mix proportion conditions-ACI 544.3Rprovides detailed information on suitable mixtureproportions for steel fiber reinforced concrete. Thewater requirement for fibrous concretes is higherthan that of normal concrete due to the high surfacearea of the fibers. The high water content providesthe basic ingredient for shrinkage cracks. Throughthe use of water reducing admixtures, the mix watercan be held to reasonable levels.6-gp ‘JO If possible,these admixtures should be used to adjust the mixproportioning for a bonded overlay so that the wa-ter/cement ratio and cement factor approach thesame values as used in the underlying material, Ifpossible, the overlay should have aggregates ofsimilar physical properties unless the original ag-gregates are unsuitable.

6.2.6 Joint overlays - Different methods of jointoverlaying have been tried; most have been unsuc-cessful.6.7 As with conventional concrete overlays, ifjoints in a base slab are overlayed with FRC withouttaking special design precautions to prevent reflec-tive cracking, the overlay will crack at joint loca-tions.

6.3 - Latex modified concrete (LMC) overlaysLatex modified mortar and concrete bonded over-

lays [3/4 to 1 l/z in. (20 to 40 mm)] have been used inthe renovation of deteriorated bridge decks and innew two-course construction to effectively resist thepenetration of chloride ions from deicing salts andprevent the subsequent corrosion of the reinforcingsteel and the spalling of the concrete deck.6.11,6.12 Some of these decks have been in use for over 10years.

Inspections of a large number of bridge decksoverlaid with LMC6.1 have indicated that there is a

CONTROL OF CRACKING 224R-25

high incidence of fine, random, shrinkage cracks in alarge portion of the renovation jobs. This type ofcracking is not as extensive in new two-course con-struction. Transverse cracks, spaced 3 to 4 ft (0.9 to1.2 m) apart, also appear on many of the bridges in-spected. However, there may be a relationship be-tween the degree of transverse cracking and the in-tensity of heavy truck traffic during reconstruction.To keep the bridges in service, traffic is normallydiverted to one lane, while renovation and applica-tion of the overlay proceed on an adjacent traffic lane.The quality of the overlay may be affected by themovement of the deck, although extensive data donot exist linking the effect of traffic-induced vibra-tions during reconstruction to deterioration or crack-ing in bridge decks. If traffic must be maintained,consideration should be given to placing overlayswhen traffic is low and/or when vehicle speed isrestricted.

To reduce the incidence of cracking and subsequent loss of latex modified concrete overlays it isrecommended 6 .1

that:

1. The surface of the underlying concrete shouldbe cleaned by sand blasting to assure adequatebonding with the overlay. To reduce air pollution,particularly in urban areas, high pressure water jetcleaning [5000 to 6000 psi (35-40 MPa) at the nozzle]may be used just prior to placement of the overlay,in lieu of sand blasting;

2. The slump of latex modified concrete mixturesshould be between 3 to 4 in. (75 to 100 mm) to re-duce differential shrinkage and the high incidence ofrandom cracking;

3. The finishing equipment should have beenproven to be effective for adequately placing the con-crete to the required density;

4. A thin coating of the overlay mixture should bethoroughly scrubbed into the surface of the under-lying clean concrete immediately before placing theoverlay mix to increase the bonding between thelayers; coarser particles of the mixture which cannotbe scrubbed into immediate contact with the surfaceof the underlying concrete, should be removed;

5. In new two-course construction, the overlayshould be placed after removing the forms from thebase concrete, so that stresses caused by the weightof the overlay are born by the underlying concrete.If placed before the forms are removed, the overlaywill have to carry a portion of its own weight andmay crack in negative moment regions;

6. Overlays should be placed only when the am-bient weather conditions are favorable, as defined inACI 308 on curing, or when appropriate actions aretaken for cold-weather concreting (ACI 306R) or hot-weather concreting (ACI 305R).

6.4 - Polymer impregnated concrete (PIC) systemsSurface impregnation and polymerization of con-

crete in place is a relatively new process but hasbeen used successfully in a number of field proj-ects.6.13,6.15 There has been considerable discussionabout this procedure due to observations of cracksduring or immediately after the drying step of thesep r o j e c t s . I n t h e c a s e s t h a t h a v e b e e n eval-uated,6.14,6.15 the cracks were determined to eitherhave been in the concrete prior to the impregnationor they were caused by improperly controlled dryingduring initial stages of the impregnation procedure.Temperatures during drying are usually in the rangeof 120 C (240 F) to 150 C (310 F) for about 4 to 12 hr.To some extent, thermal expansion will offset dryingshrinkage until the concrete cools. Ideally, duringthe soak period and after cooling, the monomerfill any cracks that have been created in the topface of the concrete due to drying. The cracksbe mended when the monomer is polymerized.crack is open and can drain (as is the case withtical surfaces and cracks through the full depth

willsur-willIf aver-of a

slab), the monomer can run out of the crack before itis polymerized, and no mending will occur. If a moreviscous monomer is used, so that it does not drainfrom the crack, the depth of penetration into theconcrete will be adversely affected. If there is a wa-ter source behind the material to be polymerized itis possible for moisture to re-enter the crack, afterdrying has been completed, but before the monomersoak starts. In this case, the presence of moistureprevents the monomer from entering the concreteadjacent to the crack, and the crack will not mend.

The engineer should thoroughly evaluate all ef-fects of the drying cycle in a PIC project and planthe drying temperatures and duration, the coolingcycle, and the monomer system to prevent the oc-currence of unmended cracks. The strain capacity,thermal expansion, and specific heat of the materialshould be considered. Restraints, preventing move-ment at the perimeter of the concrete to be poly-merized, should be avoided.

The long-term influence of polymer impregnationon the behavior of cracking in concrete is not knownat this time but will be established by the evaluationof currently completed field projects.

References6.1. Bishara, A. G., “Latex Modified Concrete Bridge

Deck Overlays - Field Performance Analysis,” ReportNo. FHWA/OH/79/004, Federal Highway Administration,Washington, D.C., Oct. 1979, 97 pp.

6.2. Gray, B. H., “Fiber Reinforced Concrete - A Gen-eral Discussion of Field Problems and Applications,” Tech-nical Manuscript M-12, U.S. Army Construction Engineer-ing Research Laboratory, Champaign, Apr. 1972.

6.3. Schrader, Ernest K., and Munch, Anthony V. “Deck

224R-26 ACI COMMITTEE REPORT

Chapter 7 - Control of cracking in massconcrete*

Slab Repaired by Fibrous Concrete Overlay,” Proceedings,ASCE, V. 102, C01, Mar. 1976, pp. 179-196.

6.4. Gray, B. H.; Williamson, G. R.; and Batson, G. B.,“Fibrous Concrete - Construction Material for the Seven-ties,” Conference Proceedings M-28, U.S. Army Construc-tion Engineering Research Laboratory, Champaign, May1972, 238 pp.

6.5. Hefner, S., “Fibrous Concrete McCarran Inter-national Airport,” Las Vegas, Nevada, Dec. 1974.

6.6. Rice, John L., “Fibrous Concrete Pavement DesignSummary,” Technical Report No. M-134, U.S. Army Con-struction Engineering Research Laboratory, Champaign,June 1975, 13 pp.

6.7. Gray, B. H., and Rice, John L., “Fibrous Concretefor Pavement Applications,” Report No. M-13, U.S. ArmyConstruction Engineering Research Laboratory,Champaign, Apr. 1972, 9 pp.

6.8. Shah, S. P., and Naaman, A. E., “Mechanical Prop-erties of Glass and Steel Fiber Reinforced Mortar,” De-partment of Materials Engineering, University of Illinois,Chicago, Aug. 1975.

6.9. “Utilization of ‘Wirand’ Concrete in Bridge Decks,”Report by General Analytics, Monroeville, Pa., for BattelleMemorial Institute, May 1971.

6.10. Walker, A. J., and Lankard, D. R., “Bridge DeckRehabilitation with Steel Fibrous Concrete,” Presented atthe Third International Exposition on Concrete Construc-tion (New Orleans, Jan. 1977), Battelle Columbus Labora-tories, 1977.

6.11. Bishara, A. G., and Tantayanondkul, P., “Use ofLatex in Concrete Bridges Decks,” Report No. EES 435(ODOT-12-74) Ohio Department of Transportation, TheOhio State University, 1974.

6.12. Clear, K. C., “Time to Corrosion of ReinforcingSteel in Concrete Slabs,” Transportation Research Record,No. 500, Transportation Research Board, 1974, pp. 16-24.

6.13. Schrader, Ernest K.; Fowler, David W.; Kaden,Richard A., and Stebbins, Rodney J., “Polymer Impregna-tion Used in Concrete Repairs on Cavitation/Erosion Dam-age,” Polymers in Concrete, SP-58, American Concrete In-stitute, Detroit, 1978, pp. 225-248.

6.14. Depuy, G. W., “Recent Developments in Concrete-Polymer Materials,” Second International Symposium onConcrete Technology (Monterrey, Mexico, Mar. 19751, U.S.Bureau of Reclamation, Denver, 1975.

6.15. Smoak, W. G., “Polymer Impregnation of NewConcrete Bridge Deck Surfaces,” Interim Report No.FHWA-RD-75-72, U.S. Bureau of Reclamation, Denver,Prepared for Federal Highway Administration, Washing-ton, D.C., June 1975.

7.1 - IntroductionTemperature induced cracking in a large mass of

concrete can be prevented if proper measures aretaken to reduce the amount and rate of temperaturechange. Measures commonly used include precooling,post-cooling or a combination of the two, and more

*Principal authors: Donald L. Houghton and Roy W. Carlson.

recently, thermal insulation has been used to protectexposed surfaces. The degree of temperature controlnecessary to prevent cracking varies greatly withsuch factors as the location, the height and thicknessof the structure, the character of the aggregate, theproperties of the concrete and the external re-straints. Although a large amount of the data forthis chapter has been obtained by experience gainedfrom the use of mass concrete in dams, it appliesequally well in mass concrete used in other struc-tures such as steam power plants, powerhouses,bridge and building foundations, navigation locks,etc. Tremie concrete, a specialized type of mass con-crete, has been amply covered in Chapter 8 of ACI304 and will not be discussed in this report.

The location of the structure affects the degree oftemperature control which will be required. Gener-ally at high altitudes the daily variations in temper-ature are greater than at low altitudes. Often athigh altitudes, the ambient temperature variationalone may be sufficient to cause cracks to form atexposed surfaces. These surface cracks continue in-ward with only approximately half the stress whichis necessary to cause internal cracking. A similarcondition is likely to be found when a structure is lo-cated at a high latitude; only in this case the temper-ature variations are seasonal, rather than daily.

In the case of a dam, the height affects the needfor crack control. If the dam is very high, the designstresses will be high and more cement must be usedto provide the stipulated factor of safety. Thismakes for more heat generation and a consequenttendency toward higher internal temperatures. Also,the higher dam will have greater horizontal dimen-sions which cause greater restraint and the need forstill closer temperature control.

The properties of the concrete affect the problemof crack control. Concretes differ widely in theamount of tensile strain they can withstand beforecracking. For strain which is applied rapidly, thetwo factors which govern the strain capacity are themodulus of elasticity and the tensile strength. Forstrain which is applied slowly, the creep (or re-laxation) of the concrete is important. The factors af-fecting strain capacity and creep rate are discussedmore fully in Section 7.2.

Another important property of concrete is thecoefficient of thermal expansion. The amount ofstrain which a temperature change will produce isdirectly proportional to the coefficient of thermal ex-pansion of the concrete. The average coefficient ofthermal expansion of mass concrete is about 9 mil-lionths per deg C (5 millionths/F), but with some ag-gregates, the coefficient may be as high as 15 mil-lionths or as low as 7 millionths (4 to 8 millionths/F).Thus, in the extreme case, where a concrete has alow tensile strength, a high modulus o f elasticity, ahigh coefficient of thermal expansion, and is fully re-strained, it may crack when there is a quick drop in

CONTROL OF CRACKING 224R-27

7.2 - Crack resistanceThe tensile strain which concrete can withstand

varies greatly with the composition of the concreteand the strain rate. When strain is applied slowly,the strain capacity is far greater than when the ac-tion is rapid. Thus, concrete in the interior of a largemass which must cool slowly, can undergo a largestrain before failure. If concrete contains rough tex-tured aggregate of small maximum size, the straincapacity will be high. However, there is an optimumwith respect to the aggregate size. Smaller aggre-gate requires more cement for a given strengthwhich results in more heat, a higher maximum tem-perature, and greater subsequent strain due to cool-ing. Thus, the gain through greater strain capacityof the richer concrete with smaller aggregate maybe more than offset by the greater strain that mustbe withstood, if the size is reduced too much.

temperature of only 3 C (6 F). On the other hand,some concretes can withstand a quick drop in tem-perature of as much as 10 C (20 F), even when fullyrestrained. More data on the thermal expansion ofconcrete may be found in the reports of ACI Com-mittee 207 (ACI 207.1R and ACI 207.2R).

From these considerations, it is apparent that thedegree of crack control necessary for the safe elimi-nation of joints may vary from nothing at all, for adam near the equator with favorable aggregates, tovery costly measures, in a location where temper-ature variations are great and where the only eco-nomical aggregates have high elastic moduli andhigh thermal expansion. In the latter case, presentpractice calls for both precooling and post-cooling,and for the application of thermal insulation to ex-posed surfaces during cold weather. The insulation isleft in place long enough to permit the concrete tem-perature at the surfaces to slowly approach the am-bient, or until additional concrete is placed on oragainst the surface being protected. Additional re-search into the most effective use of thermal in-sulation is needed particularly for regions having se-vere or sub-arctic climates.

There are two measures which can be taken toprovide safety against cracking. The first is to mod-ify the materials and mix proportions to produceconcrete having the best cracking resistance, or thegreatest tensile strain capacity. This may requirecareful aggregate selection, using the minimum ce-ment content for interior concrete, restricting themaximum aggregate size, or using other specializedprocedures. The second measure to prevent crackingis to control the factors which produce tensile strain.This may mean precooling, post-cooling, insulating(and possibly heating) the exposed surfaces of theconcrete during cold weather and designing to min-imize strains around galleries and other openings.

As stated above, the two factors governing thetensile strain which a concrete can withstand are thetensile strength and the modulus of elasticity. Manytests on very lean concretes, such as are used forthe interior of large dams, have shown that tensilefailure occurs without much “plastic” strain whenloading is applied rapidly. For such concrete, thetensile strain which the concrete can withstand isapproximately equal to the tensile strength dividedby the modulus of elasticity of the concrete. Formany purposes, then, it is sufficiently accurate to as-sume that the tensile strain capacity is inverselyproportional to the modulus of elasticity of the con-crete. It follows that the modulus of elasticity of theaggregate is important because of its large effect onthe deformability of the concrete. Tensile strength isalso important, and for this reason, crushed aggre-gates are apt to be superior to natural aggregatesfor crack prevention.

Strain capacity can be measured directly oncylindrical specimens loaded in tension, or it can bedetermined on concrete beams located at the thirdpoints.7.1

A high creep rate of concrete is helpful in pre-venting cracking when the tensile strain is appliedgradually. Since the tensile strength of concrete isnearly independent of prior loading, creep tends toincrease the strain capacity. In the case of DworshakDam, for example, the strain to failure was almostthree times as great for strain applied over 2 monthsas for quickly applied strain.7.1

The creep of concrete under sustained stress is af-fected by the stiffness of the aggregate. When themodulus is high, the creep is low and vice versa. Theimportance of aggregate rigidity on creep of con-crete may be illustrated by two examples. First, as-sume that the aggregate and the cement paste havethe same modulus of elasticity. When compressivestress is applied, the stress and the correspondingstrain will be the same in the aggregate as in the ce-ment paste. The aggregate does not creep undermoderate stress but the paste does, and the pastewhich is between aggregate particles relaxes andloses stress. The lost stress must be shifted to theaggregate to maintain equilibrium. This imposes anelastic strain on the aggregate which accounts for alarge part of the creep of the concrete. The amountof this elastic strain is directly related to the modu-lus of elasticity of the aggregate; the more rigid theaggregate, the lower the creep. Next, assume thatthe aggregate has a much higher modulus than thecement paste. When compressive stress is applied,the average stress in the aggregate will be higherthan that in the cement paste and the paste willcreep less than it did when the moduli were equal.The elastic strain in the aggregate due to the creepof the paste will then be less than it was when themoduli were equal. Thus, an increase in the rigidity

224-28 ACI COMMITTEE REPORT

of the aggregate acts in two ways to reduce thecreep of the concrete.

7.3 - Determination of temperatures and tensilestrains

Tensile strain in mass concrete results mainlyfrom the restraint of thermal contraction, and to alesser degree from autogenous shrinkage. Dryingshrinkage is important only because it may causeshallow cracks to occur at surfaces. Thus, temper-ature change is the main contributor to tensile strainin mass concrete. The prediction of probable strainrequires the prediction of the temperature to be ex-pected. This prediction can be made quite reliably ifthe adiabatic temperature curve for the concrete isknown, as well as the thermal diffusivity, boundarytemperatures and dimensions. The finite elementmethod can be used for the prediction of temper-ature distribution.7.3a 7.4 The main problem is that ofchoosing the correct boundary temperatures, whichoften depend upon the ambient temperatures. It isoften satisfactory to use air temperatures found inweather reports as the surface temperatures to beused in the computations. For information on othermethods of predicting temperatures in mass con-crete, see the report ACI 207.lR.

After the predicted temperature history is known,the determination of probable tensile strain is thenext step. This can be accomplished using finite ele-ment computer programs.7.5a 7.6 Even with the finiteelement method, a thorough analysis is laboriousbecause of the time-dependent variables. The analy-sis must include many steps of time to properlyaccount for the creep (or relaxation) and the differ-ent and changing properties of every lift of concrete.On the other hand, strains near a boundary due tobrief thermal shocks can be computed quite readilybecause in such cases the concrete can be assumed tobe fully restrained. In this case, the strain is simplythe temperature drop multiplied by the coefficient ofexpansion. This is important, because in manycases, the control of boundary strain is sufficient toprevent cracking. Internal strains usually developslowly enough to be tolerable, even if large. De-scriptions of test methods suitable for measuring thephysical properties necessary for the prediction oftemperatures and strains are given in Section 7.5.

7.4 - Control of crackingGiven the probable temperatures and strains, the

designer must determine what measures are mostpracticable to provide ample safety against cracking.The preventative measures will vary from nothingwhere weather and materials are favorable, to veryexpensive measures, where conditions are unfavor-able. Some of the conditions which facilitate crackprevention are:

1. Concrete with large tensile strain capacity.2. Small daily and seasonal temperature varia-

tions.3. Low cement content (permitted by low design

stresses).4. Cement of low heat generation.5. Short blocks.6. Slow rate of construction when no cooling is

used.7. Low degree of restraint, as with yielding foun-

dation, or in portions of structure well removed fromrestraining foundation.

8. High yearly average temperature.9. Absence of stress raisers, such as galleries.

10. Low casting temperature.

This list suggests many of the measures which canbe taken to prevent cracking. First, an attemptshould be made to produce a concrete with largetensile strain capacity. This may mean limiting themaximum aggregate size to a value somewhat belowthat which might be the most economical otherwise.Where several sources of aggregate are available ec-onomically, preference should be given to that whichyields best crack resistance; usually this will be acrushed material of low thermal expansion and lowmodulus of elasticity.

The heat producing characteristics of cement playan important role in the amount of temperature rise.ASTM Type II (moderate heat) cement should beused for mass concrete construction (Note: Type IV,low heat cement is, also, recommended, but is notreadily available). Pozzolans can be used to replacea portion of the cement to reduce the peak temper-ature due to the heat of hydration (207.2R). In somecases, up to 35 percent or more of the cement can bereplaced by an equal volume of a suitable pozzolanand still produce the same strength at 90 days or1 year. Some of the more common pozzolans used inmass concrete include calcined clays, diatomaceousearth, volcanic tuffs and pumicites and fly ash. Theactual type of pozzolan to be used and its appropriatereplacement percentage are normally determined bytest, cost, and availability.

The lowest practical cement content permitted bythe strength and durability requirements should beused to reduce the heat of hydration and the con-sequent thermal stresses and strains. More than thenecessary amount of cement is a detriment ratherthan an advantage.

In general, a reduction in the water content ofconcrete permits a corresponding reduction in the cement content. The concrete with less water andcement is superior in two important ways: it under-goes less temperature change and less dryingshrinkage. Minimum water content can be achievedby such measures as specifying powerful vibratorswhich permit low slump, by using a water-reducing

CONTROL OF CRACKING 224R-29

7.5 - Testing methods and typical data

agent, and by placing the concrete at a low temper-ature.

Precooling the concrete during its production andpost-cooling it with embedded pipe systems after itis placed are especially effective measures. Detailson pipe cooling are given in Section 7.6.

One measure which offers promise is that of plac-ing crack resistant concrete at boundaries (sides andtop of lifts). Even though the more crack resistantconcrete may be too costly to be used throughoutthe structure, it can be used to this limited extentwithout serious effect on economy. But thin layers ofconcrete next to the forms cannot be placed easilywith present-day construction methods, which makeuse of very large buckets. Therefore, it appearsmore promising to use precast concrete panels forforms and to leave these panels as a permanent partof the structure. These panels should be of goodquality for durability, and preferably lightweight soas to provide good thermal insulation. Since mostcracks originate at boundaries, this partial measuremay make the whole structure crack free. More in-formation on the use of precast panels for protectionof mass concrete can be found in ACI 347.1R.

Thermal insulation on exposed surfaces duringcold weather can protect concrete from cracking, ifenough insulation is used and it is left in place longenough. If the insulation is sufficient to allow slowcooling, the tensile strain need never exceed thedangerpoint. The concrete can relax as rapidly asthe tensile stress tends to develop, until finally,stable temperatures are reached. However, if theconcrete has a very slow relaxation rate (or creeprate) the amount of insulation and the long protectiontime required may make this measure impractical.

In extreme environments, where large amounts ofinsulation will be required during severely coldmonths, it may be necessary to remove the in-sulation in stages as the warmer months approach,Temperatures within the concrete just below the in-sulation should be allowed to slowly approach theenvironmental temperature. This is to prevent theoccurrence of thermal shock which could inducecracking at the surface with possible, subsequent,deeper propagation into the mass. Precautions mustbe taken against using too much insulation or leav-ing it in place too long, which could result in stop-ping the desired cooling of the interior mass, and, insome cases, cause the interior temperature to beginto increase again.

Insulation, as currently used for concrete, can beobtained in a variety of forms and materials havingpractical installed conductances ranging from 3.6 to0.5 kg cal/m’/hr/C (0.75 to 0.10 BTU/hr/sq ft/F). Itcan be obtained in semirigid board type panels, roll-on flexible rubber type material, and foamed spray-on material which becomes semirigid in place. Thesemirigid panels are usually installed on the inside

face of the forms. Temporary anchors embedded inthe newly placed lift of concrete retain the insulationon the concrete surface when the forms are lifted.The insulation is easily removed from the surfacewhen desired. Roll-on insulation is particularly appli-cable for use on horizontal lift joints. It is easy to in-stall and remove and can be reused many times.Spray-on insulation can be used on either horizontalor vertical surfaces. This type of insulation is partic-ularly useful for increasing the thickness and effec-tiveness of insulation already in place and for in-sulating forms. Experience has shown that insulationwhich permits transmission of light rays should notbe used because a temperature rise occurs betweenthe insulation and the concrete when the insulationis subjected to direct sunlight. Spray-on insulation oftimed longevity for frost protection of agriculturalplants and trees, also, appears to have potential forthe insulation of concrete lift joints during the activeconstruction season. This insulation can be formu-lated to disintegrate at a given time after appli-cation. Thus, it can be timed to remain effective onthe lift joints for approximately the period of timebetween successive placements and be easily re-moved by a final washing prior to placement of thenew lift. Precast panels made of low conductancelightweight concrete or regular weight concrete castwith laminated or sandwich layers of low con-ductance cellular concrete also are acceptable as ameans of insulating the interior concrete. The panelswould then serve as both forms and face concrete.

7.5.1 Adiabatic temperature rise - The temperaturerise which would occur if there were no heat loss isdefined as adiabatic temperature rise. The reader isreferred to ACI 207.1R for methods of test. That re-port gives data on adiabatic temperature rise of con-cretes having a single cement content but havingdifferent types of portland cement. Fig. 7.1 gives

typical adiabatic curves for Type II cement and var-ious quantities of cement and pozzolan. Curves Aand B in Fig. 7.1 represent data from mixes contain-ing equal volumes of cementitious materials (ce-ment plus pozzolan) thereby showing the effect ofpozzolan replacement of cement on temperaturereduction.7.5.2 Thermal properties of concrete - Thermal dif-fusivity and thermal expansion are important in thecontrol of cracking due to temperature change, andtheir determination is detailed in References ACI207.1R and 7.8 through 7.10. The approximate range of thermal properties is shown in Table 7.1. 7.5.3 Creep of concrete - Creep may be defined asthe continued deformation of concrete under sus-tained stress. A standard test for creep of concretein compression is detailed in ASTM C 512-76.’ l5Creep of concrete in tension is difficult to measure;

224R-30 ACI COMMITTEE REPORT

50

40

IO

0

- A

c _,1

DI

LEGEND .

Curve A - Portland Cement l 306Ib/cu yd(l8l kg/~);Pozzolan-None

Curve B - Portland Cement l 214 Ib/cuyd (127 kg/m3);l%zzoIon-74lb/cuyd(44kg/m)

Curve C - Portland Cement . I81 Ib/cu yd (107kg/m3);Pozzolan-63Ib/cuyd (37 kg/m)

Curve D - Portland Cement l 148 lb/w yd (88 kg/d); PDZroh-50 Ibhu yd(30 kg/m)

Type II Cement

_ 200

-5

- 0

0 4 8 I2 I6 20 24 28

Age , Days

Fig. 7.1 - Typical adiabatic temperature curves for mass concrete (Reference 7.7)

TABLE 7.1 - Illustrative range of thermal and elasticproperties of mass concrete

__-___-____~_-__~_--__-Coefficient of linear

expansion, millionths

Per O F Per O C _ --__ -_4 7.2

to to8 14.5-_. _~____ _ _. _ ~_----_ ----- --

Thermal properties

Conductivity__ _____--_-

_----__ft x hr x O F m x hr x O C-_-

- -

Diffusivity Specific heat-__-----

l---- I

pppBTU/lb O F

ft’ orhr Cal/g O C

- - -~-~0.040

to 0.220.067____------- __~--~-- - -

Elastic properties- ____- -_- -_------ ~_~_____~---_~-----_Static modulus of elasticity (E) for age of test indicated_ __--__~-~-_--_ _----____-----p,

1 day 3 days 7 days 28 days 90 days---___-_--- -_- _- --_----_, __~---_.----~- ___--- __YPWpsi kg/cm2 psi kg/cm’ psi kg/cm’ psi kg/cm’ psi, kg/cm’ Poisson’s

x 10-6 x 10-3 x 10-6 x 10-3 x 10-6 x 10

-3 x 10-6 x10-3 x 10-6 x 10-3 Ratio

-_----_L--_----_.~-___- --_-- - - - - - _,__---_ ~-_.__--_.__---0.15

0.66 46.4 2.00 141 2.56 180 4.00 281 5.00 352 to0.25

1

CONTROL OF CRACKING 224R-31

thus, creep as measured in compression is assumedto apply to tension as well. Such an assumption canbe considered as reasonable when the stress is low.When the stress exceeds about 60 percent of the ul-timate and microcracking occurs, not only does theinstantaneous deformation increase, but the rate ofcreep increases, also. However, since the measuredstrain in a beam which is gradually loaded from theage of 1 month, to failure-at about 3 months, is onlyabout 10 percent more than that computed usingcreep data as obtained from similar concrete in com-pression, it appears permissible to apply com-pression creep data to concrete stressed in tensionin cases where approximate results will suffice.

Creep of concrete is measured on carefully sealedspecimens stored at a constant temperature andloaded to a constant stress. The measurement is usu-ally made by means of embedded strain meters, al-though any reliable method of measuring strain canbe employed. Butyl rubber is satisfactory for sealingthe specimens, but neoprene should be avoided be-cause it allows some moisture to escape. Specimensshould be loaded at the same ages as specified forthe modulus of elasticity tests, but loading at theearly age of 1 day is not always practical. Again, thespecimens should be large enough to permit concretevery nearly like that to be used in the structure.Cylinders of 9 x 18 in. (28 x 56 cm) size and with 3in. (76 mm) maximum sized aggregate or 6 x 16 in(15 x 40 cm) cylinders with 11/2 in. (38 mm) maximumaggregate are frequently used. The symposium oncreep of concrete,7.1’ gives useful coefficients for con-verting creep of smaller aggregate concrete to creepfor mass concrete. Fig. 7.2 shows typical creep data

2.8T o t a l Strain 1

2 . 4 E=l481+00547~ (I+11 ---Icqc =0.521+0 0 7 0 0 LOG. (1+1) -3OayrE =0384+0.0579 L O G . (1+1) -7Dg

2 . 0E =0.231+0.0500 Log (1+1) b -28asDays 30E =0.209+0.0294 LOG. (1+1 ) -9OCOyS

1 Day “EY2

1’ 20 ;

1 . 2

0

I”

0 . 6

0 .5

0 . 4

Time,(t+l) Days

Specific Creep Only

Fig. 7.2 - Typical concrete creep curves for massconcrete.

obtained from laboratory investigations.7.1” Table7.2 illustrates important computations that can be

made using the data in the Fig. 7.2. Shown in Table7.2 are values for sustained modulus of elasticity E,which in turn are used to develop tensile stress

coefficients per degree temperature drop for the con-dition of full restraint. For example, concrete 2 daysof age loaded at age 1 day would have a sustainedmodulus of elasticity (E,) of l/1.5 = 0.66 psi x lo6(46.4 kg/cm2 x lo31 (see Fig. 7.2 and Table 7.2A),and if fully restrained would be stressed 0.66 x 5.5psi per F = 3.6 psi/F (0.46 kg/cm2/C) for each degreedrop in temperature (see Table 7.2B).

7.5.4 Modulus of elasticity - This subject is treatedin detail in ACI 304. Table 7.1 shows values of themodulus of elasticity of a particular concrete aftervarious ages of curing.

7.5.5 Autogenous volume change - Autogenousvolume change7.7, 7.13 is the expansion or contractionof the concrete due to causes other than changes intemperature, moisture or stress. Thus, it is a self-induced expansion or contraction. Expansion can behelpful in preventing cracks, but a contraction in-creases in tendency to crack. Autogenous volume

change is usually measured by strain metersembedded in concrete cylinders which are carefullysealed (to insure that there is no loss in moisture) andkept at constant temperature. Measurements arebegun as soon as the specimens are hardened andsealed, and continued periodically for months.7.5.6 Tensile strain capacity-- The tensile strain ca-pacity tests are generally performed on unreinforcedconcrete beams under third-point flexural loading.Relatively large beams ranging from 12 x 12 in. (30x 30 cm) to 24 x 24 in. (60 x 60 cm) in cross sectionand 64 to 130 in. (160 to 325 cm) long are generallyused.7.2 Strain capacity is determined from thesetests under rapid and slow loading to simulate bothrapid and slow temperature changes in the concrete.The loading rates are generally 40 psi (0.28 MPa) fi-ber stress per minute and 25 psi (0.17 MPa) fiberstress per week for rapid and slow loading tests, re-spectively. The strain for rapid loading can be mea-sured using either surface or embedded strain gagesor meters.7.1, 7.7 For long-term tests, embedded me-ters are best. The strain can also be determinedfrom deflection measurements. The concrete testbeam used for determining the strain capacityshould be protected during the test to prevent lossof moisture by wrapping it with an impermeable ma-terial. Testing should be conducted at a constanttemperature for maximum accuracy in measurement.Detailed test procedures can be found in References7.1 and 7.14 Fig. 7.3 shows the unit strain values

224R-32 ACI COMMITTEE REPORT

ITloading days

7.6 - Artificial cooling by embedded pipe systemsThe overall program for cooling concrete, includ-

ing important field control criteria, should be deter-mined during the design stage. Precooling concreteprior to placement is accomplished by a variety ofmethods, including cooling all ingredients of the mix

e

TABLE 7.2 - Illustration of computation of sustained modulusof elasticity (Es) and stress coefficients

A. Sustained modulusEs at age of concrete at time of loading, days

Time afterloading days

1 day

psi kg/cm2

x 10-6 - 10-3

0.68 47.60.66 46.20.64 44.80.63 44.1

psi kg /cm2

x 10-6 x 10-3

1.92 1341.76 1231.62 1131.35 95

3 days

t

7 days

psi Ix 10-6b t x10-3 x 10 -6 I x 10 -3

t t-- ------

2.61 183 I 4.33 I 3032.46 I 172 3.76 ’ 2632.15 151

I3.34 234

1.98 ! 139 2.991 I- 210._~

(1) Sustained modulus of elasticity IE / values are based on data given inFig. 7.2

E,z ______ ______ ~._______-___.___

unit elastic strain/psi + Vz specific creep for time of loading

R. Tensile stress coefficients for condition of full restraint and decreasing temperature

Age of concrete at time of loading

1 day 3 days 7 days I

- --- -- -- - imlb/in.‘/F kg/cm’/C TGkg,em’)C Ib/in.‘/F 1 kglcmJ/C lb/in.,,:”

0 3.7 0.47

~~g!~~/~

11.0 ’ 1.33 1 4 ; 1.81

i

’ 241 3.6

I

0.46 ’ 9.7 1.22 14 i 1.70 21I 3.00

2.603 3.5 0.45 8.9 1.12 12 I,7 3.5 0.44 7.4 j 0.94 I

1.5011 1.38

/ 18 I 2.3116 ! 2.08__

( 2 ) Coefficient of lineal thermal expansion of concrete assumed to be 5.5 millionths/F (9.9 millionths/C,

BEAM STRESS OF OUTER FIBERS I

Fig. 7.3 - Unit tensile strain versus beam stress(References 7.1 and 7.7).

versus beam stress at outer fibers for a typical labora-tory investigation.7.1* 7.11

In the preliminary studies of temperature and con-struction control plans for mass concrete projects,approximate methods for estimating tensile straincapacity under rapid and slow loadings given in Ref-erences 7.5 and 7.20 may be used.

and using small ice particles as a replacement ofpart of the mixing water. Post-cooling of concrete isaccomplished by circulating cool liquids (usually wa-ter) through pipes embedded in the concrete.

Studies made during the design stage will estab-lish such items as lift height, pipe spacing, watertemperature and rate of flow, acceptable rate oftemperature drop (for both rapid and slow drops),and approximate duration of cooling.

In general, the duration of cooling and the heat re-moved by the pipe cooling should be sufficient to in-sure that a secondary internal temperature rise inthe mass does not exceed the primary rise. It is,however, important that steep cooling gradients,which can result in cracking the mass, be avoided.This is particularly true in smaller masses where cir-culation of cooling water should be stopped when themaximum temperature has been reached and justbegins to drop. A vulnerable location in pipe coolingsystems is centered at the cooling coils where sharpgradients and cracking can be induced if terminationof cooling water circulation is not timely.

Resistance thermometers should be used in suf-ficient numbers to permit adequate monitoring andcontrol of the internal concrete temperatures.

Construction drawings should show basic pipe lay-out and spacing including minimum spacing, and thelayout at dam faces, transverse construction joints,interior openings and in sloping, partial, and isolatedconcrete lifts. A pipe layout for a typical concretelift is shown in Fig. 7.4.

CONTROL OF CRACKING 224R-33

8s 4@2'-O"

W

Multiply B y To Obtain

3 Inches 0 0 2 5 4 M e t e r s 8

wFeet 0 3 0 4 8 Meters W. ~_

COIL LIN FEET

PLAN ELEV. I I351 " = 30'-0" +_FlDw 47

Detail " B "

r-- -&--- Elev. 1140

K-1 Elev 1135

Section A-A

Fig. 7.4 - Typical cooling coil layout (Reference 7.11).

Fig. 7.5 - Schematic of embedded pipe cooling embedment system in massconcrete.

In most areas of the dam, a uniform spacing canbe maintained for the cooling pipe, but isolated areasalways exist in all dams which tend to result in aconcentration of pipes. These concentrations tend tooccur at the downstream face of the dam where in-lets and outlets to cooling pipes are located, adjacentto openings in the dam, and at isolated and slopinglifts of concrete. Proper planning will alleviate manyof the undesirable conditions that can result fromthese concentrations. For example, it must be deter-mined to what extent the cost saving procedure ofconcentrating cooling pipe inlets and outlets nearcontraction joints can be permitted at the face of thedam. Also, it must be decided if cooling pipes to iso-lated areas in the foundation and at openings such as

galleries can extend from the downstream face ofthe dam or if a vertical riser must be used.

For ease of installation, the pipe used for post-cooling should be thin wall tubing. Aluminum tubingis lightweight and easy to handle. However, break-down from corrosion inducing elements of the con-crete is a potential problem for aluminum pipe ifcooling activities must be carried on over a period ofseveral months. In this case, steel tubing is pre-ferred.

Compression type couplings are used because thinwall tubing cannot be threaded satisfactorily.

Surface connections to the cooling pipe should beremovable to a depth of 4 to 6 in. (102 to 152 mm) so

224R-34 ACI COMMITTEE REPORT

that holes can be reamed and dry packed when con-nections are removed.

Forms should be designed and constructed so thatshutdown of cooling activities is not necessary whenforms are raised.

Wire tiedowns embedded at the top of the con-crete lift at about 10 ft (3 m) spacing satisfactorilysecure the pipe during concrete placing.

Coils must be pressure tested for leaks at themaximum pressure they will receive from the cool-ing system prior to placing concrete. Pressure mustalso be maintained during concrete placement to pre-vent crushing and permit early detection of damage,should it occur.

After cooling is completed and the pipe is nolonger needed, it should be thoroughly flushed withwater at a high enough pressure to remove foreignmatter and grouted full with a grout mixture com-pensated for plastic shrinkage or settlement. Thegrout should remain under pressure until final set isattained.

Fig. 7.5 shows the schematic layout of a typicalpipe cooling system.

Sight flow indicators should be installed at the endof each embedded pipe coil to permit ready obser-vance of cooling water flow. In addition to regularobservance of flows, water temperatures and pres-sures and concrete temperatures should be observedand recorded at least once daily while the lift isbeing cooled.

The refrigeration plant for cooling water may becentrally located, or several smaller complete por-table plants may be used to permit moving the re-frigeration system as the dam progresses upward.Sufficient standby components, equal in capacity tothe largest individual refrigeration units should beprovided.

7.7 - Summary - Basic considerations for construc-tion controls and specifications

The construction controls and specifications formass concrete must be such that the structures willbe safe, economical, durable, and pleasing in appear-ance. Each of these requirements in turn affects thecrack resistance. Safety will be assured if the con-crete has sufficient strength and continuity (absenceof cracks). Economy will depend upon such featuresas the best choice of aggregates, adequate but notexcessive temperature control, low cement content,etc. Durability will depend upon the quality of theconcrete, exposure conditions, and freedom fromchemical reactions of a deteriorating nature. Pleas-ing appearance will come from good workmanship,freedom from cracks and stains, absence of leakageand leaching, etc. The importance of a com-prehensive materials test program to establish nec-

essary control prior to preparation of constructioncon t ro l s and spec i f i ca t ions canno t be over -emphasized.7.7.1 Safety7.7.1.1 Safety against crushing-concrete strength. Astrength should be specified which will provide anadequate factor of safety against crushing of the con-crete. The “nominal” factor of safety is merely thecompressive strength divided by the maximumstress to be expected in the structure. However, nei-ther the strength nor the maximum stress can be ac-curately determined. The strength is usually derivedfrom tests on cylindrical specimens which are notcompletely representative of the structure. The max-imum stress is usually taken as the design stresswhich is based upon assumed concrete properties.For such reasons, it is considered good practice touse a safety factor as high as three or four, meaningthat the strength should be three or four times theexpected maximum stress. The 90-day strength is of-ten used and is derived from tests of job cylinders.Since the cylinders are made from wet screened con-crete, the measured strength is corrected to a mass-concrete equivalent by applying a reduction factor ofabout 0.80 for typical conditions. For specific data onappropriate reduction factors, the reader should re-fer to the U.S. Burau of Reclamation, Concrete Man-ual, 8th Edition. 7.

The “factor of safety,” as defined above, is subjectto a number of additional factors which, more orless, balance one another. Since the averagestrength of the job cylinders is used, half of thetests will be weaker. The strength at 90 days is notthe ultimate strength. There can be a large gain af-ter 90 days depending upon the composition of thecement. However, even a “factor of safety” of threeis far more than enough to cover any likely differ-ences between plus and minus corrections.

For in te r io r concre te , the lowes t p rac t i ca lstrength should be specified so as to reduce the ce-ment content. This, in turn, will reduce the heat ofhydration and the consequent thermal stresses, thusincreasing the crack resistance of the concrete. Morethan the necessary amount of cement is detrimentalrather than advantageous.

7.7.1.2 Safety against sliding. Sound, uncracked con-crete provides a very large factor of safety againstsliding. However, hardened horizontal lift joints mayimpair the safety. Therefore, the specificationsshould require care in the preparation of lift surfacesand in the placement and compaction of concretethereon. Also, the lift surfaces should slope slightlyupward toward the downstream edge (in the case ofa dam) such that the downstream edge is higherthan the upstream edge. It is not necessary to use amortar layer on lift surfaces prior to the placementof the next lift.

7.72 Economy - Many factors which affect the

CONTROL OF CRACKING 224R-35

7.8. “Method of Test for Thermal Diffusivity of MassConcrete,” (CRD-C 37-73), Handbook for Concrete andCement, U.S. Army Corps of Engineers, Vicksburg, Dec.1973, 3 pp.

7.1. Houk, Ivan E., Jr.; Paxton, James A.; and Hough-ton, Donald L., “Prediction of Thermal Stress and StrainCapacity of Concrete by Tests on Small Beams,” ACIJOURNAL, Proceedings V. 67, No. 3, Mar. 1970, pp.253-261.

7.5. Sandhu, R. S.; Wilson, E. L.; and Raphael, J. M.,“Two-Dimensional Stress Analysis with IncrementalConstruction and Creep,” Report No. 67-34, StructuralEngineering Laboratory, University of California,Berkeley, Dec. 1967.

economy also affect crack resistance. For example,the least expensive aggregate may have bad thermalproperties and thus require expensive temperaturecontrol to prevent cracking. The aggregate whichmakes concrete of highest tensile-strain capacitymay increase the water requirement and, therefore,also the cement requirement, thus offsetting thebenefits of high strain capacity. Some of the factorswhich affect economy are discussed below.7.7.2.1 Selection of aggregate. Aggregate should bechosen that makes good concrete with the lowestoverall cost. If natural aggregate near the site hasunfavorable properties for crack prevention, crush-ing to increase crack resistance may be an economi-cal expedient because of the consequent saving intemperature control. When crushing is either advan-tageous or necessary, rock which has the most favor-able properties should be chosen. The rock shouldhave a low coefficient of thermal expansion, a lowmodulus of elasticity, and it should produce particlesof good shape and surface texture. All of these fac-tors are important in increasing the resistance of theconcrete to cracking.7.7.2.2 Aggregate size. The largest maximum size ofaggregate, up to approximately 6 in. (150 mm) in di-ameter, should be specified as can be placed prop-erly in the structure, except for concrete which mustresist high-velocity water flow. Larger aggregatepermits the use of less water and cement per cubicyard, resulting in savings in both the amount of ce-ment and the amount of temperature control neces-sary for required crack resistance.7.7.2.3 Water content. A reduction in the water con-tent of concrete permits a corresponding reductionin the cement content. The concrete with less waterand cement is superior in many ways: it undergoesless temperature change, less drying shrinkage, andas a result is more durable and crack resistant. Asindicated in Section 7.4, minimum water content canbe achieved by specifying adequately powerful vibra-tors which permit the use of low slump concrete, byusing a water-reducing agent when appropriate, andby producing and placing the concrete at low tem-perature.7.7.2.4 Use of pozzolan In most locations, good poz-zolans such as fly ash are available, and they can beused to replace a portion of the cement. This can re-sult in a considerable saving in cost, and possiblymore important, it can reduce the heat generationand improve the resistance against cracking. An-other advantage of using pozzolan is that when usedin adequate amounts, it reduces the expansion dueto reactive aggregates when such are encountered.The appropriate amount of pozzolan for a reactiveaggregate should be based upon test data obtainedwith the pozzolan and cement being used.7.7.3 Durability - Durability of concrete is closelyrelated to the exposure conditions. In tropical cli-

mates, for example, there may be no deterioratinginfluences acting on the concrete except that whichis subject to high-velocity water flow. For the mainstructure in such a case, any concrete which has therequired strength can be expected to last in-definitely, and the cement content should be keptlow to minimize heat generation and resultant poten-tial cracking.

Where the climate is severe, such that there ismuch freezing and thawing in winter, the water-ce-ment ratio of surface concrete should be kept lowerthan that necessary for strength alone. Air entrain-ment should be mandatory. For any concrete whichmight be subject to both alternations of freezing andwater pressure, the water-cement ratio should beless than 0.40 by weight. The effect of the richboundary concrete on thermally induced crackingwill be minimized by keeping the thickness of theboundary layer to a minimum, probably 2 ft (0.6 m)or less.7.7.4 Control of cracking - A detailed discussion ofthe control of cracking in massive structures hasbeen presented in this chapter. With proper plan-ning and execution, the procedures presented willserve as useful tools in developing a crack controlprogram for mass concrete structures.

References

7.2. Houghton, Donald L., “Determining Tensile StrainCapacity of Mass Concrete,” ACI JOURNAL, Proceedings V.73, No. 12, Dec. 1976, pp. 691-700.

7.3. Wilson, E. L., “The Determination of Temperatureswithin Mass Concrete Structures,” Report No. 68-17,Structural Engineering Laboratory, University ofCalifornia, Berkeley, Dec. 1968.

7.4. Polivka, R. M., and Wilson, E. L., “Finite ElementAnalysis of Nonlinear Heat Transfer Problems,” ReportNo. UC SESM 76-2, Department of Civil Engineering,University of California, Berkeley, June 1976.

7.6 Liu, Tony C.; Campbell, R. L.; and Bombich, A. A.,“Verification of Temperature and Thermal StressAnalysis Computer Programs for Mass ConcreteStructures,” Miscellaneous Paper No. SL-79-7, U.S. ArmyEngineer Waterways Experiment Station, Vicksburg,Apr. 1979.

7.7. Houghton, Donald L., “Concrete Volume Changefor Dworshak Dam,” Proceedings, ASCE, V. 95, P02, Oct.1969, pp. 153-166.

224R-36 ACI COMMITTEE REPORT

Chapter 8 - Control of cracking by correctconstruction practices*

7.9. “Method of Test for Coefficient of Linear ThermalExpansion of Concrete,” (CRD-C 39-55), Handbook forConcrete and Cement, U.S. Army Corps of Engineers,Vicksburg, 1939, 2 pp.

7.10. “Method of Test for Coefficient of Linear ThermalExpansion of Coarse Aggregate, Strain Gage Method,”(CRD-C 125-63), Handbook for Concrete and Cement, U.S.Army Corps of Engineers, Vicksburg, June 1963, 5 pp.

7.14. McDonald, J. E.; Bombich, A. A.; and Sullivan,B. R., “Ultimate Strain Capacity and Temperature RiseStudies, Trumbull Pond Dam,” Miscellaneous PaperC-72-20, U.S. Army Engineer Waterways ExperimentStation, Vicksburg, Aug. 1972.

7.11. Symposium on Creep of Concrete, SP-9, AmericanConcrete Institute, Detroit, 1964, 160 pp.

7.12. McCoy, E. E., Jr.; Thorton, H. T.; and Allgood,J. K., “Concrete Laboratory Studies, Dworshak (Bruce’sEddy) Dam, North Fork Clearwater River Near Orofino,Idaho: Creek Tests,” Miscellaneous Paper No. 6-613,Report 2, U.S. Army Engineer Waterways ExperimentStation, Vicksburg, Dec. 1964.

7.13. Houk, Ivan E., Jr.; Borge, Orville E.; andHoughton, Donald, “Studies of Autogenous VolumeChange in Concrete for Dworshak Dam,” ACI JOURNAL,

Proceedings V. 66, No. 7, July 1969, pp. 560-568.

7.15. Liu, Tony C., and McDonald, James E.,“Prediction of Tensile Strain Capacity of Mass Concrete,”ACI JOURNAL, Proceedings V. 75, No. 5, May 1978, pp.192-197.

7.16. Concrete Manual, 8th Edition, U.S. Bureau ofReclamation, Denver, 1975, 627 pp.

8.1 - IntroductionConstruction practices, as used in this chapter, in-

clude designs, specifications, materials, and mix con-siderations, as well as on-the-job construction perfor-mance. Before discussing control of constructionpractices which affect cracking, it is worthwhile tomention the basic cause of cracking. It is restraint.If all parts of the concrete in a concrete structureare free to move as concrete expands or contracts,particularly the latter, there will be no cracking dueto volume change.

Obviously, however, all parts of concrete struc-tures are not free, and inherently, cannot be free torespond to the same degree to volume changes. Con-sequently, differential strains develop and tensilestresses are induced. When these differential re-sponses exceed the capability of the concrete towithstand them at that time, cracking occurs. Thispoints to the importance of protecting new concretefor as long as practicable from the loss of moistureor a drop in temperature. These considerations mayresult in stresses capable of causing cracks at an

*Principal author: Lewis H. Tuthi l l .

early age but which might be sustained at greatermaturity. Preferably, concrete should have a hightensile strain-to-failure capacity. This is influencedgreatly by the aggregate, and a low modulus ofelasticity in tension is desirable.

8.2 - RestraintRestraint exists in many circumstances under

which the structure and its concrete elements mustperform. Typical examples will illustrate how re-straint will cause cracking, if the concrete is notstrong enough to withstand the tensile stresses de-veloped.8.21 - A wall or parapet anchored along its base tothe foundation or to lower structural elements lesssubject or responsive to volume change, will be re-strained from shrinking when its upper portionsshorten due to drying or cooling. Cracking is usuallyinevitable unless contraction joints (or at leastgrooves of a depth not less than 10% of the wallthickness on both sides, in which the cracks will oc-cur and be hidden) are provided at intervals rangingfrom one (for high walls) to three (for low walls)times the height of the wall.

8.2.2 - Exterior and interior concrete, particularlyin heavier sections, will change temperature or mois-ture content at different rates and to different de-grees. When this happens, the interior concrete re-strains the exterior concrete from shrinking, andtensile strains develop which may cause the exteriorto crack. This occurs when the surface cools, whilethe interior is still warm from the heat of hydration,or when the surface concrete dries faster than theinterior concrete. As noted earlier, it is often fea-sible to protect the surface for a time at early agesso that such stress-inducing differentials cannot de-velop before the concrete is strong enough to with-stand the strain without cracking.

8.2.3 - Acting similarly to the interior concrete inthe foregoing example, temperature reinforcementcan restrain the shrinkage of surface concrete, butmore and narrower cracks may result.

8.2.4 - Restraint will occur at sharp changes in sec-tion, since the effect of temperature change ordrying shrinkage will be different in the two sec-tions. If feasible, a contraction joint can be used torelieve the restraint.

8.2.5 - Restraint of flat work results from anchor-age of slab reinforcement in perimeter slabs or foot-ings. When a slab is free to shrink from all sides to-ward its center, there is a minimum of cracking.Contraction joints and perimeter supports should bedesigned accordingly (see Section 3.5.3).

8.2.6 - Wall, slabs, and tunnel linings placed againstthe irregular surface of a rock excavation are re-strained from moving when the surface expands orcontracts in response to changes in temperature or

CONTROL OF CRACKING 224R-37

8.3 - ShrinkageThe following sections discuss the major causes of

shrinkage, which is a key contributor to the forma-tion of cracks in concrete.

moisture content. As discussed in Section 8.2.1,closely spaced contraction joints or deep groovesmust be provided to prevent or hide the crackswhich often disfigure such surfaces. In tunnel lin-ings, the shrinkage in the first few weeks is primar-ily thermal, and the use of cold concrete (50 F or10 C) has reduced cracking materially. By the timedrying is significant, the concrete lining is muchstronger and better able to resist shrinkage crack-ing. However, circumferential cracks in tunnel lin-ings and other cast-in-place concrete conduits andpipe lines can be greatly reduced in number andwidth. As shown in the Bureau of Reclamation Con-crete Manual,8.1 this can be done if a bulkhead isused to prevent air movement through the tunnel,and shallow ponds of water are placed in the invertas soon as possible after lining, and left until thetunnel goes into service. If the tunnel carries water,there will be no further drying shrinkage. If it doesnot, the concrete will have become much stronger inthe humid environment and will be better able to re-sist shrinkage-induced tensile stresses.8.2.7 - The typical examples presented aboveclearly indicate that many crack control proceduresmust be considered by the engineer during design.While proper construction performance can contrib-ute a great deal (as will be discussed below), the con-tractor cannot be expected to utilize the best pro-cedures, unless these procedures are included in thedesigns and specifications on which the bid price isbased.

8.3.1 Effect of water content - The greater the wa-ter content of concrete, the more it will shrink ondrying. Such a hypothesis is clearly indicated in Fig.3.2, as well as in Reference 8.1. The use of the low-

est practical slump is important. Of major impor-tance is the selection of mix proportions that requirethe least amount of water per cubic yard for the de-sired concrete strength. This means avoiding over-sanded mixes (the richer the concrete, the coarserthe sand should be and the less there should be of itin the mix); using the largest maximum aggregatesize practical; using aggregate with the most favor-able shape and grading conducive to best work-ability; and using well-graded sand with a minimumof fines passing the l00-mesh and free of clay, suchthat its sand equivalent value is not less than 80percent AASHTO T176.

Contrary to common belief, increasing the cementcontent of concrete, per se, does not necessarilycause an increase in shrinkage. This is because thewater requirement of concrete does not changemuch with a change in cement content. Dryingshrinkage is proportional to water content (Fig. 3.2),

not cement content. Moreover, the reduction of theamount of fine aggregate to compensate for theadded cement, in accordance with correct principlesof concrete proportioning, will offset any tendency toincrease the water requirement.8.3.2 Surface drying - Surface drying will ulti-mately occur except when the surface is submergedor backfilled. It will cause shrinkage strains of up to600 millionths or more. The amount of shrinkagecracking depends on 1. how dry the surface concretebecomes, 2. how much mixing water was in the con-crete, 3. the character and degree of restraint in-volved, and 4. the extensibility of the concrete. Theextensibility represents how much the concrete canbe strained (stretched), without exceeding its tensilestrength and is the sum of creep plus elastic straincapacity. The latter is largely related to the composi-tion of the aggregate and may vary widely. Typi-cally, some concretes of highly quartzitic gravelshave a low strain capacity and a high modulus ofelasticity, while some concretes of granitic andgneissic aggregate have a high strain capacity and alow modulus of elasticity. Concretes having a lowstrain capacity are much more sensitive to shrinkagedue to drying (and to drop in temperature) and willbe subject to a greater amount of cracking.

Accordingly, as mentioned in connection with tun-nel linings and conduits, a prime objective of crackcontrol procedures is to keep the concrete wet aslong as feasible, so that it will have time to developmore strength to resist cracking forces. The impor-tance of this will vary with the weather and thetime of year. Cold concrete (below 50 F, 10 C) driesvery slowly, provided the relative humidity is above40 percent. At some depth, concrete loses moistureslowly, as shown in Fig. 3.5. Where surface dryingmay be rapid, more care must be devoted to uninter-rupted curing to get good surface strength. Crackingstresses will be further reduced by creep, if the sur-face is prevented from drying quickly at the end ofthe curing period. To accomplish this, the wet curingcover can be allowed to remain several days withoutwetting after the specified curing period (preferably7 to 10 days), until the cover and the concrete underit appear to be dry. If job conditions are likely to besuch that these measures will be worthwhile, theyshould be required in the specifications for the work.

8.3.3 Plastic shrinkage - Plastic shrinkage cracksoccur most commonly, and objectionably, in the sur-faces of floors and slabs when the ambient job condi-tions are so arid that moisture is removed from theconcrete surface faster than it is replaced by bleedwater from below. These cracks occur prior to finalfinishing and commencement of the curing process.As the moisture is removed, the surface concretecontracts, resulting in tensile stresses in the essen-tially strengthless, stiffening plastic concrete, thatcause short random cracks or openings in the sur-

224R-38 ACI COMMITTEE REPORT

face. These cracks are usually rather wide at thesurface but only a few inches in depth. The cracksgenerally range from a few inches to a few feet inlength and are a few inches to two feet apart.

Sometimes plastic shrinkage cracks appear earlyenough to be worked out in later floating or firsttrowelling operations. When this is successful, it isadvisable to postpone these operations as long aspossible to get their maximum benefit without therecurrence of cracking.

In other cases, an earlier than normal floating maydestroy the growing tension by reworking the sur-face mortar and prevent plastic cracking that wouldotherwise occur. At the first appearance of crackingwhile the concrete is still responsive, a vigorous ef-fort should be made to close the cracks by tampingor beating with a float. If firmly closed, they will bemonolithic and are unlikely to reappear. However,they may reappear if they are merely trowelledover. In any event, curing should be started at theearliest possible time.

Conditions most likely to cause plastic shrinkagecracking are high temperatures and dry winds. Ac-cordingly, specifications should stipulate that effec-tive moisture control precautions should be taken toprevent a serious loss of surface moisture undersuch conditions. Principal among these precautionsare the use of fog (not spray) nozzles to maintain asheen of moisture on the surface between the finish-ing operations. Plastic sheeting can be rolled on andoff before and after floating, preferably exposingonly the area being worked on at that time. Leasteffective but helpful are certain sprayed mono-molecular films which inhibit evaporation. Wind-breaks are desirable, and as such, it is desirable toschedule flatwork after the walls are up (ACI 305R,ACI 302.1R).

Other helpful practices that may augment thebleeding and counteract the excessive loss of surfacemoisture, are 1. using a well dampened sub-grade, 2.cooling the aggregates by dampening and shadingthem, and 3. using cold mixing water or chipped iceas mixing water to lower the temperature of thefresh concrete.

8.3.4 Surface cooling - Surface cooling will shrinkthe surface of average unrestrained concrete about10 millionths for each deg C (5.5 millionths per degF) the temperature goes down. This amounts to 9mm in a 30 m length with a drop of 30 C (l/3 in. in100 ft with a drop of 50 F). The amount of shrinkageis reduced by restraint and creep, but tensilestresses are induced. The earlier the age and theslower the rate at which cooling or drying occur, thelower the tensile stresses will be. This is due to therelaxing influence of creep, which imparts more ex-tensibility to concrete at early ages.

In ordinary concrete work, the winter protectionrequired for the development of adequate strengthwill prevent the most critical effects of cooling. Thesystem of contraction joints and grooves previouslydiscussed for control of shrinkage cracking will servethe same purpose against substantial later drops insurface temperature. In addition to Chapter 7 ofthis report, Chapters 4 and 5 of ACI 207.1R discusstemperature controls for mass concrete to minimizethe early temperature differences between interiorand exterior concrete. Primarily, these controlslower the interior temperature rise caused by theheat of hydration by using 1. no more cement thannecessary, 2. pozzolans for a portion of the cement, 3.water reducing admixtures, 4. air-entrainment, 5.large aggregate, 6. low slump, and 7. last but by nomeans least, where at all practicable, chipped ice formixing water to reduce the temperature of the freshconcrete as much as possible. See Fig. 3.4 and Fig.3.1 of ACI 207.2R. At no time should forms beremoved to expose warm surfaces to low tempera-tures. As mentioned in Section 8.3.2, the extensibil-ity, or strain the concrete will withstand beforetensile failure, is a function of the aggregate andshould be evaluated, especially on larger projects.What applies to one will not necessarily apply toanother.8.4 - Settlement

Settlement or subsidence cracks develop whileconcrete is in the plastic stage, after the initial vi-bration. They are not due to any of the causes dis-cussed above, but are the natural result of heavysolids settling in a liquid medium. Settlement cracksoccur opposite rigidly supported horizontal re-inforcement, form bolts or other embedments. Some-times concrete will tend to adhere to the forms. Acheck will appear at these locations, if the forms arehot at the top or are partially absorbent. Cracks of-ten appear in horizontal construction joints and inbridge deck slabs over reinforcing or form bolts withonly a few inches cover. The cracks in bridge deckscan be reduced by increasing the concrete cover.8.2Properly executed late revibration can be used toclose settlement cracks and improve the quality andappearance of the concrete in the upper portion ofsuch placements, even though settlement has takenplace and slump has been lost.

8.5 - ConstructionA great deal can be done during construction to

minimize cracking, or in many cases to eliminate it.But, as noted in Section 8.2.7, such actions must berequired by the specifications and by the engineer-ing forces which administer them. Such actionsinclude the following:8.5.1 Concrete aggregates - The aggregate shouldbe one which makes concrete of high strain capacity,if reasonably available (see Section 7.2). Fine and

CONTROL OF CRACKING 224R-39

coarse aggregates have to be clean and free of un-necessary fine material, particularly clays. The sandshould have a sand equivalent value in excess of 80percent, and this should be verified frequently(AASHTO T176). The sand should have sufficienttime in storage for the moisture content to stabilizeat a level of less than 7 percent on an oven-dry basis.8.5.2 Expansive cement - Expansive cement can beused to delay shrinkage during the setting of con-crete in restrained elements reinforced with the min-imum shrinkage steel required by ACI 318. Theprincipal property of these cements is that theexpansion induced in the concrete while setting andhardening is designed to offset the normal dryingshrinkage. With correct usage (particularly withearly and ample water curing on which maximumexpansion depends), the distance between joints cansometimes be tripled without increasing the level ofshrinkage cracking. Details on the types and correctusage of shrinkage compensating cements are givenin ACI 223-83.

8.5.3 “Non-shrink” grout, mortar, or concrete - Ordin-arily, the solids in grout, mortar, and concretemixtures will settle before hardening, and water willrise, some of it to the top surface. This settlementcan be objectionable if a space is to be filled uptightly without leaving a void at the top, such as un-der machine bases. Measures taken to prevent suchsubsidence have produced what is known in thetrade as “Non-shrink” grout, mortar, or concrete.Some of the materials merely prevent settlement;others in addition, provide a slight expansion as themixture hardens.

The most widely used materials contain unpolishedaluminum powder. These should contain no stearates,palmitates, or fatty acids. In an alkaline solution, suchas exists in portland cement mixtures, the aluminumreacts to form aluminum oxide and hydrogen. Thehydrogen gas tends to expand the mixture and thusprevents subsidence and may even cause expansion.The amount of aluminum powder used varies widelywith conditions, but is usually in the neighborhood of0.005 to 0.01 percent by weight of the cement. It isnot possible to specify an exact percentage becausethe amount to be used varies with such factors astemperature, alkali content of the cement, and therichness of the mix. Therefore, it is advisable to maketrial mixes with various percentages of aluminumpowder to find which percentage gives the desired(slight) expansion under the prevailing conditions.The amount of aluminum powder used is so small thatit is advisable to dilute it by blending with 50 partsof sand or fly ash. This diluted mixture will haveenough bulk so that it can be easily measured andproperly dispersed in the mix.

Among the admixtures that merely preventment, a number of different mechanisms are

settle-in op-

eration. One commercial grout is so highly acceler-

ated that it starts setting before settlement takesplace. Another is composed of organic gelling com-pounds of soluble cellulose which increase in viscosityso that the solid particles remain in suspension. Stillanother contains a form of carbon with a very largesurface area. In the dry form, it contains a largeamount of adsorbed air, which is released graduallyinto the mix producing an expansion.

Gas forming agents and air releasing agents pro-duce the same net effect, although all grouts, mortarsand concretes employing these agents have no ex-pansive properties after hardening, and have adrying shrinkage at least equal to similar plaingrouts, mortars and concretes not employing them.Grouts which expand (if unconfined) after hardeningcan function as nonshrink grouts, as opposed togrouts that expand only in the plastic state and latersuffer drying shrinkage.

Among the commercial admixtures, there is onecontaining a metallic aggregate which, in addition toopposing settlement during hardening, provides amodest expansion after hardening. This acts to holdthe grout tightly up under base plates, etc., and alsotends to offset the effect of drying shrinkage.

Where feasible, the problem of settlement can besolved by the use of dry tamped mortar, instead ofa fluid grout or mortar. Grout mixed in a colloid millwill not readily settle.

It should be noted that prepackaged “Non-shrink”grouts, like any portland cement grouts and mor-tars, are subject to shrinkage if exposed to dryingand may deteriorate and lose serviceability if ex-posed to an aggressive environment (weathering,salt spray, etc.).

8.5.4 Handling and batching - Should be done withall practical care to avoid contamination, overlap ofsizes, segregation, and breakage, so tha t ex t raamounts of fines are not needed in the mixes to ac-count for variations in grading without a serious lossof workability. This is best done by finish screeningand rinsing as a combination of coarse aggregatesizes goes to the batch plant bins. Every effortshould be made to uniformly batch and mix the con-crete so that there will be a minimum of trouble-some variation in slump and workability. These, in-variably, lead to demands for a greater margin ofworkability, with more sand and more water in theconcrete.

8.5.5 Excessive workability - Whether it is achievedwith unneeded higher slump, oversanding,small aggregate, or even higher air content (whichmay reduce strength), is always popular and in de-mand on the job. It must be discouraged if the bestconcrete for the work (having adequate workabilitywith proper handling and vibration, and having min-imum shrinkage factors) is to be obtained.

8.5.6 Cold concrete- Cold concrete, when com-

224R-40 ACI COMMITTEE REPORT

8.6 - Specifications to minimize drying shrinkageActions during construction to obtain the lowest

possible drying shrinkage must be supported by thespecifications. Unless bids are taken on this basis,the contractor cannot be expected to provide otherthan ordinary materials, mixes, and procedures. Thefollowing items should be carefully spelled out in thespecifications.

bined with factors to reduce water and cementcontent to a practical minimum will reduce temper-ature differentials which cause cracking. Cold con-crete is particularly useful for massive concretes. Itrequires less mixing water and thus reduces dryingshrinkage. In warm weather it expedites the workby reducing slump loss, increasing pumpability, andby improving the response to vibration. It is ob-tained by substituting chipped ice for all or a part ofthe batched mixing water. In cold weather, concreteis naturally cold and every effort should be made touse it as cold as possible without inviting damagefrom freezing. It is pointless to expect to protectsurfaces, edges, and corners by placing needlesslywarm concrete in cold weather. These vulnerableparts must be protected with insulation or protectiveenclosures (ACI 306R).8.5.7 Revibration - When done as late as theformed concrete will respond to the vibrator, willeliminate cracks and checks where something rigidlyfixed in the placement prevents a part of the con-crete from settling with the rest of it. Settlementcracks are most apparent in the upper part of walland column placements where revibration can bereadily used. Deep revibration corrects crackscaused by differential settlement around blockoutand window forms, and where slabs and walls areplaced monolithically.8.5.8 Finishing - Flatwork finishing can make agreat difference in the degree of freedom from alltypes of cracking (ACI 302.1R). Low-slump concreteshould be used. More than a 3 in. (76 mm) slump israrely necessary except perhaps in very hot weatherin which both slump and moisture are lost quiterapidly. Finishing should not be done in thepresence of surface water. Precautions (see Section8.3.4) should be taken to prevent plastic shrinkage.Any required marking and grooving should becarefully cut to the f u l l depth specified. Curingshould be prompt, of full duration, and the wet covershould be allowed to dry before it is removed.8.5.9 Curing and protection - Newly placed con-crete must be brought to a level of strength maturityand protected from low temperatures and dryingconditions which would otherwise cause cracking.The curing and protection should not be discontin-ued abruptly. If the new concrete is given a few daysto gradually dry or cool, creep will have an opportu-nity to reduce the possibility of cracking when thecuring and protection are fully discontinued.8.5.10 Miscellaneous - Some items normally cov-ered in specifications (or certainly which should becovered where appropriate) require special attentionduring construction because of their potential effectson cracking.

1. Reinforcement and embedments must be prop-erly positioned with the designated thickness ofcover in order to prevent corrosion, expansion andcracking.

2. Concrete should not be placed against hot re-inforcement or forms.

3. Formwork support should be strong enough tobe free of early failures and distortion causing crack-ing.

4. Subgrade and other supports must not settleunevenly, to prevent cracks due to overstress in thestructure.

5. Contact between aluminum and steel embeddedin the concrete must be eliminated, particularly ifuse of calcium chloride is permitted. If it is used, cal-cium chloride must be limited to the absolute min-imum (see Section 3.4.4).

6. Special care is needed in handling precast unitsto prevent overstress due to handling.

7. Unvented salamanders in cold weather (ACI306R) or gasoline operated equipment must beavoided where adequate ventilation is notfurnished, because of the danger of carbonationshrinkage surface cracking.

8. Control joints, discussed in Sections 3.5.3 and8.2.6, must not be omitted and grooves must be ofthe specified depth and well within the maximumpermitted spacing.

9. In addition to cleanliness of aggregate, stipu-lated in Section 8.3.1, any reactive elements ofaggregate should be neutralized through the use oflow alkali cement or a suitable pozzolan, or prefer-ably both. Certain cherts and other expansive ag-gregates and lignite can cause cracks at popouts. Jobspecifications should cover these aggregate proper-ties and constructors should ensure observance ofthese requirements.

10. Correct amounts of entrained air should bespecified and used to prevent cracking due to freez-ing and thawing and exposure to calcium or sodiumchloride.

8.6.1 Concrete materials - They can have an impor-tant influence on drying shrinkage.

1. Cement should be Types I, II, V, or IS, prefera-bly not Type III.

2. Aggregates favorable to low mixing water con-tent are (a) well graded, (b) well shaped (not elon-gated, flat, or splintery), and (c) free of clay, dirt,and excess fines.

3. Aggregate should consist of rock types whichwill produce low-shrinkage concrete (see Section3.4.2).

4. Calcium chloride should be prohibited.

CONTROL OF CRACKING 224R-41

8.1. Concrete M a n u a l 8th Edition, U.S. Bureau ofReclamation, Denver, 1975, 627 pp.

8.6.2 Concrete mixes - For least shrinkage, the mixproportioning should incorporate those factors thatcontribute to the lowest water content. This means:

1. The largest practical maximum size of aggre-gate (MSA).

2. The lowest practical sand content.3. The lowest practical slump.4. The lowest practical temperature.5. Less than half the smooth grading curve

amount of small coarse aggregate, No. 4 to 3/8 or3/4 in. (4.75 mm to 9.5 or 19 mm), especially if it iscrushed material.

8.6.3 Concrete handling and placing - Equipment(chutes, belts, conveyors, pumps, hoppers, andbucket openings) should be capable of working effec-tively with lower slump, larger MSA concrete wher-ever it is appropriate and feasible to use. (It is cau-tioned that too often, in order to expedite pumping,the actions taken are those which increase dryingshrinkage and resultant cracking: more sand, morefines, more water, more slump, smaller aggregate.When pumping is to be permitted and freedom fromshrinkage cracking is important, special emphasismust be placed on obtaining effective locations andan adequate number of contraction joints. Moreover,the use of pumping equipment capable of handlingmixes favorable to least cracking should be re-quired.)

Vibrators should be the largest and most powerfulthat can be operated in the placement.

Upper lifts of formed concrete should be re-vibrated as late as the running vibrator will pene-trate under its own weight.

8.6.4 Finishing - Finishing should follow therecommendations of ACI 302.1R to minimize oravoid all forms of surface cracking.

It is particularly important that flatwork jointgrooves have a depth of at least l/5 of slab thick-ness, but not less than 1 in. (25.4 mm) deep.

8.6.5 Forms - Forms should have ample strength tosustain strong vibration of low slump concretes.

Exposure of warm concrete surfaces to fast dryingconditions or to low temperatures prior to curing,should be avoided during form removal, if dryingand thermal shrinkage cracking is to be prevented.

8.6.6 Contraction joints - Plans should include anadequate system of contraction joints to provide forshrinkage. Formed grooves should be constructed inboth sides of parapet, retaining, and other walls atthe depth and spacing indicated in Sec. 8.2.1.

8.6.7 Curing and protection - These proceduresshould insure the presence of adequate moisture tosustain hydration and strength development in thesurface concrete. Rapid drying of the surfaces at theconclusion of the specified curing period should be

avoided. Providing time for adjustment and gradual,slow elongation will minimize cracking.

Water curing should use a wet cover in contactwith the concrete surfaces. At the end of the wetcuring period, preferably at least 7 days, the covershould be left in place until it and the concrete sur-face appear to be dry, especially in arid weather.

In less arid areas and for interiors, the forms willprovide adequate curing if exposed surfaces are pro-tected from drying and provided they can be left incontact with the concrete for at least 7 days. There-after, the forms should be left on with loosened boltslong enough to allow the concrete surfaces to drygradually.

Ponding is not a desirable method of curing in anarid climate because of the quick drying that occurswhen it is discontinued.

Because drying is slow and prolonged, a properlyapplied sealing compound provides good curing forflatwork placed on a well-wetted subgrade and pro-vides adequate curing for massive sections. In anarid climate, sealing compounds are not adequate forthinner structural sections. When used on formedsurfaces, they should be applied when the thor-oughly wetted surface is still damp but no longerwet.

8.7 - ConclusionAs noted early in this chapter, it is the responsi-

bility of the engineer to develop effective designs andclear and specific specifications. To assure both theowner’s and the engineer’s satisfaction with the re-sults, the engineer should have the owner arrange forinspection by either the owner’s personnel, the en-gineer, or a reliable professional inspection servicewho will insure that the construction is performed onthe same basis as it was bid. Without the full andfirm intent to confirm the specified character and de-gree of performance, there is a serious chance thatundesirable results will be obtained. Without firm in-spection and controls, and a clear understanding ofthe job requirements by the contractor, it is likelythat concrete will contain more water than it should,finishing operations will be expedited with the waterbrush (or hose), and curing will be interrupted or ab-breviated (not to mention other less obvious itemswhich influence the later appearance of unsightlycracks). When properly applied, the procedures dis-cussed in this chapter can be used to produce a highquality concrete with the least probable amount ofcracking.

References

8.2. Dakhil, Fadh H.; Cady, Philip D.; and Carrier,Roger, E., “Cracking in Fresh Concrete as Related to

224R-42 ACI COMMITTEE REPORT

Reinforcement,” ACI JOURNAL, Proceedings V. 72, No.8, Aug. 1975, pp. 421-428.

Chapter 9 - References9.1- Recommended references

The documents of the various standards producingorganizations referred to in this document are listedbelow with their serial designation.

American Association of State Highway and Transporta-tion Officials

T176 Plastic Fines in Graded Aggregate andSoils By Use of the Sand Equivalent Test

American Concrete Institute201.2R207.1R207.2R

Guide to Durable ConcreteMass ConcreteEffect of Restraint, Volume Change, and

Reinforcement on Cracking of MassiveConcrete

211.1

212.1R/212.2R

223

302.1R

304R

305R306R308313

318

340.lR

347.1R

504R

Standard Practice for Selecting Propor-tions for Normal, Heavyweight, andMass Concrete

Admixtures for Concreteand Guide for Use of Admixtures in Con-

creteStandard Practice for the Use of Shrink-

age-compensating ConcreteGuide for Concrete Floor and Slab Con-

structionGuide for Measuring, Mixing, Transpor-

tating, and Placing ConcreteHot Weather ConcretingCold Weather ConcretingStandard Practice for Curing ConcreteRecommended Practice for Design and

Construction of Concrete Bins, Silos,and Bunkers for Storing Granular Ma-terials

Building Code Requirements for Rein-forced Concrete

Design Handbook in Accordance with theStrength Design Method of ACI 318-83,Volume 1 - Beams, Slabs, Brackets,Footings, and Pile Caps (SP-17)

Precast Concrete Units Used as Formsfor Cast-in-Place Concrete

Guide to Joint Sealants for ConcreteStructures

517.2R

544.3R

Accelerated Curing of Concrete at Atmo-spheric Pressure - State of the Art

Guide for Specifying, Mixing, Placingand Finishing Steel Fiber ReinforcedConcrete

ASTMC 512

E 399

Test Method for Creep of Concrete inCompression

Test Method for Plane-Strain FractureToughness of Metallic Materials

lowing organizations:

Cornit Euro-International du B&ton and F&i&&m Inter-nationale de la PrkcontrainteCEB-FIP Model Code for Concrete Structures

The above publications may be obtained from the fol-

American Association of State Highway and Transporta-tion Officials444 North Capital St., N.W.Suite 225Washington, DC 20001

American Concrete InstituteP.O. Box 19150Detroit, MI 48219

ASTM1916 Race StreetPhiladelphia, PA 19103

Cornit Euro-International du B&on and Federation In-ternationale de la Precontrainte - English edition avail-able from:British Cement AssociationWexham SpringsSlough SL# 6PLENGLAND

9.2 - Cited referencesCited references are provided at the end of each chapter.

This report was submitted to letter ballot of the committee whichconsists of 24 members; 21 were affirmative, 2 were not returned,and 1 abstained. It has been processed in accordance with theInstitute procedure and is approved for publication and discus-sion.

CONTROL OF CRACKING 224R-43

ACI Committee 224

David DarwinChairman

R. S. Barneyback, Jr.Eduardo Santos BasilioAlfred G. BisharaRoy W. CarlsonNoel J. EverardJ. Ferry-BorgesPeter Gergely

Grant T. Halvorsen*Chairman

Florian G. BarthAlfred G. BisharaHoward L. BoggsMerle E. BranderDavid Darwin*Fouad H. Fouad*Peter Gergely

Cracking

Bernard L. MeyersPast Chairman

Donald L. HoughtonPaul H. KaarTony C. LiuJ. P. LloydLeRoy LutzV. M. MalhotraDan NausEdward G. Nawy

Robert E. PhilleoMilos PolivkaJulius G. PotyondyRobert E. PriceErnest K. SchraderLewis H. TuthillRobert L. Yuan

The committee voting on the 1990 revisions was as follows:

Randall W. PostonSecretary

Will HansenTony C. LiuEdward G. NawyJohn D. NicholasHarry PalmbaumArnfinn RustenAndrew Scanlon

Ernest K. SchraderWimal SuarisLewis H. Tuthill*Thomas D. VertiZenon Zielinski

*Members contributing to these revisions.