Вы находитесь на странице: 1из 22

Research and Development of the


Portland Cement Association




Influence ggregates

of Physical



on Frost Resistance

of Concret(

BY George Verbeck and Robert Landgren


Reprint from the Copyrighted of the



SOCIETY FOR TESTING MATERIALS Philadelphia 3, Pa. Volume 60, pp. 1063-1079, 1960




By George Verbeck and Robert Landgren




AND DEVELOPMENT LABORATORIES 5420 O]CI Orcl,ard Roa(l Skokie, Illinois




ib( need ior aIl undcrstandiug of the influcmcc of aggregates on the rcsis~an cc of ccmcrcte to freezing and thawing is becoming incrcasingl y important. 1hc durabilitj~ of concretes made with different. aggregates (Iepcnds upon the rate at which the aggregates become critically saturated in the concrctc au(l upon the different physical responses of the aggregates to freezing. An aggregate i~,coucrcte for which significant summertime drying is presumed can giv( :1.(,isfwctory performance if the Icngtb of time rcquimd for the :l,~~lT/@l1?to 1)<! ,(ICcn! imli ~,saturated is longer than its winters exposure to

wcLi.iiig au(l fI cx:xmg cum


the aggregate

would cause dimrpLion


iromn while sat.uratcd. The time requirements for satumlion of +qgrcgatcs in concrete arc anal yzed in terms of the physical charactcris tics of the aggregates and the coucfetc Important factors arc the porosi(y and pore size distribu tion of the aggregate and th c permcabil ity ancl thickness of the mortar cover pro tecting the aggregates from water. fhc (1iffcren t responses of aggregates to freezing when satu rated depend upon Lhe pore characteristics of the aggregates and the cetncnt. pask. %turalcci

:Iw;r(g:lW of low porosity may accommodak pore water frcxzing l)y simple dastic expansion. Saturated aggrcgalcs of moderate or high porosity may fail in tcrnally bccausc th c parf icle dimension CXCCCCIS a certain cri lica] size or may cause failurc in the paslc immediately ad jaccn t L() the tiggregat c parlicle because of aggregate pom water displacement. The magnitude of {he hydraulic pressures (fevclopc(l is significantly influenced by the size of tk aggregate particle and the permeability and air content 01 the surrounding paste. 1he applicability of certain types of laboratory tests of tbc frost rcsisbancc of aggregates ao(i concretes is questioned on the basis of these mechanisms.

A better understanding of the influence of aggregates on the resistance of concrete to freezing and thawing and deicer scaling is becoming increasingly essential. Sources of aggregates of known satisfactory performance are being de* Pmsentcd at the Sixty-third Annual Meeting of the Society, June z7July 1, 1960. 1 Manager, Applied Research Section, Research and Development Laboratories, Portland Cement Assn. ~Associate Research Engineer, Applied Research Section, Research and Development Laboratories, Portland Cement Assn.

pleted, particularly near some large melropo]itan areas, and some so-called marginal aggregates are being rejected for use in concrele. In areas new to the use of concrele in pavements, performance records are nol available for the local aggregates. It is believed that some of Lhe so-called marginal aggregates are rejected for use in concrete because of inappropriate testing techniques and that oLher aggregates now considered as marginal or unacceptable can be successfully used pro-




vialed certain simple and frequently attainable requirements are imposed upon the aggregate and the concrete. It is the purpose of this paper to consider the mechanisms by which the pores in aggregates influence concrete durability, to assist in providing a fundamental and rational basis for proper evaluation of aggregates, and to establish some principles whereby aggregates of moderate frost susceptibility may be successfully used in concrete. To do this requires an understanding both of the mechanisms involved and of the influence of the various physical characteristics of the aggregates on these mechanisms and the resulting performance of the concrete. Special cases in which concrete durability may be directly influenced by chemical reactions of the aggregate or possibly by differences in the thermal expansivities of the components of the concrete will not be considered.

The general problem of the resistance of concrete to freezing and thawing and the related action of de-icing chemicals has been presented in numerous publications (I8) ? The effactiveness of air entrainment in providing a high order of durability to concretes made with good quality aggregates has been adequately demonstrated. The general concepts that have successfully provided an understanding of the frost resistance of cement paste (3) are also applicable to aggregates. Aggregates may cause disruption of concrete by the generation of high destructive hydraulic pressures during freezing. By analogy with paste, the magnitude of the hydraulic pressure developed in aggregates depends upon their degree of
$The boldface numbers in parentheses refer to the list of references appended to this paper.

saturation (proportion of total void space filled with water) and the permeability and size of the aggregate particle. Furthermore, if the degree of saturation of the aggregate is sufficiently high (above the critical, 91.7 percent) so that the remaining air-filled void space cannot accommodate the 9 per cent expansion of water during freezing, then water will be expelled into the paste surrounding the aggregate particle and potentially destructive hydraulic pressures may develop there as well as in the aggregate. Thus, the properties of paste, its permeability, air content, and porosity, are also involved inthe problem of aggregate durability. In addition, the paste can significantly influence the degree of saturation of the aggregate in concrete by limiting the ingress of water; the protection afforded in this manner depends upon the permeability and thickness of thepaste or mortar cover separating the aggregate from the wet surface of the concrete. A more detailed consideration of the various mechanisms by which the various physical characteristics of the aggregate and of the paste can influence the durability of the concrete reveals that the effects of some of these characteristics are closely interrelated and that some may have several and per,haps opposing effects. Analysis of the mechanisms is considerably simplified by separating the two distinct over-all problem into aspects within which the individual roles of the pertinent physical characteristics of the aggregate and the paste can be developed more clearly. These major aspects and the important factors involved are as follows: 1. The time required for an aggregate to become critically saturated when in concrete exposed to water as influenced by: (a) Pore size and porosity of aggregate.



(b) Thickness and permeability of protective mortar cover. 2. The various phenomena in the freezing of fully waturated aggregate demonstrating: (a) Elastic accommodation by aggregate. size of aggregate (in(b) Critical ternal hydraulic pressures). (c) Influence of confining mortar (external expulsion distance and external hydraulic pressures). (d) The influence of various factors modifying these effects of freezing, that. is, soluble materials, degree of saturation and freezing point depression in fine aggregate pores. These various phenomena in freezing as influenced by the physical properties of aggregate and paste will now be considered in detail and supporting experimental data obtained by this laboratory presented where appropriate.
Time Required joy Critical Saturation:

Most concrete in above-ground structures or in pavements placed on free draining subbases does not normally remain saturated but instead is subjected to drying, particularly during the summer. A single drying period occurring some time prior to freezing produces a marked improvement in concrete durability as has been demonstrated in numerous laboratory tests. Therefore for most purposes, we must be concerned not only about the durability of the concrete were it to become critically saturated but more importantly about the length of time required for concrete to attain such a state of vulnerability to freezing when exposed to the wettingand-freezing conditions of winter. Following the beneficial drying of summer, a few aggregates may require only a few weeks of wetting to become critically and damagingly saturated. Most aggregates, because of their porosity and pore

size or because they are used in concretes of low permeability (low water-cement ratio), may require many months to attain a critical saturation and hence successfully pass through the freezing season to the next drying and recuperating season. Consequently, from this realistic point of view, legitimate arguments have been made against indiscriminate use of test data obtained by freezing of continually wetted concretes made with presaturated aggregates as the sole criterion for aggregate acceptability (6). On the basis of unrealistic tests, an aggregate may be condemned as non-durable when, in fact, it might give satisfactory performance in structures above ground or in pavemeilts which are on free draining subbases and are exposed to summertime drying. The two major factors that appear to determine the time required for an aggregate in concrete to become critically saturated are: (a) the nature of the aggregate-that is, the porosity and the size distribution of pores in the aggregate, and (b) the nature of the mortar surrounding, and in a very real sense protecting, the aggregate in the concrete from the supply of water external to the concrete-its permeability and its thickness, that is, the distance from a piece of aggregate to the surface of the concrete. These factors, currently under study, are not as yet understood sufficiently well to permit an exact quantitative calculation of their individual influences; however, present understanding is sufficient to permit a qualitative and semiquantitative interpretation of their influences. Pore Size and Porosity: These characteristics of aggregates in relation to the durability of concrete have been studied by many investigators. Rhoades and Mielenz (9) compared aggregates having pores of small or large diameter when cast in similar



cement pastes and concluded that the aggregate with the smaller pores would attain and retain a higher degree of saturation and hence be most likely to fail when frozen, Blanks (2) reported tests for capillary water retention of aggregates and in addition showed that certain aggregates had pores that were small enough to compete successfully with some hydrated cement pastes for available pore water. In order to consider this problem in greater detail we will use, as a simple working model, a system consisting of a piece of dry aggregate, of known porosity andpore size, separated from asupplyof free water bya membrane of mortar of known thickness and permeability. This model simulates a wetted concrete pavement surface with a piece of dry aggregate a known distance beneath the surface. For convenience we will assume that the mortar membrane is saturated and serves .. cmlv . to transmit water from the surfacq~, of the concrete into the aggregate. T#~rate at which water moves through t@mortar membrane depends upon the cliaracteristics of the membrane difference resulting and a pressure from capillary tension across the membrane. The messure difference. ., AP. is approximatel~ expressed by the followingrelati~~ship (10):

rate of water movement into the aggregate under the postulated conditions depends significantly upon the relative humidity within the aggregate and is reduced rapidly as the relative humidity within the aggregate approaches 100 per cent. For example, the pressure gradient causing water movement into aggregate at 90 per cent relative humidity is only 15 per cent of that for aggregate at 50 per cent relative humidity.
I ,00 -r Foprock -025 per cent

0.80 g $ 0,60 3 %




j I

74 16

90 95 {loo QR

Relative Humidity, per cent

FIG. 1,The Degree of Saturation Attained by Aggregates at Various Humidities Differs Wide~y and Reflects Their Different Pore Size Distributions. For discrimination the scale for relative humidity has been plotted as proportional to (~. RH)Z.

where: = density of water at temperad ture T, gpercu cm, = gas constant = 82.05 R cu cm X atmos mole X deg K T = absolute temperature, deg K, = molecular weight of water = M 18.02, and % RH = relative humidity in aggregate, per cent. This relationship indicates that the

In order to express these rates of water movement in terms of degree of saturation of an aggregate, it is necessary to know the water vapor adsorption characteristics of the aggregate. As shown in Fig. 1, the relationships between degree of saturation and relative humidity differ significantly among aggregates; aggregates with fine pores such as the traprock adsorb considerable water vapor and attain a higher degree of saturation at any particular humidity than aggregates with a coarse pore system such as the dolomite.




From these relationships and from the logarithmic dependence of rate of water movement into the aggregate on the relative humidity within the aggregate, it is apparent that an aggregate with a fine pore structure similar to the traprock would reach a high degree of saturation much more rapidly than an aggregate with the coarse pore structure of 10000
$ 9 .5

i7aprock-Q25 per cent Absorption - 20

$! 0 g
h z z g .Q ~ g < z ~ z % .= g L?

1000 Gronite -Q 45 per cent Absorption

40 ~
60 80 ; : .t


90 g 95 : .+ 0

-~! 10-Groywacke-2.34

98 L? 99

per cent Absorption Dofomite-6.68 per cent Absorption ,,,>.

0.20 0.40 0.60 0.00 Oegree of Saturation


FIG. 2.Aggregates Encased in Mortar Absorb Water at Different Rates Because of Their Different Pore Size Distributions. Note the logarithmic dependence of this absorption rate up:n the relative humidity in the aggregate at various degrees of saturation.

the dolomite, even if the aggregates had the same porosity. Moreover, as is also shown in Fig. 1, these particular aggregates differ not only in pore size distribution but also in total porosity or absorption. These porosity differences serve further to accentuate the differences in behavior of these aggregates. For example, at 52 per cent relative humidity both the traprock and dolomite aggregates would be subject to equal pressures and hence have equal rates of water uptake, but the

traprock would already be 82 per cent saturated and would require only 0.045 per cent by weight of additional water for total saturation, whereas the dolomite would be only 6 per cent saturated and would require 6.28 per cent by weight of additional water for total saturation. Figure 2 shows the rates at which these aggregates tend to absorb water at various stages of saturation and at the various relative humidities. It may readily be observed that the traprock has a much higher rate of absorption than the dolomite, even when the aggregates are compared at equal degrees of saturation. From the above, it should be apparent that for aggregates with similar pore size distributions, the one with high porosity should require more time to attain any particular degree of saturation than the one of low porosity. For example, for two such aggregates, one of 15 per cent porosity would require three times as long to increase in internal humidity from 50 to 51 per cent than would one of 5 per cent porosity. This is a matter of very practical importance. With any aggregate containing coarse pores and a sufficiently thick mortar cover of low permeability, the time required for critical saturation will have considerable significance. If the time required to attain a disruptive state of saturation (assume 96 per cent) were 3 months for the aggregate having 5 per cent porosity, the aggregate with 15 per cent porosity would require 9 months. This additional 6-month period would be more than sufficient to take concrete with the 15 per cent porosity aggregate through the critical wetting-and-freezing season and satisfactory performance would be expected. Hence it maybe seen that under some circumstances it may be beneficial to the performance of concrete for the aggre-



to have a high porosity, and that ~ggregates of some intermediate porosity in concrete may actually show the poorest performance. These considerations of the time required for critical saturation have qualitative practical significance. A complete quantitative evaluation requires additional information about uniformity of distribution of absorbed water in aswregates, the process of drying of aggre~~tes in concrete, and the influence of the mortar other than that of transmitting

Mortar Cover


Waterermmbility, K,, cm per sec %-in, Thick M-in. Thick %-in. Tlnck

by We]ght 0.70 0.45

3000 x 1


111 477

879 885

792 980

water directly to the aggregate on the rate c saturation of the aggregate. Thickness Cover: and Permeability oj Mortar

It is reasonable to assume that the mortar cover transmits water by capillary tension in a manner approximating Darcys law for flow under hydrostatic pressure and therefore that the rate of flow wouid decrease as the thickness of the membrane increases, and that a highly permeable membrane would transmit water more rapidly than a membrane of low permeability. The permeability of paste or mortar depends significantly upon the watercement ratio and the degree of hydration of the cement. For well hydrated pastes the permeability may be increased by as

much as 100 fold as the water-cement ratio is increased from 0.40 to 0.70 by weight (II). Such differences in permeability have a significant influence on the rate at which the mortar membrane will transmit water into an aggregate. That the permeability and thickness of the mortar cover does significantly affect the time required for an aggregate to become destructively saturated in concrete is demonstrated in Table 1. These data, which are typical of results of our various laboratory evaluations, were obtained by embedding l-in. cubes of a dolomite aggregate (6.68 per cent absorption) to the depths shown in mortar slabs having water-cement ratios of 0.70 and 0.45 by weight. After a preliminary 28 days of moist curing, the slabs were air dried to constant weight at 73 F and 50 per cent relative humidity and then subjected to wetting by continuous ponding of water on the upper surface. The slabs were subjected to freezing every third day (sides and bottom insulated) to determine whether any of the aggregate cubes had become destructively saturated as evidenced by the occurrence of a popout. It may be seen that the length of time required to destructively saturate the aggregate was beneficially lengthened by the use of the low water-cement ratio mortar and a greater thickness of cover. In field concrete, the use of low watercement ratios could extend the time required for critical saturation of aggregate sufficiently to provide satisfactory performance even though the aggregate would have marginal or poor performance in air-entrained concrete of high watercement ratio. In the case of concrete pavements, loss of protective mortar by wear or by grinding (to meet surface smoothness tolerances) may alter the performance of aggregate particles near the concrete surface.




The Various Phenomena Saturated Aggregate:

in the Freezing

In addition to the effect on the length of time the concrete can withstand exposure to water before the aggregate becomes critically saturated and potentially vulnerable to freezing action, the pore characteristics of aggregates and the mortar in concrete significantly influence the response of the critically saturated aggregate to freezing. The influence of pore characteristics of
A (hmrtzlte

traprocks. Such an aggregate will demonelastic accommodation. (b) An aggregate of moderate or high porosity but of low Permeabildy, perhaps typical of cherts and other absorptive aggregates with a fine pore structure, which can cause failure because of high internal pressures, i.e., critikal size ejects. (c) An aggregate of moderate or high porosity but of relatively high permeability, perhaps typical of many limestones, dolomites, sandstones, and other
A Chert
Abs3rpt,on:21 percent

A Dolom,le
Abwrpi 800:6,7 Wm. +

Abwpt,on= 0,1 per cd


per sec Kl=300xlo-Cocm per *ec

~. 1/ -.h<

Elastic Oilot ion

I.st anlo.eo.s


, /> ,,,

p ,

:./1 m L_____, critical Lmo, Size

0.09 W, E 3(1- 2,U)

780 PSI P= (Accommodated)

P= 17,000 psi IF0,15)

P = 35,000 PSI (Fo,IsI

Unconfined Free z(ng ot 15 m per h,

Lm,,: (No Oornoge) .,0* Expulsion Distance F<eezmg Wh#le Conf med Cube Size In, 1 AL. 0,09 W, L 513 [No Oomogel 1/4 (/2 I


Lm,,= 33


IFOIIS ,f Larger)

[No 0.mogel

4ZZ?% Required

concrete S.rfoce Freeze ~i; ;,,;~ > T%je

AL ~ 0.0015 0.0030 O,0060

Cube S(2e ~ 1/4 (/2 I

AL ~ 0.0040 0.0080 0.0160 l

:/?; ~ ~,age~ la -;,-~r+e 3

FIG. 3.Summary

of Various Effects of Freezing Saturated Aggregat.ee.

aggregates and mortar on the reaction of saturated aggregate when frozen under various conditions is summarized in Fig, 3. Three different types of aggregates have been used to demonstrate three different responses to freezing, depending upon the pore characteristics of the aggregate as measured by total porosity (absorption) and permeability. The different types of aggregates typifying the different types of response on freezing are: (a) An aggregate of tow porosity, about 0.3 per cent by volume, and perhaps typical of some quartzites, marbles, or

absorptive aggregates with a relatively coarse pore structure which can cause failure because of high external pressures in the surrounding mortar. Elastic Accommodation: Under conditions of normal use, the concrete is not subjected to extremely rapid freezing, and therefore a significant portion of the excess volume of water created in the freezing aggregate has time to escape and thus keep internal pressures below the tensile strength of the aggregate. It is not necessary that all of the excess volume be expelled in order to




avoid rupture of the aggregate as the aggregate can elastically accommodate a portion of this excess volume without rupture. The principle of elastic accommodations is clearly revealed by considering the limiting case of instantaneous freezing of completely saturated aggregate. Under such circumstances, none of the water within the particle has time to escape or be expelled. For avoidance of failure, all of thevolume increase which accompanies the formation of ice from liquid water (9 per cent of the water volume) must be accommodated elastically within the aggregate.

It may be noted in Fig. 3 that the calculated tensile stress created on instantaneous freezing of a saturated aggregate having an absorption of 0.1 per cent by weight (porosity = 0.3 percent) is only about 780 psi, a stress capable of being withstood by this aggregate. There is a class of aggregates, those that have porositiesup to about 0.30 per cent by volume, that are completely invulnerable to freezing action and can be frozen instantaneously even when completely saturated, Obviously, saturated aggregates containing significantly greater amounts of freezable water would require cor-

TABLE 11,FULLY SATURATED AGGREGATES OF LOW POROSITY DO NOT FAIL WHEN FROZEN RAPIDLY BECAUSE OF INTERNAL ELASTIC ACCOMMODATION. 1 by }~by ~-in. aggregate prisms frozen at 6Fper sec. Aggregate Abs::pt$tn, by weight ....... ....... ....... ....... ...... ....... 1.7 1.1 0.7 0.17 0.10 0.05 Modulus of Ela#ic::y, CsJ&d&d psi 7600 4400 5500 2000 1500 700 Results of Freezing

Limestone . . . Sandstone . . . Limestone . . . Traprock . . . . Limestone, . . Marble, . . . . . .

.......... .......... .......... .......... .......... ..........

3 x 106
2 : 6


Cracked Cracked Cracked Cracked No distress No distress

The ice pressure, or its counterpart the tensile stress which is created within an aggregate, can be estimated from therelation: ~ = 0.09WfE 3(1 2#) where: P = internal ice pressure or aggregate tensile stress required to increase volume of aggregate sufficiently to accommodate expansion, psi, Wf= volumetric fraction of freezable water in aggregate, cu cm per cu cm = aggregate porosity for saturated aggregate, and E = modulusof elasticity of aggregate, psi, and P = Poissons ratio of aggregate.

despondingly greater dilations in order to accommodate the internal volume increase of the water on freezing. Estimates of the tensile stresses corresponding to these required dilations are shown in Fig, 3 for the aggregates having 2,1 and 6.7 per cent absorption by weight. It is seen that the required tensile stresses aremuch beyond the tensile strengths of these aggregates and hence that rupture would be expected, Experimental verification of elastic accommodation during very rapid freezing of saturated aggregates having porosities of about 0.3 per cent by volume (0.1 per cent absorption by weight) or less is shown in Table 11. For such aggregates, vacuum saturated and rapidly frozen at 6 F per see, the calculated tensile stresses




required for elastic accommodation were 1500 psi or less. Stresses of this magnitude are approximately equal to or below the tensile strength of these aggregates which withstood freezing under these most critical conditions of test. Aggregates with higher porosities and calculated tensile stresses greater than their tensile strength failed by rupture. Of course, under conditions of normal use in concrete, aggregates are not subjected to such very rapid freezing, and a significant portion of the excess volume of water created in the aggregate may be expelled during freezing. Uncon$ned Freezing and Critical Size oj Aggregate: The magnitude of the hydraulic pressure developed in a saturated aggregate particle during freezing depends upon the rate of freezing and the porosity, permeability, and size of the aggregate particle (12). If for simplicity it is assumed that both the propagation of freezing and the expulsion of water are unithe maximum hydraulic directional, pressure can be estimated, using Darcys law, as:
Pm,. =

can be used to estimate the maximum permissible size or critical size of the aggregate as follows:
L In&x 27.7 KIT

0.09 dWf/dt

0.09 dWf/dt

where: P max dWj/dt maximpm pressure, psi, = rate of freezing of water, = aggregate porosity X rate of linear propagation of freezing zone, cm per see, = dimension of aggregate in L direction of freezing and expulsion of water, in., = permeability coefficient of agKI gregate, cm per see, and 27.7 = conversion factor, in. hydraulic head per psi. Since P~,~ cannot exceed the tensile strength of the aggregate, this equation

where: permissible sizeL max = maximum (critical size, in., and T = tensile strength of aggregate, psi. The results of calculations based on this equation and assuming a freezing rate of 1.5 in. per hour (1.06 X 103 cm per see) are shown in Fig. 3. This freezing rate is considered to approximate rates of freezing encountered in-natural exposures and in some laboratory freezing tests. It should be noted that the chert type aggregate of moderate absorption, but of low permeability (fine pores), has a calculated critical size at this freezing rate of only 0.5 in. However, aggregates of high porosity can have a very large critical size if they also have a high permeability. For the dolomite aggregate shown in Fig. 3 and for the assum;d, rate of freezing, a critical dimension of approximately 33 in. is calculated. A saturated aggregate of this type should ,~~t fail by the development of high l$@nal hydraulic pressure when unidri%ctionally frozen at the prescribed rate,. provided the aggregate is unconfined and the excess volume of water created during the freezing can be expelled. The preceding calculations were based on the assumption of unidirectional expulsion of water, an assumption which permits simple mathematical expression. During the actual unidirectional freezing, some of the excess volume of water would be expelled laterally and would tend to relieve the internal hydraulic pressure and hence increase the practical critical size of the aggregate, Complete mathematical expressions including the lateral expulsion of water have not been obtained.



The concept of critical size of aggregate can be demonstrated by laboratory tests such as those shown in Table 111. These results were obtained by the freezing and thawing of various size fractions of a particular chert (3.6 per cent absorption) previously vacuum saturated. These data clearly show that large particles of this aggregate are much more vulnerable to freezing than small particles.

the expelled water must move into the surrounding paste for volumetric accommodation. We will assume unidirectional freezing of an aggregate cube with the freezing zone entering one face and water being expelled from the other five faces. As a simplifying and reasonably justifiable condition, we will further assume that equal amounts of water are expelled from all five surfaces and that the expelled water completely fills the air voids for a uniform distance, AL, around the five sides of the cube. AL can be calculated as follows: *L = 0.09 W,L

Per tint passjng ~-m. .meve Original . . . . . . . . . 12 cycles, freezing and thawing . . . 51 cycles, freezing and thawing. o 33 61

l Per cent passing %-in. sieve l 0 12


Per cent pas:: :g sieve


0 1

Chert aggregate, 3.6 per cent absorption by weight, saturated by 3 hours evacuation and 5 days under water. Freezing and thawing under water.

Freezing While Con$ned by Mortar: To avoid failure during freezing, a saturated aggregate particle having an absorption greater than that which can be elastically accommodated (about 0.1 per cent) and a size smaller than its critical size must be able to expel water into the surrounding air-entrained mortar without the development of hydraulic pressures that exceed the tensile strength of the concrete. The factors of primary importance to this mechanism are the freezing rate, aggregate size and porosity, and the permeability and air content of the surrounding paste. Let us first consider the distance that

where: AL = distance required for volumetric accommodation of expelled water, in., L= dimension of aggregate cube, in,, fraction of freezable Wf = volumetric water in aggregate, cu cm per cu cm, . porosity of saturated aggregate, and, air content of paste, fractional = A= 0.18 (estimated). Figure 3 contains the results of calculations of expulsion distances for various size cubes of the chert and dolomite. It can be seen that the expulsion distances calculated on the above basis are in general relatively small. For small size particles of the chert having intermediate porosity, the calculated expulsion distances of 0.0015 and 0.0030 in. would appear to be sufficiently small to be accommodated by the paste without the development of distress, particularly when these distances are compared to the allowable distance, the bubble spacing factor, L, of 0.0100 in. ($ which is considered satisfactory to protect airentrained concrete. For the l-in. chert cube, the calculated distance of 0,0060 in. would also appear to be within this




normally ascribed limit, but, as has been previously shown, failure of this particular size of chert could result solely from considerations of critical size if this aggregate were frozen at a rate of 1.5 in. per hour. The expulsion distances calculated for the dolomite aggregate of high porosity are large, perhaps excessive for the larger sizes of the aggregate. The required expulsion distance depends significantly on the aggregate size and air content of the surrounding paste. Air entrainment can significantly reduce the required exTABLE IV. TIME REQUIRED FOR DESTRUCTIVE SATURATION OF AGGREGATES IN MORTAR IS SIGNIFICANTLY LONGER FOR SMALLER SIZE AGGREGATE PARTICLES. 0.70 W/C mortar~-in. mortar cover. Days of Wetting Sustained Before Failure(Popout) Aggr#:n!ube


Absorption, per tint by weight

1 Dolomite No. 1.. , 6.68 Dolomite No. 2, . . . 3.59 111 90

213 >980 177 309

pulsion to aggregate The

distance the




serve of when



vulnerability as unsound concrete.


used in non-air-entrained advantageous

effect of usin~ a small maximum size of su~:h aggrcgxte in concrete on the required expul sirm distance and potential destructive action appears obvious. Experimental laboratory verification of the beneficial effect of reduced aggregaw size is shown in Table IV. These da! a were obtained in tests similar to those described in connection with Table I. It may be observed that the time required for destructive saturation is significantly lengthened by the use of smaller size of aggregate. As previously discussed, the distances

shown in Fig. 3 do not appear disruptively high when compared to calculated protected thicknesses of pastes of about 0,0100 in. in air-entrained concrete, However the magnitude of the. hydraulic pressures associated with these distances is of interest. The hydraulic pressures developed at three important stages of the freezing are shown as stages 1, 2, and 3 in Fig. 3 as follows: Stage l.The instant of freezing of the exterior face of the aggregate (that face closest to the freezing surface of the concrete). At this instant, the permeability of the aggregate, its total porosity and th c freezing rate govern the pressures generated by movement of water through the aggregate into paste air voids present. at the aggrcga.te surfaces. Stage 2.The instant of freezing of the interior surface of the aggregate when pressures are determined by the rate at which water is ejected from the aggregate upon freezing, the depth of paste saturated by the aggregate (A-L) and the paste permeability. Stage 3.The freezing of the paste plus air void system immediately adjacent to this interior surface which is completely saturated for a distance AL. This pressure is a function of the freezing rate, paste air content, and the permeability of the paste. Accurate measurements can be made of the pertinent physical characteristics of aggregates, and reasonable assumptions can be made regarding rates of freezing of aggregates and the air content of the surrounding paste. However, the least known factor is the permeability of the paste, which has a very significant and direct effect on hydraulic pressure. This permeability depends upon the watercement ratio of the paste, the degree of hydration of the cement, and apparently is greatly influenced by any prior drying of the paste. Calculated hydraulic pressures at the



three different stages of freezing of the chert and dolomite aggregates as shown in Table V are based on the several possible paste permeabilities (which vary over 10,000 fold) rather than on a single assumed and perhaps misleading representative permeability y. The calculations are based on directly measured aggregate characteristics and estimated factors for rate of freezing (1.5 in. per hr) and air content of paste (18 percent). The results shown in Table Vclearly

would also have an opposite and beneficial effect of decreasing the rate of saturation of theaggregate as previously discussed. As shown in Table V the most damaging stage of freezing, at least that stage developing the highest hydraulic pressure, does not occur during the freezing of the aggregate itself but rather during the subsequent freezing of the completely saturated paste (paste plus air voids) adjacent to the aggregate.
PRESSURES DEVELOPED AT THE THE PORE CHARACTERISTICS OF THE SURROUNDING PASTE. paete = 18 per cent. Hydraulic Pressure Develo ed at Various Freezing Stages, f psi For a Dolomite For a Chert Permeability = 1 X 10-10 Permeability = 30il X 10-10 cm per sec cm per sec AL = 0.003in. AL = 0.008in. Stage la Stage 2 6100 41 5 1 Stage 3 Stage 1 Stage 2
18 600 8 43 000


WaterC;:t;:t by weight PriorTreatment

Permeabilityof MaturePaste, K1 , cm per sec (11

Stage 3
49 500

0.30 0.50 0.70 0.50

Never dried Never dried Never dried Dried to 79 per cent RH and resaturated

1 x 10-13 150 1200 10 000

1020 1020 1020 1020

200 44 4

8 8 8

290 36 4

700 117 11

0 Unidirectional b See Fig. 3.

expulsion of water assumed.

demonstrate that the permeability of the paste is a major factor influencing the hydraulic pressures developed. These data also indicate that the permeability of a mature paste of 0.50 water-cement ratio can be increased by as much as 60 to 70 fold by prior partial drying. Such changes can have very large and significant effects on the hydraulic pressures developed, and more information on the influence of drying on permeability is desirable. It should be noted that whereas low paste permeabilities would create high hydraulic pressures during freezing of saturated aggregate, low permeability

pressures shown for of freezing of the two aggregates were calculated, assuming unidirectional expulsion of water. As previously discussed, the hydraulic pressures actually created during the first stage of freezing may be several fold less than the pressures shown because of lateral expulsion of water. However, the relative pressures calculated for the chert and dolomite materials should be correct; that is, the pressure created at the beginning of freezing of the low-permeability chert should be approximately 100 times greater than for the high-permeability dolomite.
the first stage








It would appear that a saturated porous aggregate of low permeability would be more likely to cause popouts in a concrete surface than a saturated aggregate of the same porosity but of high permeability. In the former case, a high initial hydraulic pressure is created close to the freezing surface of the concrete, precisely at the location where the lowest pressure is required to cause
o u
J ~

saturated, or if all of the water does not freeze, either because of the presence of soluble material or because of the fineness of aggregate pores, then the maximum calculated parameters for hydraulic pressure, expulsion distance, and critical size must all be modified by appropriate factors. In most instances, the freezing of aggregates in concrete involves aggregates that are not totally saturated, total
Relative Humidity, per cent


000 -

I .00



i70prock-025 percent

0.80 0.60 .0.40 0.20 o

w Q60 $ s u. o ~ 040z 3 3

020 g !5 #o







Minimum Freezing Temperature, deg Fohr

FIG. 4.The Relative Amount of Water Expelled and Hence the Disruptive Effect of Freezing Depend Upon the Degree of Saturation of the Aggregate,

rupture renceof Mod##ng

of the concrete a popout. Factors:

and the occur-

FIG. 5.The Fraction of Pore Water Remaining Unfrozen at Various Temperatures Depends IJpon the Size of the Aggregate Pores and Can Be Estimated from the Adsorption Isotherm. Fordiscrimination, the scale for relative humidity has been plotted as proportional to (% RH).

In the preceding discussions of the various aspects of freezing of aggregates it has been assumed, for convenience and simplicity of presentation, that the various porous materials were completely saturated and that all of the water contained in the pores of the aggregate would freeze at the normal freezing point of water, 32 F, In an actual case of freezing of an aggregate, neither of these assumptions is necessarily true. If the aggregate is not

saturation being extremely difficult to attain. The degree to which total saturation is approached influences significantly the damaging effects of the freezing. The hydraulic pressure created during freezing is a direct function of the amount of water which must be expelled from the freezing zone. If the aggregate is totally saturated, the amount of water expelled is approximately 9 per cent of the total volume of water contained. If the porous system is only 91.7 per cent saturated, the remaining unfilled 8.3 per cent of pore space is sufficient to accommodate the 9 per cent expansion that accompanies the




formation of ice. Since no water need be expelled from the porous system, macroscopic hydraulic pressures are not created. Figure 4 shows the degree to which the hydraulic pressures and exmdsion distances ~reviouslv described . are decreased in instances where the aggregate is less than totally saturated. In addition, other factors modify the effects of freezing. Some of the water contained in the pores of an aggregate will not freeze at 32 F. Water-soluble salts and cement alkalies lower the freezing point of the pore liquid. The freezing point depression due to this effect varies with the materials present and their concentration but usually is not great enough to prevent freezing of the pore liquid at moderate freezing temperatures. Depression of freezing point also occurs with water in fine capillaries or surface adsorbed. Powers and Brownyard (I 3) reviewed the mechanisms involved in these physical processes and calculated the resulting theoretical freezing point depression. Figure 5 shows such theoretical freezing point depressions, calculated as a function of equilibrium relative humidity, superimposed on the water adsorption isotherms previously presented. It is apparent from this figure that significant amounts of pore water will not freeze in certain aggregates. In the case of the traprock, only about 10 per cent of the water absorbed by vacuum saturation should freeze at 40F. It should be noted, however, that those aggregates having higher total absorption have the lower reductions in amount of freezable water due to the effects of capillarity and adsorption. Such reductions in freezable water result in smaller volumes of water being expelled and thereby moderate the disruptive actions.

The actual field durability of concretes made with various aggregates depends

upon the rate of saturation of the aggregates and their response to freezing. These two phenomena involve various physical characteristics of the aggregates particle size, permeability, porosity, and pore size distributionand of the mortar or paste component of the concreteits air content, permeability and its thickness as protective cover. It has been shown that certain characteristics of the aggregates or mortar can play a dual role with resulting opposite effects on concrete performance. Aggregates of high porosity and mortars of low permeability beneficially reduce the rate of saturation of aggregates in concrete but also have deleterious effects during the freezing of saturated aggregates. For this reason, simple correlations or effects are not to be expected. The mechanisms and mathematical expressions presented are applicable in a. qualitative and semi-quantitative sense. The fundamental principles involved are sufficiently well defined to permit examination of some of the physical tests used to evaluate the so-called [durability of aggregates in concrete. The need for rapid and convenient tests for the durabilit y of aggregates is apparent, and a number of test methods have been reported in the technical literature. Some of these tests yield information regarding the frost susceptibility of various aggregates, but the limitations of such information and the degree to which it mayor may not apply to actual field conditions warrant consideration. It would appear reasonable that a test method should incorporate the determination, or at least consideration, of those physical characteristics of aggregates actually involved in the mechanisms of saturation and response to freezing, In tests such as the sodium or magnesium sulfate soundness tests, the mechanism of disruption is different from




that resulting from the freezing of water. Although results of this type of test must reflect something about the pore characteristics of the aggregates, the results cannot be directly and rationally interpretedin terms of any of the factors of importance to the actual mechanism of freezing in concrete. Such test results have only rough and uninterpretable empirical correlation with concrete performance. A somewhat more realistic type of test involves the freezing of unconfined particlesof aggregate, In this type of test the observed deterioration of the aggregate is caused by the appropriate agent freezing. However, such tests incorporate unrealistic conditions when the aggregates are presoaked in water or subjected saturation procedure to a vacuum prior to freezing. Such efforts to have the aggregate at a high state of saturation prior to test do not take account of the time period required for the aggregate to become critically saturated when in field concrete. In addition, such unconfined tests do not directly evaluate the type of concrete distress caused by expulsion of water into the mortar surrounding the aggregate. Although. the test may appear to incorporate some aspect of saturability of the aggregate, the primary disruptive action involves only critical size effects since the, aggregate is unconfined. The results obtained by such tests depend upon aggregate particle size, permeability, degree of saturation and porosity anli rate of freezing and can be altered by the presence of soluble materials such as salts or alcohol. Whether an aggregate does or does not fail in tests of this type depends upon several arbitrarily selected test conditions andsuch tests may yield results of little importance to actual concrete performance. Laboratory tests using concrete specimens containing the actual aggregates, cement, and mix design under considera-

tion are by far the most satisfactory. However, such tests must be conducted in a manner consistent with the principles here discussed. To test concrete proposed for above grade structures or for pavements placed on free draining subbases and exposed to seasonal drying, a suitable procedure would include a period of air drying prior to wetting and subsequent freezing and thawing. Test methods using concretes containing presaturated aggregates without provision for preliminary drying can produce rapid and misleading deterioration in concretes not comparable with the actual service results. Various examples of the inordinately severe conditions that have been incorporated in laboratory tests can be cited. In a recent series of cooperative freezingand-thawing tests conducted by the Highway Research Board (14), the aggregates were presoaked for 7 days prior to use and the concretes were moist cured for 14 days without a period of air drying prior to freezing. This procedure was useful for the purpose of tie test program, namely, to improve reproducibility but was unrealistically severe, A particular aggregate, selected because it was poor, deteriorated rapid] y during freezing-and-thawing tests even in air-entrained concrete, showing a 40 per cent loss in dynamic modulus of elasticity in about 20 to 35 rapid cycles of freezing and thawing in water. In contrast, the actual field performance of concretes containing this aggregate has been excellent in above grade structures (grain bins apparently showing no deterioration in 30 years) and at least marginal in pavements. Non-air-entrained pavements, particularly those laid directly on plastic subgrade soil, began to show signs that the concrete was not durable when they had attained an age of 10 to 15 years (1s). The frost resistance of a concrete pavement con-


VmmEcIC AND LANDGREN drying of the concrete prior to the freezing-and-thawing test, improved the performance of the concrete by about 80 fold, This chert has an excellent field service record in pavements placed on sandy soil or on a granular base where partial drying of the concrete occurs but has a poor service record in concretes placed on poorly draining heavy clay soils.

taining this aggregate would undoubtedly be improved by air entrainment, a free-draining subbase, a smaller maximum size of aggregate, and the use of lower water-cement ratio concrete. These conditions, which are easily attainable, would be expected to improve the frost resistance of the concrete probably several fold. Satisfactory performance of such concrete as regards frost resistance might then be realized. Laboratory freezing-and-thawing tests of concretes made with a particular chert aggregate also have demonstrated that results obtained with presaturated aggregate do not reflect actual performance in properly designed pavements. Non-airentrained concretes made with saturated chert gravel and cured continuously moist failed rapidly on freezing and showed a 7.5 per cent loss in dynamic modulus in only 5 cycles of freezing and Air-entrained concretes also thawing. failed rapidly, showing a 70 per cent loss in dynamic modulus in 5 cycles. The ineffectiveness of air entrainment under such circumstances would be expected as the saturated chert aggregates fail because of the high internal hydraulic pressures and critical size effects previously discussed. Testing of this aggregate in a more appropriate way, that is, air drying of the chert prior to use and a period of

Based on the preceding discussions it is believed that the following major conclusions can properly be made: 1. The influence of aggregates on the durability of concrete depends upon the physical characteristics of the aggregates and certain properties of the mortar component of the concrete in a complex but understandable manner. 2. Certain test methods commonly used to evaluate the durabilit y of aggregates for use in concrete are frequently inappropriate and misleading as regards actual field performance of the concrete. 3, Through proper design, based .on the principles discussed, much can be done to improve significantly the actual performance of field concrete made with many aggregates rejected by commonly used tests.

REFERENCES (1) H. S. Sweet, Research on Concrete Durability as Affected by Coarse Aggregate, Proceedings, Am. Sot. Testing Mats., Vol. 48, p. 988 (1948). (2) R. F. Blanks, Modern Concepts Applied to Concrete Aggregate, Proceedings, Am. Sot. Civil Engrs., Vol. 75, p. 441, April (1949). (3) m. C. Powers, The Air Requirement of Frost-Resistant Concrete, Proceedings, Highway Research Board, Vol. 29, p. 184 (1949). (4) R, L. Blaine, C. M. Hunt, and L. A. Tomes, The Sorption Characteristics of Aggregates, Proceedings, Concrete Highway Research Board, Vol. 32, p. 298 (1953). (5) D. W. Lewis, W. L. Dolch, and K. B. Woods, Porosity Determinations and the Significance of Pore Characteristics of Aggregates, Proceedings, Am, Sot. Testing Mats., Vol. 53, p. 949 (1953). (6) T. C. Powers, Resistance to Weathering Freezing and Thawing, ASTM Special Techtrica! Publication No. 169, p. 182, 1956, Am. Sot. Testing Mats.


(7) G. Verbeck and P. Klieger, Studies of Salt Scaling of Concrete, Highway Research Board, BsdZetin150, p. 1 (19S7). (8) W. L. Dolch, Studies of Lhnestone Aggregates by Fluid Flow Methods, Pioceedin.gs, Am. Sot. Testing Mats, Vol. S9, p.



1204 (1959).
(9) R. Rhoades and R. Mielenz, Petrography


of Concrete Aggregate, Jozwnal, Am. Concrete Inst., p. 581, June, 1946, Pro(lo) ceedings, Vol. 42. A. E. Scheidegger, The Physics of Flow Through Porous Med~a, p. 150, University of Toronto Press (1957). (14)


T. C. PoWers,L. E. Coeeland, J. C. HaYes,

and H. M. Mann, (Permeability of Portland Cement Paste, Journal, Am. Con-


crete Inst., p. 285, Nov. 1954, Proceedings, Vol. 51. T. C. Powers, Basic Considerations Pertaining to Freezing and Thawing Tests, Proceedings, Am. Sot. Testing Mats, Vol. 55, p. 1132 (1955). T. C. Powers and T. L. Brownyard, Physical Properties of Hardened Portland Cement Paste, Journal, Am. Concrete Inst., p. 933, April (1947), Proceedings, Vol. 43. Report on Cooperative Freezing and Thawing Tests of Concrete, Highway Research Board, Special Repori No. 47 (1959). Bert Myers, Iowa State Highway Commission, Private Communication, April 24, 19.56.

Bulletins Published by the Research Department Research and Development Laboratories of the Portland Cement Association
100. List of Published Bulletins and Papers of the Research Department, May, 1959 (Also lists earlier research papers of the Portland Cement Association). 101. Determination of the Apparent Density of Hydraulic Cement in Water Using a Vacuum Pycnometer, by C. L. FORD.
Reprinted from ASTM Bulletin, No.
231, 81-04 (July,


102. Long-Time Study of Cement Performance in Concrete-Chapter 11. Report on Condition of Three Test Pavements After 15 Years of Service, by FRANK H. JACKSON.
Reprinted from Journal of the American ceedings, 54, 1017-1032 (1957-1958 ). Concrete

Institute (June, 1958); Pro-

103. Effect

of Mixing and Curing Temperature

Reprinted from Journul of tite American ceecttngs, 54, 1063-1081 ( 1957-1958).

on Concrete Strength,
Institute (June,






Successive Determination of Manganese, Sodium and Potassium in Cement by Flame Photometry, by C. L. FORD.
Reprinted from ASTM Bulletin, No. 233, 5%62 (October, 195S).

105. The Surface


of Tobermorite,

37, 714-734 (MM, 1959).

KANTRO and C. H. WEISE. Reprinted from Canadian

of Che?ntstw,

106. The Flow of Water in Hardened


Cement Paste,
40, 308-323

by T. C.

POWERS, H. M. MANN and L. E, COPELAND. Reprinted from HighwaV Research Board Special

107. The Ball-Mill Hydration of Tricalcium Silicate At Room Temperature, by D. L. KANTRO,STEPHEN BRUNAUER and C, H, WEISE.
Reprinted from Journal of Collofd Science,
14, 3133-3~13 0959).

108. Quantitative Determination Cement by Combined X-Ray

of the Four Major Phases of Portland and Chemical Analysis, by L. E. CoPE-

LAND, STEPHEN BRUNAUSR, D. L. KANTRO, EDITH G, SCHULZ and C. H. WEISE. Reprinted from Anatvticat Che?rdstw, 31, 1521-1530 (September, 1959).

109. Function

of New PCA Fire Research Laboratory,

dnd DeueCopment Labora-

Reprinted from the Journal of the PCA Research

tories, 1, No. 2, 2-12 (May, 1959).

110, Capillary

Continuity or Discontinuityy in Cement Pastes,

1, No. 2, 30-40 (May, 1959).

by T. C.

POWERS, L. E. COPELANDand H. M. MANN. Reprinted from the Jowrtat of the PCA Research

and Devetop?nentLabora-

111. Petrography

of Cement and Concrete, by L. S. BROWN. Reprinted from the Journa[ of the PCA Research and Development
torfes, 1, No. 3, 23.34 (September, 1959),


tThe Gravimetrie Determination 112. \ Cement, by C. L. FORD.

Reprinted from ASTM Bulletin,






No. 245, 71-75 (April,

113. Quantitative Determination of the Four Major Phases in Portland Cement by X-Ray Analysis, by STEPHEN BRUNAU~, L. E. COPELAND,
1091-1100 (1959) .

EDITH G. SCHULZ. Societu for Testing Materials, 59,

Reprinted from Proceedings of the American

114. Long-Time Study of Cement Performance in Concrete-Chapter Concrete Exposed to Sea Water and Fresh Water, by I. L. TYLER.
Reprinted from Journal of the American Proceedings, 56, 826-336 (1960). Concrete Institute (March,


115. A Gravimetric Method for the Determination land Cement, by C. L. FORD.

Reprinted from ASTM Bulletin,

of Barium Oxide in Port.


No. 247, 77-S0 (July,

116. The Thermodynamic Functions for the Solution of Calcium Hydroxide in Water, by S, A. GREENBERG and L. E. COPELAND.
Reprinted from Journal of Ph@cal Chemistry, 64, 1057-1059 (August, 1960).

117. Investigation

of Colloidal Hydrated
from Journat of Ph@cal


I. Volubility Products,
ANDERSON. 64, 1151-1156 (September, 1960).



118. Some Aspects of Durability and Volume Change of Concrete for Prestressing, by PAUL KLIEGER.
Reprinted from the Journal of the PCA Research tortes, 2, No. 3, 2-12 (September, 1660). and Development Labora.

119. Concrete

Mix WaterHow Impure Can It Be? by HAROLD H, STEINOUR.

and Development Laboraand 1960);

Reprinted from the Journal of the PCA Research tories, 2, No. 3, 32-50 (September, 1960).

120. Corrosion
G. J.

of Prestressed

Wire in Concrete,

by G. E.

MONFORE (November,

V~RB~CK. Reprinted from Journal of the Amertcan Concrete Proceedings, 57, 491-515 (September, 1960).

121. Freezing

and Thawing Tests of Lightweight Aggregate

J, A. HANSON. Concrete Institute Reprinted from Journal of the American Proceedings, 57, 779-796 (1961).




122. A Cement-Aggregate Reaction That Occurs With Certain Sand-Gravel Aggregates, by WILLIAM LERCH.
Reprinted from the Journal of the PCA Research ties, 1, No. 3, 42-50 (September, 1959). and Development Laborato.

123. Volume









HAROLD ROPER. Reprfnted from the JournQl of the PCA Research tories, 2, No. 3, 13-19 (September, 1960).

and Development


124. A Short Method for the Flame

sium, Manganic,
by C. L. FORD.

Photometric Determination of MagneSodium, and Potassium Oxides in Portland Cement,

No. 250, 25-29, (December, 1980).

from ASTM BuUetin,

125. Some



of the Hydration

of Portland



Reprinted from the Journal of tories, 3, No. 1, 47-58 (January,

thePCA Research

and Development


126. Influence of Physical Characteristics of Aggregates on Frost sistance of Concrete, by GEORGE VERBECK and ROBERT LANDGREN.


from Proceedings

of the American

SocWiV Jor resting