in the first few days. That there is a variation with that petrography and mineralogy are of funda-
water/cement ratio is confirmed by theory and experi- mental importance and that, without a knowledge
ment[16] from which it appears that calculation by of these, materials from different sources but of
the method of mixtures applied to a simple mixture similar composition cannot be compared. Descrip-
of cement powder and water is not adequate: tion, source, thermal expansion coefficients, texture
judging by Siegmann's figures[16] it is likely to and mineral composition are given for all the
lead to an over-estimate, particularly during materials tested. In many cases the proportions of
hardening. the various mineral constituents are given. It is
stated that rocks from various sources but of similar
Thermal diffusivity composition have approximately the same expan-
This does not appear to have been measured sivities. It seems, however, that this is true only if the
directly on cement paste although there are a number of minerals present is small because, in the
number of studies on concrete to which reference case of igneous and volcanic rocks particularly,
will be made later. Since it is a function of con- there is considerable variation (these "have several
ductivity, specific heat and density, it will vary minerals as part of their essential compositions
according to different influences on these. Since and a large number of possible accessory and altera-
both conductivity and specific heat increase with tion minerals")[19]. Furthermore, "the cubical
moisture content then one might expect a reduced expansion of rocks and minerals, in general, is not
influence on diffusivity. There are no figures directly related to the linear expansion."[20]
available to the author for such relative effects but The general study of petrography is dealt with
some concerning the influence of temperature[14] of course in many appropriate geological textbooks
show that diffusivity decreases with increase in but its application to aggregates in particular is
temperature. dealt with in two papers at least[21, 22]. More
recently, the contribution of petrography to the
Conclusion study of the physical and mechanical properties of
A general conclusion on the thermal behaviour rocks has been assessed[23] and an index of quality
of cement paste is difficult to arrive at but it is defined which i s intended to facilitate the inter-
evident that statement of thermal properties is in- pretation of qualitative data. This might be useful
adequate unless it is related to age, temperature and in the consideration of crushed rock aggregate
moisture content at the least. The thermal behaviour sources.
of hydrating or hydrated cement is complex and A summary of thermal coefficients is given in
any values used, based on published figures, would Table 3. This gives an indication of the range of
need to be carefully qualified. values which might be expected. Other lists have
also been published[2, 25].
AGGREGATES--THERMAL BEHAVIOUR Table 3. (after Skinner[24]). Thermal expansion of rocks.
There is at least one general survey[17] of the Temperature interval 20-100°C
thermal behaviour of aggregates. The quoted range Average linear
of thermal expansion is 0.9-16.0x 10-6/°C but expansion No. of
typical values are not given for the other thermal R o c k type coefficient determinations
properties. Emphasis is given to possible aniso- Granites a n d rhyolites 8 + 3 × 10-6 21
tropy, particularly in rocks containing calcite, a Andesites a n d diorites 7+ 2 4
feature demonstrated most effectively elsewhere[18]. Basalts, gabbros a n d
diabases 5.4 _+ 1 10
It is a factor which cannot be ignored although it
Sandstones 10 ___2 I0
might be expected that any such tendency will be Quartzites 11 2
reduced in concrete since the aggregate particles Limestones 8+ 4 20
Marbles 7 _+2 9
will probably have a random orientation. On the Slates 9+ 1 3
Other hand particle shape may result in a "pre-
ferred" orientation--flaky or elongated particles
might tend to behave in such a manner. The remarks above are generally applicable to
It is again convenient to deal with the thermal crushed rock aggregates and the quoted references
properties separately. deal with these in particular. However, a great deal
of concrete is made with gravel aggregates and
Thermal expansion relevant information is difficult to obtain. Gravel
The results of a comprehensive study on the sources may be more variable but arguments
thermal expansion of materials for concrete analagous to the mineral composition of crushed
aggregates have been reported[19]. It is obvious rocks can perhaps be applied to the influence of
170 A. L. Marshall
gravel composition. A method has in fact been Values measured for some American dams[30]
described[26] to facilitate the measurement of the are given in Table 4 to indicate the possible range
cubical expansion of materials (such as gravels and while some low temperature values have also been
sands in the present context)--the properties of reported [ 10].
crushed rock can be directly measured,
It is not clear to what extent porosity might Specific heat
affect the thermal expansion of aggregates but it Values appear to be less readily available than
does affect, with permeability and absorption, for the other thermal properties, but the values
other properties such as chemical stability[22]. given in Table 4 suggest that, although there i~
Micro-cracking may also have some influence since some variation, it is less dependent on composition
it can affect the behaviour of the material-- than are other properties.
apparently similar rocks from similar sources can be
quite different in behaviour, particularly in freezing Table 4. (Bureau of Reclamation data[30]). Thermal conduc-
and thawing tests[21 ]. Porosity and micro-cracking tivity and specific heat of some rocks for concrete aggregate
also, as one might expect, have an influence on the
Thermal
index of quality[23]. A table has been published[27] Specific heat Conductivity
(based on a survey of a number of references) which Aggregate (kcal/kg °C) (kcal m/m z h (3
shows a tendency to a reduction in coefficient of
Quartz 0'175 4.45
expansion with a reduction in silica content, but Granite 0"171 2.50
the relationship may be more complex since the Dolomite 0.192 3-70
coefficient of expansion of silicates varies[24] and, Limestone 0.202 2-78
Basalt 0.183 1.47
in some cases, they are also anisotropic. Marble 0.209 2-11
If one can summarise, the following quotation Feldspar 0.194 2-00
in ay serve: [18 ]
" . . . the following features are of decisive im-
portance for thermal dilatation : CONCRETE--THERMAL BEHAVIOUR
I. mineral composition--the thermal dilatation Having discussed the thermal properties of tile
of rock is actually the resultant of thermal main constituents of concrete it is appropriate n o ~
dilatations of minerals of which the rock to consider the composite material. As a guide, it
consists, would be difficult to do better than to reiterate
2. rock structure, some of the main conclusions reached in the
Boulder Canyon Project report :[14]
3. rock texture."
I. Thermal properties of concrete are not
affected by age.
Thermal conductivity 2. Rock type of coarse aggregate is the most
The results of tests on a number of rocks have important variable affecting conductivity.
been reported[28]. Some display a reduction in 3. Original water content has much less effect
conductivity with increase in temperature (for on conductivity than the type of coarse
example calcite, marble, limestone, dolomite and aggregate. Increased (mixing) water content
some holocrystalline rocks) while some others produces decreased conductivity.
exhibit little change or a slight increase with increase
4. High conductivities decrease with an increase
in temperature (anorthosites and diabasic rocks).
of temperature, average conductivities show
Elsewhere[29] values are listed from a number of
little change with variation of temperature and
sources and it is pointed out that such factors as
low conductivities increase with an increase
bedding, schistosity and fracturing may introduce
of temperature.
marked differences in different directions. It also
appears that "even for relatively well-defined sub- 5. Rock type of coarse aggregate produces small
stances, measurements by qualified investigators variation in specific heat.
show startling discrepancies"[29] some of which 6. Specific heat varies with original water
must be attributed to experimental error. From content. This variation, per unit of weight, i~
further tabulated values it is also evident that several times greater than for other materials.
wetting and compression have a significant effect However, since the variation in weight of
on the thermal conductivity of certain rocks water is limited, the total effect of varied
although the compressive load quoted is fairly high water content is limited to about ten per
(68.95 MN/m2). Increase due to wetting appears cent. Specific heat increases as water content
to be not necessarily due to porosity. increases.
The Thermal Properties o f Concrete 171
7. Specific heat increases as temperature in- A few mathematical models have been suggested.
creases. Two proposed by Dougill[34] are:
8. Diffusivity varies over a wide range due to Ea
change in rock type.
°co geta + (1 -- g)ctp 4 ~p get, + (1 -- g)ctp
9. Diffusivity decreases as (original) water con- = 2 Ea (1)
tent increases. 2 ~ g+2(l--g)
/~p
10. Diffusivity decreases as temperature in- and
creases. (~c - ~ta) = (~-- °ta)(1 - g)" (2)
By and large these conclusions confirm some of A third, postulated by Hobbs[35] is:
the comments made previously and the report is an (ot~ - cQg2k a
excellent guide in the absence of specific informa- ~c = ~tm k,, + k , + g ( k , - k,,) (3)
tion. Some qualification is necessary, however,
and this relates principally to the influence of age. is the coefficient of thermal expansion, E is
The tests reported were carried out on relatively Young's modulus, k the bulk modulus, g the
mature concrete and it is possible, as will be shown, aggregate volume concentration and n an exponent.
that during the first few days there are significant The suffices c, m, p and a refer to concrete, mortar,
changes. Furthermore, the report does not deal paste and aggregate respectively.
with thermal expansion. Consequently it should be At very early ages, of course, there is a rapid
useful to consider some of the other available change in the properties of concrete and this is
literature. thought to apply to thermal expansion[27]. It
might be expected that creep (or stress relaxation)
Thermal expansion in the cement paste or mortar would be fairly high,
If concrete is regarded as a cement paste matrix but at the same time strengths are low and rigidity
with aggregate filler then one might expect the of the paste is very low relative to the aggregate.
behaviour of the paste to be modified by the filler. There might then be some internal disruption of
The degree of modification will depend on the the material where there are significant temperature
relative coefficients of expansion of the two (or changes and differences in coefficients of expansion.
more) materials and their relative rigidities. For Hence the relationships given above may not be
example,[1] the thermal expansion of concrete applicable at early ages, although some modification
relative to paste is reduced relatively more with to the elastic modulus might take account of it.
aggregates of low coefficients of expansion than
with those of high expansion. Bonnell and Harper Thermal conductivity and diffusivity
[25] confirm the influence of aggregate on the Useful guidance may be obtained from the
thermal expansion of concrete. Unfortunately they Boulder Canyon Project report[14] on the relative
did not study the influence of water/cement ratio influences of the major variables on the composite
and, in their oft-quoted work, used concretes of the and Table 5 is therefore reproduced from it. One
same slump. There may have been some masking quoted range of values of diffusivity is 0.00186 m2/h
of effect, therefore, particularly at intermediate (basalt aggregate) to 0.00744 m2/h (chert aggregate)
humidities. The Boulder Canyon Report gives an empirical
The fine aggregate has an important influence[31 ] method for predicting conductivity and specific heat
but, as an approximate indication, there is some based on "the verified assumption that various con-
evidence[32] to show that aggregates affect the crete materials contribute to the conductivity or
thermal coefficients of mortar and concrete roughly specific heat of concrete in proportion to the amount
in proportion to their quantity in the mixture of material present." The proportions are by weight
(absolute volume of coarse aggregate expressed as and percentage. Reference should be made to the
a percentage of the volume of concrete). It has also report for the relevant values but it is obvious that
been concluded[2] that "The coefficient of thermal rock type has the major influence: the factors for
expansion of concrete can be determined from the all the rock types quoted are greater than the
weighted averages of the coefficient of thermal cement factors apart from their greater proportion
expansion of the various constituents if the correct by weight.
moisture effect can be determined." It is important This is evident from theoretical considerations
to reduce thermal expansion differences between [11] although the relationship to weight is not as
aggregate and matrix[33] and it is therefore recom- simple as is expressed above:
mended that the coefficients be determined for the k c = km(2M-M2)q kmka(l--M) 2
constituents as well as the composite. k . M +k,.(l - M )
172 A. L. Marshall
Table 5. (Boulder Canyon Project report[14]). Relative effec'ts of major cariables on thermal
properties of concrete
Thermal properties
Specific
Test conditions Conductivity heat Diffusivity
where M = 1 - v " ( l - p ) and p is the volume of period when heat is being generated in the cement
mortar per unit volume of concrete, k is thermal water reaction. However, it is perhaps more true
conductivity and the suffices m and a refer to to say that, within the normal range, it is the cement
mortar and aggregate respectively. It has also been paste which is affected by these, particularly where
demonstrated theoretically[36] that such relation- a non-absorbent aggregate is being used.
ships are largely independent of variations in pore The most significant influence on concrete appears
or particle geometry, provided the inclusions do to be variations in the type of aggregate. Con-
not extend as continuous layers or columns sequently, if one is looking for accuracy in estimat-
through the material. ing particular thermal properties, it must be related
During the first few days of the life of the material to the accuracy of measurement of the properties
it can be inferred that the important chemical of the aggregate. Inspection of Table 3 demon-
changes taking place which convert the material strates that there are wide variations within in-
from a viscous liquid or "soil" to a (relatively) rigid dividual rock types. Hence it follows that the type
solid will have a significant effect on thermal of aggregate must be very carefully defined before
properties. This appears to be borne out by some proper use can be made of published values. Here
experimental work[37, 38] which indicates that it can be argued, of course, that reasonable estimates
thermal conductivity decreases by the order of can be made but one must then define "reasonable".
30% during the first seven days after casting. For example, is 5~o error, or 10~, or even 20°,,
acceptable? Because, apart from attempting to
Specific heat understand the behaviour of the material, the chief
The Boulder Canyon Report[14] gives factors value of figures is in attempting to estimate par-
for calculating specific heat. It appears, J37, 38] ticular properties. Even the use of formulae such
however, that there is a reduction of over 2 0 ~ in as those quoted in the paper is dependent on the
the first few days. Reference[39] has been made to actual values fed into the formulae.
rapid changes in specific heat of adiabatically Despite the body of information already avail-
curing concrete at ages earlier than 14 days, a able there appears to be no substitute for experi-
feature which has been echoed[40]. There is thus mental determination of the required properties,
some doubt as to what the values are during the using the actual materials to be employed in con-
early stages. This may be due in part to the " n o r m a l " struction. Here again there must be reservations
practice of using the method of simple mixtures because environmental influences are extremely
For composite calculations, a practice which may important and it is difficult in many cases to predict
be doubtful[16]. the natural environment with real accuracy al-
though for design purposes the extreme values may
be of more concern. The size of structural member
CONCLUSION
is also significant here: for example, the smaller it
It is evident that a number of factors affect the is, the more rapid is the heat transfer and so the
thermal behaviour of concrete, in particular, age, more rapid is the movement and the less the
temperature and moisture content. The two latter possible benefits of creep.
are determined by the environment in which the There is still much to be learned and this is
material is situated, other than during the initial especially true of early-age behaviour. Very little
The Thermal Properties of Concrete 173
evidence is available but what there is suggests Acknowledgement--This paper results from a review of the
substantial variation in the first few days. Evidently early-age behaviour of concrete prepared at the suggestion
of The Science Research Council whose award of a travel
there is scope for further work and the author's grant in the United Kingdom has made it possible and to
own p r o g r a m m e is directed towards throwing whom the author's thanks are due. His thanks are also due
further light on the subject. However, even with to the many people in this country and overseas who provided
information and references on which the review is based.
mature concrete there are substantial gaps in the
available data which could usefully be filled.
REFERENCES
I. S. L. MEYERS, Thermal expansion characteristics of hardened cement paste and of
concrete, Highway Rsch. Board, Proc. 30th Annual Meeting, Washington, 1951,
pp. 193-203.
2. L . J . MITCHELL,Thermal expansion tests on aggregates, neat cements and concretes,
Proc. Amer° Soc. for Testing Materials, 53, 963 (1953).
3. R. A. HELMU'm, Dimensional changes of hardened portland cement paste caused by
temperature changes, Proc. Highway Rsch. Board, 40, 315 (1961).
4. G . J . V~RBECKand R. H. H~LMU~, Structure and physical properties of cement paste,
Proc. 5th Int. Symp. on Chem. of Cement, Tokyo 1968, Part III Properties of Cement
Paste and Concrete Vol. III, pp. 1--44. Cement Assocn. of Japan, 1969.
5. M. JAKOn, Heat Transfer, Vol. I, pp. 758, Wiley, New York (1949).
6. P. J. ARNOLD, Thermal conductivity of masonry materials, Current Paper 1/70, Bldg.
Rsch. Station, reprinted from Journal 111VE, 37, 101 and 117 (1969).
7. H.B. JESPERSON,Thermal conductivity of moist materials and its measurement, J. IHVE,
21, 157 (1953).
8. R. CADIER~tW_S,and J. GENEVAV,The thermal conductivity of materials, Library Com-
munication No. 633, Bldg. Rsch. Station, April 1953, translated from Ann. I'ITBTP,
5(52-53), 471 (1952).
9. O. Kazsc/-mR and H. ESOORN, The heat transfer in damp, porous substances of different
structures,' Lib. Comm. No. 763, Bldg. Rsch. Station, July 1956, translated from Forsch.
Geb. Ing., 22(1) (1956).
10. A . E . LEtcrz and G. E. MONFORE, Thermal conductivities of portland cement paste,
aggregate and concrete down to very low temperatures, J. PCA Res. Devel. Labs., 8,
27 (1966).
11. D. CAMPnELL-ALLENand C. P. THORNE, The thermal conductivity of concrete, Mag.
Concr. Res., 15, 39 (1963), Discuss. MCR, 16, 233 (1964).
12. L. J. MITCrtELL, Hardened Concrete--Thermal properties, Significance of tests and
properties of concrete and concrete-making materials, ASTM, STP 169-A, 202 (1966).
13. D. L. KAWrRO, C. H. WEISE, and S. BRtmAUER, Paste hydration of beta-dicalcium
silicate, tricalcium silicate and alite, HRB Special Report 90, Symp. on Structure of
Portland Cement Paste and Concrete, Washington, D.C. 1966. pp. 309-327.
14. Thermal properties of concrete, Boulder Canyon Project Final Report, Bulletin No. 1,
Part VII, 1940.
15. U. DANIELSSON,Conduction calorimeter studies of the heat of hydration of a portland
cement, Swed. Cement. and Concr. Res. Inst., Handlingar NR38, 1966.
16. J. SIEGMA~CN,Spezifische W/irme von Zementpasten und Reaktionsw/irme des Betons
w/ihrend des Abbindens und Erh/irtens, Zement Kalk Gips, Oct. 1954, pp. 390-395.
17. H. K. CooK, Concrete Aggregates--Thermal properties, Significance of tests and
properties of concrete and concrete-making materials, ASTM STP 169-A, 476(1966).
18. I. DUDKOVA, Correlation between texture and thermal dilatation of crystalline lime-
stones, Acta polytech-Pr. CSL. V.U.T. Vf~zk.Praze, L 2, 5 (1966).
19. W. H. JOHNSON, and W. H. PARSONS, Thermal expansion of concrete aggregate
materials, J. Res., Natl Bur. Stds, 32, 101 (1944).
20. H. S. SWEET, Physical and chemical tests on mineral aggregates and their significance,
Symp. on MineralAggregates, ASTM, STP No. 83, 1948, pp. 49-73.
21. K. MATHER,and J. W. SILVERMAN,The petrography of aggregates for mass concrete in
Aggregates for Mass Concrete, Bull. No. 39, Waterways Experiment Station, Corps
of Engrs., US Army, August 1954, pp. 1-17.
22. R. RHOADES and R. C. MIELENZ, Petrographic and Mineraiogic characteristics of
aggregates, ASTM, STP No. 83, 1948, pp. 20~8.
23. C. TOURENQ and D. FOURMAINTRAUX,Contribution de le p6trographie a l'6tude des
propridt6s physiques et m6caniques des roches, Bull. Liais. Labs. Pants et Chauss6es,
No. 50, 157 (1971).
24. B.J. SKINNER, Thermal expansion, Section 6, Handbook of Physical Constants, Geol.
Soc. of America, Memoir 97, 1966, pp. 75-95.
25. D . G . R . BONNELLand F. C. HARPER, The thermal expansion of concrete, Nat. Bldg.
Studies, Tech. Paper No. 7, HMSO 1951.
174 A. L. Marshal[
Die thermischen Eigenschaften yon Beton werden besprochen. Der Einfluss von
F a k t o r e n wie Alter, Verh/iltnis Wasser zu Zement, T e m p e r a t u r und Feuchtigkeitsgehalt
hinsichtlich der spezifischen W/irme, thermische A u s d e h n u n g und Diffusionskoeffizient
yon Zementpaste, Zuschl/ige und Beton werden behandelt.