Heat Loss Compensation in Semi-Adiabatic Curing Test of Concrete

(ProQuest: ... denotes formulae omitted.)

INTRODUCTION

During curing of concrete, heat is generated by the chemical reactions between the cementitious materials and water. Accumulation of heat within the concrete mass results in a substantial temperature rise, which could be as high as 60 °C (108 °F).1-3 Due to heat dissipation through the concrete surfaces to the surroundings, the temperature of the concrete mass is not uniform and would eventually drop to the ambient level, leading to uneven thermal expansion and contraction of the concrete. Restraints against such thermal expansion and contraction, which could be external if the concrete structure is connected to movement restraints or internal if the temperature difference within the concrete mass is substantial,4-6 would induce tensile strains in the concrete that might be large enough to cause early thermal cracking. The thermal cracks so caused would adversely affect the appearance, water-tightness, integrity, and durability of the concrete structure. To avoid or at least alleviate this problem, limits are usually imposed on the maximum temperature of the concrete (typical limits range from 70 to 80 °C [158 to 176 °F]) and the maximum temperature difference in the concrete (typical limits range from 20 to 30 °C [36 to 54 °F]).

The two key factors governing the temperature rise of the concrete during curing are the heat generation from the concrete and the heat dissipation to the surroundings. Hence, the most effective means of reducing the maximum temperature and/or maximum temperature difference are to minimize the heat generation from the concrete by redesigning its mixture composition and to exercise appropriate temperature control measures such as installation of internal cooling pipes and application of external heat insulation.4-6 In any case, first of all, it is necessary to determine the amount of heat generation from the concrete before any decision can be made on how to satisfy the imposed temperature and temperature difference limits.

The amount of heat generation from a concrete mixture is usually expressed in terms of the adiabatic temperature rise of the concrete, that is, the temperature rise of the concrete under adiabatic condition. There are simple guidelines for estimating the adiabatic temperature rise of typical concrete mixtures.4,5 The adiabatic temperature rise, however, is influenced by many factors, including the mixture proportions; the types, sources, and fineness of the cementitious materials; and the placing temperature and is therefore difficult to be estimated accurately using simple guidelines. For accurate determination of the adiabatic temperature rise, measurement by means of an adiabatic curing test is preferred. In fact, in some construction projects, adiabatic curing tests are required and limits are imposed on the adiabatic temperature rise of the concrete to keep the amount of heat generation at an acceptable level.

To perform the adiabatic curing test, a curing chamber capable of preventing heat dissipation from the concrete specimen to the surroundings is needed. Because, in reality, it is impossible to provide perfect heat insulation to completely stop heat dissipation, the usual practice is to provide good heat insulation and at the same time control the temperature around the insulated concrete specimen. This is done by controlling the temperature in the curing chamber to be the same as that of the concrete specimen so that there is no or very little temperature difference and the heat dissipation becomes minimal. A curing chamber specially designed to perform such function is generally referred to as an adiabatic curing chamber or adiabatic calorimeter. Water is the most commonly used medium to control the temperature around the concrete specimen7-13 but air has also been used.14,15 Basically, the concrete specimen wrapped with insulating materials is placed inside the temperature control medium (the water or air), a thermometer or thermocouple is installed in the concrete specimen to measure its temperature, and the temperature reading is fed into a temperature control system to control the temperature of the surrounding medium. To ensure uniformity of the temperature around the concrete specimen, the water or air is kept in active circulation.

An adiabatic curing chamber fitted with a temperature control system, whether analogue-controlled by an electronic circuit or digital-controlled by a microprocessor (the more advanced ones like those used by the authors are fully computerized), is fairly sophisticated. It is best for use in an air-conditioned and well-equipped laboratory. On construction sites, for convenience and simplicity, a curing chamber not fitted with any temperature control system is often used instead. As a result, the curing tests conducted on site with no temperature control applied are not truly adiabatic. These curing tests are called semi-adiabatic or quasi-adiabatic curing tests. Several different versions of semi-adiabatic curing tests have been developed.15-18 In general, due to heat loss, that is, heat dissipation from the concrete specimen to the surroundings, the measured temperature rise is lower than the adiabatic temperature rise by an error dependent on the specimen size and the effectiveness of the heat insulation provided. With similar heat insulation provided, the error is larger for a smaller specimen size and smaller for a larger specimen size, while for the same specimen size, the error is larger with less heat insulation provided and smaller with better heat insulation provided. Consequently, the measured temperature rise by a semi-adiabatic curing test is highly dependent on the test setup.

Based on the authors' own experience, the measured temperature rise by a semi-adiabatic curing test could be lower than the adiabatic temperature rise by more than 12 °C (22 °F) if the specimen size is smaller than 1.0 m3 (1.3 yd 3 ) or the heat insulation provided is no good. It has therefore been suggested to compensate for the heat loss when estimating the adiabatic temperature rise from the results of a semi-adiabatic curing test.15,17 However, there is still no established method for heat loss compensation. In the present study, a new heat loss compensation method of estimating the heat loss from the spatial variation of the measured temperature (that is, the temperature profile across the concrete mass) and the time variation of the measured temperature (that is, the temperature-time curve), and compensating for the heat loss accordingly is proposed. Parallel adiabatic and semi-adiabatic curing tests were carried out to verify the applicability of the proposed heat loss compensation method.

RESEARCH SIGNIFICANCE

Semi-adiabatic curing tests are commonly adopted in construction projects to estimate the temperature rise of curing concrete and to check compliance with any imposed limit on the adiabatic temperature rise of the concrete to be cast. Due to heat loss, however, the measured temperature rise is always lower than the adiabatic temperature rise by an error dependent on the test setup. The proposed heat loss compensation method, which has been shown in this study to be capable of reducing such errors to less than 1.3 °C (2.3 °F), can help to resolve the problem and render the semi-adiabatic curing test a much more reliable test than before.

HEAT LOSS AND COMPENSATION

Consider a curing concrete specimen of volume V wrapped all round by insulating materials. Let the actual measured volumetric mean temperature of the concrete be TV , the placing temperature be TP , and the true adiabatic temperature rise be TG . The actual measured temperature rise (TV - TP ) is lower than the true adiabatic temperature rise TG by an amount proportional to the heat loss HL , as given by

... (1)

where . and c are the density and specific heat of the concrete, respectively. From this equation, the adiabatic temperature rise TG may be obtained as

... (2)

Applying the Fourier Law19 to the heat conduction through the insulating materials, the rate of heat loss may be determined as

... (3)

where k is the overall thermal conductivity of the insulating materials, TS is the surface mean temperature of the concrete, TA is the temperature of the surrounding medium (in the case of a semi-adiabatic curing test, TA is the same as the ambient temperature), and t is the time after placing of the concrete. Integrating with respect to the time t, the heat loss HL may be calculated as

... (4)

Substituting the previous formulas for the heat loss into Eq. (2), the adiabatic temperature rise TG is derived as

... (5)

The term (k/V. c ) in Eq. (5) represents the heat loss characteristic of the test setup. Denoting this term by ., Eq. (5) becomes

... (6)

Hence, to evaluate the heat loss compensation to be applied, it is necessary to determine the heat loss characteristic . of the test setup and measure the mean temperature values TV and TS of the concrete during the test.

Even in an adiabatic curing test, because it is difficult to control the temperature of the surrounding medium to be exactly the same as that of the concrete specimen, there may still be some heat loss to the surroundings. Higher accuracy would be obtained if heat loss compensation were applied in the adiabatic curing test. Morabito and Barberis8 suggested to determine the heat loss characteristic after the adiabatic curing test has been completed by setting the temperature of the surrounding medium at an arbitrary low level to allow cooling of the concrete specimen and measuring the corresponding rate of temperature drop with time. Having determined the heat loss characteristic, they assumed a constant temperature difference between the concrete specimen and the surrounding medium during the adiabatic curing test and compensated for the heat loss in the estimation of the adiabatic temperature rise.

It is suggested herein that for a semi-adiabatic curing test, a similar method to that of Morabito and Barberis8 may be applied to determine the heat loss characteristic. Differentiating Eq. (6) with respect to the time t

... (7)

At a sufficiently long time after placing of the concrete, there should be no further heat generation from the concrete and the adiabatic temperature rise TG would become fairly constant. Based on the authors' own experience, this happens at approximately 120 hours after placing of the concrete. Hence, when t is greater than 120 hours, the left side of Eq. (7), that is, the rate of adiabatic temperature rise, would become negligibly small. Neglecting the further adiabatic temperature rise when t is greater than 120 hours and moving . to the left side, Eq. (7) becomes

... (8)

Therefore, the heat loss characteristic . may be determined from the temperature difference (TS - TA ) and the corresponding rate of drop of TV with t when there is no further heat generation from the concrete. In practice, the authors always conduct the semi-adiabatic curing tests up to at least 144 hours after placing of the concrete and calculate three estimated values of . within the time interval of 120 to 144 hours after placing of the concrete to obtain an average value of .

Regarding the volumetric mean temperature TV and surface mean temperature TS of the concrete, if the temperature distribution within the concrete specimen is known, they may be obtained as

... (9a)

... (9b)

where V and A are the volume and surface area of the concrete specimen and T(x, y, z) is the temperature at the point with spatial coordinates (x, y, z ).

Due to heat dissipation through the concrete surfaces, the temperature distribution is not uniform. In some tests conducted on site, only the temperature at the center of the concrete specimen was measured and both TV and TS were taken to be the same as the temperature at the center (this practice may be called one-point temperature measurement). The authors have measured the temperature distributions in a number of specimens, however, and found that in some cases, the temperature variation within the specimen could be quite large (some of the results are presented in this paper). Hence, the practice of one-point temperature measurement is rather crude.

The authors always use cubical specimens in both adiabatic and semi-adiabatic curing tests. The use of cubical specimens has the advantage that if the test setup is similar in all the three coordinate axis of the concrete cube, the temperaturedistribution along the three coordinate axis may be assumed to be the same, thus requiring fewer measurement points to determine the temperature distribution within the concrete specimen. It is suggested herein to measure the temperature of the concrete specimen at the following four points: the center of mass, the center of one face, the midpoint of an edge, and one corner, as shown in Fig. 1 (this may be called four-point temperature measurement). Alternatively, if only the temperature at two points could be measured, the measurement should be made at the center of mass and the center of one face (this may be called two-point temperature measurement).

Let the temperature readings at the center of mass, the center of one face, the midpoint of an edge, and one corner be Tm , Tf , Te , and Tc , respectively. From these, the temperature distribution T(x, y, z) may be obtained by curve fitting as a polynomial function of the spatial coordinates x, y, and z, and then the values of TV and TS can be determined using Eq. (9a) and (9b) as linear functions of Tm , Tf , Te , and Tc , as will be shown later in the paper. Finally, the adiabatic temperature rise TG given by Eq. (6) can be evaluated by dividing the time into many small time steps and performing the required integration numerically as depicted by the following equation

... (10)

where (TS - TA)i is the temperature difference at time step i , .t i is the time interval at time step i, and n is the number of time steps. In practice because the temperature difference varies quite slowly with time, a constant time interval of 5 minutes should be small enough to yield accurate results.

TEST SETUP

Adiabatic curing test

A computer-controlled environmental chamber was used for the adiabatic curing test. It was capable of controlling the air temperature inside the chamber within ±0.2 °C (±0.36 °F) of the preset value. The concrete specimen was cubical and had a dimension of 0.25 m (0.82 ft). It was cast into a plastic bag adhered to the inside of an insulated mold, as shown in Fig. 2. The insulated mold was specially designed such that the mold, which was made of 18 mm (0.7 in.) thick plywood to provide rigidity and had a relative density of 0.98, was fabricated as the outer shell, while the insulating layer, which was made of 50 mm (2.0 in) thick phenolic polymer boards and had a relative density of 0.04, was affixed inside the mold. Because the insulating layer had smaller mass and thus smaller heat capacity than the mold, this design had the advantage that the insulated mold would absorb less heat from the concrete than the conventional design of putting the mold in contact with the concrete. The plastic bag was sealed after concrete casting to avoid leakage and the mold was then placed near the center of the chamber.

Four thermocouples were installed to measure the temperature of the concrete specimen, as shown in Fig. 1. They were connected to a computer outside the chamber, which recordedthe temperature readings and calculated the target temperature that the chamber should have been preset to. In this study, the target temperature of the chamber was calculated as Tm - 0.5 °C (0.9 °F). It was purposely set slightly lower than the temperature of the concrete to avoid feeding heat from the chamber to the concrete. The target temperature was updated at 1-minute intervals and was input to the computer inside the chamber for controlling the air temperature in the chamber.

Semi-adiabatic curing test

The test setup for the semi-adiabatic curing test is quite simple. It consists only of an insulated mold, several thermo-couples for measuring the temperature of the concrete and the ambient temperature, and a data logger for recording the temperature readings. To evaluate the effects of specimen size, two alternative designs for the insulated mold were developed, one for a 0.5 m (1.64 ft) cube specimen and the other for a 1.0 m (3.28 ft) cube specimen, as shown in Fig. 3. The insulated mold for the 0.5 m (1.64 ft) cube specimen was designed similar to that for the adiabatic curing test. It comprised a mold made of 18 mm (0.7 in.) thick plywood and an insulating layer made of 100 mm (3.9 in.) thick phenolic polymer boards affixed inside the mold. Becuase the lightweight insulating materials were not strong enough to withstand the pressure from the concrete, however, the insulated mold for the 1.0 m (3.28 ft) cube specimen could not be designed in the same way and had to be designed to have the insulation affixed outside the mold. The actual design adopted was a mold made of 18 mm (0.7 in.) thick plywood, strengthened with steel angles to withstand the lateral concrete pressure, and covered with external heat insulation. To evaluate the effects of heat insulation, two different configurations of heat insulation were adopted for the 1.0 m (3.28 ft) cube specimen, as will be explained in the following.

As before, a plastic bag was adhered to the inside of the insulated mold, the concrete was cast into the plastic bag, and the plastic bag was then sealed to avoid leakage. The insulated mold, with the concrete specimen cast inside, was placed in an indoor environment shielded from air movement and thermal disturbances. To simulate the actual site conditions, however, no air conditioning was applied. To prevent heat conduction from the bottom of the insulated mold to the ground, the insulated mold was supported on wooden spacers rather than just placed on the floor.

The thermocouples inside the concrete specimen were arranged as shown in Fig. 1. Four additional thermocouples were mounted on the external surfaces of the insulated mold, each located at the center of one side surface, to measure the ambient temperature. The temperature readings were taken at 5-minute intervals and recorded for data analysis after the curing test was completed.

EXPERIMENTAL PROGRAM

The experimental program consisted of two major parts. Part 1 consisted of the parallel adiabatic and semi-adiabatic curing tests of 10 laboratory-made concrete mixtures to verify the applicability of the proposed heat loss compensation method and to study the effects of mixture composition on adiabatic temperature rise. Part 2 consisted of the semi-adiabatic curing tests of two batches of ready mixed concrete, cast into different size specimens with different configurations of heat insulation provided, to evaluate the sensitivity of the semi adiabatic curing test to specimen size and heat insulation after the application of the proposed heat loss compensation.

In Part 1, from each concrete mixture, two cubical specimens, one 0.25 m (0.82 ft) and the other 0.5 m (1.64 ft) in size, were cast. The smaller specimen was subjected to an adiabatic curing test while the larger specimen was subjected to a semi adiabatic curing test so as to compare the temperature rise results obtained by the two test methods. For quality assurance, three 150 mm (5.9 in.) cubes were also cast to determine the 28-day cube strength of the concrete. The 10 concrete mixtures tested were designed to have the mixture compositions as presented in Table 1. They may be grouped into three series: Series A comprising Mixture A0, A1, A2, and A3 for studying the effects of pulverized fuel ash (PFA); Series B comprising Mixtures A0, B1, B2, and B3 for studying the effects of condensed silica fume (CSF); and Series C comprising Mixtures C1, A0, C2, and C3 for studying the effects of paste volume. All concrete mixtures were designed to have the same water-cementitious material ratio (w/cm) of 0.40. An ordinary portland cement of strength class 52.5 N and with a fineness of 350 m2/kg (1709 ft2/lb), a local PFA with a mean particle size of 19 µm (0.00075 in.), and a Norwegian CSF with a fineness of 20,000 m2/kg (98,000 ft2/lb) were used to produce the concrete. A high-range water reducing admixture was added to each concrete mixture at a fixed dosage of 1.0% by weight of the total cementitious materials content. Crushed granite was used for both the fine and coarse aggregates.

In Part 2, two batches of ready mixed concrete, designated as Mixtures R1 and R2, were tested. From each batch, three cubical specimens, one of size 0.25 m (0.82 ft), another of size 0.5 m (1.64 ft), and the third of size 1.0 m (3.28 ft), were cast. The 0.25 m (0.82 ft) and 0.5 m (1.64 ft) specimens were subjected to adiabatic and semi-adiabatic curing tests as in Part 1. To evaluate the effects of heat insulation, the 1.0 m (3.28 ft) specimen cast from Mixture R1 was subjected to semi-adiabatic curing tests with no heat insulation other than the mold itself provided while the 1.0 m (3.28 ft) specimen cast from Mixture R2 was subjected to semi-adiabatic curing test with the mold covered with external heat insulation. The external heat insulation provided to the specimen cast from Mixture R2 consisted of a layer of 50 mm (2.0 in.) thick extruded polystyrene boards surrounding the mold with a 50 mm (2.0 in.) wide air gap between and two layers of heat insulation blanket wrapping the extruded polystyrene boards. For quality assurance, from each batch of concrete, three 150 mm (5.9 in.) cubes were also cast to determine the 28-day cube strength. The same concrete mixture was ordered for the two batches of concrete, but in reality, as revealed by the 28-day cube strength results, the two batches of concrete were quite different in quality. An external ready mixed concrete supply was employed because the total volume of concrete to be cast was larger than the production capacity of the laboratory.

All the thermocouples used were Type K nickel/chromium thermocouples. They were all calibrated against a platinum resistance thermometer with an accuracy of ±0.015 °C (±0.027 °F) in a calibration bath before use. The adiabatic and semi-adiabatic curing tests were conducted continuously up to at least 144 hours after placing of the concrete. The temperature readings were taken automatically by data loggers and no interference of the equipment was allowed during the entire period of testing.

RESULTS AND DISCUSSIONS

Spatial temperature profiles

Due to heat loss, the temperature within each semi-adiabatic curing test specimen was not uniform. In the 0.5 m (1.64 ft) specimens, the maximum temperature difference within the concrete volume ranged from 4.9 to 9.3 °C (8.8 to 16.7 °F). In the 1.0 m (3.28 ft) specimens, the maximum temperature difference within the concrete volume ranged from 8.3 °C (14.9 °F) for the specimen cast from Mixture R2 and insulated during testing to 25.0 °C (45.0 °F) for the specimen cast from Mixture R1 and not deliberately insulated during testing. To study the spatial temperature profiles in these specimens, six additional thermocouples were installed inside each 0.5 m (1.64 ft) specimen cast from Mixtures A0, C2, and C3 and each 1.0 m (3.28 ft) specimen cast from Mixtures R1 and R2. Among the additional thermocouples, two were installed at the 1/3 points between the center of mass and the center of face, two were installed at the1/3 points between the center of face and the midpoint of edge, and the last two were installed at the 1/3 points between the midpoint of edge and the corner of the specimen.

For illustrating the spatial temperature profiles, the temperature variations from the center of mass to the center of face, from the center of face to the midpoint of edge, and from the midpoint of edge to the corner in the 1.0 m (3.28 ft) specimen cast from Mixture R1, which has the largest temperature variations, are plotted in Fig. 4. In this figure, each temperature difference is normalized as a ratio of the total temperature difference so as to compare the shapes of the spatial temperature profiles at different times after placing of the concrete. By fitting different polynomial curves, it has been found that the spatial temperature profiles may be approximated by quartic polynomial curves, as depicted by the dotted lines in the figure. The spatial temperature profiles of the 0.5 m (1.64 ft) specimens cast from Mixtures A0, C2, and C3 (not shown herein for brevity) are similar. Hence, the spatial temperature profiles within the concrete volume of the 0.5 m (1.64 ft) specimens may also be approximated by quartic polynomial curves.

Assuming quartic polynomial temperature profiles within the concrete volume, the temperature function T(x, y, z) may be expressed as

... (11)

where .1 to .4 are coefficients to be determined. At the center of mass with spatial coordinates (0, 0, 0), the center of face with spatial coordinates (0, 1, 0), the midpoint of edge with spatial coordinates (1, 1, 0), and the corner with spatial coordinates (1, 1, 1), the respective temperature should be Tm , Tf , Te , and Tc . Substituting these conditions into Eq. (11), the coefficients .1 to .4 can be determined. Then by substituting the expression for T(x, y, z) so derived into Eq. (9a) and (9b), the following formulas for TV and TS are obtained

... (12a)

... (12b)

These formulas are for four-point temperature measurement. If two-point temperature measurement was adopted, it is suggested to assume Te = Tc = Tf and apply the formulas with T e and Tc replaced by Tf . If one-point temperature measurement was adopted, however, no temperature profile could be determined and the temperature throughout the concrete volume would have to be taken as Tm , leading to TV = TS = Tm . The resulting formulas to be used are summarized in Table 2.

Adiabatic curing tests results

The temperature-time curves obtained by the adiabatic curing tests all exhibited the general trend of an initial rapid increase in temperature during the first 15 hours, followed by a further increase in temperature at a decreasing rate during the next 25 hours or so and finally ending with a slow and steady decrease in temperature. To demonstrate this trend, some typical temperature rise-time curves (those of Mixtures A2 and A3) are plotted in Fig. 5. The proposed heat loss compensation with TV and TS determined by four-point temperature measurement has been applied to all the concrete mixtures tested. After such heat loss compensation, all the temperature rise-time curves were raised slightly and became flat at the later stage, as illustrated in Fig. 5. The adiabatic temperature rise values of the laboratory-made concrete mixtures so determined after heat loss compensation are summarized in the second column of Table 3. It is seen that the adiabatic temperature rise, which varies between 47.3 and 69.7 °C (85.1 and 125.5 °F), is highly dependent on the mixture proportions of the concrete. For reference, the corresponding mean 28-day cube strengths are listed in the last column of Table 1.

It should be noted that the heat loss compensation method applied herein is not quite the same as that developed by Morabito and Barberis.8 Morabito and Barberis assumed a constant temperature difference between the concrete specimen and the surrounding medium during the curing test for the heat loss compensation, whereas in the method applied herein, the actual measured temperature difference was used to evaluate the heat loss compensation to be applied, as depicted by Eq. (10). Furthermore, the practice of four-point temperature measurement adopted herein should theoretically be more accurate and reliable.

Semi-adiabatic curing test results

For each laboratory-made concrete mixture, one semi adiabatic curing test on a 0.5 m (1.64 ft) specimen cast of the concrete was conducted. Two typical temperature rise-time curves so obtained (those of Mixtures A2 and A3) are plotted in Fig. 6. Comparing to those presented in Fig. 5 for the same mixtures obtained by the adiabatic curing tests, it can be seen that when subjected to semi-adiabatic curing tests instead of adiabatic curing tests, due to heat loss, the peak temperature reached became lower, whereas the rate of temperature decrease with time became faster. Without heat loss compensation, the adiabatic temperature rise can only be estimated as the peak temperature obtained by the semi-adiabatic curing test. Such estimated adiabatic temperature rise values of the concrete mixtures tested are tabulated in the third column of Table 3. Compared with those obtained by the adiabatic curing tests, which should be more accurate, the estimated adiabatic temperature rise values obtained by the semi adiabatic curing tests with no heat loss compensation applied are much too low to be considered acceptable.

The proposed heat loss compensation with TV and T S determined by four-point, two-point, or one-point temperature measurement has been applied to all the concrete mixtures tested. After such heat loss compensation, the temperature rise-time curves were raised significantly and became flat at the later stage, as illustrated in Fig. 6. The estimated adiabatic temperature rise values so obtained after heat loss compensation using four-point temperature measurement, two-point temperature measurement, and one point temperature measurement are tabulated in the fourth to sixth columns of Table 3.

Taking the adiabatic temperature rise obtained by the adiabatic curing test (that is, the temperature rise presented in the second column of Table 3) as the true value for comparison, it can be seen that after heat loss compensation, the estimated adiabatic temperature rise obtained by the semi-adiabatic curing test would become much more accurate. With four-point temperature measurement adopted in the heat loss compensation, the estimated adiabatic temperature rise obtained by the semi-adiabatic curing test is generally accurate to within 1.3 °C (2.3 °F) error. With two point temperature measurement adopted, the estimated adiabatic temperature rise obtained by the semi-adiabatic curing test tends to be slightly higher and less accurate, but nevertheless is still accurate to within 1.8 °C (3.2 °F) error. With one-point temperature measurement adopted, the estimated adiabatic temperature rise obtained by the semi-adiabatic curing tests tends to be even higher and the error could be as large as 2.3 °C (4.1 °F).

It is actually expected that the adoption of four-point temperature measurement in the heat loss compensation would in general yield the most accurate estimate of adiabatic temperature rise. Because, during the test, Tc < Te < Tf < T m , the assumption of Tc = Te = Tf in the two-point temperature measurement and the assumption of Tc = Te = Tf = Tm in the one-point temperature measurement would tend to overestimate TV and TS and hence overestimate the adiabatic temperature rise. It is therefore advocated that the practice of four-point temperature measurement should be preferred.

Effects of specimen size and heat insulation

From each batch of ready mixed concrete, one specimen (a 0.25 m [0.82 ft] cube) was cast for adiabatic curing tests and two specimens (a 0.5 m [1.64 ft] cube and a 1.0 m [3.28 ft] cube) were cast for semi-adiabatic curing tests. The 0.25 m (0.82 ft) and 0.5 m (1.64 ft) specimens cast from each batch of concrete were tested as before. The 1.0 m (3.28 ft) specimen cast from Mixture R1, however, was tested with no heat insulation other than the mold itself provided while the 1.0 m (3.28 ft) specimen cast from Mixture R2 was tested with the mold covered with external heat insulation. When interpreting the test results of Mixtures R1 and R2, it should be noted that the cube compression tests have revealed that Mixture R1 had a mean 28-day cube strength of 45.3 MPa (6570 psi) whereas Mixture R2 had a mean 28-day cube strength of 56.4 MPa (8180 psi). Hence, Mixtures R1 and R2 were two different mixtures and their test results should not be compared with each other. Nevertheless, the test results of the same mixture can be compared directly.

The adiabatic temperature rise values obtained by the adiabatic curing tests of the 0.25 m (0.82 ft) specimens and the estimated adiabatic temperature rise values obtained by the semi-adiabatic curing tests of the 0.5 m (1.64 ft) specimens are presented in Table 4. Heat loss compensation using four point temperature measurement has been incorporated in these test results. For Mixture R1, the adiabatic temperature rise by the adiabatic curing test was 49.9 °C (89.8 °F), whereas the estimated adiabatic temperature rise by the semi adiabatic curing test was 50.2 °C (90.4 °F); hence, the error was only 0.3 °C (0.5 °F). For Mixture R2, the adiabatic temperature rise by the adiabatic curing test was 54.2 °C (97.6 °F), whereas the estimated adiabatic temperature rise by the semi-adiabatic curing test was 53.3 °C (95.9 °F); hence, the error was only 0.9 °C (1.6 °F).

The temperature rise-time curves of the 1.0 m (3.28 ft) specimens cast from Mixtures R1 and R2 are plotted in Fig. 7. From these curves, it can be seen that because the specimen cast from Mixture R2 were provided with better heat insulation, its temperature reached a higher peak value and then decreased at a slower rate with time. The proposed heat loss compensation using four-point, two-point, or one point temperature measurement has been applied. After heat loss compensation, the temperature rise-time curves were raised significantly and became flat at the later stage, as depicted by the dotted lines in the figure. The estimated adiabatic temperature rise values without heat loss compensation and those after heat loss compensation using four-point, two-point, and one-point temperature measurement are tabulated in the third to sixth columns of Table 4. These results indicate very clearly that even with the use of a large size and well-insulated specimen like the 1.0 m (3.28 ft) cube cast from Mixture R2, the maximum temperature rise measured by the semi-adiabatic curing test could still be considerably lower than the true adiabatic temperature rise and therefore heat loss compensation should always be applied.

The results in Table 4 reveal that for a 1.0 m (3.28 ft) cube specimen, regardless of whether provided with deliberate heat insulation, the estimated adiabatic temperature rise after application of heat loss compensation using four-point temperature measurement is accurate to within 1.3 °C (2.3 °F) error. If the two-point or one-point temperature measurement were adopted, however, the error would become as large as 10.1 °C (18.2 °F) for the specimen cast from Mixture R1, which was not provided with deliberate heat insulation, and would become of the order of 2.9 °C (5.2 °F) for the specimen cast from Mixture R2, which was provided with deliberate heat insulation. Hence, the errors due to the adoption of two-point or one-point temperature measurement are larger with less effective heat insulation provided and smaller with more effective heat insulation provided. Although such errors could be reduced by the adoption of four-point temperature measurement, it should be advisable to provide good heat insulation in all curing tests.

Finally, there is the question of whether the 0.5 m (1.64 ft) or 1.0 m (3.28 ft) specimen is better for the semi-adiabatic curing test. With the proposed heat loss compensation using four-point temperature measurement applied, the test would become independent of specimen size and much less affected by imperfect insulation. Hence, both the 0.5 m (1.64 ft) and 1.0 m (3.28 ft) specimens are acceptable. Actually, what really matters is the heat loss characteristic of the test setup .. The . values of the 0.5 m (1.64 ft) specimens, the 1.0 m (3.28 ft) specimen used for Mixture R1, and the 1.0 m (3.28 ft) specimen used for Mixture R2 have been found to be 2.18 × 10-6 s-1, 6.54 × 10-6 s-1, and 1.34 × 10-6 s-1 , respectively. The higher the . value is, the faster the rate of heat loss will be and the larger the error will be even after heat loss compensation. It is recommended that the test setup should be designed such that its . value is not higher than 2.18 × 10-6 s-1 .

Effect of mixture proportions

From the results of Mixtures A0, A1, A2, and A3, it can be seen that at a constant w/cm and paste volume, the adiabatic temperature rise decreased with increasing PFA content. This trend is the same as that reported in the literature.4,5 Compared with Mixture A0 in which no PFA was added, at PFA contents of 25% (Mixture A2) and 35% (Mixture A3), the reductions in adiabatic temperature rise were 8.4 °C (15.1 °F) and 14.2 °C (25.6 °F), respectively. From the results of Mixtures A0, B1, B2, and B3, it can be seen that at a constant w/cm and paste volume, the adiabatic temperature rise also decreased with increasing CSF content. Compared with Mixture A0 in which no CSF was added, at CSF contents of 10% (Mixture B2) and 15% (Mixture B3), the reductions in adiabatic temperature rise were 6.0 °C (10.8 °F) and 7.5 °C (13.5 °F), respectively. Hence, both PFA and CSF are effective cement replacement materials for reducing the adiabatic temperature rise.

From the results of Mixtures C1, A0, C2, and C3, it can be seen that at the same w/cm, the adiabatic temperature rise of concrete increased in direct proportion with the paste volume. Compared with Mixture A0, which had a paste volume of 33%, decreasing the paste volume to 30% would reduce the adiabatic temperature rise by 4.5 °C (8.1 °F), whereas increasing the paste volume to 36% would increase the adiabatic temperature rise by 4.0 °C (7.2 °F). Hence, keeping the paste volume small is also an effective means of avoiding high adiabatic temperature rise.

CONCLUSIONS

A heat loss compensation method for determining the heat loss characteristic of the test setup from the rate of temperature drop with time when the heat generation from the concrete becomes insignificant, estimating the heat loss from the heat loss characteristic so determined and the measured temperature difference between the concrete and the surrounding medium, and compensating for the heat loss accordingly has been developed for use with both the adiabatic and semi-adiabatic curing tests of concrete. To take into account the spatial variation of temperature within the concrete volume, four-point temperature measurement (at center of mass, center of one face, midpoint of an edge, and one corner of the cubical specimen) is adopted.

To verify the feasibility and accuracy of the proposed method, parallel adiabatic and semi-adiabatic curing tests of 10 laboratory made concrete mixtures were conducted. It was found that with the proposed heat loss compensation applied, the estimated adiabatic temperature rise values obtained by the semi-adiabatic curing tests were all accurate to within 1.3 °C (2.3 °F) error. The simpler two-point temperature measurement (at center of mass and center of one face) and one-point temperature measurement (only at center of mass) had also been attempted but the resulting estimated adiabatic temperature rise values were generally less accurate.

To study the effects of specimen size and heat insulation, two batches of ready mixed concrete were each cast into 0.5 m (1.64 ft) and 1.0 m (3.28 ft) cubes and provided with different heat insulation for semi-adiabatic curing tests. The results revealed that with the proposed heat loss compensation applied, the semi-adiabatic curing test would become independent of specimen size and much less affected by imperfect heat insulation. Nevertheless, the test setup should be designed such that its heat loss characteristic (. value) is not higher than 2.18 × 10-6 s-1 .

Comparing the adiabatic temperature rise values of concrete with different mixture proportions, it is evident that both PFA and CSF are effective cement replacement materials for reducing the adiabatic temperature rise and that the paste volume of the concrete mixture should be kept small to avoid a high adiabatic temperature rise.