You are on page 1of 9

ACI MATERIALS JOURNAL TECHNICAL PAPER

Title No. 112-M55

Improved Schmidt Method for Predicting Temperature


Development in Mass Concrete
by Christopher P. Bobko, Vahid Zanjani Zadeh, and Rudolf Seracino

Designing mass concrete structural elements to avoid early-age maturity method. Second, the use of outdated curves with
thermal cracking requires good predictions of temperatures within limited mixture proportion inputs, or reliance on calorimetry
the mass concrete. An improved method for predicting temperature experiments, is avoided by modeling the heat of hydration
in mass concrete structural elements is proposed and validated. The using empirical equations. Results from the new modeling
new method combines empirical methods for predicting tempera-
approach are much closer to temperature measurements
ture rise associated with heat of hydration with the Schmidt method,
from the laboratory and the field, and to predictions from
a simplified numerical tool for solving the heat transfer problem.
Methods for modeling thermal insulation with the Schmidt method detailed finite element models, compared to the previous
are also discussed. The new method is simple enough to implement implementation of the Schmidt method. Improved analysis
in a spreadsheet analysis. Three case studies are modeled with the using this simple but accurate tool will aid designers of mass
previous implementation of the Schmidt method and the proposed concrete structural elements.
new implementation. The model predictions are compared with
temperature measurements and predictions from detailed finite Early-age thermal cracking in concrete
element modeling. In all cases, the new implementation provides Early-age cracking in mass concrete occurs because of
much better predictions than previous versions of the Schmidt temperature gradients that develop during the cement hydra-
method and nearly matches the predictions made by finite element tion reactions. As cement hydrates, heat is released and
modeling.
concrete expands with the increase in temperature. At initial
Keywords: finite element analysis; heat of hydration; mass concrete; stages of hydration, the concrete has a relatively low modulus
Schmidt method; thermal cracking. of elasticity, and any internal stresses are easily relieved due
to the high rate of creep.6 As the process continues, less
INTRODUCTION heat is released and the temperature reduces as the struc-
Cement hydration is an exothermic reaction that can ture reaches thermal equilibrium with its environment. The
produce high amounts of heat during curing, especially concrete contracts with temperature reduction, but the mate-
in the first few days or weeks after casting. This heating rial is now much stiffer and the internal volume of concrete
leads to temperature gradients within the concrete. If the resists the contraction of the concrete surface. Tensile
temperature gradients are too great, resulting tensile thermal stresses then develop along the surface of the concrete. If the
stresses can exceed the concrete tensile strength and cause tensile stresses exceed the tensile strength of the concrete,
cracking, especially at early ages when the concrete is still which may still be rapidly developing, cracking may occur.
developing its full strength. Large volumes of concrete, Temperature gradients are therefore a key driver of
known as mass concrete, have the potential to experi- early-age thermal cracking. Factors that control tempera-
ence temperature gradients during curing that can lead to ture gradients in mass concrete include the initial placement
cracking. Analysis of mass concrete structures requires an temperature, heat generation, and heat dissipation.2 Although
accurate analysis of the temperature development based initial temperature can be modified to some extent, amounts
on concrete mixture proportions and boundary conditions. and rates of heat generation and heat dissipation are the most
Various analytical methods including graphical methods, important. Heat generation depends on cement content, the
simplified analytical methods, and full finite element simu- Blaine fineness of the cement, use of supplementary cementi-
lations1-5 have been proposed to estimate the temperature tious materials, and the addition of different admixtures such
distributions within mass concrete. Some models are easier as retarders or accelerators.7 Heat dissipation is a function
to implement but are less accurate and have less flexibility, of the thermal properties of the concrete, including thermal
while others are more complex but provide greater accuracy diffusivity, heat capacity, and thermal conductivity, and of the
and flexibility in model inputs. thermal conditions surrounding the concrete. When concrete
This manuscript proposes and validates a new version structures have large mass-surface ratios, the heat generated
of the Schmidt method for temperature prediction. This in the interior is not easily dissipated through the concrete
new version can be implemented in a spreadsheet and can
model a wide variety of cement mixtures. Two primary ACI Materials Journal, V. 112, No. 4, July-August 2015.
drawbacks of the previous implementation of the Schmidt MS No. M-2014-103.R2, doi: 10.14359/51687454, received September 25, 2014,
and reviewed under Institute publication policies. Copyright © 2015, American
method are overcome. First, the effect of elevated tempera- Concrete Institute. All rights reserved, including the making of copies unless
permission is obtained from the copyright proprietors. Pertinent discussion including
ture on the reaction rate is modeled by the equivalent age author’s closure, if any, will be published ten months from this journal’s date if the
discussion is received within four months of the paper’s print publication.

ACI Materials Journal/July-August 2015 579


due to its low thermal conductivity, and the structure almost ficients for the empirical model were developed from more
becomes adiabatic, resulting in high internal temperatures. than 300 calorimetry tests, considering variations across
The increased heat on the surface of the concrete structure, cement chemistry, fly ash, slag, and silica fume content, and
however, can dissipate much more rapidly into the environ- dosing of some common chemical admixtures. Researchers
ment, resulting in a much lower temperature rise unless the have begun to incorporate this empirical model in finite
heat loss is controlled.6 When the surface temperature is element models. It is implemented directly in Concrete-
much lower than the interior temperature, large temperature Works4 and has been implemented through a user subroutine
differentials occur. in DIANA.3
The primary methods of controlling temperature gradi- Three commonly used methods for analysis of thermal
ents in mass concrete include modifications to the concrete rise without using finite element models are the Portland
mixture proportions, precooling methods, thermal insulation Cement Association’s temperature estimation method
blankets or sheets, and post-cooling methods. Modifying (PCA method), the ACI graphical method, and the Schmidt
the concrete mixture proportions and applying thermal insu- method.2 The PCA method is a quick method for estimating
lation are among the most cost-effective and widely used the maximum temperature reached in a concrete, but it does
methods for controlling temperature in mass concrete. not consider information about the type of cement used and
does not give information about the time that the maximum
Current standards and specifications temperature is reached. The graphical method in ACI
ACI 301-108 suggests thermal limits including a 207.2R-072 uses several empirical equations and charts to
maximum temperature after placement of 158°F (70°C), and predict temperature rise for a variety of conditions, but is
a maximum temperature differential of 33°F (19.4°C). In limited by the constraints of the empirical equations. The
addition, ACI 301-108 specifies that hydraulic cement with Schmidt method—one of the more widely used methods—is
a low heat of hydration (but not a Type III or HE cement) based on a numerical solution for the heat transfer equation
or portland cement with Class F fly ash or slag, should be where the temperature is calculated for discrete nodes at
used for mass concrete construction. The concrete is then discrete time steps, as described in ACI 207.2R-07.2 Typi-
specified to be cured and protected for a minimum of 7 days cally the temperature rise for this method is calculated using
unless otherwise specified. ACI 301-108 also requires charts in ACI 207.2R-07,2 but using the adiabatic tempera-
temperatures to be measured hourly at the center of the ture measured from semi-adiabatic calorimetry tests has also
concrete and the nearest surface of the concrete. Tempera- been proposed.12
ture is monitored from the time the concrete is placed until Riding et al.12 assessed the effectiveness of these
the difference between internal concrete temperatures, commonly used temperature prediction methods for
and the average daily ambient temperatures are below the calculating the maximum temperature and the time of the
maximum temperature differential, after the peak tempera- maximum temperature in mass concrete structures. Riding
ture has been reached. et al.12 used these three methods and Schmidt’s method with
the adiabatic temperature results to model the temperatures
Analyses of temperature development of eight mass concrete bridge members with varying size,
The most detailed analyses of temperature development shape, formwork, environmental exposure conditions, and
can be made using finite element models, where full three-di- mixture proportions. They found that all three methods
mensional geometries, boundary conditions, heat release, and underestimated the maximum temperature of the concrete
thermal properties can be explicitly modeled. Commercial structures. The graphical method in ACI and the Schmidt
finite element codes including TNO DIANA, FEMASSE, method both also gave poor predictions of the time at which
and CESAR have built-in features that enable modeling of the peak temperature took place. The Schmidt method results
thermal rise and dissipation in mass concrete structures. Anal- were found to be greatly improved; however, when the adia-
ysis codes designed specifically for studying mass concrete, batic temperature results from semi-adiabatic calorimetry
such as ConcreteWorks,4 have also been developed. Advan- tests for each concrete mixture was used in place of the ACI
tages of finite element models include flexibility of inputs, charts. Riding et al.12 attributed the errors in these results
accuracy of results, and an ability to perform a stress analysis to be largely caused by the differences in modern concretes
based on the thermal results. Drawbacks of using specialized versus that which was used when the PCA and ACI guide-
finite element models include the cost of the programs and the lines were developed. Modern concrete tends to use much
complexity of the model approach. finer cements and consist of more chemical admixtures and
Finite element methods generally require results from more supplementary cementitious materials (SCMs), such
isothermal or semi-adiabatic calorimetry experiments to as fly ash, compared to older concrete mixtures.
model the heat of hydration associated with specific concrete
mixture proportions. To avoid the need for experiments, an RESEARCH SIGNIFICANCE
empirical model based on the cement constituent material The aim of this paper is to present a new implementa-
properties as well as the concrete mixture proportions was tion of the Schmidt method that incorporates the concept
proposed to model the early-age heat-of-hydration develop- of equivalent age and empirical equations for the heat of
ment.9,10 The model is based on the equivalent age matu- hydration. The goal is to develop temperature predictions
rity method,11 which allows the current temperature of the with accuracy similar to predictions made using finite
concrete to affect the estimation of heat generation. Coef- element models, by using a method that can be implemented

580 ACI Materials Journal/July-August 2015


in a spreadsheet analysis. This implementation provides where 50% of the heat loss at the bottom boundary is retained
an accurate tool that will allow researchers and engineers due to insulating effects of the ground. No justification for
to easily study the effects of size, insulation, construction this estimate was presented. Alternatively, insulation can be
processes, and mixture proportions on the temperatures in modeled by including extra nodes of concrete that have an
mass concrete structural elements. equivalent thermal resistance as the insulation. A rational
method for implementing insulation in the Schmidt method
ANALYTICAL PROCEDURE requires calculating the equivalent thickness of concrete,
Schmidt Method with ACI heat-of-hydration curves given the thickness of the insulation and its R-value. The
The Schmidt method, first published in 1934 and applied R-value is a measure of thermal resistance of material to heat
to the analysis of thermal rise in mass concrete in 1936,13 flow through a given thickness. It is the reciprocal of thermal
is a simplified computational approach for solving the heat conductivity and is given by
transfer problem in a body with a uniform thermal diffu-
sivity. Implementing the Schmidt method requires assuming R-value = L/k (2)
a one-dimensional discretization of a structural element.
Typically, the minimum dimension of the structural element where L is the material’s thickness; and k is its thermal
is modeled because it represents the direction of maximum conductivity. By assuming the same R-value for insulation
thermal gradients. The length of each discrete element and and concrete, the equivalent concrete thickness can be found
the thermal diffusivity of the material determine the size of
the time step, ∆t LInsulation × kConcrete
LConcrete = (3)
k Insulation
2
(∆x)
∆t = (1)
2h 2 New Schmidt Method with equivalent age and
empirical heat of hydration
where ∆ x is the distance between nodes; and h2 is the diffu- Two primary drawbacks for using the Schmidt method
sion constant.2,13 The method works by determining the new with the ACI 207.2R-072 heat-of-hydration curves are
temperature at a node at the current time as the average of evident. First, the adiabatic temperature rise is outdated. It
the temperature of the neighboring nodes in the prior time is based on cement that is very coarse compared to most
step, plus any temperature rise associated with heat added to modern cements, and does not incorporate the effects of
the node during the time step. commonly used pozzolans or chemical admixtures. This
The next component of the Schmidt method is modeling the inaccuracy is expected to lead to overall lower temperature
addition of heat as the concrete hydration reaction proceeds. predictions. Second, the temperature rise is presented as a
ACI 207.2R-072 presents graphs of adiabatic tempera- function of time, even though it is known that increasing the
ture rise for concrete containing 375 lb/yd3 (223 kg/m3) temperature increases the rate of reaction, thereby increasing
of Types I, II, III, and IV cement. ACI 207.2R-072 also the rate of temperature rise.14 This inaccuracy is expected
suggests scaling the temperature rise based on the amount of to lead to lower temperature predictions at the center of
cement and pozzolans actually used in the concrete mixture the mass concrete structural element. In turn, predictions
to be modeled. It is suggested that pozzolan produces only of temperature difference are expected to be lower than the
approximately 50% as much heat as the cement it replaces, actual differences.
so an equivalent cement content should be the actual cement To improve predictions, a new implementation of the
weight per cubic yard plus one half of the pozzolan weight Schmidt method was developed, incorporating the equiva-
per cubic yard. The ratio of the equivalent cement content to lent age concept and Riding et al.’s10 empirical model for
the 375 lb/yd3 (223 kg/m3) baseline is the scaling factor for heat of hydration as a function of cement chemistry and
the data from the temperature-rise graphs. ACI 207.2R-072 concrete mixture proportions. The empirical model describes
notes that the provided graphs are typical of cement produced the rate of heat generation per unit volume, with coefficients
before 1960. It is likely that temperature rises much more for parameters defined in SI units
quickly in modern cements that are much finer. Because the
adiabatic temperature-rise graphs are outdated and do not
β
β τ E 1 1
 τ  β − 
t
a
 − 
R T T
include pozzolanic materials, ACI 207.2R-072 recommends Qh (te ) = H u ⋅ Cc ⋅   ⋅   ⋅ α u ⋅ e  e  ⋅ e  r c  (4)
that adiabatic temperature-rise tests be conducted on a  te   te 
sample of the concrete to provide a more accurate represen-
tation of the anticipated thermal rise. where Qh is the rate of heat generation per unit volume
The final component of the Schmidt method involves (J/m3·h); Ea is the apparent activation energy (J/mol); R is
modeling the starting values and boundary conditions. the universal gas constant (8.314 J/mol/K); te is the equiv-
Starting temperatures and outside air conditions can be alent age (h); Hu is the total heat available for reaction at
assumed based on average local weather conditions. Many 100% hydration (J/g); Cc is the total amount of cementitious
mass concrete structural elements, however, are insulated to material (kg/m3); αu is the ultimate degree of hydration; τ is
control their temperatures. ACI 207.2R-072 gives some guid- the hydration time parameter (h); β is the hydration slope
ance on how to handle insulation in an illustrative example, parameter; Tr is the equivalent age reference temperature

ACI Materials Journal/July-August 2015 581


(K); and Tc is the temperature of the concrete (K).10 The Table 1—Dimensions of studied structural
equivalent age is calculated by elements, ft (m)
Height Length Width
− Ea  1 1 
t  − 
te = ∑e
R  Tc Tr 
⋅ ∆t (5) Mass concrete block
5.0 (1.52) 5.0 (1.52) 5.0 (1.52)
0 experiment
Sunset Beach Bridge 6.0 (1.83) 42.0 (23.8) 27.0 (8.23)
where Δt is a time interval (h); and Tc is the average concrete Wilmington Bypass Bridge 14.76 (4.50) 93.18 (28.4) 61.02 (18.6)
temperature during that time interval (K).10 The variables Ea,
Hu, αu, τ, and β are found through empirical relationships as dosings are considered, where WRRET is the mass percent
a function of mixture proportions and cement chemistry solids of ASTM Type B and D water reducer/retarder per
gram of cementitious material, ACCL is the mass percent
( )
Ea = 41, 230 + 1, 416, 000 ⋅  pC3 A + pC4 AF ⋅ pcem ⋅ pSO3 ⋅ pcem  − 347, 000 ⋅ pNa2Oeq
  solids of ASTM Type C calcium-nitrate-based accelerator
−19.8 ⋅ Blaine + 29, 600 ⋅ pFA ⋅ pFA − CaO + 16, 200 ⋅ pslag
per gram of cementitious material, LRWR is the ASTM Type
−51, 600 ⋅ pSF − 3, 090, 000 ⋅ WRRET − 345, 000 ⋅ ACCL A low-range water reducer dosage, MRWR is the mid-range
water reducer dosage, PCHRWR is the ASTM Type F poly-
(6) carboxylate-based high-range water reducer dosage, and
NHRWR is the ASTM Type F naphthalene- or melamine-
based high-range water reducer dosage.10 Note that Eq. (4)
Hu = Hcem · pcem + 461 · pslag + 1800 · pFA–CaO · pFA + 330 · pSF (7)
has been corrected from Riding et al.,10 where the last expo-
nential term involves subtraction rather than addition.
To implement the equivalent age and Riding et al.’s10
H cem = 500 ⋅ pC3 S + 260 ⋅ pC3 A + 420 ⋅ pC4 AF empirical model in the Schmidt method, calculations were
(8)
+ 624 ⋅ PSO3 + 1186 ⋅ pFreeCa + 850 ⋅ pMgO performed using a spreadsheet, with calculations repeating
at each time interval Δt as calculated by Eq. (1). First, the
average temperature of the two neighboring elements is
 −0.0885 − 13.7 ⋅ pC4 AF ⋅ pcem 
calculated as in the traditional implementation of the Schmidt
 −283 ⋅ pNa2Oeq ⋅ pcem 
1.031⋅ w/cm   method. Then, at each discrete element, the equivalent age
αu = + exp
0.194 + w/cm  −9.90 ⋅ pFA ⋅ pFA − Ca O  in the current time step is calculated based on the equiva-
  lent age and temperature of the prior time step. Next, the
 −339 ⋅ WRRET − 95.4 ⋅ PCHRWR
temperature rise generated during the time interval is calcu-
(9) lated using the heat generation rate Q(te) provided by Eq. (4)
multiplied by the time interval Δt and divided by the heat
capacity of the concrete, C. This temperature rise is added to
 2.92 − 0.757 ⋅ pC3 S ⋅ pcem + 98.8 ⋅ pNa2O ⋅ pcem + 1.44 ⋅ pslag  the average temperature of the neighboring nodes from the
τ = exp   prior time step, giving the new temperature.
 + 4.12⋅⋅ pFA ⋅ pFA − CaO − 11.4 ⋅ ACCL + 98.1⋅ WRRET 
(10) Structures studied
Temperature development was analyzed for three mass
concrete structural elements: a mass concrete block that
 −0.464 + 3.41⋅ pC3 S ⋅ pcem − 0.846 ⋅ pslag + 107 ⋅ WRRET + 33.8 ⋅ LRWR was cast as a laboratory experiment, a footing on the Sunset
β = exp  
 + 15.7 ⋅ MRWR + 38.3 ⋅ PCHRWR + 8.97 ⋅ NHRWR  Beach Bridge in North Carolina, and a footing on the Wilm-
(11) ington Bypass Bridge in North Carolina. Early-age tempera-
ture measurements within the mass concrete block, and finite
where Hcem is the total heat of hydration of the cement (J/kg); element model predictions of early-age temperatures for all
pcem is the cement mass to total cementitious content mass three structures were reported by Edwards.3 Dimensions
ratio; pFA is the fly ash mass to total cementitious material of the block and footings are given in Table 1. The general
mass ratio; pFA-CaO is the mass percent of CaO in fly ash; pslag orientation of the blocks and locations of temperature sensors
is the granulated slag mass to total cementitious material and important calculation nodes are shown in Fig. 1. The
mass ratio; pSF is the silica fume mass to total cementitious shortest dimension of each structural element was consid-
material mass ratio; and “Blaine” is the Blaine fineness of ered for modeling with the Schmidt method because it was
cement (m2/kg).10 The chemistry of the portland cement is expected to generate the largest temperature gradients. The
identified, where pC3A is the percent aluminate content, pC4AF mass concrete block experiment was uninsulated, except for
is the percent ferrite content, pC3S is the percent alite content, any possible insulating effect of the plywood forms, which
pSO3 is the percent SO3 content, pNa2O is the percent Na2O was neglected in the Schmidt method analyses. The footings
content, pNa2Oeq is the percent sodium equivalent alkalis from the Sunset Beach Bridge and the Wilmington Bypass
content (given by 0.658 × %K2O + %Na2O), pMgO is the Bridge were insulated with 0.75 in. (19 mm) thick blankets.
percent MgO content, and pFreeCa is the percent CaO content Concrete mixture proportions are given in Table 2, and
in the portland cement.10 Finally, chemical admixture relevant additional input values for Riding et al.’s10 empir-

582 ACI Materials Journal/July-August 2015


Table 2—Mixture proportions, lb/yd3 (kg/m3)
Mass concrete block experiment Wilmington
and Sunset Beach Bridge Bypass Bridge
Cement—Type I/II 448 (266) 521 (309)
Class F fly ash 207 (123) 207 (123)
Silica fume 35 (21) 0 (0)
Water 258 (153) 260 (154)
Sand 1194 (708) 1144 (679)
No. 57 Stone 1800 (1068) 1560 (926)
Fig. 1—General orientation of mass concrete blocks and loca-
Table 3—Empirical model cement chemistry and tions of temperature sensors and important calculation nodes.
chemical admixture input values
Mass concrete block experi- Wilmington Bypass
Value ment and Sunset Beach Bridge Bridge
pC3S 0.61 0.60
pC2S 0.12 0.13
pC3A 0.07 0.06
pC4AF 0.11 0.11
pSO3 0.028 0.027
pFreeCa 0.009 0.009
pMgO 0.013 0.017
pNa2Oeq 0.0047 0.0043
pNa2O 0.001 0.001
pFA-CaO 0.01 0.01
Blaine 400.6 m3/kg 371.5 m3/kg
ACCL 0 0
LRWL 0 0
MRWR 0.0065 0.0065
NHRWR 0 0
PCHRWR 0.00345 0.00345
WRRET 0 0

ical model for heat of hydration are presented in Table 3.


Thermal properties of the concrete were assumed constant for
all concrete mixture proportions and all ages. Conductivity of
concrete was assumed to be 12.8 Btu/h·ft·°F (2.21 W/m·K), Fig. 2—(a) Measured and predicted temperatures; and (b)
and heat capacity of concrete was assumed to be 0.18 Btu/ measured and predicted temperature differences for mass
lb·°F (0.75 J/g·K), while conductivity of the insulation blan- concrete block experiments.
kets was assumed to be 0.42 Btu/h·ft·°F (0.072 W/m·K).3
Predictions at the first node deep and measurements
COMPARISON OF PREDICTIONS AND at the top and bottom of the block were also compared. It
EXPERIMENTAL RESULTS was noted that outer elements in all the Schmidt method
Comparison with temperature measurements models always under-predicted temperatures. The first node
Figure 2(a) shows a comparison between the Schmidt deep in the concrete seemed to give a better match to the
method predictions and the temperature measurements made measured temperatures at the outside edges of the concrete.
on the mass concrete block experiment. Predictions at the Note that no insulation was modeled for the mass concrete
center node and measurements at the center of the block block experiment in the Schmidt method models or the finite
are compared. The new Schmidt method prediction with element models because the boundaries in the experiment
the empirical equations for heat of hydration gave a good were uncontrolled. The measured temperature at the top was
prediction for the center temperature, while the Schmidt influenced by solar heating, while the measured temperature
method using the ACI heat-of-hydration curves under- at the bottom was influenced by the ground temperature.
predicted the temperature for the first 4 days. Considering the limitations, the new Schmidt method predic-
tion at one node deep adequately predicted the measured

ACI Materials Journal/July-August 2015 583


Fig. 3—(a) Predicted temperatures from Schmidt
Method and FEM; and (b) predicted temperature differ-
ences from Schmidt Method and FEM for mass concrete Fig. 4—(a) Predicted temperatures from Schmidt Method
block experiments. and FEM; and (b) predicted temperature differences from
the Schmidt Method and FEM for Sunset Beach Bridge.
temperature at the top, while the Schmidt method with ACI
heat-of-hydration curves under-predicted the temperature. and those made with the finite element model for the mass
The measured temperatures at the bottom were consistently concrete block experiment. The finite element model closely
under-predicted, highlighting the importance of accurately matched the temperature measurements, so the same obser-
modeling insulation. vations made in the comparison with temperature measure-
Figure 2(b) shows a comparison between the tempera- ments on the mass concrete block experiment are relevant in
ture difference predictions using the Schmidt method and this comparison.
the measured temperature differences. Recall that tempera- Figure 4(a) shows a comparison of temperature predictions,
ture differences are responsible for thermal gradients that and Fig. 4(b) shows a comparison of temperature differen-
can result in early-age thermal cracking, and temperature tial predictions for the modeled Sunset Beach mass concrete
differences are typically limited to 35°F (19.4°C). Measure- footing. The new Schmidt method predicted temperatures
ment differences between the center and the top and the that matched well with temperatures from finite element
center and the bottom are presented because of the varying modeling. At the center, the maximum temperature predic-
boundary conditions in the experiment. For the Schmidt tion with the Schmidt method was reached approximately 16
method predictions, temperature one node deep from the hours earlier than the value predicted by the finite element
top and bottom was used as the basis for comparisons. The model, but the maximum temperature value matched very
measured center-top temperature difference oscillated with well. The new Schmidt Method model predicted a faster
the sunlight around the new Schmidt method prediction with rate of cooling compared to the finite element model. The
the empirical equations for heat of hydration. The measured Schmidt method model with the ACI heat-of-hydration
center-bottom temperature difference was lower because curves under-predicted temperatures. Similarly, the new
of the influence of the ground temperature. The Schmidt Schmidt method model predicted a very similar maximum
method prediction using the ACI heat-of-hydration curves temperature differential compared to the finite element
predicted a much lower temperature difference. model, although the temperature difference decreased
at a faster rate. The Schmidt method model with the ACI
Comparison with finite element model predictions heat-of-hydration curves under-predicted the temperature
Figure 3(a) shows a comparison of temperature predictions, differentials.
and Fig. 3(b) shows a comparison of temperature differential Figure 5(a) shows a comparison of temperature predic-
predictions between those made with the Schmidt method tions, and Fig. 5(b) shows a comparison of temperature

584 ACI Materials Journal/July-August 2015


35°F (19.4°C), while the new Schmidt method model and
the finite element model predicted maximum temperature
differentials well above this limit.

Limitations of Schmidt method


Although many of the previously identified limitations
of the Schmidt method were overcome, some limitations
remain. The Schmidt method is better suited to rectangular
shapes with little variation in other dimensions and should
be considered a first rough estimate for more complicated
geometries. The Schmidt method cannot model cooling pipes
that may be necessary to control temperatures, although
many smaller mass concrete elements do not require post-
cooling systems. The method does not consider the effects of
solar radiation, convection, or other sources of heat or heat
dissipation on boundary elements, although the presence of
insulation blankets often limits those effects in practice.
Finally, although maximum temperature and tempera-
ture differentials are an important factor for design of mass
concrete structural elements, early-age thermal cracking
may still occur even if typical limits are met. Other mate-
rial properties, including the evolution of strength, stiffness,
and creep properties with equivalent age, play a role in
determining the potential for early-age thermal cracking.3,5
Designers, contractors, and owners should consider these
factors as a complement to the thermal analysis that can be
performed using the proposed method.

Fig. 5—(a) Predicted temperatures from Schmidt Method CONCLUSIONS


and FEM; and (b) predicted temperature differences from Based on the findings of this study, the following conclu-
Schmidt Method and FEM for Wilmington Bypass Bridge. sions can be made:
1. A new implementation of the Schmidt method was
differential predictions for the modeled Wilmington Bypass
developed, incorporating the effect of temperature on heat of
mass concrete footing. Recall that this footing was much
hydration and using updated empirical equations that predict
larger overall than the Sunset Beach footing or the mass
heat of hydration from knowledge of the concrete mixture
concrete block experiment, and used slightly different
proportions and cement chemistry.
mixture proportions. The new Schmidt method predicted
2. The new method is able to more accurately predict
temperatures that matched well with temperatures from
maximum temperatures, times to maximum temperatures,
finite element modeling, although the temperatures were
and temperature differentials compared to the imple-
slight under-predicted. The high temperatures experienced
mentation of the Schmidt method currently suggested by
in this mass concrete footing change the rate of reaction,
ACI 207.2R-07.
which can be captured through the equivalent age prin-
3. Many of the previously identified limitations of the
ciple implemented in the new Schmidt method model. Once
Schmidt Method were overcome, while maintaining a rela-
again, the Schmidt method model with the ACI heat-of-
tively simple analysis procedure where calculations may be
hydration curves greatly under-predicted temperatures.
performed using a spreadsheet.
In this case, the times when maximum temperatures were
4. Remaining limitations of the Schmidt method were
reached were severely delayed in the Schmidt method model
discussed, and designers, contractors, and owners should be
with the ACI heat-of-hydration curves.
aware of these when using the proposed method.
Similarly, for the Wilmington Bypass mass concrete
footing, the new Schmidt method model predicted a very
AUTHOR BIOS
similar maximum temperature differential compared to the Christopher P. Bobko is an Associate Professor at North Carolina State
finite element model, although the temperature difference in University, Raleigh, NC. He received his PhD and MS from the Massa-
the new Schmidt method model did not reach the maximum chusetts Institute of Technology, Cambridge, MA, and his BSE from Princ-
eton University, Princeton, NJ. His research interests include the material
temperature difference predicted by the finite element model. behavior of complex porous composites in civil engineering.
The Schmidt method model with the ACI heat-of-hydration
curves under-predicted the temperature differentials. In Vahid Zanjani Zadeh was a Researcher at North Carolina State Univer-
sity. He received his PhD from North Carolina State University; his MSc
this case, the existing Schmidt method model predicted a from the University of Akron, Akron, OH; his MS from the International
temperature differential just above the maximum limit of Institute of Earthquake Engineering and Seismology, Tehran, Iran; and his
BSc from the University of Science and Technology, Tehran, Iran.

ACI Materials Journal/July-August 2015 585


ACI member Rudolf Seracino is a Professor at North Carolina State mentary Cementitious Materials: Experimental and Finite-Element Inves-
University, Raleigh, NC. He received his PhD from the University of tigation,” Journal of Materials in Civil Engineering, ASCE, V. 24, No. 4,
Adelaide, Adelaide, Australia, and his MASc and BASc from the University 2012, pp. 362-372. doi: 10.1061/(ASCE)MT.1943-5533.0000389
of Toronto, Toronto, ON, Canada. He is Co-Chair of ACI Subcommittee 6. Lu, H. R.; Swaddiwudhipong, S.; and Wee, T. H., “Evaluation of
440-M, FRP-Repair of Masonry Structures. His research interests include Internal Restrained Strain in Concrete Members at Early Age,” ACI Mate-
the application of advanced materials and systems to enhance the resilience rials Journal, V. 97, No. 5, Sept.-Oct. 2000, pp. 612-618.
of critical civil infrastructure. 7. Mehta, P. K., and Monteiro, P. J. M., Concrete: Microstructure, Prop-
erties and Materials, third edition, McGraw-Hill, New York, 2006, 659 pp.
8. ACI Committee 301, “Specifications for Structural Concrete (ACI
ACKNOWLEDGMENTS 301-10),” American Concrete Institute, Farmington Hills, MI, 2010, 77 pp.
The support of the North Carolina Department of Transportation, under 9. Riding, K. A., “Early Age Concrete Thermal Stress Measurement and
RP2012-09 and TA2014-07, is gratefully acknowledged. Modeling,” PhD dissertation, University of Texas at Austin, Austin, TX,
2007, 612 pp.
REFERENCES 10. Riding, K. A.; Poole, J. L.; Folliard, K. J.; Juenger, M. C. G.; and
1. ACI Committee 207, “Guide to Mass Concrete (ACI 207.1R-05),” Schindler, A. K., “Modeling Hydration of Cementitious Systems,” ACI
American Concrete Institute, Farmington Hills, MI, 2005, 30 pp. Materials Journal, V. 109, No. 2, Mar.-Apr. 2012, pp. 225-234.
2. ACI Committee 207, “Report on Thermal and Volume Change Effects 11. Carino, N., and Lew, H., “The Maturity Method: From Theory to
on Cracking of Mass Concrete (ACI 207.2R-07),” American Concrete Insti- Application,” Structures 2001: A Structural Engineering Odyssey, 2001,
tute. Farmington Hills, MI, 2007, 32 pp. pp. 1-19.
3. Edwards, A., “Early Age Thermal Cracking of Mass Concrete Foot- 12. Riding, K. A.; Poole, J. L.; Schindler, A. K.; Juenger, M. C. G.; and
ings on Bridges in Coastal Environments,” MS thesis, North Carolina State Folliard, K. J., “Evaluation of Temperature Prediction Methods for Mass
University, Raleigh, NC, 2013, 278 pp. Concrete Members,” ACI Materials Journal, V. 103, No. 5, Sept.-Oct.
4. Folliard, K. J.; Juenger, M.; Schindler, A.; Riding, K.; Poole, J.; Kalli- 2006, pp. 357-365.
vokas, L. F.; Slatnick, S.; Whigham, J.; and Meadows, J. L., “Prediction 13. Rawhouser, C., “Cracking and Temperature Control of Mass
Model for Concrete Behavior—Final Report,” Report No. FHWA/TX-08/0- Concrete,” ACI Journal Proceedings, V. 42, No. 4, 1945, pp. 304-346.
4563-1, 2008, 78 pp. 14. Schindler, A. K., “Effect of Temperature on Hydration of Cemen-
5. Lawrence, A.; Tia, M.; Ferraro, C.; and Bergin, M., “Effect of Early titious Materials,” ACI Materials Journal, V. 101, No. 1, Jan.-Feb. 2004,
Age Strength on Cracking in Mass Concrete Containing Different Supple- pp. 72-81.

586 ACI Materials Journal/July-August 2015


Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.

You might also like