EARLY-AGE RESPONSE OF CONCRETE PAVEMENTS TO TEMPERATURE AND MOISTURE VARIATIONS

. In this paper, the early-age response of a Jointed Plain Concrete Pavement (JPCP) to temperature and moisture variations at the time of paving and immediately following construction is discussed. A newly constructed JPCP on US-30 near Marshalltown, Iowa, USA was instrumented and monitored during the critical time immediately following construction to identify its early-age behavior with respect to pavement deformation due to temperature and moisture variations. Th e instrumentation consisted of Linear Variable Diff erential Transducers (LVDTs) at the slab corner, center, and edges, and thermocouples and humidity sensors installed within the slab depth. Th e slab deformation associated with temperature and moisture variations were quantifi ed using fi eld-measured vertical displacements and pavement surface profi les. Th e positive temperature gradients during setting times and the negative moisture diff erence aft er setting times caused permanent upward curling and warping in the instrumented pavement. Th e relative corner defl ection of the slab to center or mid-edge calculated using the slab profi le and LVDT measurements show similar trends.


Introduction
Th e temperature and moisture variations across the depth of the Portland Cement Concrete (PCC) pavements due to changes in the climate result in a unique defl ection behavior which has been recognized as curling and warping of the pavements since mid 1920 (Westergaard 1927).According to 2004 National Cooperative Highway Research Program (NCHRP) "Guide for Mechanistic-Empirical Design of New and Rehabilitated Pavement Structures", Project 1-37A in the USA, in general, temperature diff erences across the depth of the concrete pavement result in curling while moisture diff erences result in warping behavior.Both temperature and moisture gradients can cause either upward or downward distortion of pavement slabs, and pavement slabs are not necessarily fl at at rest (i.e., under no external forces that cause slab distortion (Yu et al. 2004).
Curling and warping of PCC slab infl uences the degree of support by subgrade and the stiff ness along the joint (Armaghani et al. 1986(Armaghani et al. , 1987)).Th e weight of the slab tends to hinder the curling and warping deformation from taking place and as a result restraint stresses are induced within the concrete slab (Huang 1993).
Th e early age behavior of PCC is signifi cantly infl uenced by temperature, moisture, and creep of concrete (Rao et al. 2001).Based on profi lograph records of concrete pavements in California, Hveem (1951) concluded that slab curling was due to the combined eff ect of temperature and moisture, both of which change non-uniformly through the depth of the slab.Many signifi cant research eff orts in the past have tried to address the combined effects of temperature, moisture, and creep on the early-aged slab behavior (Jeong, Zollinger 2005).A positive temperature diff erence between the top and the bottom surfaces of the concrete slab in daytime causes the slab corners to curl downwards, while a negative temperature diff erence during night time results in the upward curling of PCC.Th e moisture diff erence through the slab depth because of weather condition results in non-uniform concrete shrinkage and non-uniform volume changing through depth (Rao et al. 2001).However, curling and warping behavior of early aged concrete is aff ected by not only temperature and moisture diff erences due to weather conditions but also early age curing conditions and temperature conditions during pavement construction (Rao et al. 2001;Rao, Roesler 2005;Yu et al. 1998) A signifi cantly irreversible drying shrinkage of concrete near the top of the slab and a positive temperature gradient at the time of concrete setting can cause permanent upward curling and warping at zero temperature gradient (Yu et al. 2004).Th is permanent curling and warping (built-in curling and warping) is partially recovered by the creep of the slab aft er hardening of the concrete over time (Rao et al. 2001;Rao, Roesler 2005).Once the pavement attains permanent curling and warping aft er setting, the upward curling of the slab for the fi rst few nights aft er the placement of concrete is the critical condition for early age cracking because the tensile stresses at the top due to upward curling and slab weight are greater than incompletely developed concrete strength (Lim, Tayabji 2005).
Limited experimental research studies (Armaghani et al. 1987;Jeong, Zollinger 2005;Lim et al. 2009;Rao et al. 2001;Wells et al. 2006;Yu et al. 1998) have been undertaken to better understand the actual behavior of the concrete under pure environmental loading.2004 National Cooperative Highway Research Program (NCHRP) "Guide for Mechanistic-Empirical Design of New and Rehabilitated Pavement Structures", Project 1-37A note, that in addition, the new Mechanistic-Empirical Pavement Design Guide (MEPDG) for the design of new and rehabilitated pavement structures in the USA require quantifying the permanent curling and warping in terms of temperature diff erence.
In spite of many research eff orts, the early-age curling and warping behavior of PCC pavements under environmental conditions has not been fully understood.Th e primary objective of this study was to measure and analyze the early-age Jointed Plain Concrete Pavement (JPCP) behavior in terms of changes in pavement defl ections to temperature and moisture variations.To achieve this objective, a newly constructed JPCP slab on US-30 near Marshalltown, Iowa was instrumented to monitor the pavement response to temperature and moisture variations during the fi rst seven days aft er the construction in the summer of 2005.A series of laboratory tests were undertaken to characterize the properties of paving material during the controlled fi eld evaluation.Th e instrumentation installed within the pavement is described.Th e procedure and the results of data analysis using the collected data from the instrumented pavement are discussed in this paper highlighting the important fi ndings regarding the early-age curling and warping behavior of JPCP slabs.

Instrumentation and data collection
Th e test JCPC pavement was constructed on an opengraded granular base.Th e transverse joint spacing was approximately 6 m (20 ft ).Th e passing lane was approx 3.7 m (12 ft ) in width, and the travel lane was approx 4.3 m (14 ft ) in width.A Hot Mix Asphalt (HMA) shoulder was added approx 2 months aft er initial construction.
Tie-bars of 914 mm (36 in) length and 12.7 mm (0.5 in) diameter were inserted approx every 76 mm (30 in) across the longitudinal joints.Dowel bars of 457 mm (18 in) length and 38 mm (1.5 in) diameter were inserted approximately every 305 mm (12 in) across the transverse joints.
Two test sections, one corresponding to late morning (11:00 AM CST) construction conditions and the other representative of aft ernoon (3:30 PM CST) construction, were selected for surface profi le measurements.Temperature sensors, relative humidity sensors, and Linear Variable Diff erential Transducers (LVDTs) were placed in each section to observe the environmental eff ects on the slab behavior during early age (7 day aft er construction) without traffi c loading.Iowa State University's (ISU's) PCC mobile laboratory parked near the test section monitored the weather conditions such as ambient temperature, ambient relative humidity, wind direction and rainfall on special days.During the fi eld evaluation periods, sky was clear and sunny.
To obtain the fundamental physical properties of the paving material, a series of laboratory tests at various ages were conducted in ISU's PCC mobile laboratory and ISU's PCC laboratory using in-situ samples obtained from the paving site.Th e split tensile test (ASTM C 496), the compressive strength test (ASTM C 39), and the elastic modulus test (ASTM C 469) was performed on PCC samples obtained during construction.Th e results of laboratory tests are available in Kim (2006).In addition, the coeffi cient of thermal expansion (CTE, AASHTO TP 60) was measured to be 9.63 × 10 -6 /°C (5.35 × 10 -6 /°F).

Pavement temperature and relative humidity instrumentation
Temperature and humidity sensors installed within the test sections recorded the slab temperature and moisture data at fi ve-minute intervals throughout the fi eld evaluation periods.Temperature instrumentation consisted of Th ermochron I-buttons attached to a stake at diff erent depths from pavement surface (64 mm, 89 mm, 114 mm, 165 mm, 190 mm, 266 mm, 419 mm) and placed at 0.9 m (3 ft ) from the pavement edge before the paving.Humidity instrumentation consisted of Hygrochron I-Buttons inserted into small Poly Vinyl Chloride (PVC) pipes which were placed side by side at diff erent depths from pavement surface (38.1 mm, 88.9 mm, 127 mm, 165.1 mm).

Measurement of vertical slab movements using LVDT
Two slabs (slabs 19 and 20) which were paved in the afternoon were selected as representative slabs to study the pavement vertical movements entirely due to environmental loads.As shown in Fig. 1, LVDTs were installed in special locations on each slab to capture the vertical movements of the slab.In the test slab19, nine LVDTs were installed at corners, the mid-slab edges and the center of the slab.In the test slab20, seven LVDTs were installed at the corners, the mid-slab edges near longitudinal joints and transverse joints.All the sensors were placed only aft er the concrete hardened (1 day aft er paving).LVDTs were held by a bracket fastened to the steel rod inserted in subgrade and placed on a smooth glass on the PCC pavement.Th e LVDTs were connected to data loggers, which collected data at 10 min interval throughout the fi eld evaluation periods.

Pavement surface profi le measurement
Rollingprofi ler by an International Cybernetics Corporation was used for surface profi le measurements at diff erent times (morning and the aft ernoon) in both test sections.Rollingprofi ler, a kind of inclinometer profi ler, can measure true unfi ltered elevation profi le of surface along the line being profi led (ICC 2006) Many researchers have used the inclinometer profi ler measurements to quantify slab curvature (Rao et al. 2001;Vandenbossche 2003).Rollingprofi ler measurements in this study followed transverse and diagonal traces to capture the slab curvature.Measurements were made on four individual slabs in both test sections at diff erent times.Each profi ling segment was measured independently.

Analysis of temperature and moisture
Th e temperature and moisture variations within the PCC pavement during early-age (7 days aft er construction) could be obtained from the installed temperature and moisture sensors.In addition, average pavement temperatures, diff erences in temperature and moisture between the top and bottom of the pavement, and temperature distributions with depth could be obtained from the measured temperature data.
Average pavement temperatures were calculated from temperature readings of six temperature I-button sensors (Sensor 1, 2, 3, 4, 5 and 6).Temperature diff erences were calculated by subtracting the temperature sensor reading at 266.7 mm below the slab surface (Sensor 1) from the sensor reading at 63.5 mm below the slab surface (Sen-sor 6).Note that the closest temperature sensor to the top of the pavement surface was located at 63.5 mm below the slab surface.Moisture diff erences were computed by subtracting the moisture sensor reading at 165.1 mm below the slab surface (Sensor 1) from the sensor reading at 38.1 mm below the slab surface (Sensor 4).Air temperature, average pavement temperature and subgrade tempe rature variations during the initial day (day 0 aft er paving) and during the early aged days (6 and 7 days aft er paving) are compared in Figs 2 and 3. Weather conditions were clear and sunny during both days.In both Figs 2 and 3, average pavement temperatures are higher than ambient temperature.
From Fig. 2, it can be observed that the average pavement temperature within 8 h of paving increases and reaches a max value, while air temperature decreases.Th e same trend is observed during the fi rst 8 h of paving for both the test sections.Aft er 8 h of paving, the pavement temperature follows a pattern that is similar to that of air temperature as reported by previous research study ( Armaghani et al. 1987).Th is increase in pavement temperature within the fi rst 8 h of paving may be due to the heat of concrete hydration at the time of setting.Note that aft er 8 h of paving, the max and min pavement temperatures occurred normally 1 to 2 h aft er air temperature reached their maxima and minima.Armaghani et al. (1987) reported that this trend was observed in the majority of the samples that were randomly selected from the collected temperature data obtained over a period of 3 years in Florida.Both from Figs 2 and 3, the subgrade temperature variation is not high and usually follows the pattern of pavement temperature.

Fig. 2. Temperature variation with time during paving
In-depth temperature distributions within 12 h and 7 days of paving in test section 1 are plotted in Figs 4 and 5, respectively.It can be observed from Fig. 4 that within 12 h of paving, temperature distributions shift ed towards the right.Th is means that the pavement temperatures at night time were higher than those of day time without increase in air temperature.Also the mitigation of temperature due to heat of hydration of concrete occurred through the thickness.From Fig. 5, the max positive temperature diff erence decreased with depth whereas the max negative temperature diff erence increased with depth with the air temperatures changing.
PCC to acquire a certain degree of hardening, a fl at slab condition in this test section could be associated with a positive temperature gradient rather than a zero temperature gradient.
Th e variations in PCC slab curvature were infl uenced not only by temperature diff erence but also moisture difference between the top and the bottom of the slab surface.Th e variations in temperature and moisture diff erences with time are plotted in Fig. 6.In general, temperature differences are positive during daytime and early night time and negative during late night time and early morning.In contrast, moisture diff erences presented as "RH.Diff " in Fig. 6 show the reverse trend.Especially between day 0 and day 2 aft er paving, moisture diff erences are negative for most part, i.e., higher moisture at the bottom of the slab compared to the top.Th is indicates higher drying shrinkage of concrete near the top of the slab causing the slab corner to warp upward between day 0 and day 2 aft er paving.

Changes in LVDT measurement response to temperature and moisture variations
Th e collected data from the LVDTs were voltage variations corresponding to the slab vertical displacement.To get the relative slab vertical displacement, each LVDT voltage reading was subtracted from the reference voltage reading which represents the fl at slab condition.However, it's quite diffi cult to decide the time of occurrence of fl at slab condition.So the LVDT voltage reading corresponding to fi rst zero-temperature gradient during the evaluation periods was selected as the reference reading.Th us, the actual pavement behavior could be studied based on the shape of the PCC slab at zero temperature diff erence.Th e subtracted voltage readings were then converted to displacement values based on the equation provided by the LVDT manufacturer (Omega Engineering Inc 2006).
Th e vertical displacements at diff erent slab locations (corner, edge, and center) were obtained from the corresponding LVDTs.Th e vertical displacements of corner relative to center were computed to examine the movement of slab (downward or upward).Fig. 7 illustrates the Within 12 h of paving, the concrete hardened at the positive temperature diff erence condition (78%) rather than the negative temperature diff erence condition (11%).Th e laboratory test results showed that within 12 h, the PCC achieved 50% of the 28-day compressive strength.Th us, if 12 h aft er paving is assumed to be suffi cient for the Fig. 6.Pavement temperature and moisture diff erence between the top and the bottom of slab with time relative vertical displacements of corner to center in slab 19.In this fi gure, a negative displacement value indicates downward movement of the slab while a positive value indicates upward movement of the slab.From the relative vertical displacement of corner to center, an upward movement of the slab is observed for negative temperature gradients (slab curls upward) while a positive temperature diff erence results in downward movement of the slab (slab curls downward).
Taking into consideration that the average max positive and negative temperature differences during field evaluation period were 5.8 °C and -3.0 °C, respectively, the relative vertical displacement at the max positive temperature difference should be higher than that at the max negative temperature difference.However, this could not be observed in this study.Therefore, the upward curling of the slab associated with negative temperature gradient appears to be more obvious in this study compared to the downward curl of the slab which is associated with a positive temperature gradient.This phenomenon may be related to a certain positive temperature gradient which results in flat slab condition.
A positive temperature gradient occurred between the top and the bottom of the pavement due to daytime construction and heat of hydration.Due to rapid drying of moisture in the exposed slab top, there might have been drying shrinkage of concrete near the slab top and a higher saturated condition at the slab bottom.Th is in combination with slower moisture movement through slab depth compared to temperature led to a fl at slab condition at positive temperature gradient.Th is phenomenon has been commonly observed in previous research studies on PCC early age behavior and is referred (Rao et al. 2001;Rao, Roesler 2005;Yu et al. 1998).In addition, the concrete is still plastic and hence it is quite diffi cult to support the whole weight just by the slab corners (Byrum 2001).Th erefore, when a zero temperature gradient occurs, the slab tends to curl upwards (Rao et al. 2001;Rao, Roesler 2005;Yu et al. 1998).

Changes in profi le measurement response to temperature and moisture changes
Th e Rollingprofi ler profi le measurements were analyzed to confi rm the trend in LVDT vertical displacements.Th e curvature of the slab measured by the Rollingprofi ler, called as slab curvature profi le in this study, was confounded with the construction slope and surface irregularities in the raw data of surface profi le measurements.Currently, there does not seem to be a standard method to identify the curvature of the slab due to curling and warping from the raw surface profi ling data.However, several procedures have been proposed to detect the slab curvature profi le (Byrum 2001;Marsey, Dong 2004;Sixbey et al. 2001;Vandenbossche 2003) from raw surface profi ling data.Among them, the similar procedure suggested by Sixbey et al. (2001) and Vandenbossche (2003) was used in this study.
A straight line from the fi rst reading to the end reading of the raw surface profi le curve was plotted.Each raw surface profi le data point was subtracted from this linear line to remove the construction slope, and then normalized to the fi rst measured profi le data point to eliminate the eff ect of surface irregularities.In this manner, the slab curvature profi les were zeroed to fi rst reading and end reading in a measured trace.
Th e diagonal slab curvature profi le in test section 1 constructed using this procedure is displayed in Fig. 8 for illustration.Th e slab curvature profi le measured in test section 1 clearly showed upward curling for the morning measurements and almost fl at shape for the aft ernoon measurements.Th is behavior could be attributed to the permanent upward curling and warping resulting from the positive temperature gradients during setting time and to the negative moisture diff erences aft er setting time.Th e profi le results for test section 2 could not be discussed here due to space constraints.
Comparisons between LVDT readings and slab curvature profi le readings in test section 1 were conducted.Th e relative displacements of the corner to the center or the mid-edge in measured direction (R c ) were calculated as following similar procedure by previous researchers (Marsey, Dong 2004) and plotted with time as shown in Fig. 9. Th e upward movement at the slab corner is posi- tive in Fig. 9. Th e R c trends with time are similar for both LVDT measurements and slab curvature profi les.Especially, the upward curling of the slab is evident in both measured directions, indicative of the presence of permanent upward curling and warping during the fi eld evaluation periods.Even though diff erent slabs were used for LVDT measurements and surface profi ling, it is interesting to note that the magnitudes of R c calculated from slab curvature profi le are higher compared to the LVDT measurements.However, this trend has been indirectly observed in previous research studies (Armaghani et al. 1987;Beckemeyer et al. 2002;Rao et al. 2001;Rao, Roesler 2005;Yu et al. 1998) which estimated temperature diff erence associated with fl at slab condition with diff erent slab curvature measurement techniques (using either LVDT or surface profi le data).
In general, research studies (Armaghani et al. 1987;Beckemeyer et al. 2002;Byrum 2001;Rao et al. 2001) have reported higher temperature diff erences associated with fl at slab condition based on surface profi le measurements compared to those estimated from the LVDT measurements.Th e trends observed in this study are in agreement with the fi ndings from previous research studies.

Summary of fi ndings
A newly constructed JPCP on US30 near Marshalltown, Iowa, USA was instrumented to evaluate and study the early-age JPCP behavior in terms of pavement defl ection with respect to temperature and moisture variations.Th e temperature and moisture data obtained from the fi eld were analyzed.Th e slab deformation associated with temperature and moisture were measured and analyzed through vertical displacement and pavement surface profi les.Th e following are the fi ndings of this study: during the fi rst 8 h of paving, the average pavement  temperature trends do not follow the air temperature trends.Although the air temperature decreases, the pavement temperature increases possibly due to the heat of concrete hydration.Aft er the fi rst 8 h of paving, the pavement temperature follows the air temperature with some phase lag; the temperature diff erences usually are positive at  daytime and early night time and negative at late night time and early morning while moisture differences show the reverse trend.Especially, between day 0 and day 2 aft er paving, the moisture diff erences (between the top and bottom of the slab) are negative for most of the times resulting in a higher drying shrinkage near the top slab and then causing the corner of the slab to warp upward; the relative corner displacements from center or  mid-edge (R c ) calculated from both slab curvature profi le measurements and the LVDT measurements show similar trend.Both measurements show that the slab behavior during fi eld evaluation periods tend to be mostly upward at the corner.Th is behavior can be attributed to the permanent upward curling and warping resulting from positive temperature gradients during setting time and negative moisture diff erences aft er setting time.

Fig. 9 .
Fig. 9. Th e comparison of relative displacement of the corner (R c ) with time between LVDT measurement and slab curvature profi le: a -diagonal direction; b -transverse direction a b