- Email: [email protected]
Field monitoring study of an integral abutment bridge supported by prestressed precast concrete piles on soft soils
Engineering Structures 104 (2015) 18–31
Contents lists available at ScienceDirect
Engineering Structures journal homepage: www.elsevier.com/locate/engstruct
Field monitoring study of an integral abutment bridge supported by prestressed precast concrete piles on soft soils B. Kong, C.S. Cai ⇑, X. Kong Dept. of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, LA 70803, United States
a r t i c l e
i n f o
Article history: Received 10 November 2014 Revised 19 August 2015 Accepted 6 September 2015
Keywords: Integral bridge Field monitoring Thermal performance Slab strain Backfill pressure
a b s t r a c t Integral abutment bridges (IABs) have been constructed during the past several decades around the world. The purpose was to eliminate expansion joints and minimize joints induced problems. Even though IABs have been widely accepted due to the satisfying performances, yet they have not been largely applied in practice. Some of the reasons may be attributed to the uncertainties of these bridges under different loading conditions, especially the daily and yearly varying temperature effects. In this paper, the behavior of the first IAB constructed on the soft soil condition in Louisiana is discussed. A field monitoring program is introduced and the measured results from 08/11/2011 to 03/15/2014 are presented. The field monitoring program leads to the following observations, (1) significant seasonal and daily temperature variations are observed on the bridge slabs but within the AASHTO temperature design specification; (2) the displacements and rotations of bridge components are well correlated with the temperature variations; (3) the thermal stresses generated in the slabs may exceed the allowable material cracking capacity; (4) the soil behavior behind the abutments is complicated and long term monitoring program is needed; (5) the integral abutment primarily behaves in translation rather than rotation; and (6) the pile inflection (zero bending) point is observed and the strong and weak axels bending are all important due to the bridge skewness. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Expansion and contraction are two basic responses of bridges induced by temperature variations. Traditionally, a relief system, consisting of expansion joints, bearing supports, or other devices, are designed to accommodate these movements. After years of services, however, this system, especially the expansion joint, may become one of the vulnerable elements affecting the sustainability of bridges [18,21]. Bridge engineers have been trying to eliminate expansion joints whenever possible; and the concept of integral bridges without joints was inspired. Generally speaking, integral abutment bridges (IABs) can be categorized into three types [24]: (1) a full IAB, as discussed in this paper, refers to a single or multi-span bridge as shown in Fig. 1. The superstructure of the bridge, i.e., concrete slabs, prestressed concrete beams, steel girders, and approach slabs, is casted monolithically with a stub type abutment and supported on a row of piles; (2) a semi-integral bridge is similar to the full IAB, but the abutment is not rigidly connected to the substructure; and (3) a deck-extension bridge extends its deck slab over the ⇑ Corresponding author. E-mail address: [email protected] (C.S. Cai). http://dx.doi.org/10.1016/j.engstruct.2015.09.004 0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
abutment into the approach pavement, but the main beams or girders are not fixed on the abutment. In an integral configuration, expansion joints are eliminated and the corresponding jointrelated issues can be minimized. However, a new challenge arises since the horizontal movements from the superstructure are transferred to the substructure, which have to be well accommodated through the soil–structure interaction or other special designed mechanisms. The benefits of IABs have been widely accepted in the past several decades around the world. Survey has been conducted on the construction experience of IABs in American [16] and European countries [4,25], where many similarities were observed even though significant differences existed in some aspects. Field monitoring study methods have been adopted to investigate and justify the design and construction concepts, including (a) the maximum allowable design criteria (e.g., total and individual bridge’s span lengths and skews); (b) the structure design parameters (e.g., orientations of the pile, abutment, and wingwall); (c) the soil–structural interaction behaviors (e.g., between the soil–pile, abutment-backfill, and approach slab-backfill); (d) the joint connection effects (e.g., at the interfacial locations between the abutment-deck-girder, abutment-pile cap, approach slababutment, and intermediate pier-girder); (e) the stress relief
B. Kong et al. / Engineering Structures 104 (2015) 18–31
19
Fig. 1. Schematic view of a full integral abutment bridge (IAB).
mechanisms (e.g., diameters, depths, and filling materials of the pre-sized holes surrounding the piles, and the compacting degree of the backfill materials behind the abutments); and (g) the long term effects (e.g., the temperature, shrinkage, creep, and steel relaxation) [3,13,2,10,11,15,5,8,12,19,7,25,14,20,26,4,6]. In the state of Louisiana, no full IABs have ever been built before the year of 2011. The reasons are partly due to the lack of references from the previous studies regarding on the behaviors of bridges on the unique soil conditions in Louisiana. To this end, the present paper reports the field monitoring program from 08/11/2011 to 03/15/2014 for the first full IAB, the Caminada Bay Bridge, designed by the Louisiana State Department of Transportation and Development (LADOTD). This bridge also has some special features that have not been focused in the previous investigations on integral bridges, such as the long continuous slab spans, deep precast prestressed concrete piles, shallow integral abutment, and very soft soil conditions. Based on the monitoring results, the temperature induced effects on the superstructure and substructure of such a bridge are demonstrated. Specifically, (1) significant seasonal and daily temperature variations are observed on the bridge slabs but within the AASHTO temperature design specification; (2) the displacements and rotations of bridge components are well correlated with the temperature variations; (3) the thermal stresses generated in the slabs may exceed the allowable material cracking capacity; (4) the soil behavior behind abutments is complicated and long term monitoring program is needed; (5) the integral abutment primarily behaves in translation rather than rotation; and (6) the pile inflection (zero bending) point is observed and the strong and weak axels bending are all important due to the bridge skewness.
2. Bridge descriptions The Caminada Bay Bridge is located at the Grand Isle, LA (29°150 4800 N, 89°570 2400 W), about 160 km to the south of New Orleans, LA. While the total length of the bridge is 1202 m (3945 ft), the monitoring program is conducted on the first eleven spans, as shown in Fig. 2, including a 3 m (10 ft) sleeper slab, a 12 m (40 ft) approach slab, a 91 m (300 ft) continuous concrete slab, and also the abutment, bent, pile, and soil. The width of the bridge is 15 m (50 ft) consisting of two 6.4 m (21 ft) lanes and a 2 m (7 ft) sidewalk on the northern side. For the parts that are monitored, the slabs are fully integrated with the first bent (Bent1) at the left end, simply supported on the eleventh bent (Bent11) at the right end, and rigidly connected with all the interior bents in between, where the expansion and fixed connection joints are designated as ‘‘E” and ‘‘F” in Fig. 2. Each bent is rigidly supported on a single row of four prestressed precast concrete (PPC) piles. The soil types, referred to the boring log information near Bent1, can be approximately subdivided into two layers, including a medium
sandy soil layer from the ground to the depth of 18.9 m (62 ft), and a medium clay layer through the rest of the depth. The materials and the corresponding properties for this Caminada Bay IAB are listed in Table 1. 3. Instrumentations Bridge Diagnostics, Inc. (BDI) was contracted by the Louisiana Transportation Research Center (LTRC) and Louisiana State University (LSU) to install the bridge monitoring system. In this project, a total of 81 instruments were applied on the bridge, as listed in Table 2, including the vibrating wire strain gages, vibrating wire tiltmeters, vibrating borehole wire extensometers, vibrating wire pressure cells, piezometers, and vibrating wire thermistors. The large application of vibrating wire gages is due to their good performances, without drifts, for the long-term monitoring. Meanwhile, each sensor is provided with an extra temperature thermistor so that the temperature information of the bridge elements can be simultaneously obtained. 3.1. Superstructure instrumentation For the superstructure, a total of 22 sensors, with 14 embedded sisterbars and 8 surface strain gages, as shown in Fig. 3, were applied on the 46 cm (18 in.) depth concrete deck. They were adopted to measure the positive and negative strains due to the temperature changes. Specifically, the embedded sisterbars were placed at the rebar locations before the pouring of concrete with 8 cm (3 in.) above the bottom surfaces on the approach slab, Span1, Span3, and Span5, and with 5 cm (2 in.) below the top surfaces on the Bent1, Bent2, and Bent5. The surface strain gages, otherwise, were mounted under the bottom surfaces from Span3 to Span6 after the completion of the concrete pouring. 3.2. Substructure instrumentation For the substructure, a total of 59 instrumentations were installed as shown in Fig. 4. The monitoring program related to sensors and instrumentations in the abutment and the foundation of the bridge was conducted by the research groups of Drs. G.Z. Voyiadjis and K.A. Alshibli at LSU. Some of the important information is briefly described here for the convenience of readers, and the detailed information can be referred to the report by Voyiadjis et al. [23]. For example, (a) 32 sisterbars were installed at the four corners of two 24 m (80 ft) long PPC piles at the easternmost of Bent1 to measure the pile strains. The distances from the sisterbars at each sensor sections to the bottom surface of Bent1 were 1.2 m (4 ft), 3.7 m (12 ft), 6.1 m (20 ft), and 8.5 m (28 ft), respectively; (b) 2 tiltmeters were attached at the middle section of the 1.2 m (4 ft) high Bent1 and Bent11 to record the bents’ rotations; (c) two rows
20
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 2. Plan and elevation view of the first eleven spans of Caminada Bay IAB.
Table 1 Material properties of Caminada Bay Bridge. Structure items
Material types
Material properties
Slabs, bents
Class AA (M) concrete Class P (M) high performance concrete
Strength of 28 MPa (4000 psi)
Reinforcing rebar in the bents and slabs
Type 316LN stainless steel
Reinforcing rebar in other structures Prestressing strands
Grade 60 steel
Minimum compressive strength of 41 MPa (6000 psi) at 28 days, and average compressive strength of 69 MPa (10,000 psi) at 56 days Elastic modulus of 200 GPa (30,000 ksi), tensile strength of 515 MPa (75 ksi), and yield strength of 205 MPa (30 ksi) Yield strength of 414 MPa (60 ksi)
Grade 270 steel
Yield strength of 1860 MPa (270 ksi)
PPC piles
of 9 soil pressure gages, with one row placed about 30 cm (1 ft) below the slab’s bottom surfaces and the other at the bent’s middle height, were mounted on the backwall of Bent1 to measure the soil pressures on the abutment. The pressure gages in the transverse direction from the easternmost end of Bent1 were 1.4 m (56 in.), 3.5 m (138 in.), 5.6 m (221 in.), 7.7 m (303 in.), and 11.9 m (468 in.), respectively; (d) 4 soil strain gages, with 2 connected to Bent1, and another 2 floating with about 1.2 m (4 ft) and 2.1 m (7 ft) distance away from Bent1, respectively, were applied to investigate the soil’s deformation or Bent1’s movement; and (e) several other gages were embedded in Bent1 and through the surrounding soil depths to obtain the corresponding structure temperatures and pore water pressures.
3.3. Data acquisition system
Table 2 Instrumentations applied on Caminada Bay Bridge.
The data acquisition system, as shown in Fig. 5, consists of a CR1000 datalogger, an AVW200 interface, an AM16-32 multiplexer, a wireless cell modem, and a solar battery. In addition, it has been outfitted with a wireless communication link so that the operation and data collection can be handled remotely from a computer through a LoggerNet program at LSU. It is worthy to note that the acquisition system has been equipped with a battery back-up mechanism. It allows continuous data collection throughout a typical power outage and alert emails will be sent to researchers when the system is working on battery.
Structure
Gages
Location
Numbers
Superstructure
Sisterbar
Approach slab bottom (embedded) Span 1, 3, & 5 bottom (embedded) Bent 1, 2, & 5 top (embedded) Span 3, 4, 5, & 6 bottom (surface)
2
Two easternmost piles at Bent1 Bent1 and Bent11 (surfaces) Bent1 (embedded) Bent1 back face Backfill behind Bent1
16 2 = 32
3.4. Data post-processing
12=2 4 9 4
Backfill behind Bent1 Backfill behind Bent1
2 6
The data acquisition was conducted from 08/11/2011 to 03/15/2014 in which the raw data of each sensor were recorded every 3 min and 20 s. Based on the assessment of the collected data, most of the instruments functioned well. Some of the failed sensors include 3 surface strain gages on the slabs, 4 sisterbars in the piles, 2 pressure gages on the abutment, and 1 soil strain meter.
Strain gage Substructure
Sisterbar Tiltmeter Thermistor Pressure cell Soil strain meter Thermistor Piezometers
23=6 23=6 24=8
B. Kong et al. / Engineering Structures 104 (2015) 18–31
21
Fig. 3. Plan and elevation views of sensors on the approach slab and the first six-span slabs.
However, these missed data does not affect the investigations since sensors are redundantly applied for most of the concerning areas. Two post-processing steps were conducted on the raw data before further analysis. First, the outlier data have to be picked out and removed. Otherwise, they will affect the accuracy of the calculated hourly or daily average temperature values, especially when one or several data within one hour is mistakenly too large or small. Secondly, complying with the requirements of the sensor suppliers, a proper temperature correction process should be performed for certain sensor types. It is due to the differences of the thermal expansion coefficients between the steel wire in the sensor gages and the thermal coefficients of the measured concrete elements, e.g., the strain gages in concrete slabs and PPC piles. Some basic algorithms have been proposed for the data outlier removal in the statistic field, such as the Standard Deviation Method, Z-score, Modified Z-score, Tukey’s, Adjusted Boxplot, MAD and Median Rule [22]. While each method has its own advantages and restrictions, most methods are based on the assumption that the data is normally distributed. For example, as shown in Fig. 6, about 68%, 95%, and 99.7% of the data fall in the area within 1, 2, and 3 times the standard deviation from the mean value if the data follows a normal distribution. In this sense, if a data falls with a distance, e.g., 2 or 3 times of the standard deviation away from the mean value, it can be considered as an outlier. In this paper, the raw data from a representative strain gage, i.e., located in the concrete slab of Span5 as shown in Fig. 3, is taken out as an example to discuss the outlier removal algorithm adopted in this study. Four sets of sample data within one hour, i.e., 08/11/2011 17:00, 09/11/2011 24:00, 10/11/2011 03:00, and 11/11/2011 09:00, are randomly selected from the raw readings, and the normality test is conducted on these data. Fig. 7 shows the representative normality test plots of the strain and temperature readings at 08/11/2011 17:00, respectively. The other results are calculated and listed in Table 3. Based on the statistic theory, the distribution is non-normal if the P-value is smaller than 0.01 or the A-squared value is larger than the critical values, i.e., 0.787 and 1.072 for 95% and 99% confidence level, respectively. It can be observed based on the calculations that while most strain data within an hour period can be considered as normally distributed, the temperature data are not always normally distributed. For example, the temperature data in the case of 08/11/2011 17:00 is more uniformly distributed within an hour. In addition, it can be reasonably predicted that for sensors embedded in the deep soil, e.g., strain gages in the piles under the ground, the temperatures may be more constantly varying throughout years. Therefore, through a comprehensive comparison between different outlier removal methods, together with the sensitivity
analysis, the SD and Tukey’s method are tentatively adopted and coded in MATLAB [17]. The SD method using the mean and standard deviation may not be appropriate for some cases since the extreme values may affect the standard deviation; however the Tukey’s method is robust, which is less sensitive to the extreme values. Therefore, the outlier removal is conducted on the measured data within every 1-h time frame in three steps: (1) performing normality test, (2) using the SD method on the normal distributed data, and (3) using the Tukey’s method on the non-normal distributed data. It is worthy to note that most of the outliers in the present measured data are extreme values, thus the simple visual removal method using the filter function in Excel was adopted as well. Detailed information regarding on these methods is not elaborated in details, since they are the basic theory and can be referred to the statistical textbook in the public domain [22]. The discussions in the following sections are based on the data after outlier removals and temperature corrections. 4. Field monitoring results The monitored results reported in this study were from 08/11/2011 to 03/15/2014. The initial reference condition for all the instrumentations was set up on 08/11/2011, i.e., all values were set as zeroes. Therefore, the results discussed below are the changes of bridge thermal responses corresponding to the bridge behaviors on 08/11/2011. It should be noted that some data are missed during the period from 03/10/2013 to 7/31/2013. However, it does not affect the discussions in the following sections, since the bridge responses show a yearly returning pattern, and the extreme values and distribution trends are still obtained. The following discussion is aiming to qualitatively describe the thermal response of the IAB under temperature effects. All the results are based on the assumption that the bridge response is primarily correlated to the temperature variation. The Pearson’s correlation coefficient is firstly calculated which is a measure of the linear correlation of dependence between two variables. Values between +1 and 1 are provided in the following tables in which +1 is for a total positive correlation, 0 is for no correlation, and 1 is for a total negative correlation. It should be noted that results in the tables only serve to provide a general overview of the correlation degree, and more detailed discussion and measurement locations are elaborated in the following sections. Table 4 shows the temperature correlation between the ambient air and bridge components, where the correlation degree is high at the bridge slab and decreases through the bridge depth to the pile bottom. Table 5 shows the correlation between the temperature change and bridge response. For most of the cases, the correlation degree is high,
22
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 4. Instrumentations applied on substructure Bent1, (a) plan view, (b) side view, (c) elevation view [23].
especially for the measurements on the rotation and displacement. However, the strain results show comparatively low correlation degree, especially at the bottom of slab surface, and that may be attributed to the environmental effects. Based on the Pearson’s correlation method, the correlation assumption between the temperature variations and bridge responses is generally valid in the following discussion. 4.1. Environmental conditions During the monitoring period, no instrumentations were particularly installed to measure the environmental conditions,
e.g., ambient temperatures and wind speeds. However, this information is of great importance especially for the thermal behaviors of IABs. It is because the uncertainties of IABs are mainly caused by the temperature changes within the bridges, and those variations in turn are determined by the environmental conditions. In this sense, the ambient temperatures and wind speeds, recorded at the bridge site from a nearest weather station located at the Grand Isle, LA (http://tidesandcurrents.noaa.gov), are firstly referred and discussed. The measured hourly-varying air temperatures and wind speeds at the weather station are shown in Fig. 8. Firstly, the periodical and cycling varying trends can be clearly observed.
23
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 5. Acquisition system applied on the pole next to the bridge.
Specifically, the sinusoidal fluctuating patterns are approximately shown in the yearly and hourly varying temperatures. This observation will be helpful if the temperature data is not available. For example, some of the missing temperature data are predicted using the curve fitting method in the following section. Secondly, the minimum and maximum ambient temperatures during the monitoring period, as marked in Fig. 8, are approximately 0 °C (32 °F) and 35 °C (95 °F), respectively, with a difference of 35 °C (63 °F). Finally, the number of the freezing days during the three monitored years is less than 14. According to the AASHTO LRFD [1] specification, this bridge can be considered being located in a moderate climate region; and the corresponding bridge design
Fig. 6. Normal distribution curve.
Normality Plot
Histogram 3 5
2
4
1
Z Score
Number
6
3 2
y = 4.0452x - 9820.8 R² = 0.986
0
1
-1
0
-2 -3 2,427.2
Strain Reading (Digital)
2,427.6
2,428.0
2,428.4
Strain Reading (Digital)
(a) Histogram
3
20
2
15
1
Z Score
Number
25
10 5 0
Normality Plot
0
y = -0.0039x R² = -8E-17
-1 -2
88.9
89
89.1
89.2
-3 88.9
Temperature Data (ºF)
Temperature Data (ºF)
ºC=(ºF-32º)x5/9
(b)
ºC=(ºF-32º)x5/9
Fig. 7. Normality test data measured at Span5 on 08/11/2011 17:00: (a) strain, (b) temperature.
88.9
24
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Table 3 Normality test for the strain and temperature readings of the strain gage at Span5. Date
08/11/2011 09/11/2011 10/11/2011 11/11/2011
Strain reading
17:00 24:00 03:00 09:00
Temperature reading
A-squared
P value
A-squared
P value
0.122 0.231 0.213 0.437
0.984 0.77 0.827 0.264
17.627 0.714 0.346 0.638
0 0.051 0.442 0.081
temperature specification for this region is compared and correlated with the field measurements in the following discussion. The variations of the wind speeds are also shown in Fig. 8. The wind speeds and air temperatures generally do not show obvious correlations. The occurrences of the peak wind speeds, however, are almost in coincidence with the conditions when the temperatures are at sudden drops. In addition, the measured wind speed values are generally wandering around 5 m/s, with some days larger than 10 m/s but smaller than 17 m/s. According to Elbadry and Ghali [9], the wind speeds greatly determine the thermal convection coefficients, and those coefficients in turn affect the heat transfer mechanisms at the surfaces of the bridge slabs. The effects of the present wind speed on the temperature variations at the bridge’s surfaces will be discussed in the next section based on the field measurements. 4.2. Bridge temperatures The thermal performances of bridges are directly determined by the variations of bridge temperatures. These temperatures are not uniformly but varyingly distributed within different structural components throughout the bridge. In this section, with the help of the largely applied instrumentations, where each gage is attached with a thermistor, the measured temperatures in terms of the seasonal variations, daily gradients, and distributions from the slab’s top surface to the pile’s bottom section are observed and discussed. Fig. 9 shows the measured hourly-varying temperatures at the bent’s top surface, approach slab’s bottom surface, and deck slab’s bottom surface, respectively. The measured ambient temperature is also plotted in the figure for a reference. It can be observed that the top surface of the bent or slab has significant temperature variations, especially during the summer seasons; while during the winter season, these high variations decrease and the temperatures at the top and bottom surfaces are close to each other. This temperature distributing phenomenon can be explained by the heat transfer mechanisms. The temperatures at the top surfaces are primarily affected by the air temperatures, solar radiations, and surface convection behaviors. For example, the high temperatures at the top surfaces often appear in the hot time of the day and with no or very slight winds. For the hourly temperature fluctuation, they are attributed to the change of wind speeds. In the present bridge, temperature fluctuations perform actively when the wind speeds are larger than 5 m/s based on the field measurements
shown in Fig. 8. Moreover, after the top surface is heated by the solar, the high energy transfers and conducts through the slab’s depth to the bottom surface. When arriving at the bottom surface, the energy is either blown away by the adjacent air through convection or transferred to the soil underneath through conduction. It can be observed in Fig. 9 that the temperatures at the approach slabs are in between to that of the bent’s top and slab’s bottom surface since the approach slab contacts with the soil where the thermal convection or conduction may be less significant than the ambient air or concrete. For the seasonally and uniformly varying slab’s temperatures during the monitored period, as marked in Fig. 9, the maximum and minimum values are approximately 35 °C (95 °F) and 0 °C (32 °F) which is calculated as the average temperatures at the bridge’s top and bottom surfaces, respectively. Therefore, for the present IAB, if assuming the base construction temperature equivalent to the measured minimum air temperature of 0 °C (32 °F), the effective temperature difference that will induce seasonal movements is 35 °C (63 °F). This value is within the temperature design values of 38 °C (70 °F) according to the AASHTO LRFD [1] specification. Besides the uniform temperature variations, the gradient distributions are also important in bridge designs. For temperature distributions along and perpendicular to the traffic directions, the temperature differences are reasonably negligible based on the field measurements; thus they are not demonstrated here. For the vertical gradients through the slab depths, however, the differences are apparent. Fig. 10 shows the temperatures measured at the top and bottom surfaces of the slab and the ambient air during the hottest week, i.e., 08/19/2011 to 08/26/2011, and the coldest week, i.e., 12/29/2011 to 01/06/2012, respectively. During the hottest week, the temperatures at the top surfaces and air show a high correlation with almost the same variation trends. In addition, a certain lagging effect, though not significant, is observed between the temperatures at the top and bottom surfaces due to the thermal inertia for the present 46 cm (18 in.) depth slab. During the coldest week, however, the temperature differences are not significant and both the temperatures at the top and bottom surfaces are close to each other. According to the AASHTO LRFD [1] specifications, the positive and negative thermal gradients are specified based on four subdivided radiation zones, together with the consideration of the superstructure’s materials, geometries, and overlays types. Then, if following the code guidelines, the positive and negative temperature difference at the top and bottom surfaces of the current slab are supposed to be 10.2 °C (18.5 °F) and 3.1 °C (5.6 °F), respectively. Therefore, the measured positive and negative temperature differences, i.e., 10 °C (18 °F) on 08/20/11 17:00 and 1.11 °C (2 °F) on 1/6/12 07:00, respectively, are within those values calculated from the design specification.
Table 5 Pearson’s correlation coefficient between bridge temperature and structure response. Slab average temperature
Table 4 Pearson’s correlation coefficient between ambient air and bridge temperature. Ambient air temperature Slab average temperature Bent top surface temperature Slab bottom surface temperature Pile B-B section temperature Pile C-C section temperature Pile D-D section temperature Pile E-E section temperature
0.963 0.863 0.797 0.598 0.028 0.683 0.578
Bent top surface strain Slab bottom surface strain Bent11 rotation Bent1 rotation Soil displacement at B location Soil displacement at A location Bent bottom surface soil pressure Bent top surface soil pressure Exterior pile B-B section strain (4) Exterior pile B-B section strain (27) Exterior pile B-B section strain (35) Exterior pile B-B section strain (44)
0.629 0.010 0.958 0.893 0.871 0.841 0.841 0.386 0.884 0.709 0.016 0.681
B. Kong et al. / Engineering Structures 104 (2015) 18–31
25
4.3. Slab strains
Fig. 8. Measured environmental conditions at the weather station.
Fig. 11 shows all the temperature variations from the top surfaces of the slabs to the bottom parts of the piles, and some of the observations are described as follows. First, the temperatures generated by the solar radiations can be transferred through about 1.7 m height deck and abutment to the top part of the pile. However, the temperatures show negligible variations along the piles to the bottom section. Second, the temperature variations in the soils, e.g., curve No. 7, are more constant with the soil going deeper, even though they may be slightly influenced by the environments near the ground. For example, the temperatures in soils could be higher than those in the slabs and bents during the winter seasons, such as curve No.4. Fig. 12 shows the best curve fitting relationships between the measured bridge and air temperatures, where the former one is considered as the average results between the readings from the sensors at the bent’s top and slab’s bottom surfaces. It can be observed that the bridge and air temperatures are almost linearly correlated in the middle temperature range, while the polynomial curve shows better results in the lower and higher temperature ranges. This conclusion justifies one of the statements discussed above that it is able to predict the bridge’s uniformly varying temperatures using the air temperatures.
Fig. 9. Measured hourly-varying temperatures of bridge slabs.
For reinforced concrete structures, the strains in the steel rebar and surrounding concrete should be the same prior to the cracking of the concrete due to the strain compatibility. Generally speaking, the concrete is weak in tension, and the corresponding strength is about 10% of its compressive strength. When in compression, however, the concrete can sustain up to about 300 microstrains before failures for general concrete materials adopted in bridge engineering. Fig. 13 shows one representative readings of a strain gage at the Bent5 top rebar location. The highly-varying strain variations can be observed, and the long term seasonal trends show a negative correlation with the temperature changes. In addition, the calculated thermal stresses change during the temperature decrease of 35 °C (63 °F), if using the maximum measurement of 100 microstrain, would induce a stress change of 2.49 MPa (360 psi). This value is about 9% of the concrete compressive strength. In this sense, the top surface of the bent has a possibility of cracking due to the temperature variations alone since the slab–bent–pile connection joint is integrally constructed. Similarly, another reading of the representative gage embedded in the bottom parts of the Span5 slab is shown in Fig. 13. The induced compressive thermal stress variation, if using the 50 microstrains, is about 4% of the compression strength during the monitored periods.
Fig. 10. Bridge and air temperatures: (a) hottest week from 08/19/2011 to 08/26/ 2011, (b) coldest week from 12/29/2011 to 01/06/2012.
26
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 11. Measured bridge temperatures through the depth of bridge.
4.4. Abutment behaviors The displacements and rotations of the abutments are measured using the soil strain meters and tiltmeters [23], where the positive sign convention is defined for bents moving or rotating inwards, i.e., toward the water between the bridge two ends. Fig. 14 shows the soil deformations at two locations, i.e., right behind the Bent1 backwall surface and 2.1 m away from Bent1, designated as locations A and B in the plot, respectively. The displacements at these two locations are observed showing good correlations with the bridge temperature variations, where the yearly variations of the soil deformations are approximately 5.5 mm at location A and 2 mm at location B, respectively. The relationships between the variations of the bridge temperatures and soil deformations at these two locations A and B are plotted in Fig. 15. Specifically, three temperature decreasing periods are selected in three consecutive years, which are from 08/11/2011 to 01/14/2012 (1st year), 08/16/2012 to 01/14/2013 (2nd year), and 10/6/2013 to 01/29/2014 (3rd year), respectively. The soil deformations are plotted with respect to the temperature variations, and the relationships between them are linearly fitted and shown in the figure. First, it can be observed that, for the soils at location A right behind abutment, the slops of the fitting curves in the 2nd and 3rd years are larger than that in the 1st year. The
Fig. 12. Best fitting of the air temperatures and the bridge slab average temperatures.
Fig. 13. Measured strains and temperatures: (a) top surface of Bent5, (b) bottom surface of Span5.
reason behind this scenario may be that the soils are pushed away and disturbed due to the slab’s thermal displacements during the first year, so that the soils become soft and are more readily to be pushed away by the thermal deformations in the following 2nd and 3rd years. Secondly, the slop of the 3rd year curve is smaller than that of the 2nd year one which may be attributed to the soil settlement behaviors around gap or the soil property change due to the repeated compaction. During the winter seasons, the bridge stops expanding and begin to contract; thus a gap is generated behind the abutment since soils are not elastic materials
Fig. 14. Measured bridge temperatures and soil deformations.
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 15. Linear fitting of the bridge temperature vs. soil deformation changes: (a) location A, (b) location B.
so that they cannot be fully returned to their initial positions. Under this circumstance, the soils at above or near around will fill the gap. As time goes, the soils become more compacted, which impedes the bent displacements in the years after. Thirdly, for the soils at location B, the slops of the fitting curves in the three years almost do not change. It may be implied that the bridge thermal deformations have negligible effects on the soils at 2.1 m away from the abutments under the current bridge length and temperature variations. Besides the displacement behaviors, the rotations of the bents are measured on the Bent1 and Bent11, as shown in Fig. 16. Similar to the displacement behaviors, rotations are also negatively correlated with the bridge temperature variations. For the Bent1, with the temperature changing of 35 °C (63 °F) during the monitored period, the rotation of the abutment varies about 0.028°. This small value is due to the integral structural configuration and soil– structure interaction effects. For the Bent11, however, the rotations are almost perfectly correlated with the temperature variations, with the values approximately 10 times larger than that at Bent1; and that is attributed to the simply supported, not integral, connection detail at the slab–abutment. In addition, compared with the displacement of Bent1, the variations of the rotations of Bent1 is insignificant and only contributes about 10% of the total displacements at Bent1. In this sense, Bent1 should primarily move in translations rather than rotations. Among all the recorded data, the measured backfill pressures on Bent1 show the most significant randomness. There seems no clear and regular trends for all the pressure sensor readings; and that may be due to the uneven soil properties and compaction levels especially during the first year. However, the backfill behaviors
27
Fig. 16. Measured bridge temperatures and rotations: (a) Bent1, (b) Bent11.
become clear when the soils are stable in the long run. Fig. 17 shows the representative top and bottom pressure readings at the middle location of Bent1. The pressures have negative correlations with the soil displacements, and the variation of the soil pressure through the monitored years is about 20 kN/m2 when it becomes stable. In addition, the top and bottom pressure cells show the same sign, which justifies one of the statements above that the abutment moves primarily in translation instead of rotations. It should be noted that the observations and conclusions obtained on the behaviors of integral abutment in this section,
Fig. 17. Measured backfill pressures at the middle location of Bent1.
28
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 18. Plan view of the Bent1 and pile diagram.
Fig. 19. Measured pile strains at cross sections from B-B to E-E.
including the deformation, rotation, and earth pressure, need to be further justified with the long term monitoring results. 4.5. Pile strains Strains at the four corners of the exterior pile (EP) and interior pile (IP), as shown in Fig. 18 [23], are measured through the pile’s depths at four elevations from the top to bottom sections. For the convenience of discussion, these four sections from the top to bottom are designated as B-B, C-C, D-D, and E-E as shown in Fig. 4, respectively, where the positive x and y axes are defined as perpendicular and parallel to the traffic directions as shown in Fig. 18. All the measured strains at the four pile sections are shown in Fig. 19, where the numbers at the four corners are the sensor names. It can be observed that the strains measured at the top parts of the piles, e.g., B-B and C-C sections, are larger than those at the bottom, e.g., D-D and E-E sections. For example, the measured maximum positive and negative strain variations, appearing at the C-C section, are 40 and 34 microstrains, respectively, though they are only less than 3% of the concrete compressive strength of the piles. In addition, the bridge contracts inward were corresponding to the temperature variations, such as during the temperature decrease period from 06/11/2011 to 12/11/2011. Under this condition, the outer surfaces of the piles should have been extended with tensile behaviors. This performance has been justified from the measured strains at sections C-C to E-E, where the pile is bending with respect to the negative x-axis. At the B-B
section, however, the pile is differently bending with respect to the positive y-axis, which may be due to the bridge skew effects or the rigid connections between the pile head and bent, where tensile steel rebars are constructed by extending from the pile heads to the bents. All the measured strains are further calculated and decomposed into four strain components, i.e., the axial strain, x-axis bending, y-axis bending, and torsional strains, according to the basic knowledge of the materials mechanics. Fig. 20 shows the strains at two representative sections of both the EP and IP, i.e., sections B-B and D-D, and some of the behaviors are observed as: (a) for the axial strains, they show good correlations with the temperature variations at both sections of the two piles. For example, with the increase of temperatures, from December to August, the bridge expanded and the axial strains were accordingly increased. For the daily varying trends, the two piles show opposite axial strain signs. This scenario may be due to the different abutment thermal behaviors in the transverse direction, where the decrease of temperatures induced sagging movements at the middle parts of the bents but hogging ones at the two far ends; (b) for the x-axis bending strains, they are directly affected by the longitudinal thermal movements of the bridge; (c) for the y-axis bending strains, they reflect the behavior of the bridge bending transversely with respect to the longitudinal y-axis of the bridge. It can be found that the y-axis bending strains at the top B-B section are larger than that of the x-axis strains, and that may also be attributed to the skew effects or the rigid pile-bent connections of the bridge; (d) for
B. Kong et al. / Engineering Structures 104 (2015) 18–31
29
Fig. 20. Calculated strain components: (a) B-B cross section, (b) D-D cross section.
the torsional strain, the measured values are reasonably as small as expected and can be ignored. Fig. 21 shows the thermal strains along the depth of the EP at three representative days, i.e., 08/14/2011, 11/02/2011, and 02/13/2012. They refer to the cases with the temperature increase of 5 °C (10 °F), 17 °C (30 °F), and 28 °C (50 °F), respectively. Through comparisons, the results in the axial strain and torsional strain plots do not show much difference with the change of temperatures. However, the x-axis bending strain, induced from the bridge longitudinal deformations, is most sensitive to the temperature variations, and the y-axis bending strain cannot be ignored, especially at the top parts of the pile, due to the bridge skew and fixed connection effects. In addition, in the x-bending strain and
y-bending strain plots shown in Fig. 21, zero bending points (or inflection points) appear in both cases. 5. Conclusion This paper reports the monitoring results for the first IAB, Caminada Bay Bridge, on the unique soft soil conditions in Louisiana from 08/11/2011 to 03/15/2014. Based on the assessment on the measured results, about 10 out of 81 gages were failed. Some of the sensors show outlier data which can be either filtered by Excel or the algorism proposed in the paper. The Pearson’s correlation coefficient is calculated and it demonstrated that there is a certain correlation degree between the temperature variations and bridge
30
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 21. Measured exterior pile strains under three temperature variations.
responses in general. Some of the most important observations are concluded as follows: (1) The air temperatures, measured at the weather station near the bridge site, show periodical trends, i.e., in an approximately sinusoidal varying pattern in this case. In addition, based on the measured wind speeds in between 10 m/s and 17 m/s, the surface convection behaviors are expected to be significant, and this phenomenon is proven from the field measurements where highly varying behaviors at the slabs are observed at the top surfaces of the bridge. (2) The measured seasonal temperatures variations of 35 °C (63 °F) and daily positive and negative gradients of the slabs, i.e., 10 °C (18 °F) and 1.11 °C (2 °F), respectively, are within the design values that are specified by the AASHTO LRFD [1] specification. The average temperature within the slabs can be predicted using the ambient air temperature by a linear or polynomial function. In addition, through a comprehensive study on the temperature distributions of the whole bridge, i.e., from the slab’s top surface to the pile’s bottom part, the bridge temperatures are found primarily varying within the superstructure. However, the temperature variations at the top part of the pile are also significant. (3) The strains measured at the surfaces of the top bents and bottom slabs generally show opposite variation trends. Based on the measurements during the monitored period, the maximum temperatures induced tensile stresses at the top bents of 2.49 MPa (360 psi) may possibly crack the concrete, while the induced compressive stresses are negligible and the maximum value is approximately 4% of the concrete material strength. (4) The soils behind the abutment affect the behaviors of the integral abutments in terms of their displacements and rotations. These effects are complicated, and the soil restraints to the abutment deformations will accumulate with time due to the plastic behaviors of soils. However, at the locations with 2.1 m away from the integral abutments or when the expansion joints are provided between the slab and abutment, these soil restraining effects become negligible.
During the monitoring period, the variations of the soil displacements right behind the integral abutment and 2.1 m away are 5.5 mm and 2 mm, respectively. The abutment primarily behavior in translation rather than rotation. (5) The bending strains of the piles with respect to the two main axes, i.e., parallel (y-axis) and perpendicular (x-axis) to the traffic direction, are both important, especially at the upper parts of the piles, due to the skew effects and rigid pile–bent connections. However, these strain values, with a maximum variation of 40 microstrains, are a small portion of the concrete compression capacities under the current temperature variations alone. In addition, for the pile bending profiles induced from the thermal movements of the superstructure, the zero bending points (or inflection points) appear at both bending directions.
Acknowledgments The investigators are thankful to the Innovative Bridge Research and Deployment (IBRD) program, Federal Highway Administration and Louisiana Transportation Research Center (LTRC) for funding this project. The contents presented reflect only the views of the writers who are responsible for the facts and the accuracy of the data presented herein. We would like to also express thankfulness to those who provided help during the development of these initial tasks of this research program. Special thanks go to the project manager Dr. Walid Alaywan, Louisiana Department of Transportation and Development (LADOTD).
References [1] AASHTO LRFD. Bridge design specifications. 4th ed. Washington, DC, USA; 2014. [2] Abendroth RE, Greimann LF, Lim KH, Thomas ME, Sayers BH, Kirkpatrick CL, et al. Field testing of integral abutments (No. HR-399); 2005. [3] Arockiasamy M, Butrieng N, Sivakumar M. State-of-the-art of integral abutment bridges: design and practice. J Bridge Eng 2004;9(5):497–506. [4] Barr P, Halling MW, Huffaker C, Boyle H. Behavior and analysis of an integral abutment bridge (No. UT-13.12); 2013.
B. Kong et al. / Engineering Structures 104 (2015) 18–31 [5] Breña SF, Bonczar CH, Civjan SA, DeJong JT, Crovo DS. Evaluation of seasonal and yearly behavior of an integral abutment bridge. J Bridge Eng 2007;12 (3):296–305. [6] Civjan SA, Kalayci E, Quinn BH, Breña SF, Allen CA. Observed integral abutment bridge substructure response. Eng Struct 2013;56:1177–91. [7] Davids WG, Sandford T, Ashley S, DeLano J, Lyons C. Field-measured response of an integral abutment bridge with short steel H-piles. J Bridge Eng 2009;15 (1):32–43. [8] Dunker KF, Liu D. Foundations for integral abutments. Pract Period Struct Des Constr 2007;12(1):22–30. [9] Elbadry MM, Ghali A. Temperature variations in concrete bridges. J Struct Eng 1983;109(10):2355–74. [10] Frosch RJ, Chovichien V, Durbin K, Fedroff D. Jointless and smoother bridges: behavior and design of piles (No. FHWA/IN/JTRP-2004/24); 2006. [11] Hassiotis S, Khodair Y, Roman E, Dehne Y. Evaluation of integral abutments (No. FHWA-NJ-2005-025); 2006. [12] Hoppe EJ, Bagnall TM. Performance of a skewed semi-integral bridge: volume I: field monitoring (No. FHWA/VTRC 08-R20); 2008. [13] Huang J, French CE, Shield CK. Behavior of concrete integral abutment bridges (No. MN/RC-2004-43); 2004. [14] Kim W, Laman JA. Numerical analysis method for long-term behavior of integral abutment bridges. Eng Struct 2010;32(8):2247–57. [15] Laman JA, Pugasap K, Kim W. Field monitoring of integral abutment bridges (No. FHWA-PA-2006-006-510401-01); 2006.
31
[16] Maruri RF, Petro SH. Integral abutments and jointless bridges (IAJB) 2004 survey summary. In: Integral abutment and jointless bridges (IAJB 2005); 2005. p. 12–29. [17] MATLAB (R2010b). The MathWorks Inc. [18] Mistry VC. Integral abutment and jointless bridges. In: Integral abutment and jointless bridges (IAJB 2005); 2005. p. 3–11. [19] Ooi PS, Lin X, Hamada HS. Field behavior of an integral abutment bridge supported on drilled shafts. J Bridge Eng 2009;15(1):4–18. [20] Pétursson H, Collin P, Veljkovic M, Andersson J. Monitoring of a Swedish integral abutment bridge. Struct Eng Int 2011;21(2):175–80. [21] Thippeswamy HK, GangaRao HV, Franco JM. Performance evaluation of jointless bridges. J Bridge Eng 2002;7(5):276–89. [22] Seo S. A review and comparison of methods for detecting outliers in univariate data sets (doctoral dissertation, University of Pittsburgh); 2006. [23] Voyiadjis GZ, Cai CS, Alshibli K, Faghihi D, Kong B, Yang Yi. Integral abutment bridge for Louisiana’s Soft and Stiff Soils. Report. Department of Civil and Environmental Engineering, Louisiana State University, LA; 2014. [24] Weakley K. VDOT integral bridge design guidelines. In: Integral abutment and jointless bridges (IAJB 2005); 2005. [25] White H, Pétursson H, Collin P. Integral abutment bridges: the European way. Pract Period Struct Des Constr 2010;15(3):201–8. [26] William GW, Shoukry SN, Riad MY. Study of thermal stresses in skewed integral abutment steel girder bridges. Struct Eng Int 2012;22(3):308–17.
Contents lists available at ScienceDirect
Engineering Structures journal homepage: www.elsevier.com/locate/engstruct
Field monitoring study of an integral abutment bridge supported by prestressed precast concrete piles on soft soils B. Kong, C.S. Cai ⇑, X. Kong Dept. of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, LA 70803, United States
a r t i c l e
i n f o
Article history: Received 10 November 2014 Revised 19 August 2015 Accepted 6 September 2015
Keywords: Integral bridge Field monitoring Thermal performance Slab strain Backfill pressure
a b s t r a c t Integral abutment bridges (IABs) have been constructed during the past several decades around the world. The purpose was to eliminate expansion joints and minimize joints induced problems. Even though IABs have been widely accepted due to the satisfying performances, yet they have not been largely applied in practice. Some of the reasons may be attributed to the uncertainties of these bridges under different loading conditions, especially the daily and yearly varying temperature effects. In this paper, the behavior of the first IAB constructed on the soft soil condition in Louisiana is discussed. A field monitoring program is introduced and the measured results from 08/11/2011 to 03/15/2014 are presented. The field monitoring program leads to the following observations, (1) significant seasonal and daily temperature variations are observed on the bridge slabs but within the AASHTO temperature design specification; (2) the displacements and rotations of bridge components are well correlated with the temperature variations; (3) the thermal stresses generated in the slabs may exceed the allowable material cracking capacity; (4) the soil behavior behind the abutments is complicated and long term monitoring program is needed; (5) the integral abutment primarily behaves in translation rather than rotation; and (6) the pile inflection (zero bending) point is observed and the strong and weak axels bending are all important due to the bridge skewness. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Expansion and contraction are two basic responses of bridges induced by temperature variations. Traditionally, a relief system, consisting of expansion joints, bearing supports, or other devices, are designed to accommodate these movements. After years of services, however, this system, especially the expansion joint, may become one of the vulnerable elements affecting the sustainability of bridges [18,21]. Bridge engineers have been trying to eliminate expansion joints whenever possible; and the concept of integral bridges without joints was inspired. Generally speaking, integral abutment bridges (IABs) can be categorized into three types [24]: (1) a full IAB, as discussed in this paper, refers to a single or multi-span bridge as shown in Fig. 1. The superstructure of the bridge, i.e., concrete slabs, prestressed concrete beams, steel girders, and approach slabs, is casted monolithically with a stub type abutment and supported on a row of piles; (2) a semi-integral bridge is similar to the full IAB, but the abutment is not rigidly connected to the substructure; and (3) a deck-extension bridge extends its deck slab over the ⇑ Corresponding author. E-mail address: [email protected] (C.S. Cai). http://dx.doi.org/10.1016/j.engstruct.2015.09.004 0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
abutment into the approach pavement, but the main beams or girders are not fixed on the abutment. In an integral configuration, expansion joints are eliminated and the corresponding jointrelated issues can be minimized. However, a new challenge arises since the horizontal movements from the superstructure are transferred to the substructure, which have to be well accommodated through the soil–structure interaction or other special designed mechanisms. The benefits of IABs have been widely accepted in the past several decades around the world. Survey has been conducted on the construction experience of IABs in American [16] and European countries [4,25], where many similarities were observed even though significant differences existed in some aspects. Field monitoring study methods have been adopted to investigate and justify the design and construction concepts, including (a) the maximum allowable design criteria (e.g., total and individual bridge’s span lengths and skews); (b) the structure design parameters (e.g., orientations of the pile, abutment, and wingwall); (c) the soil–structural interaction behaviors (e.g., between the soil–pile, abutment-backfill, and approach slab-backfill); (d) the joint connection effects (e.g., at the interfacial locations between the abutment-deck-girder, abutment-pile cap, approach slababutment, and intermediate pier-girder); (e) the stress relief
B. Kong et al. / Engineering Structures 104 (2015) 18–31
19
Fig. 1. Schematic view of a full integral abutment bridge (IAB).
mechanisms (e.g., diameters, depths, and filling materials of the pre-sized holes surrounding the piles, and the compacting degree of the backfill materials behind the abutments); and (g) the long term effects (e.g., the temperature, shrinkage, creep, and steel relaxation) [3,13,2,10,11,15,5,8,12,19,7,25,14,20,26,4,6]. In the state of Louisiana, no full IABs have ever been built before the year of 2011. The reasons are partly due to the lack of references from the previous studies regarding on the behaviors of bridges on the unique soil conditions in Louisiana. To this end, the present paper reports the field monitoring program from 08/11/2011 to 03/15/2014 for the first full IAB, the Caminada Bay Bridge, designed by the Louisiana State Department of Transportation and Development (LADOTD). This bridge also has some special features that have not been focused in the previous investigations on integral bridges, such as the long continuous slab spans, deep precast prestressed concrete piles, shallow integral abutment, and very soft soil conditions. Based on the monitoring results, the temperature induced effects on the superstructure and substructure of such a bridge are demonstrated. Specifically, (1) significant seasonal and daily temperature variations are observed on the bridge slabs but within the AASHTO temperature design specification; (2) the displacements and rotations of bridge components are well correlated with the temperature variations; (3) the thermal stresses generated in the slabs may exceed the allowable material cracking capacity; (4) the soil behavior behind abutments is complicated and long term monitoring program is needed; (5) the integral abutment primarily behaves in translation rather than rotation; and (6) the pile inflection (zero bending) point is observed and the strong and weak axels bending are all important due to the bridge skewness.
2. Bridge descriptions The Caminada Bay Bridge is located at the Grand Isle, LA (29°150 4800 N, 89°570 2400 W), about 160 km to the south of New Orleans, LA. While the total length of the bridge is 1202 m (3945 ft), the monitoring program is conducted on the first eleven spans, as shown in Fig. 2, including a 3 m (10 ft) sleeper slab, a 12 m (40 ft) approach slab, a 91 m (300 ft) continuous concrete slab, and also the abutment, bent, pile, and soil. The width of the bridge is 15 m (50 ft) consisting of two 6.4 m (21 ft) lanes and a 2 m (7 ft) sidewalk on the northern side. For the parts that are monitored, the slabs are fully integrated with the first bent (Bent1) at the left end, simply supported on the eleventh bent (Bent11) at the right end, and rigidly connected with all the interior bents in between, where the expansion and fixed connection joints are designated as ‘‘E” and ‘‘F” in Fig. 2. Each bent is rigidly supported on a single row of four prestressed precast concrete (PPC) piles. The soil types, referred to the boring log information near Bent1, can be approximately subdivided into two layers, including a medium
sandy soil layer from the ground to the depth of 18.9 m (62 ft), and a medium clay layer through the rest of the depth. The materials and the corresponding properties for this Caminada Bay IAB are listed in Table 1. 3. Instrumentations Bridge Diagnostics, Inc. (BDI) was contracted by the Louisiana Transportation Research Center (LTRC) and Louisiana State University (LSU) to install the bridge monitoring system. In this project, a total of 81 instruments were applied on the bridge, as listed in Table 2, including the vibrating wire strain gages, vibrating wire tiltmeters, vibrating borehole wire extensometers, vibrating wire pressure cells, piezometers, and vibrating wire thermistors. The large application of vibrating wire gages is due to their good performances, without drifts, for the long-term monitoring. Meanwhile, each sensor is provided with an extra temperature thermistor so that the temperature information of the bridge elements can be simultaneously obtained. 3.1. Superstructure instrumentation For the superstructure, a total of 22 sensors, with 14 embedded sisterbars and 8 surface strain gages, as shown in Fig. 3, were applied on the 46 cm (18 in.) depth concrete deck. They were adopted to measure the positive and negative strains due to the temperature changes. Specifically, the embedded sisterbars were placed at the rebar locations before the pouring of concrete with 8 cm (3 in.) above the bottom surfaces on the approach slab, Span1, Span3, and Span5, and with 5 cm (2 in.) below the top surfaces on the Bent1, Bent2, and Bent5. The surface strain gages, otherwise, were mounted under the bottom surfaces from Span3 to Span6 after the completion of the concrete pouring. 3.2. Substructure instrumentation For the substructure, a total of 59 instrumentations were installed as shown in Fig. 4. The monitoring program related to sensors and instrumentations in the abutment and the foundation of the bridge was conducted by the research groups of Drs. G.Z. Voyiadjis and K.A. Alshibli at LSU. Some of the important information is briefly described here for the convenience of readers, and the detailed information can be referred to the report by Voyiadjis et al. [23]. For example, (a) 32 sisterbars were installed at the four corners of two 24 m (80 ft) long PPC piles at the easternmost of Bent1 to measure the pile strains. The distances from the sisterbars at each sensor sections to the bottom surface of Bent1 were 1.2 m (4 ft), 3.7 m (12 ft), 6.1 m (20 ft), and 8.5 m (28 ft), respectively; (b) 2 tiltmeters were attached at the middle section of the 1.2 m (4 ft) high Bent1 and Bent11 to record the bents’ rotations; (c) two rows
20
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 2. Plan and elevation view of the first eleven spans of Caminada Bay IAB.
Table 1 Material properties of Caminada Bay Bridge. Structure items
Material types
Material properties
Slabs, bents
Class AA (M) concrete Class P (M) high performance concrete
Strength of 28 MPa (4000 psi)
Reinforcing rebar in the bents and slabs
Type 316LN stainless steel
Reinforcing rebar in other structures Prestressing strands
Grade 60 steel
Minimum compressive strength of 41 MPa (6000 psi) at 28 days, and average compressive strength of 69 MPa (10,000 psi) at 56 days Elastic modulus of 200 GPa (30,000 ksi), tensile strength of 515 MPa (75 ksi), and yield strength of 205 MPa (30 ksi) Yield strength of 414 MPa (60 ksi)
Grade 270 steel
Yield strength of 1860 MPa (270 ksi)
PPC piles
of 9 soil pressure gages, with one row placed about 30 cm (1 ft) below the slab’s bottom surfaces and the other at the bent’s middle height, were mounted on the backwall of Bent1 to measure the soil pressures on the abutment. The pressure gages in the transverse direction from the easternmost end of Bent1 were 1.4 m (56 in.), 3.5 m (138 in.), 5.6 m (221 in.), 7.7 m (303 in.), and 11.9 m (468 in.), respectively; (d) 4 soil strain gages, with 2 connected to Bent1, and another 2 floating with about 1.2 m (4 ft) and 2.1 m (7 ft) distance away from Bent1, respectively, were applied to investigate the soil’s deformation or Bent1’s movement; and (e) several other gages were embedded in Bent1 and through the surrounding soil depths to obtain the corresponding structure temperatures and pore water pressures.
3.3. Data acquisition system
Table 2 Instrumentations applied on Caminada Bay Bridge.
The data acquisition system, as shown in Fig. 5, consists of a CR1000 datalogger, an AVW200 interface, an AM16-32 multiplexer, a wireless cell modem, and a solar battery. In addition, it has been outfitted with a wireless communication link so that the operation and data collection can be handled remotely from a computer through a LoggerNet program at LSU. It is worthy to note that the acquisition system has been equipped with a battery back-up mechanism. It allows continuous data collection throughout a typical power outage and alert emails will be sent to researchers when the system is working on battery.
Structure
Gages
Location
Numbers
Superstructure
Sisterbar
Approach slab bottom (embedded) Span 1, 3, & 5 bottom (embedded) Bent 1, 2, & 5 top (embedded) Span 3, 4, 5, & 6 bottom (surface)
2
Two easternmost piles at Bent1 Bent1 and Bent11 (surfaces) Bent1 (embedded) Bent1 back face Backfill behind Bent1
16 2 = 32
3.4. Data post-processing
12=2 4 9 4
Backfill behind Bent1 Backfill behind Bent1
2 6
The data acquisition was conducted from 08/11/2011 to 03/15/2014 in which the raw data of each sensor were recorded every 3 min and 20 s. Based on the assessment of the collected data, most of the instruments functioned well. Some of the failed sensors include 3 surface strain gages on the slabs, 4 sisterbars in the piles, 2 pressure gages on the abutment, and 1 soil strain meter.
Strain gage Substructure
Sisterbar Tiltmeter Thermistor Pressure cell Soil strain meter Thermistor Piezometers
23=6 23=6 24=8
B. Kong et al. / Engineering Structures 104 (2015) 18–31
21
Fig. 3. Plan and elevation views of sensors on the approach slab and the first six-span slabs.
However, these missed data does not affect the investigations since sensors are redundantly applied for most of the concerning areas. Two post-processing steps were conducted on the raw data before further analysis. First, the outlier data have to be picked out and removed. Otherwise, they will affect the accuracy of the calculated hourly or daily average temperature values, especially when one or several data within one hour is mistakenly too large or small. Secondly, complying with the requirements of the sensor suppliers, a proper temperature correction process should be performed for certain sensor types. It is due to the differences of the thermal expansion coefficients between the steel wire in the sensor gages and the thermal coefficients of the measured concrete elements, e.g., the strain gages in concrete slabs and PPC piles. Some basic algorithms have been proposed for the data outlier removal in the statistic field, such as the Standard Deviation Method, Z-score, Modified Z-score, Tukey’s, Adjusted Boxplot, MAD and Median Rule [22]. While each method has its own advantages and restrictions, most methods are based on the assumption that the data is normally distributed. For example, as shown in Fig. 6, about 68%, 95%, and 99.7% of the data fall in the area within 1, 2, and 3 times the standard deviation from the mean value if the data follows a normal distribution. In this sense, if a data falls with a distance, e.g., 2 or 3 times of the standard deviation away from the mean value, it can be considered as an outlier. In this paper, the raw data from a representative strain gage, i.e., located in the concrete slab of Span5 as shown in Fig. 3, is taken out as an example to discuss the outlier removal algorithm adopted in this study. Four sets of sample data within one hour, i.e., 08/11/2011 17:00, 09/11/2011 24:00, 10/11/2011 03:00, and 11/11/2011 09:00, are randomly selected from the raw readings, and the normality test is conducted on these data. Fig. 7 shows the representative normality test plots of the strain and temperature readings at 08/11/2011 17:00, respectively. The other results are calculated and listed in Table 3. Based on the statistic theory, the distribution is non-normal if the P-value is smaller than 0.01 or the A-squared value is larger than the critical values, i.e., 0.787 and 1.072 for 95% and 99% confidence level, respectively. It can be observed based on the calculations that while most strain data within an hour period can be considered as normally distributed, the temperature data are not always normally distributed. For example, the temperature data in the case of 08/11/2011 17:00 is more uniformly distributed within an hour. In addition, it can be reasonably predicted that for sensors embedded in the deep soil, e.g., strain gages in the piles under the ground, the temperatures may be more constantly varying throughout years. Therefore, through a comprehensive comparison between different outlier removal methods, together with the sensitivity
analysis, the SD and Tukey’s method are tentatively adopted and coded in MATLAB [17]. The SD method using the mean and standard deviation may not be appropriate for some cases since the extreme values may affect the standard deviation; however the Tukey’s method is robust, which is less sensitive to the extreme values. Therefore, the outlier removal is conducted on the measured data within every 1-h time frame in three steps: (1) performing normality test, (2) using the SD method on the normal distributed data, and (3) using the Tukey’s method on the non-normal distributed data. It is worthy to note that most of the outliers in the present measured data are extreme values, thus the simple visual removal method using the filter function in Excel was adopted as well. Detailed information regarding on these methods is not elaborated in details, since they are the basic theory and can be referred to the statistical textbook in the public domain [22]. The discussions in the following sections are based on the data after outlier removals and temperature corrections. 4. Field monitoring results The monitored results reported in this study were from 08/11/2011 to 03/15/2014. The initial reference condition for all the instrumentations was set up on 08/11/2011, i.e., all values were set as zeroes. Therefore, the results discussed below are the changes of bridge thermal responses corresponding to the bridge behaviors on 08/11/2011. It should be noted that some data are missed during the period from 03/10/2013 to 7/31/2013. However, it does not affect the discussions in the following sections, since the bridge responses show a yearly returning pattern, and the extreme values and distribution trends are still obtained. The following discussion is aiming to qualitatively describe the thermal response of the IAB under temperature effects. All the results are based on the assumption that the bridge response is primarily correlated to the temperature variation. The Pearson’s correlation coefficient is firstly calculated which is a measure of the linear correlation of dependence between two variables. Values between +1 and 1 are provided in the following tables in which +1 is for a total positive correlation, 0 is for no correlation, and 1 is for a total negative correlation. It should be noted that results in the tables only serve to provide a general overview of the correlation degree, and more detailed discussion and measurement locations are elaborated in the following sections. Table 4 shows the temperature correlation between the ambient air and bridge components, where the correlation degree is high at the bridge slab and decreases through the bridge depth to the pile bottom. Table 5 shows the correlation between the temperature change and bridge response. For most of the cases, the correlation degree is high,
22
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 4. Instrumentations applied on substructure Bent1, (a) plan view, (b) side view, (c) elevation view [23].
especially for the measurements on the rotation and displacement. However, the strain results show comparatively low correlation degree, especially at the bottom of slab surface, and that may be attributed to the environmental effects. Based on the Pearson’s correlation method, the correlation assumption between the temperature variations and bridge responses is generally valid in the following discussion. 4.1. Environmental conditions During the monitoring period, no instrumentations were particularly installed to measure the environmental conditions,
e.g., ambient temperatures and wind speeds. However, this information is of great importance especially for the thermal behaviors of IABs. It is because the uncertainties of IABs are mainly caused by the temperature changes within the bridges, and those variations in turn are determined by the environmental conditions. In this sense, the ambient temperatures and wind speeds, recorded at the bridge site from a nearest weather station located at the Grand Isle, LA (http://tidesandcurrents.noaa.gov), are firstly referred and discussed. The measured hourly-varying air temperatures and wind speeds at the weather station are shown in Fig. 8. Firstly, the periodical and cycling varying trends can be clearly observed.
23
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 5. Acquisition system applied on the pole next to the bridge.
Specifically, the sinusoidal fluctuating patterns are approximately shown in the yearly and hourly varying temperatures. This observation will be helpful if the temperature data is not available. For example, some of the missing temperature data are predicted using the curve fitting method in the following section. Secondly, the minimum and maximum ambient temperatures during the monitoring period, as marked in Fig. 8, are approximately 0 °C (32 °F) and 35 °C (95 °F), respectively, with a difference of 35 °C (63 °F). Finally, the number of the freezing days during the three monitored years is less than 14. According to the AASHTO LRFD [1] specification, this bridge can be considered being located in a moderate climate region; and the corresponding bridge design
Fig. 6. Normal distribution curve.
Normality Plot
Histogram 3 5
2
4
1
Z Score
Number
6
3 2
y = 4.0452x - 9820.8 R² = 0.986
0
1
-1
0
-2 -3 2,427.2
Strain Reading (Digital)
2,427.6
2,428.0
2,428.4
Strain Reading (Digital)
(a) Histogram
3
20
2
15
1
Z Score
Number
25
10 5 0
Normality Plot
0
y = -0.0039x R² = -8E-17
-1 -2
88.9
89
89.1
89.2
-3 88.9
Temperature Data (ºF)
Temperature Data (ºF)
ºC=(ºF-32º)x5/9
(b)
ºC=(ºF-32º)x5/9
Fig. 7. Normality test data measured at Span5 on 08/11/2011 17:00: (a) strain, (b) temperature.
88.9
24
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Table 3 Normality test for the strain and temperature readings of the strain gage at Span5. Date
08/11/2011 09/11/2011 10/11/2011 11/11/2011
Strain reading
17:00 24:00 03:00 09:00
Temperature reading
A-squared
P value
A-squared
P value
0.122 0.231 0.213 0.437
0.984 0.77 0.827 0.264
17.627 0.714 0.346 0.638
0 0.051 0.442 0.081
temperature specification for this region is compared and correlated with the field measurements in the following discussion. The variations of the wind speeds are also shown in Fig. 8. The wind speeds and air temperatures generally do not show obvious correlations. The occurrences of the peak wind speeds, however, are almost in coincidence with the conditions when the temperatures are at sudden drops. In addition, the measured wind speed values are generally wandering around 5 m/s, with some days larger than 10 m/s but smaller than 17 m/s. According to Elbadry and Ghali [9], the wind speeds greatly determine the thermal convection coefficients, and those coefficients in turn affect the heat transfer mechanisms at the surfaces of the bridge slabs. The effects of the present wind speed on the temperature variations at the bridge’s surfaces will be discussed in the next section based on the field measurements. 4.2. Bridge temperatures The thermal performances of bridges are directly determined by the variations of bridge temperatures. These temperatures are not uniformly but varyingly distributed within different structural components throughout the bridge. In this section, with the help of the largely applied instrumentations, where each gage is attached with a thermistor, the measured temperatures in terms of the seasonal variations, daily gradients, and distributions from the slab’s top surface to the pile’s bottom section are observed and discussed. Fig. 9 shows the measured hourly-varying temperatures at the bent’s top surface, approach slab’s bottom surface, and deck slab’s bottom surface, respectively. The measured ambient temperature is also plotted in the figure for a reference. It can be observed that the top surface of the bent or slab has significant temperature variations, especially during the summer seasons; while during the winter season, these high variations decrease and the temperatures at the top and bottom surfaces are close to each other. This temperature distributing phenomenon can be explained by the heat transfer mechanisms. The temperatures at the top surfaces are primarily affected by the air temperatures, solar radiations, and surface convection behaviors. For example, the high temperatures at the top surfaces often appear in the hot time of the day and with no or very slight winds. For the hourly temperature fluctuation, they are attributed to the change of wind speeds. In the present bridge, temperature fluctuations perform actively when the wind speeds are larger than 5 m/s based on the field measurements
shown in Fig. 8. Moreover, after the top surface is heated by the solar, the high energy transfers and conducts through the slab’s depth to the bottom surface. When arriving at the bottom surface, the energy is either blown away by the adjacent air through convection or transferred to the soil underneath through conduction. It can be observed in Fig. 9 that the temperatures at the approach slabs are in between to that of the bent’s top and slab’s bottom surface since the approach slab contacts with the soil where the thermal convection or conduction may be less significant than the ambient air or concrete. For the seasonally and uniformly varying slab’s temperatures during the monitored period, as marked in Fig. 9, the maximum and minimum values are approximately 35 °C (95 °F) and 0 °C (32 °F) which is calculated as the average temperatures at the bridge’s top and bottom surfaces, respectively. Therefore, for the present IAB, if assuming the base construction temperature equivalent to the measured minimum air temperature of 0 °C (32 °F), the effective temperature difference that will induce seasonal movements is 35 °C (63 °F). This value is within the temperature design values of 38 °C (70 °F) according to the AASHTO LRFD [1] specification. Besides the uniform temperature variations, the gradient distributions are also important in bridge designs. For temperature distributions along and perpendicular to the traffic directions, the temperature differences are reasonably negligible based on the field measurements; thus they are not demonstrated here. For the vertical gradients through the slab depths, however, the differences are apparent. Fig. 10 shows the temperatures measured at the top and bottom surfaces of the slab and the ambient air during the hottest week, i.e., 08/19/2011 to 08/26/2011, and the coldest week, i.e., 12/29/2011 to 01/06/2012, respectively. During the hottest week, the temperatures at the top surfaces and air show a high correlation with almost the same variation trends. In addition, a certain lagging effect, though not significant, is observed between the temperatures at the top and bottom surfaces due to the thermal inertia for the present 46 cm (18 in.) depth slab. During the coldest week, however, the temperature differences are not significant and both the temperatures at the top and bottom surfaces are close to each other. According to the AASHTO LRFD [1] specifications, the positive and negative thermal gradients are specified based on four subdivided radiation zones, together with the consideration of the superstructure’s materials, geometries, and overlays types. Then, if following the code guidelines, the positive and negative temperature difference at the top and bottom surfaces of the current slab are supposed to be 10.2 °C (18.5 °F) and 3.1 °C (5.6 °F), respectively. Therefore, the measured positive and negative temperature differences, i.e., 10 °C (18 °F) on 08/20/11 17:00 and 1.11 °C (2 °F) on 1/6/12 07:00, respectively, are within those values calculated from the design specification.
Table 5 Pearson’s correlation coefficient between bridge temperature and structure response. Slab average temperature
Table 4 Pearson’s correlation coefficient between ambient air and bridge temperature. Ambient air temperature Slab average temperature Bent top surface temperature Slab bottom surface temperature Pile B-B section temperature Pile C-C section temperature Pile D-D section temperature Pile E-E section temperature
0.963 0.863 0.797 0.598 0.028 0.683 0.578
Bent top surface strain Slab bottom surface strain Bent11 rotation Bent1 rotation Soil displacement at B location Soil displacement at A location Bent bottom surface soil pressure Bent top surface soil pressure Exterior pile B-B section strain (4) Exterior pile B-B section strain (27) Exterior pile B-B section strain (35) Exterior pile B-B section strain (44)
0.629 0.010 0.958 0.893 0.871 0.841 0.841 0.386 0.884 0.709 0.016 0.681
B. Kong et al. / Engineering Structures 104 (2015) 18–31
25
4.3. Slab strains
Fig. 8. Measured environmental conditions at the weather station.
Fig. 11 shows all the temperature variations from the top surfaces of the slabs to the bottom parts of the piles, and some of the observations are described as follows. First, the temperatures generated by the solar radiations can be transferred through about 1.7 m height deck and abutment to the top part of the pile. However, the temperatures show negligible variations along the piles to the bottom section. Second, the temperature variations in the soils, e.g., curve No. 7, are more constant with the soil going deeper, even though they may be slightly influenced by the environments near the ground. For example, the temperatures in soils could be higher than those in the slabs and bents during the winter seasons, such as curve No.4. Fig. 12 shows the best curve fitting relationships between the measured bridge and air temperatures, where the former one is considered as the average results between the readings from the sensors at the bent’s top and slab’s bottom surfaces. It can be observed that the bridge and air temperatures are almost linearly correlated in the middle temperature range, while the polynomial curve shows better results in the lower and higher temperature ranges. This conclusion justifies one of the statements discussed above that it is able to predict the bridge’s uniformly varying temperatures using the air temperatures.
Fig. 9. Measured hourly-varying temperatures of bridge slabs.
For reinforced concrete structures, the strains in the steel rebar and surrounding concrete should be the same prior to the cracking of the concrete due to the strain compatibility. Generally speaking, the concrete is weak in tension, and the corresponding strength is about 10% of its compressive strength. When in compression, however, the concrete can sustain up to about 300 microstrains before failures for general concrete materials adopted in bridge engineering. Fig. 13 shows one representative readings of a strain gage at the Bent5 top rebar location. The highly-varying strain variations can be observed, and the long term seasonal trends show a negative correlation with the temperature changes. In addition, the calculated thermal stresses change during the temperature decrease of 35 °C (63 °F), if using the maximum measurement of 100 microstrain, would induce a stress change of 2.49 MPa (360 psi). This value is about 9% of the concrete compressive strength. In this sense, the top surface of the bent has a possibility of cracking due to the temperature variations alone since the slab–bent–pile connection joint is integrally constructed. Similarly, another reading of the representative gage embedded in the bottom parts of the Span5 slab is shown in Fig. 13. The induced compressive thermal stress variation, if using the 50 microstrains, is about 4% of the compression strength during the monitored periods.
Fig. 10. Bridge and air temperatures: (a) hottest week from 08/19/2011 to 08/26/ 2011, (b) coldest week from 12/29/2011 to 01/06/2012.
26
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 11. Measured bridge temperatures through the depth of bridge.
4.4. Abutment behaviors The displacements and rotations of the abutments are measured using the soil strain meters and tiltmeters [23], where the positive sign convention is defined for bents moving or rotating inwards, i.e., toward the water between the bridge two ends. Fig. 14 shows the soil deformations at two locations, i.e., right behind the Bent1 backwall surface and 2.1 m away from Bent1, designated as locations A and B in the plot, respectively. The displacements at these two locations are observed showing good correlations with the bridge temperature variations, where the yearly variations of the soil deformations are approximately 5.5 mm at location A and 2 mm at location B, respectively. The relationships between the variations of the bridge temperatures and soil deformations at these two locations A and B are plotted in Fig. 15. Specifically, three temperature decreasing periods are selected in three consecutive years, which are from 08/11/2011 to 01/14/2012 (1st year), 08/16/2012 to 01/14/2013 (2nd year), and 10/6/2013 to 01/29/2014 (3rd year), respectively. The soil deformations are plotted with respect to the temperature variations, and the relationships between them are linearly fitted and shown in the figure. First, it can be observed that, for the soils at location A right behind abutment, the slops of the fitting curves in the 2nd and 3rd years are larger than that in the 1st year. The
Fig. 12. Best fitting of the air temperatures and the bridge slab average temperatures.
Fig. 13. Measured strains and temperatures: (a) top surface of Bent5, (b) bottom surface of Span5.
reason behind this scenario may be that the soils are pushed away and disturbed due to the slab’s thermal displacements during the first year, so that the soils become soft and are more readily to be pushed away by the thermal deformations in the following 2nd and 3rd years. Secondly, the slop of the 3rd year curve is smaller than that of the 2nd year one which may be attributed to the soil settlement behaviors around gap or the soil property change due to the repeated compaction. During the winter seasons, the bridge stops expanding and begin to contract; thus a gap is generated behind the abutment since soils are not elastic materials
Fig. 14. Measured bridge temperatures and soil deformations.
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 15. Linear fitting of the bridge temperature vs. soil deformation changes: (a) location A, (b) location B.
so that they cannot be fully returned to their initial positions. Under this circumstance, the soils at above or near around will fill the gap. As time goes, the soils become more compacted, which impedes the bent displacements in the years after. Thirdly, for the soils at location B, the slops of the fitting curves in the three years almost do not change. It may be implied that the bridge thermal deformations have negligible effects on the soils at 2.1 m away from the abutments under the current bridge length and temperature variations. Besides the displacement behaviors, the rotations of the bents are measured on the Bent1 and Bent11, as shown in Fig. 16. Similar to the displacement behaviors, rotations are also negatively correlated with the bridge temperature variations. For the Bent1, with the temperature changing of 35 °C (63 °F) during the monitored period, the rotation of the abutment varies about 0.028°. This small value is due to the integral structural configuration and soil– structure interaction effects. For the Bent11, however, the rotations are almost perfectly correlated with the temperature variations, with the values approximately 10 times larger than that at Bent1; and that is attributed to the simply supported, not integral, connection detail at the slab–abutment. In addition, compared with the displacement of Bent1, the variations of the rotations of Bent1 is insignificant and only contributes about 10% of the total displacements at Bent1. In this sense, Bent1 should primarily move in translations rather than rotations. Among all the recorded data, the measured backfill pressures on Bent1 show the most significant randomness. There seems no clear and regular trends for all the pressure sensor readings; and that may be due to the uneven soil properties and compaction levels especially during the first year. However, the backfill behaviors
27
Fig. 16. Measured bridge temperatures and rotations: (a) Bent1, (b) Bent11.
become clear when the soils are stable in the long run. Fig. 17 shows the representative top and bottom pressure readings at the middle location of Bent1. The pressures have negative correlations with the soil displacements, and the variation of the soil pressure through the monitored years is about 20 kN/m2 when it becomes stable. In addition, the top and bottom pressure cells show the same sign, which justifies one of the statements above that the abutment moves primarily in translation instead of rotations. It should be noted that the observations and conclusions obtained on the behaviors of integral abutment in this section,
Fig. 17. Measured backfill pressures at the middle location of Bent1.
28
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 18. Plan view of the Bent1 and pile diagram.
Fig. 19. Measured pile strains at cross sections from B-B to E-E.
including the deformation, rotation, and earth pressure, need to be further justified with the long term monitoring results. 4.5. Pile strains Strains at the four corners of the exterior pile (EP) and interior pile (IP), as shown in Fig. 18 [23], are measured through the pile’s depths at four elevations from the top to bottom sections. For the convenience of discussion, these four sections from the top to bottom are designated as B-B, C-C, D-D, and E-E as shown in Fig. 4, respectively, where the positive x and y axes are defined as perpendicular and parallel to the traffic directions as shown in Fig. 18. All the measured strains at the four pile sections are shown in Fig. 19, where the numbers at the four corners are the sensor names. It can be observed that the strains measured at the top parts of the piles, e.g., B-B and C-C sections, are larger than those at the bottom, e.g., D-D and E-E sections. For example, the measured maximum positive and negative strain variations, appearing at the C-C section, are 40 and 34 microstrains, respectively, though they are only less than 3% of the concrete compressive strength of the piles. In addition, the bridge contracts inward were corresponding to the temperature variations, such as during the temperature decrease period from 06/11/2011 to 12/11/2011. Under this condition, the outer surfaces of the piles should have been extended with tensile behaviors. This performance has been justified from the measured strains at sections C-C to E-E, where the pile is bending with respect to the negative x-axis. At the B-B
section, however, the pile is differently bending with respect to the positive y-axis, which may be due to the bridge skew effects or the rigid connections between the pile head and bent, where tensile steel rebars are constructed by extending from the pile heads to the bents. All the measured strains are further calculated and decomposed into four strain components, i.e., the axial strain, x-axis bending, y-axis bending, and torsional strains, according to the basic knowledge of the materials mechanics. Fig. 20 shows the strains at two representative sections of both the EP and IP, i.e., sections B-B and D-D, and some of the behaviors are observed as: (a) for the axial strains, they show good correlations with the temperature variations at both sections of the two piles. For example, with the increase of temperatures, from December to August, the bridge expanded and the axial strains were accordingly increased. For the daily varying trends, the two piles show opposite axial strain signs. This scenario may be due to the different abutment thermal behaviors in the transverse direction, where the decrease of temperatures induced sagging movements at the middle parts of the bents but hogging ones at the two far ends; (b) for the x-axis bending strains, they are directly affected by the longitudinal thermal movements of the bridge; (c) for the y-axis bending strains, they reflect the behavior of the bridge bending transversely with respect to the longitudinal y-axis of the bridge. It can be found that the y-axis bending strains at the top B-B section are larger than that of the x-axis strains, and that may also be attributed to the skew effects or the rigid pile-bent connections of the bridge; (d) for
B. Kong et al. / Engineering Structures 104 (2015) 18–31
29
Fig. 20. Calculated strain components: (a) B-B cross section, (b) D-D cross section.
the torsional strain, the measured values are reasonably as small as expected and can be ignored. Fig. 21 shows the thermal strains along the depth of the EP at three representative days, i.e., 08/14/2011, 11/02/2011, and 02/13/2012. They refer to the cases with the temperature increase of 5 °C (10 °F), 17 °C (30 °F), and 28 °C (50 °F), respectively. Through comparisons, the results in the axial strain and torsional strain plots do not show much difference with the change of temperatures. However, the x-axis bending strain, induced from the bridge longitudinal deformations, is most sensitive to the temperature variations, and the y-axis bending strain cannot be ignored, especially at the top parts of the pile, due to the bridge skew and fixed connection effects. In addition, in the x-bending strain and
y-bending strain plots shown in Fig. 21, zero bending points (or inflection points) appear in both cases. 5. Conclusion This paper reports the monitoring results for the first IAB, Caminada Bay Bridge, on the unique soft soil conditions in Louisiana from 08/11/2011 to 03/15/2014. Based on the assessment on the measured results, about 10 out of 81 gages were failed. Some of the sensors show outlier data which can be either filtered by Excel or the algorism proposed in the paper. The Pearson’s correlation coefficient is calculated and it demonstrated that there is a certain correlation degree between the temperature variations and bridge
30
B. Kong et al. / Engineering Structures 104 (2015) 18–31
Fig. 21. Measured exterior pile strains under three temperature variations.
responses in general. Some of the most important observations are concluded as follows: (1) The air temperatures, measured at the weather station near the bridge site, show periodical trends, i.e., in an approximately sinusoidal varying pattern in this case. In addition, based on the measured wind speeds in between 10 m/s and 17 m/s, the surface convection behaviors are expected to be significant, and this phenomenon is proven from the field measurements where highly varying behaviors at the slabs are observed at the top surfaces of the bridge. (2) The measured seasonal temperatures variations of 35 °C (63 °F) and daily positive and negative gradients of the slabs, i.e., 10 °C (18 °F) and 1.11 °C (2 °F), respectively, are within the design values that are specified by the AASHTO LRFD [1] specification. The average temperature within the slabs can be predicted using the ambient air temperature by a linear or polynomial function. In addition, through a comprehensive study on the temperature distributions of the whole bridge, i.e., from the slab’s top surface to the pile’s bottom part, the bridge temperatures are found primarily varying within the superstructure. However, the temperature variations at the top part of the pile are also significant. (3) The strains measured at the surfaces of the top bents and bottom slabs generally show opposite variation trends. Based on the measurements during the monitored period, the maximum temperatures induced tensile stresses at the top bents of 2.49 MPa (360 psi) may possibly crack the concrete, while the induced compressive stresses are negligible and the maximum value is approximately 4% of the concrete material strength. (4) The soils behind the abutment affect the behaviors of the integral abutments in terms of their displacements and rotations. These effects are complicated, and the soil restraints to the abutment deformations will accumulate with time due to the plastic behaviors of soils. However, at the locations with 2.1 m away from the integral abutments or when the expansion joints are provided between the slab and abutment, these soil restraining effects become negligible.
During the monitoring period, the variations of the soil displacements right behind the integral abutment and 2.1 m away are 5.5 mm and 2 mm, respectively. The abutment primarily behavior in translation rather than rotation. (5) The bending strains of the piles with respect to the two main axes, i.e., parallel (y-axis) and perpendicular (x-axis) to the traffic direction, are both important, especially at the upper parts of the piles, due to the skew effects and rigid pile–bent connections. However, these strain values, with a maximum variation of 40 microstrains, are a small portion of the concrete compression capacities under the current temperature variations alone. In addition, for the pile bending profiles induced from the thermal movements of the superstructure, the zero bending points (or inflection points) appear at both bending directions.
Acknowledgments The investigators are thankful to the Innovative Bridge Research and Deployment (IBRD) program, Federal Highway Administration and Louisiana Transportation Research Center (LTRC) for funding this project. The contents presented reflect only the views of the writers who are responsible for the facts and the accuracy of the data presented herein. We would like to also express thankfulness to those who provided help during the development of these initial tasks of this research program. Special thanks go to the project manager Dr. Walid Alaywan, Louisiana Department of Transportation and Development (LADOTD).
References [1] AASHTO LRFD. Bridge design specifications. 4th ed. Washington, DC, USA; 2014. [2] Abendroth RE, Greimann LF, Lim KH, Thomas ME, Sayers BH, Kirkpatrick CL, et al. Field testing of integral abutments (No. HR-399); 2005. [3] Arockiasamy M, Butrieng N, Sivakumar M. State-of-the-art of integral abutment bridges: design and practice. J Bridge Eng 2004;9(5):497–506. [4] Barr P, Halling MW, Huffaker C, Boyle H. Behavior and analysis of an integral abutment bridge (No. UT-13.12); 2013.
B. Kong et al. / Engineering Structures 104 (2015) 18–31 [5] Breña SF, Bonczar CH, Civjan SA, DeJong JT, Crovo DS. Evaluation of seasonal and yearly behavior of an integral abutment bridge. J Bridge Eng 2007;12 (3):296–305. [6] Civjan SA, Kalayci E, Quinn BH, Breña SF, Allen CA. Observed integral abutment bridge substructure response. Eng Struct 2013;56:1177–91. [7] Davids WG, Sandford T, Ashley S, DeLano J, Lyons C. Field-measured response of an integral abutment bridge with short steel H-piles. J Bridge Eng 2009;15 (1):32–43. [8] Dunker KF, Liu D. Foundations for integral abutments. Pract Period Struct Des Constr 2007;12(1):22–30. [9] Elbadry MM, Ghali A. Temperature variations in concrete bridges. J Struct Eng 1983;109(10):2355–74. [10] Frosch RJ, Chovichien V, Durbin K, Fedroff D. Jointless and smoother bridges: behavior and design of piles (No. FHWA/IN/JTRP-2004/24); 2006. [11] Hassiotis S, Khodair Y, Roman E, Dehne Y. Evaluation of integral abutments (No. FHWA-NJ-2005-025); 2006. [12] Hoppe EJ, Bagnall TM. Performance of a skewed semi-integral bridge: volume I: field monitoring (No. FHWA/VTRC 08-R20); 2008. [13] Huang J, French CE, Shield CK. Behavior of concrete integral abutment bridges (No. MN/RC-2004-43); 2004. [14] Kim W, Laman JA. Numerical analysis method for long-term behavior of integral abutment bridges. Eng Struct 2010;32(8):2247–57. [15] Laman JA, Pugasap K, Kim W. Field monitoring of integral abutment bridges (No. FHWA-PA-2006-006-510401-01); 2006.
31
[16] Maruri RF, Petro SH. Integral abutments and jointless bridges (IAJB) 2004 survey summary. In: Integral abutment and jointless bridges (IAJB 2005); 2005. p. 12–29. [17] MATLAB (R2010b). The MathWorks Inc. [18] Mistry VC. Integral abutment and jointless bridges. In: Integral abutment and jointless bridges (IAJB 2005); 2005. p. 3–11. [19] Ooi PS, Lin X, Hamada HS. Field behavior of an integral abutment bridge supported on drilled shafts. J Bridge Eng 2009;15(1):4–18. [20] Pétursson H, Collin P, Veljkovic M, Andersson J. Monitoring of a Swedish integral abutment bridge. Struct Eng Int 2011;21(2):175–80. [21] Thippeswamy HK, GangaRao HV, Franco JM. Performance evaluation of jointless bridges. J Bridge Eng 2002;7(5):276–89. [22] Seo S. A review and comparison of methods for detecting outliers in univariate data sets (doctoral dissertation, University of Pittsburgh); 2006. [23] Voyiadjis GZ, Cai CS, Alshibli K, Faghihi D, Kong B, Yang Yi. Integral abutment bridge for Louisiana’s Soft and Stiff Soils. Report. Department of Civil and Environmental Engineering, Louisiana State University, LA; 2014. [24] Weakley K. VDOT integral bridge design guidelines. In: Integral abutment and jointless bridges (IAJB 2005); 2005. [25] White H, Pétursson H, Collin P. Integral abutment bridges: the European way. Pract Period Struct Des Constr 2010;15(3):201–8. [26] William GW, Shoukry SN, Riad MY. Study of thermal stresses in skewed integral abutment steel girder bridges. Struct Eng Int 2012;22(3):308–17.