PULLOUT RESISTANCE FACTORS FOR INEXTENSIBLE MSE REINFORCEMENTS EMBEDDED IN SANDY BACKFILL November 16, 2012 Word Count: Abstract: 216 Text: 4361 Figures 7@250: 2250 Tables 3@250 750 Total Words: 7577 William D. Lawson Associate Professor (CORRESPONDING AUTHOR) Department of Civil and Environmental Engineering Texas Tech University P.O. Box 41023 Lubbock, TX 79409 Ph: (806) 742-3521 / (Fax: 806-742-0444) E-mail: william.d.lawson@ttu.edu Priyantha W. Jayawickrama Associate Professor Texas Tech University P.O. Box 41023 Lubbock, TX 79409 Ph: (806) 742-3523 ext. 245 / (Fax: 806-742-3488) E-mail: priyantha.jayawickrama@ttu.edu Timothy A. Wood Research Associate Texas Tech University P.O. Box 41023 Lubbock, TX 79409 Ph: (806) 742-3523 ext 260/ (Fax: 806-742-3488) E-mail: timothy.a.wood@ttu.edu and James G. Surles Associate Professor Texas Tech University P.O. Box 41042 Lubbock, TX 79409 Ph: (806) 742-2580 ext. 238 / (Fax: 806-742-1112) E-mail: james.surles@ttu.edu 13-2684 Page 1
Pullout Resistance Factors for Inextensible MSE Reinforcements Embedded in Sandy Backfill William D. Lawson, Priyantha W. Jayawickrama, Timothy A. Wood, and James G. Surles ABSTRACT: This paper presents results from a laboratory program of 402 pullout tests of inextensible reinforcements used for Mechanically Stabilized Earth (MSE) walls. Results focus on evaluation of pullout resistance factors for ribbed steel strip and welded steel grid reinforcement embedded in sandy backfill that marginally met AASHTO requirements for select granular fill. This project used Texas Tech University s large-scale MSE Test Box with dimensions of 12ft x 12ft x 4ft and an applied overburden capacity of 40 ft of backfill. This test box facilitates pullout testing at a scale not unlike typical field construction. The research design evaluated pullout resistance factors for both ribbed strip and welded grid reinforcements for a variety of independent variables including overburden pressure, reinforcement length, level of compaction, grid wire size, and grid geometry including both transverse and longitudinal wire spacing. Appropriate statistical analyses were used to interpret the data within the context of published AASHTO design guidance for inextensible MSE reinforcements. The results show that pullout behaviors of both ribbed strips and welded grids in properly compacted sandy backfill are conservative compared to the default pullout resistance factors provided by AASHTO. The data also suggest that the current AASHTO equations for pullout resistance factors for grid reinforcement do not accurately capture the influence of transverse and longitudinal bar spacings. Key Words: pullout resistance, inextensible reinforcement, sandy backfill, MSE wall 13-2684 Page 2
INTRODUCTION This paper summarizes the findings from a comprehensive research study that investigated the pullout resistance behavior of two types of inextensible MSE reinforcements, ribbed strip also known as high adherence (HA) strip reinforcements and welded steel grid reinforcements. The primary objective of the research study was to determine the pullout resistance factors (F* values) for HAstrip reinforcements and welded steel grid reinforcements embedded in sandy backfill and compare them with guidelines provided in AASHTO for F* values to be used in the design of MSE walls. This paper reports findings from Phase I of this research study which included 99 pullout tests on HA-strip reinforcement and 195 tests on welded steel grid reinforcements embedded in sandy backfill. In addition to the above, the research project also included a separate series of pullout tests to evaluate the impact of skewing of strip reinforcement and cutting and splaying of grid reinforcement on pullout resistance capacity. The findings from this test series were documented in a previous publication (1). Phase II of the research study examined pullout behavior of inextensible MSE reinforcements in gravel backfill. This work is currently ongoing. METHOD Large-Scale MSE Pullout Resistance Test System MSE pullout tests for this study were performed using the large-scale pullout resistance test system at Texas Tech University, the centerpiece of which is the MSE Test Box shown in FIG. 1. The dimensions of the MSE Test Box are 12ft by 12ft in plan and 4ft in depth. In this test system, the soil overburden pressures applied to the embedded earth reinforcement are simulated using a reaction frame assembly which consists of nine 4ft by 4ft pressure plates that are hydraulically jacked against three wide flange cross beams. This reaction frame assembly is designed to accommodate simulation of overburden pressures up to 40ft of fill. Vertical overburden pressure is monitored throughout the test using five earth pressure cells that are grouted into the bottom slab of the MSE Test Box. A second, independent measurement of the applied overburden is made using a pressure transducer that monitors the fluid pressure in the hydraulic jacks. Pullout testing of each embedded earth reinforcement was accomplished by attaching a pullout load assembly to the reinforcement and then applying the pullout force using a 60-ton hollow core hydraulic jack. HA-strip and welded steel grid reinforcements are connected to the pull rod using a gripper mechanism. Two independent systems were used to measure and record the pullout force. The first is an annular load cell mounted on the pull rod. The second is a pressure transducer that measures hydraulic pressure in the hollow-core jack. 13-2684 Page 3
FIG. 1. Large Scale MSE Pullout Resistance Test System: Civil Engineering Department, Texas Tech University. Similarly, the displacement of the reinforcement during pullout testing was determined using two independent measurements. The first is an optical measurement of the displacement using a witness marker against a graduated scale. The second is a digital measurement of displacement that is made using an electronic displacement gage. Materials Backfill The backfill material used in this test program was a fine to medium sand that marginally satisfies the requirements for MSE select backfill as specified in AASHTO LRFD Bridge Construction Specifications (2), FHWA Publication No. FHWA-10-024 (3) and TxDOT Specification Item No. 423/Type B select backfill (4). Figure 2 presents the particle size distribution curve for the sandy backfill material used in this study. The material was determined to be non-plastic (ASTM D4318) and classifies as SP-SM according to the Unified Soil Classification System (ASTM D2487). The coefficient of uniformity, C u, of this sandy backfill is 4.7, which is close to the default C u value recommended by AASHTO when a specific C u is not known for the wall backfill at the time of the wall design (5). The backfill has a maximum dry unit weight of 124.5pcf and an optimum water content of 7.8% as established through testing conducted in accordance with TxDOT standard test procedure, Tex-114-E, which uses the same compactive energy as the Standard Proctor compaction method (ASTM D698). 13-2684 Page 4
FIG. 2. Particle size distribution curve for MSE wall backfill material. During testing, backfill material was placed in lifts of loose fill with approximate thickness of 5.0-in. This material was then compacted with 10 passes of a vibratory compactor to achieve a compacted thickness of about 4.0-in. The compacted density was measured using a nuclear density gage to make sure that a target relative compaction of 95% had been achieved during compaction. The average relative compaction calculated based on nuclear density gage readings obtained for all test setups was 95.7% and the standard deviation was 2.4%. The average water content measured was 6.5% and the corresponding standard deviation was 1.3%. MSE Reinforcements. The test program for this research study evaluated HA-strip reinforcements which are produced by The Reinforced Earth Company. HA-strip reinforcements have been used in the construction of MSE walls for more than 30 years. As a result, a significant body of data exists on the pullout resistance of this type of reinforcement. More importantly, AASHTO and FHWA-NHI provide guidance on default pullout resistance factors, F* to be used with HA-strip reinforcements (3, 5). One of the contributions of this research study was to expand the body of pullout resistance data using the large-scale MSE test box, as previous data were obtained from tests in smaller test boxes. The test program also evaluated pullout resistance for welded steel grid reinforcements. Welded steel grids tested in this study were specially fabricated by 13-2684 Page 5
an approved grid manufacturer to meet the specific requirements of the research. Grid longitudinal bar sizes used were W9.5 and W20. Grid transverse bar sizes were W7.5, W11, and W15. Longitudinal bar spacings were 2in, 6in, 9in and 12in. Transverse bar spacings were 6in, 12in, 18in, and 24in. Pullout Test Procedure The pullout test procedure included preparation of the test setup, application of the overburden pressure, and application of the pullout load. Test setup preparation involved filling the MSE Test Box with the backfill and compacting the backfill material to achieve desired density while embedding the MSE reinforcements at appropriate depths within the compacted backfill. Application of the vertical overburden pressure was accomplished using hydraulic jacks to apply compressive force to a set of nine 4ft x 4ft stiffened steel plates placed on top of the backfill and covering the entire surface area of the MSE Test Box. The hydraulic jacks were connected to a main hydraulic pump via a manifold so that all nine jacks were pressurized equally and simultaneously. A hollow core jack that applied tension force to the MSE reinforcement by pushing against a stationary bearing surface, or bulkhead, attached to the MSE Test Box was used to apply the pullout load. The pull force was transmitted to the MSE reinforcement through a high tensile strength extension rod. Reinforcements were connected to the extension rod by a pin-joint gripper to facilitate uniformly-applied, pure axial load without bending. After connecting the gripper mechanism, an optical displacement marker (also called witness marker) was mounted on the wall of the MSE Test Box to provide visual measurement of the pullout displacement. The optical displacement marker consisted of a pointer with a magnetic base and this witness marker was used to establish test zero as the reference from which both optical and digital displacements during pullout testing were measured. With the entire pullout mechanism set up and the overburden pressure applied, the force and displacement sensors and the data acquisition system were activated to start recording data. The pullout test was then commenced with application of the pullout force via the hollow core hydraulic jack by pumping the jack manually. Reinforcements were pulled to a maximum displacement of 1.50 inch or until the reinforcement ruptured, which is the typical testing termination criterion. The pullout force measured at 0.75 inch displacement was used for calculation of the pullout resistance factor, F*. SCOPE OF THE TEST PROGRAM Pullout tests in sandy backfill were completed during the period from June 2010 through June 2011. The pullout test program in sandy backfill commissioned for this study included a total of 411 pullout tests on HA-strips, welded steel grids, HA ladders, and straight smooth bars, of which 9 tests were removed from the dataset due 13-2684 Page 6
to test irregularities. This yielded a total of 402 pullout tests on inextensible reinforcements as shown in Table 1. This paper discusses only part of these data, namely, pullout tests on straight HA strips (99 tests) and straight welded steel grids (195 tests). As mentioned previously, the findings on the effects of skewing and splaying where the direction of pullout, β, is other than 0 have been addressed elsewhere and are not repeated here (1). Also, this paper does not discuss test data associated with the HA ladders and smooth steel bars. Table 1. Pullout Test Matrix Reinforcement Number Detail Description of Tests HA Ribbed Strips Straight pull 99 Skewed (β = 15, 30 ) 27 Welded Steel Grids Straight pull 195 Cut and Splayed (β = 15, 30 ) 12 HA Ladders Straight pull 52 Smooth Bars Straight pull 17 TOTAL 402 DATA ANALYSIS The significant database developed through the above test program was reviewed and analyzed using appropriate statistical analysis procedures. The primary objectives in this analysis were twofold: (1) analyze the database to identify key variables that have statistically significant influence on the measured F*, and (2) to develop F* prediction models and corresponding prediction intervals based on the new data set. The prediction limits corresponding to 95% confidence level could then be compared with published AASHTO reference F* lines. The statistical procedures used in the quantitative evaluation of F* data consisted of analysis of variance (ANOVA) and non-linear regression analysis. These procedures require that the data set satisfy the homoscedasticity (uniformity of variance) condition. This was accomplished by transforming F* data into ln(f*) using the Box-Cox Transformation. Then the statistical analyses were performed on transformed F*, i.e. ln(f*). The statistical analysis procedure used with each data set followed the same basic steps. First, an ANOVA was performed with ln(f*) as the response variable and the other variables of interest as factors. The factors included those variables that are recognized in the current AASHTO F* equation as key variables. For example, the current AASHTO F* equation recognizes depth of fill as a key variable that influences pullout resistance factor of HA-strips. Similarly, it recognizes depth of fill, transverse bar diameter and transverse bar spacing as key variables influencing 13-2684 Page 7
pullout resistance factor of grid reinforcement. In addition to these variables, other variables that have potential influence on F* were also included in this analysis as factors. The embedded length of the reinforcement, longitudinal bar spacing and longitudinal bar diameter are examples of such variables. Factors used in ANOVA for both strip and grid type reinforcements are shown in Table 2. Table 2. Factors and Covariates Used in ANCOVA Type of Factors Reinforcement per AASHTO Additional Covariates HA Ribbed Strips Depth of fill Test layer (vertical position of reinforcement in the test cell) Overburden stress ratio Welded Steel Depth of fill Embedment Test layer Grids length Transverse bar Longitudinal Overburden stress ratio diameter bar diameter Transverse bar spacing Longitudinal bar spacing Also shown in Table 2 is a second set of variables identified as covariates. Covariates represent extraneous variables related to testing. The large test box used in this research could accommodate three layers of reinforcement in a single filling. Moreover, each layer of reinforcement consisted of multiple test specimens. Therefore, it was important to determine whether the position of the reinforcement within the test box (top, middle or bottom layer) had any impact on the measured F*. Another important test variable was the overburden stress ratio (OSR). Once again, this variable becomes relevant because multiple test specimens were embedded and tested in the same test setup. Ideally, a test specimen should not experience an overburden stress greater than its test overburden stress prior to being tested. However, this ideal condition could not be achieved for all test specimens in this test series. Therefore, overburden stress ratio was also introduced in the statistical analysis as a covariate. In this manner, any undue influence due to these extraneous variables on the response variable could be filtered out during statistical analysis. An ANOVA analysis that includes covariates is known as analysis of covariance, or ANCOVA. The form of the ANCOVA model used in the present analysis is shown in Eq.1. This simple model, used for illustrative purposes, involves one factor A with k levels and one covariate x, which is a numeric variable. y ij x (1) i ij ij where y ij = j th observation of the response variable at the i th level of A 13-2684 Page 8
µ = the intercept α i = the effect of the i th level of the factor A β = a slope parameter associated with the covariate x ij which is measured along with y ij. ε ij = the error term that is assumed to be normally distributed with mean zero and constant variance. Quite simply, this model allows a different linear regression line between y and x for each level of the factor A. In cases where a predictive model is desired, the preliminary ANOVA is used to identify which variables are to be included in a nonlinear regression model to relate F* to the other variables. A nonlinear regression was necessary as opposed to linear regression because of the nature of the relationship between F* and depth of fill (DOF). It was found that a model of the following form (Eq.2) provided a very good fit for the data. ln( F*) 0 1 exp( 2 DOF) linear terms (2) where β 0, β 1, and β 2 = regression coefficients DOF = depth of fill Linear terms = test layer, OSR etc. MINITAB version 16 was used to perform the nonlinear regression analysis using the Gauss-Newton optimization algorithm to find the least-squares estimators for the parameters. RESULTS AND DISCUSSION Pullout Tests on HA Ribbed Strip Reinforcements Properly Compacted HA-Strips Figure 3 is the AASHTO chart for default values for pullout resistance factors for MSE reinforcements (5). Figure 4 is a plot that has been prepared using the same format as the AASHTO chart. It shows the pullout resistance factors, F* obtained in this study for HA ribbed strips embedded in properly-compacted sandy backfill (i.e., relative compaction greater than or equal to 95% Standard Proctor). In addition to the individual test data points, Figure 4 shows two reference lines for F*. The first line, designated by a solid line, is the AASHTO reference line for default F*. This reference line was developed based on research conducted by the Florida DOT using uniform sand with a fine to medium grain size. The AASHTO reference line represents the 95 th percent lower limit of the Florida DOT data (6). The second 13-2684 Page 9
reference line, designated by the dashed line, represents the 95 th percent lower limit for the data obtained from this study. FIG. 3. Default Values for Pullout Resistance Factors for MSE Reinforcement (AASHTO, 2010) The most obvious independent variable is the depth of fill. A second variable that was of interest was the length of reinforcement. The current AASHTO specifications assume that F* is independent of the length of reinforcement. The length of reinforcement was included among the independent variables so that the validity of the assumption inherent in the current AASHTO specifications could be tested. In the statistical analysis performed, the depth of fill emerged as the variable that has the most dominant influence on F*. The effect of length of reinforcement 13-2684 Page 10
was determined to be statistically insignificant (p>0.05); therefore, the reinforcement length was not included as an independent variable in the final predictive model. F* 0.0 2.0 4.0 6.0 8.0 0 5 10 15 20 Depth of Fill (ft) 25 30 35 40 45 50 F* Data from Current Study Lower 95th Predictive Limit based on Current Study AASHTO (2010) Reference F* for Ribbed Strips FIG. 4. Pullout resistance factor, F* vs. Depth of Fill for HA strip reinforcements in sandy backfill, properly compacted (95% Rel. Comp.) Among the test condition variables examined, the test layer was identified as a significant variable influencing the measured F*. The overburden stress ratio was found to be a second important variable. To ensure that the final predictive model for F* was free from any systematic influence from such test condition variables, standard test conditions were defined and then predictions made for these conditions. Reinforcement placed in middle layer of the MSE Test Box would experience minimum interference from potential boundaries; therefore, the middle layer was selected as the standard (or reference) test position. Similarly, a reinforcement on 13-2684 Page 11
which the overburden stress never exceeded test overburden stress (maximum prior overburden stress equal to test overburden stress) was chosen as the reference. In this way, the reference line shown as a dashed line in Figure 4 represents the 95 th percent lower prediction limit for the data collected and analyzed in this study. It should be noted that, while F* data shown on Figure 4 have not been adjusted to represent standard test conditions, the lower-bound prediction limit represents F* values that would result from pullout tests performed under standard test conditions. Comparison of the two lower bound limits suggests that the AASHTO limit is conservative, especially for pullout resistance behavior associated with depth of overburden less than 12ft. Under-Compacted HA-Strips Due to a change in test protocol in the early portion of the test program, 36 pullout tests were performed on HA-strips where the backfill was under-compacted; that is, the relative compaction was about 91 percent of standard Proctor rather than the required 95 percent. These data provide an indication of how relative compaction influences F*. Table 3 presents estimated means and 95% confidence intervals for F* values from pullout tests in both properly compacted and under-compacted backfill at the depth of fill of 12ft. Table 3. Influence of Relative Compaction on Pullout Resistance Factor, F* Degree of Compaction Mean F* Lower Confidence Interval Upper Confidence Interval Under-compacted backfill 1.77 1.50 2.09 Properly compacted 2.69 2.26 3.20 ANOVA for the HA-strip data show that compaction is a statisticallysignificant independent variable (p=0.000). Comparison of the mean F* values in Table 3 indicates that compaction is also highly significant from practical construction perspective. What might reasonably be viewed as slight undercompaction (4% below specification) appears to dramatically reduce pullout resistance (34% decrease). Pullout Tests on Welded Steel Grid Reinforcements Welded Steel Grids, Undifferentiated As has been noted, this research study tested pullout resistance of welded steel grids over a range of overburden pressures, embedment lengths, longitudinal and transverse bar sizes and longitudinal and transverse bar spacings. However, a significant portion of these tests were done on grids with W20 (0.50in) longitudinal bars, W11 (0.37in) transverse bars and longitudinal bar spacing of 9.0 inches. Figure 5 is plot of pullout resistance versus depth of fill for these tests. It should be noted that, when plotting the above data, the pullout resistance factors have been normalized by dividing by t/s t. Such normalization was done because, 13-2684 Page 12
according to AASHTO, F* varies linearly with t/s t. Normalized F* is used throughout this paper because it allows all F*-data to be combined in a single plot. 0 Normalized F* (F*St/t) 0 20 40 60 80 100 5 10 Depth of Fill (ft) 15 20 25 30 35 40 Data Collected in Current Study 95% Predictive Lower Bound 45 FIG. 5. Normalized Pullout resistance factor, F* vs. Depth of Fill for grid reinforcements in sandy backfill Figure 5 shows two reference lines for normalized F*. The first line, designated by a solid line, is the AASHTO reference line for welded steel grids. The second reference line, designated by the dashed line, represents the 95 th percent lower limit for the data obtained from this study. Comparison of the two lower bound limits suggest that AASHTO limit is conservative, especially for pullout resistance behavior associated with depth of overburden less than 12ft. ANOVA performed on this data set revealed that the depth of fill is statistically significant (p= 0.011) and is negatively correlated with the normalized 13-2684 Page 13
pullout resistance factor. Accordingly, normalized F* would be highest at shallow depths and lowest at higher depths of fill, which is the expected trend. Welded Steel Grids; Effect of Embedment Length The welded grid pullout test program conducted in this research consisted of several components each designed to evaluate the effect of a specific variable. One of these was embedment length. The test program included 18 pullout tests conducted on grids with lengths of 3.0ft, 6.0ft, 9.0ft and 12.0ft. ANOVA conducted on this data set shows that embedment length is not statistically significant (p>0.05). This is consistent with the AASHTO equation for F* for grids. Welded Steel Grids; Effect of Transverse Bar Size and Spacing A second and a significant portion of grid test program was dedicated to evaluation of the effect of transverse bar size and spacing. In this portion of the test program, the longitudinal bar size, spacing and grid width were kept constant at W20 (0.50in), 9.0in and 18in respectively. The pullout tests were performed using three transverse bar sizes: W7.5 (0.31in), W11 (0.37in) and W15 (0.44in) and four transverse bar spacings: 6in, 12in, 18in and 24in. This sub-program included 111 grid pullout tests. According to AASHTO, F* for inextensible grids varies linearly with the transverse bar diameter/transverse bar spacing ratio (i.e., t/s t ), the relationship being based on data collected from a number of previous research studies (7, 8, 9). As discussed previously, this research study utilized the normalized F* parameter to facilitate comparison of test results on a single plot. Results from ANOVA analysis showed that transverse bar spacing had strong influence (p=0.000) on the normalized F* (i.e., F*S t /t). The influence of transverse bar diameter, however, was not statistically significant (p=0.113). This means that AASHTO s F* relationship for inextensible grids is consistent with the research findings relative to transverse bar diameter, but the findings from this study differ from AASHTO relative to transverse bar spacing. Accordingly, transverse bar spacing was retained as a significant variable in subsequent non-linear regression analyses and predictive models made for specified values of transverse bar spacings. Figure 6 shows normalized F* as a function of depth of fill and transverse bar spacing. As mentioned previously, the longitudinal bar spacing was maintained constant at 9.0in in this data set. According to this data normalized F* increased significantly as the transverse bar spacing increased from 6in to 12in, and continued to increase and reached a maximum at 18in bar spacing. Welded Steel Grids; Effect of Longitudinal Bar Size and Spacing The final data set discussed in this paper involves the influence of longitudinal bar size and spacing. This test program involved a total of 63 pullout tests. This test program included two longitudinal bar sizes: W9.5 (0.35in) and W20 (0.50in) and four different longitudinal bar spacings: 2.0in, 6.0in, 9.0in and 12.0in. All of the grids tested consisted of 3 longitudinal bars. The transverse bar size and spacing were kept constant at W11 (0.37in) and 12.0in respectively. ANOVA shows that the longitudinal bar spacing is statistically significant (p=0.000). The longitudinal bar 13-2684 Page 14
size was not found to be statistically significant (p=0.812). Figure 7 shows a plot of normalized F* values vs. depth of fill for the different longitudinal bar spacings tested. F*(St/t) 0.0 20.0 40.0 60.0 80.0 100.0 120.0 0 5 10 Depth of Fill (ft) 15 20 25 30 35 40 45 AASHTO Reference F* (St/t) Measured F* (St/t) St = 6 in Measured F* (St/t) St = 12 in Measured F* (St/t) St = 18 in Measured F* (St/t) St = 24 in Lower Bound F* (St/t) St = 6in Lower Bound F* (St/t) St = 12in Lower Bound F* (St/t) St = 18in Lower Bound F* (St/t) St = 24in FIG. 6. Normalized Pullout resistance factor, F* vs. Depth of Fill for grid reinforcements in sandy backfill; Effect of Transverse Bar Spacing 13-2684 Page 15
F*(St/t) 0.0 50.0 100.0 150.0 200.0 0 5 Depth of Fill (ft) 10 15 20 25 30 35 40 45 AASHTO Reference F* Measured F* Sl = 2 in Measured F* Sl = 6 in Measured F* Sl = 9 in Measured F* Sl = 12 in Lower Bound F*(St/t) Sl = 2 in Lower Bound F*(St/t) Sl = 6 in Lower Bound F*(St/t) Sl = 9 in Lower Bound F*(St/t) Sl = 12 in FIG. 7. Normalized Pullout resistance factor, F* vs. Depth of Fill for grid reinforcements in sandy backfill; Effect of Longitudinal Bar Spacing Longitudinal bar spacing is negatively correlated with the normalized pullout resistance factor. Normalized F* values are lowest at 12-inch spacing and peak at 2- inch spacing. CONCLUSIONS This paper presents findings of a research study to investigate the pullout resistance behavior of inextensible HA-strip and welded steel grid reinforcements embedded in sandy backfill and tested in Texas Tech University s large-scale MSE Test Box. Data are reported for 99 pullout tests on HA-strips and 195 pullout tests on welded steel grids. The findings of this research study support the following conclusions: 13-2684 Page 16
1. Pullout behaviors for both HA-strips and welded steel grids in sandy backfill, properly compacted, are conservative compared to the default F* vs. depth of fill relationships published by AASHTO. The margin between measured pullout behavior and the AASHTO reference lines is greatest for shallow overburden depths, and the margin decreases with increasing overburden depth. 2. Relative compaction strongly influences pullout resistance for HA strips in that a slight drop in compaction can dramatically reduce pullout resistance. This effect, unintentionally encountered in this study, warrants further analysis. 3. Consistent with AASHTO specifications, the data from this study indicate that the pullout resistance factor is independent of the length of reinforcement. 4. According to the current AASHTO specifications, the pullout resistance factor for grid reinforcement linearly increases with transverse bar diameter and inversely decreases with transverse bar spacing. The data obtained from this study support the linear increase in pullout resistance with transverse bar diameter. However, the findings suggest that AASTHO s formulation does not accurately describe for relationship between pullout resistance and transverse bar spacing. 5. Current AASHTO specifications assume that the pullout resistance factor for grid reinforcement is independent of the longitudinal bar spacing. The study findings clearly indicate a strong relationship between pullout resistance and the longitudinal bar spacing, narrower bar spacings yielding higher pullout resistance factors. In summary, this study significantly expands the body of pullout test data for inextensible reinforcements upon which current MSE design guidance is based. Insights from this study can help lead to more reliable, more efficient and more economical MSE wall designs. ACKNOWLEDGEMENT This research work described in this paper was sponsored by the Texas Department of Transportation (TxDOT). The authors gratefully acknowledge the Texas Department of Transportation for sponsoring this research. The authors thank Marie Fisk, P.E., John Delphia, P.E., and Marcus Galvan, P.E. of TxDOT for their technical guidance during this research. The authors thank Robert Gladstone, P.E., Guy Nelson, P.E., John Sankey, P.E., and Thomas Taylor, P.E. for their participation in this project as an industry advisory panel. REFERENCES 1. Jayawickrama, Priyantha W., William D. Lawson, Timothy A. Wood, and Asitha Senanayake. Effect of Skewing and Splaying on Pullout Capacity of MSE Reinforcement. Paper 12-2021, Scheduled for publication in the Transportation 13-2684 Page 17
Research Record, Journal of the Transportation Research Board, Washington DC., 2012. 2. AASHTO. AASHTO LRFD Bridge Construction Specifications. 3rd Edition. Washington, DC: American Association of State Highway and Transportation Officials, Inc., 2010. 3. Federal Highway Administration. Design and Construction of Mechanically Stabilized Earth Walls and Reinforced Soil Slopes Vol.I, Publication No. FHWA- 10-024. November 2009. 4. Texas Department of Transportation. ITEM 432: Retaining Walls. In Standard Specifications for Construction and Maintenance of Highways, Streets, and Bridges. Austin, TX: Texas Department of Transportation, 2004. 5. AASHTO. AASHTO LRFD Bridge Design Specifications, Customary U.S. Units. 5th Edition. Washington D.C.: American Assocaition of State Highway amd Transportation Officials, Inc., 2010. 6. The Reinforced Earth Company. Apparent Coefficient of Friction, f* to be Used in the Design of Reinforced Earth Structures. Technical Bulletin; MSE-6, Vienna, VA: The Reinforced Earth Company, 1995. 7. Bishop, J.A. and Anderson L.R., Performance of a Welded Wire Retaining Wall, Report Submitted to the Hilfiker Company, Utah State University, Logan, Utah, 1979. 8. Peterson, L. M. Pullout Resistance of Welded Wire Mesh Embedded in Soil. M.S. Thesis, Utah State University, Logan, Utah, 1980. 9. Nielsen, M. R., & Anderson, L. R., Pullout Resistance of Wire Mats Embedded in Soil, Report Submitted to the Hilfiker Company, Utah State University Logan, Utah 1984. 13-2684 Page 18