NCHRP-RRD-351.pdf

Research Results Digest 351 Responsible Senior Program Officer: E.T. Harrigan January 2011 INTRODUCTION The objective of NCHRP Project 9-26A was to develop or update precision state- ments of AASHTO standard methods of test designated by the technical sections of the AASHTO Highway Subcommittee on Materials (HSOM). To meet this objective, NCHRP Project 9-26A used both data min- ing techniques and interlaboratory studies (or “round robins,” as defined in ASTM D6631, Standard Guide for Committee D01 for Conducting an Interlaboratory Practice for the Purpose of Determining the Preci- sion of a Test Method). This research results digest summa- rizes the findings of four interlaboratory studies (ILS) conducted in 2009 and 2010 to develop precision statements for the AASHTO standard methods of test shown in Table 1. Reports were published in the form of NCHRP web-only documents (WODs) as tasks related to individual stan- dard methods were completed. Precision statements and supporting results were pro- vided to the AASHTO HSOM for review and possible adoption. A complete report of the development of each precision statement is presented in the four WODs (1, 2, 3, 4) shown in Table 1. FINDINGS AASHTO T 148, “Measuring Length of Drilled Concrete Cores” An ILS was conducted to prepare preci- sion estimates for AASHTO T 148, “Mea- suring Length of Drilled Concrete Cores.” Six drilled concrete cores with varying dimensions and surface roughness were obtained from several test sections in the FHWAs Long-Term Pavement Perfor- mance program. The cores were deliv- ered to 11 laboratories, where the length of each core was measured using a 3-point caliper described in AASHTO T 148. The measurements were carried out at nine different locations at the center and along the circumference of the cores. A complete set of measurements was repeated fi ve times by each laboratory for the purpose of deter- mining repeatability precision estimates. The collected data were analyzed accord- ing to ASTM E691, “Standard Practice for Conducting an Interlaboratory Study to Determine the Precision of a TestMethod.” Table 2 summarizes the test data from the participating laboratories and their statis- tical analysis. Analysis of the experimental data pro- vided the following fi ndings: PRECISION STATEMENTS FOR AASHTO STANDARD METHODS OF TEST T 148, T 265, T 267, AND T 283 This digest summarizes key fi ndings obtained in 2009 and 2010 from continuing NCHRP Project 9-26A, “Data Mining and Interlaboratory Stud- ies to Prepare Precision Statements for AASHTO Standard Test Methods.” Project 9-26A was conducted by the AASHTO Materials Reference Labora- tory under the direction of the principal investigator, Haleh Azari. This digest is based on the contractors task reports, which are available online as NCHRP Web-Only Documents 163 through 166. Responsible Senior Program Officer: E. T. Harrigan NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM 1. The variability of the measurements signif- icantly increases as the length of the cores reaches the limits of the 3-point caliper mea- suring range described in AASHTO T 148. This was indicated by the highest repeatabil- ity standard deviation of the 12-in.-long core and highest reproducibility standard devia- tion of the 4-in.-long core. 2. The repeatability standard deviation increases with the increase in surface roughness of the cores as was indicated by higher variability of one of the 4-in.-diameter cores that had more surface irregularities than the other 4 in.- diameter cores. However, the variability was not statistically significant and the within- laboratory variability of all 4-in. diameter cores could be combined. 3. The variability of the measurements was found to be the same for cores with the same diame- ter (4-in. or 6-in.) and signifi cantly different for cores with different diameters (4-in. and 6-in.). Therefore, the standard deviations of the mea- surements of the same diameter cores were combined to prepare separate sets of preci- sions for 4-in.- and 6-in.-diameter cores. Precision estimates for the length measurements of the 4-in.- and 6-in.-diameter cores were computed after combining the standard deviations that were not signifi cantly different. Based on the signifi cant dif- ference in the precision estimates of 4-in.- and 6-in.- diameter cores, repeatability and reproducibility precision estimates were reported separately for each diameter. The resulting standard deviations and the allowable range of differences between two results within one laboratory and between different labora- tories are presented in Table 3. AASHTO T 265, “Laboratory Determination of Moisture Content of Soils” An ILS was conducted to prepare precision esti- mates for AASHTO T 265, “Laboratory Determi- nationof Moisture Content of Soils.” Test data were collected for four aggregate-soil blends judged suit- able for base and subbase construction. Specifi- 2 Repeatability (S r ) Reproducibility (S R ) Sample ID # of Labs D x L (inches) Intended Height (m m) Average Measured Height (m m) STD S x (m m) CV% 1s, (m m) CV% 1s, (m m) CV% LT 659 9 6 x 12 304.80 314.47 2.40 0.76 0.29 0.09 2.41 0.77 LT 755 10 6 x 8 203.20 203.55 1.70 0.83 0.60 0.29 1.78 0.87 LT 425 10 6 x 6 152.40 152.48 1.58 1.03 0.76 0.50 1.72 1.13 LT 2894 8 4 x 9 228.60 228.23 0.43 0.19 0.28 0.12 0.50 0.22 LT 1119 8 4 x 7 177.80 177.22 1.01 0.57 0.49 0.28 1.10 0.62 LT 523 8 4 x 4 101.60 107.36 2.24 2.09 0.31 0.29 2.26 2.10 *D and L stand for diam eter and length, respectively, of the cores. Table 1 Test methods and web-only documents AASHTO Standard Method of TestNCHRP Web-Only Document T 148, Measuring Length of Drilled Concrete Cores165 T 265, Laboratory Determination of Moisture Content of Soils164 T 267, Determination of Organic Content in Soils by Loss on Ignition163 T 283, Resistance of Compacted Hot Mix Asphalt (HMA) to 166 Moisture-Induced Damage Table 2 Summary of statistics of concrete core length measurements (mm) cally, there were two coarse-graded blendsone containing clay fi ller (Blend CC) and the other silt fi ller (Blend CS), and two fi ne-graded blendsone containing clay fi ller (Blend FC) and the other silt filler (Blend FS). Blends with a limited amount of materials passing the #200 sieve were selected. The ILS samples were prepared at the AASHTO Materials Resources Laboratory (AMRL) Profi ciency Sample Facility using procedures developed for the AMRL Proficiency Sample Program. A total of 1,260 samples were prepared and sent to the 35 selected laboratories. Each laboratory received 36 samples that consisted of three replicates of each of the four soil-aggregate blends prepared at three different percentages of moisture. The coarse blend samples weighed about 350 g and the fine blends samples weighed about 150 g. The fi ne blend sam- ples were prepared with 4%, 6%, and 8% moisture (designated as below optimum,optimum,and above optimum, respectively); the coarse blend samples were prepared with 3 %, 5 %, and 7% moisture (designated below optimum, optimum, and above optimum, respectively). The experimental data were analyzed accord- ing to ASTM E691, “Standard Practice for Con- ducting an Interlaboratory Study to Determine the Precision of a Test Method.” Tables 4 through 7 summarize the test data from the participating lab- oratories and their statistical analysis for the four blends. Analysis of the data provided the following findings: 1. The standard deviations of the blends with clay were not signifi cantly different from those of the blends with silt. Therefore, the standard deviations were combined. 2. The standard deviations of the coarse blends with 3% moisture (below optimum) were not signifi cantly different from those of the blends with 5% moisture (optimum). There- fore, these standard deviations were com- bined. 3. The standard deviations of the coarse blends with 7% moisture (above optimum) were sig- nifi cantly different from those of the blends with 3% and 5% moisture content. Due to uncertainty in the results of 7% moisture con- tent, they were not included in the precision estimate analysis. 3 Condition of Test and Test Property Standard Deviation, mm Acceptable Range of Two Results, mm Repeatability (Sr) 4-in.-diameter 0.41.0 6-in.-diameter 0.71.9 Reproducibility (SR) 4-in.-diameter 0.92.4 6-in.-diameter 1.84.9 Table 3 Precision estimates for measurement of drilled concrete cores based on AASHTO T 148 Repeatability (S r ) Reproducibility (S R ) Sa mp le Type # of Labs Targe t % Average % S x CV % 1s, % d2s, % 1s, % d2s, % Coarse Aggregate w/ clay (3% moisture) 27 3.0 3.02 0.06 1.9 0.042 0.1 0.07 0.2 Coarse Aggregate w/ clay (5% moisture) 28 5.0 4.98 0.11 2.3 0.044 0.1 0.12 0.3 Coarse Aggregate w/ clay (7% moisture) 25 7.0 6.89 0.26 3.8 0.060 0.2 0.27 0.8 Table 4 Summary of statistics of % moisture content of coarse aggregate with clay (CC) 4. The standard deviations of the fi ne blends with 4% moisture content (below optimum) and those of the blends with 6% moisture content (optimum) were not significantly different. Therefore, these standard deviations were combined. 5. The standard deviations of the fi ne blends with 8% moisture content (above optimum) were signifi cantly different from those of the blends with 4% and 6% moisture content. Due to uncertainty in the results of 8% moisture con- tent, they were not included in the precision estimate analysis. 6. The bias and low precision of the moisture content data for the above optimum blends were speculated to be due to availability of excess moisture for evaporation. When the mixture is above the optimum moisture con- tent, free moisture is available to evaporate and escape from the pores of the bottles. How- ever, in mixtures below the optimum and at the optimum, moisture adheres to the soil- aggregate particles. 7. The standard deviations of the coarse blends were significantly different from those of fine blends. Therefore the computed preci- sion estimates from the two blends are pre- sented separately in the proposed precision statement. Table 8 presents the precision estimates for mois- ture content determination based on the results of the 4 Repeatability (S r ) Reproducibility (S R ) Sa mp le Type # of Labs Targe t % Average % S x CV % 1s, % d2s, % 1s, % d2s, % Coarse aggregate w/ silt (3% mo isture) 27 3.0 3.03 0.05 1.6 0.05 0.1 0.06 0.2 Coarse aggregate w/ silt (5% mo isture) 29 5.0 5.02 0.10 2.1 0.06 0.2 0.12 0.3 Coarse aggregate w/ silt (7% mo isture) 29 6.6 6.60 0.33 5.0 0.44 1.2 0.49 1.4 Table 5 Summary of statistics of % moisture content of coarse blend with silt (CS) Repeatability (S r ) Reproducibility (S R ) Sa mp le Type # of Labs Targe t % Average % S x CV % 1s, % d2s, % 1s, % d2s, % Fine Aggregate w/ clay (4% moisture) 30 4.0 4.04 0.14 3.4 0.18 0.5 0.20 0.6 Fine Aggregate w/ clay (6% moisture) 29 6.0 5.92 0.20 3.4 0.17 0.5 0.25 0.7 Fine Aggregate w/ clay (8% moisture) 30 8.0 7.39 0.63 8.5 0.73 2.0 0.87 2.4 Table 6 Summary of statistics of % moisture content of fi ne blend with clay (FC) Repeatability (S r ) Reproducibility (S R ) Sa mp le Type # of Labs Targe t % Average % S x CV % 1s, % d2s, % 1s, % d2s, % Fine Aggregate w/ silt (4% moisture) 29 4.0 3.97 0.11 2.9 0.17 0.5 0.18 0.5 Fine Aggregate w/ silt (6% moisture) 30 6.0 5.97 0.16 2.7 0.12 0.3 0.19 0.5 Fine Aggregate w/ silt (8% moisture) 30 8.0 7.69 0.46 6.0 0.60 1.7 0.68 1.9 Table 7 Summary of statistics of % moisture content of fi ne blend with silt (FS) ILS conducted in this study. The standard deviations corresponding to coarse and fi ne blends were used to compute the allowable differences between two moisture content measurements. AASHTO T 267, “Determination of Organic Content in Soils by Loss on Ignition” An interlaboratory study was conducted to pre- pare precision estimates for AASHTO T 267, Deter- mination of Organic Content in Soils by Loss on Ignition.” Samples from three types of soils (clay, silt, and sand) were each blended with three differ- ent percentages (2%, 5%, and 8%) of fi ne walnut shell grits as organic material and sent to 30 labora- tories for organic content measurement. The labora- tories were instructed to test three replicates of each organic content level of each soil type. Results were obtained from 27 different laboratories. ILS test results were analyzed for precision in accordance to ASTM E 691. Before the analysis, any outlier data were eliminated by following the procedures described in ASTM E 691 for determin- ing repeatability (Sr) and reproducibility (SR) esti- mates of precision. For each set of data, the h and k statistics, representing the between and within lab- oratory consistency, were used to identify the out- lier data. Data exceeding the critical h and k values were eliminated; once identifi ed for elimination, the same data were eliminated from any smaller subsets analyzed. Multiple sets of data in each soil type were elim- inated based on the critical h and k values. After eliminating the outlier data, the averages and the repeatabilityand reproducibility standard deviations of the data were determined. The Srand SRprecision estimates were determined using the remaining data in conformance with ASTM E 691. A summary of statistics of the measurements is shown in Table 9. The comparison of the design and measured organic content values in the table indi- catesthat every soil has a certain percentage of intrin- sic organic material; clay has the greatest amount of intrinsic organic material, whereas sand has the least amount. Upon subtracting the intrinsic organic con- tents from the measured organic contents, the aver- age of the measured values agree closely with the design values as shown in Table 10. In addition to the adjusted averages, Table 10 also provides the adjusted variability of the measurements. The table shows that the standard deviation of the measurements for sand increases with the increase in the percentage of organic material. The increased variability of the sand blend with higher percentages of organic material indicates segregation of organic material during shipment. This could be explained by the non-cohesive nature of sand that does not allow the ground walnut shell grits to adhere to sand particles. Analysis of the data provided the following findings: 1. Clay has the greatest amount of intrinsic organic material and sand has the least amount. 2. The within-laboratory and between-laboratory standard deviations were very consistent for different organic content levels of clay or silt blend. Therefore, for these two blends, the standard deviations corresponding to 2%, 5%, and 8% organic material were combined. 3. For the sand blend, the within-laboratory and between-laboratory standard deviations of 5 Material and Type Index Standard Deviations (1s) Acceptable Range of Two Results (d2s) Single-Operator Precision: Coarse blend Fine blend 0.05 0.16 0.14 0.46 M