|FHWA > HfL > Technology Partnerships > Aggregate Image Measurement System 2 (AIMS2): Final Report|
Aggregate Image Measurement System 2 (AIMS2): Final Report
Upon completion of the new design, AIMS2, the project moved into two stages of independent testing. This testing compared the AIMS2 results against the AIMS1 results and confirmed that the testing methodology was rugged.
Texas A&M University (TAMU) was selected to perform the tests on the new equipment under the direction of Eyad Masad, Ph.D., P.E. The TAMU facility was selected because of the staff's extensive expertise in construction materials and materials research and familiarity with the AIMS1 system, as well as the facility's extensive database containing a variety of well-characterized materials. Samples of these well-characterized materials were scanned with the AIMS2 to compare the results. The complete report of the TAMU study is provided in appendix A. This report contains three sections: Chapter 1—Improvements of the AIMS, Chapter 2—Ruggedness Evaluation of AIMS2, and Chapter 3—Interlaboratory Study.
Chapter 1 of the TAMU Report (appendix A) discusses the calibration of the AIMS2 results to the AIMS1 results. Materials were scanned with the new hardware, and the results were compared to the research system results. The data demonstrated that the AIMS2 system provides material characterization similar to that of the AIMS1 research equipment. With this confirmation that the characterizations are comparable, the research relating material shape characterizations to performance that was completed with the AIMS1 equipment can be considered applicable to the shape characterizations provided by the AIMS2 system. (See references 1, 2, 3, and 4.)
The calibration tests comparing the research AIMS1 to the new AIMS2 demonstrated that most of the outputs of the new system match the research edition quite well. Angularity and form data output from the new system aligned very closely with the old system's output. This was expected, as the image analysis algorithms had not been changed; only the image acquisition hardware had been updated. However, the texture output was found to be of a different magnitude on the new system when compared to the AIMS1. This difference in texture was not unexpected, as the updated equipment provided increased capabilities in both lighting and camera performance. The new system continued to rank the various materials in a similar manner as the old with respect to each other. The output was simply of a different scale. Because of research completed with the original equipment in studying texture and its relationship to in-place performance, it was desirable to maintain the texture scale of the research system. It was decided that the best solution to address this shift in scale was to apply a shift factor to the new system's data.
This shift of texture in the AIMS2 system with respect to the AIMS1 system is the result of three converging design changes: enhanced illumination, improved camera performance, and the enclosed chamber for image acquisition. The new top-light design provides oblique-angle spot lighting, which provides excellent contrast to the particle surface compared to the ring lighting of the AIMS1 system. The new camera provides improved gray-scale texture images by providing a more precise control of the image intensity. Finally, the enclosure eliminates ambient lighting to provide a controlled lighting environment for image acquisition. It is believed that the new system represents an improvement in performance, while maintaining excellent correlation to previous results that have been linked to pavement materials performance.
Ruggedness TestingThe second stage of the testing was designed to confirm that the testing methodology was rugged. Ruggedness, in this meaning, is not the ability of the equipment to stand up to abuse, but the ability of the testing methodology to handle small, anticipated, variations in the parameters used for capturing the data. It confirms that the operational parameters that occur within the equipment, laboratory environment, and system operation do not significantly impact the results. This testing followed the guidelines of ASTM C1067, "Standard Practice for Conducting a Ruggedness or Screening Program for Test Methods for Construction Materials."
Material sampling and preparation procedures are well established in the aggregate industry, and these guidelines had been applied for sample preparation with the AIMS1 research. There was no need to reevaluate these material sampling procedures, and the appropriate existing procedures were followed for the ruggedness testing.
The AIMS2 system is computer-controlled with little operator input required to characterize material samples. As such, the ruggedness factors that were selected focused on gaining an understanding of the level of control required during system operation as well as during system calibration to provide a rugged testing process. Special accommodation was made in the test software to permit imputing variability into the system to simulate the acquisition parameter variation. These controls variations are not available during normal operation of the AIMS2.
Fine-aggregate characterization requires slightly different operating parameters than coarse-aggregate characterization requires. Separate tests and factors were selected for coarse and fine characterizations, although some factors were common to both (see table 1).
The ruggedness testing included several different materials. Separate analyses were run using dark-colored coarse aggregates, light-colored coarse aggregates, dark-colored fine aggregates, and light-colored fine aggregates. Two sizes of fine samples were selected (#30 and #16). Each of the characterizations specified within the ASTM C1067 procedure (two replicates of eight unique configurations) was statistically analyzed. Fine aggregate output includes angularity and form 2D. Coarse aggregate characterizations of angularity, texture, sphericity, and flat and elongated ratio were analyzed for factor significance. During the ruggedness testing, 10 experiments were conducted, which are detailed in appendix A, chapter 2.
Conclusions of Calibration and Ruggedness Testing
The results of the comparison between AIMS1 (the research system) and AIMS2 (the system developed within this project) proved that the two systems provide the same ranking of aggregates and give comparable results. Consequently, the numerous research studies utilizing the AIMS1 system are applicable to the AIMS2 characterizations.
The ruggedness study following ASTM C1067-00 identified several factors that were found to be statistically significant in affecting the AIMS2 results. The semitransparent doors originally selected were not able to minimize the impact of ambient light; therefore, nontransparent doors were installed to provide the required isolation.
An additional ruggedness study following ASTM E1169 identified several system control parameters that could affect the AIMS2 shape characterizations. Consequently, limits were established for these parameters to minimize their influence on the results. These parameter limits were set as system defaults where appropriate. The parameters and limits listed in table 2 are recommended as settings for the parameters to ensure rugged AIMS2 results.
The final phase of this project was to conduct the ILS to evaluate the repeatability and reproducibility of the system. This experiment followed ASTM C802, "Standard Practice for Conducting an Interlaboratory Test Program to Determine the Precision of Test Methods for Construction Materials." Three different material sources with varying mineralogy were selected, each type containing material sizes #200 (0.075 mm) to 25 mm retained.
The original plan for the ILS outlined and budgeted in the project proposal was for five AIMS2 systems to be used in the ILS, with one additional control unit at the TAMU laboratory. Due to the high level of interest in the AIMS2 technology, three additional systems were manufactured to accommodate all of the interested laboratories and still meet the project schedule. These eight systems were sent to the 32 laboratories that participated in the ILS.
This ILS provides two different precision estimates for the test method: a single-operator precision (within-laboratory precision) and multilaboratory precision (between-laboratory precision). The details of the study are presented in appendix A, chapter 3.
To ensure that the ILS materials were of uniform characteristics, the TAMU laboratory sampled and fractionated each material according to established procedures, separating each material into eight samples, one for each ILS system. Each of these eight samples (all sizes) was then scanned through a single AIMS2 system at TAMU. This system was not used to generate any of the ILS data set, but was used as a control. This procedure ensured uniform material samples and provided a baseline for the material characterizations, set by an experienced lab.
The labs participating in the ILS did not have experience with the AIMS2, nor did they initially have the necessary equipment. Each lab was provided instructions and a complete AIMS2 unit including the enclosure, microscope, computer, and sample trays. Also provided with each system were the three material samples of Superpave sieve sizes (#200 to 25 mm). These materials—a granite, a gravel, and a limestone mineralogy—were selected to provide a range of material characteristics.
Each system was shipped to and set up at multiple laboratories. Each lab, after setting up the system, checked system calibration to ensure proper operation. Each of the three material samples, which contained multiple aggregate sizes, was scanned twice by the same operator, with the second, replicate scan performed on a different day to provide within-lab system repeatability. Once the scans were completed (two scans of three material samples), the system was repacked and shipped to the next participating lab, where the process was repeated.
The results of the ILS provided repeatability and reproducibility information for different shape indices and parameters provided by the AIMS2. The data also suggested that the #200 (0.075 mm) retained analysis was more variable than the other sizes. This variability is attributed to touching particles influencing this size.
Additional work is necessary to improve the performance of the system for the 0.075-mm (ASTM #200 sieve) retained material due to multiple particles (touching) being analyzed as a single particle. The determination of an improved touching particle filter value (CHPR) for this size (0.075 mm) is expected to reduce the variations in the measurements and reduce the variation reported for this size. Because of this additional work, the 0.075-mm data are not included in the precision table recommendations.
The coefficient of variation should be used to describe the precision of the results to avoid bias against materials with low AIMS2 characterization values. Each AIMS2 shape parameter requires a separate precision value. Table 3 presents the recommended values for the precision table to be included within the analysis specification for the AIMS2 characterizations. (Additional detail on the study's precision results is available in appendix A, chapter 3.) Overall, the single-operator and multilaboratory precision results are considered acceptable coefficient of variation values given the natural variation in aggregate materials.
PDF files can be viewed with the Acrobat® Reader®