Evaluation of the Repeat Load Triaxial Test and its Potential for Classifying Basecourse Aggregates

Transcription

1 Evaluation of the Repeat Load Triaxial Test and its Potential for Classifying Basecourse Introduction The Repeat Load Triaxial (RLT) test is starting to gain favour in New Zealand as a method predicting the in service performance of Basecourse aggregates. It is hoped that adoption of the test in NZ will enable increased confidence of the in service performance of existing basecourse materials whilst also allowing greater opportunity to develop high performing alternative or modified basecourse materials that may not conform to the existing specifications. In response to this Winstone (WA) undertook a six month testing programme to try to better understand both the RLT test and a comparative look at how individual basecourses performed. To achieve this WA agreed to lease the RLT equipment from PaveSpec Ltd, the Company formed by Dr Greg Arnold, formerly of Transit NZ, who has spent many years studying pavement aggregates and the application of RLT testing. The equipment was set up at WA s IANZ accredited Auckland Laboratory in November 2006 under the watchful eye of Dr Arnold. Over 85 RLT tests were completed during the period of November 2006 to May 2007 covering a wide range of natural, modified and alternative basecourse materials. This paper discusses some preliminary findings specifically related to the suitability of the RLT test for classifying basecourse aggregates and highlights areas of further work. 1

2 Background Repeat Load Triaxial (RLT) testing is not by any means a new test. However it has been recently been identified and developed as a test to simulate and determine the effect of repeated heavy traffic on basecourse aggregate. Therefore RLT testing has the potential to effectively identify those aggregates more suitable for high trafficked roads. A lot of the early development work was carried out by Dr Arnold and Transit NZ (Arnold 2004). Currently TNZ M/4 is the prescribed basecourse specification for high trafficked roads, this is an empirical, recipe based specification and there have been concerns as to the performance variability of basecourses conforming to the specification. RLT is a test that could, in conjunction with source and production testing, give further confidence of the in service performance of a basecourse. RLT testing has been developed with the aim of providing an aggregate rutting resistance measure, as detailed by Arnold 2003, The RLT apparatus (Fig 1) applies repetitive loading on cylindrical materials for a range of specified stress conditions, the output is deformation (shortening of the cylindrical sample) versus number of load cycles (usually 50,000) for a particular set of stress conditions. Multi-stage RLT tests are used to obtain deformation curves for a range of stress conditions to develop models for predicting rutting. Fig1: PaveSpec Ltd RLT equipment 2

3 Objectives The objectives of the preliminary work discussed in this paper were to investigate some of the key test parameters and understand their influence on the RLT results. There are many potential influences on the RLT test results for the purpose of this paper the focus will be on: Operator Sensitivity Repeatability Compaction Water Content Each of these will be discussed and preliminary findings presented. Procedure A Kanga 950KV Vibrating Hammer with a 146mm diameter compaction foot was used to compact the samples (fig2). A sample preparation calculator was used to achieve target compaction density by outputting the correct mass of sample material to compact to a specific height in the split mould (fig3). Compaction effort and unusual observations were recorded and explained in each individual test report. Fig:2 Fig:3 Fig:4 Fig:5 The compacted sample (fig4) is removed from the split mould, and a sealing membrane is fitted (fig5). The membrane gives an air tight seal enabling loads to be applied by means 3

4 of air pressure. The sample is mounted in the pressure cell(fig6). The cell is then sealed on to the base. The pressure cell is aligned underneath the vertical load actuator and stand(fig7). The vertical load actuator and pressure cell are both powered by compressed air, and controlled by the computer software UTS017 Cyclic load trial. This software is installed on a late model Windows PC. The physical link between the instruments and software is through a signals box(fig8). The vertical actuator applies a cyclic load to the test sample at a rate of 5Hz throughout the duration of the test. A small buffer tank(fig9) feeds compressed air directly to the apparatus. This reduces the effect caused by variable air demand on the pressure. Fig:6 Fig:7 Fig:8 Fig:9 The stresses used in each stage are used and recommended by Dr Greg Arnold. The stresses are believed to be the most representative combination in the appropriate resilient modulus envelope. 4

5 Fig 10 below displays stresses applies in all tests. Stage Confining (kpa) Deviator (kpa) Fig10:The Six Stress Stages It is recommended that RLT test samples should be compacted to a target Maximum Dry Density (MDD) of 95% and target Optimum Moisture Content (OMC) of 100%. To achieve this a sample moisture content is measured and the correct moisture to be added or evaporated is calculated by weight. A simple sample preparation calculator (fig11) was developed to help streamline this process and reduce errors. Material Description: M.D.D O.M.C. (%) Received M.C Density percentage Target Density (kg/m3) Sample Height (m) Sample volume (10^-3) Mass of Dry (kg) Mass of H2O (kg) Mass Received (kg) H2O to Add (kg) Total Weight (kg) Weight for each Layer (kg) Mass Before Oven (kg) Mass After Oven (kg) Fig11 :Sample Preparation Calculator 5

6 Preliminary Results To achieve actual rut depth performance of a sample, results data is run through finite element analysis software developed by Dr Arnold. The magnitude of rut deformation can then be determined. It is still possible for the raw data to be analysed and comparisons can be drawn however it is more difficult to determine magnitudes. A spreadsheet results template was created in to which the raw RLT data can be entered. The template adjusts the raw data in relation to other data collected, accumulates and uses regression analysis as a means of comparing the data. The ouput is an accumulative strain plot (fig12) and results table(fig13). These were created and documented for every individual test, and this allowed a uniform method for displaying test data. Accumulative Strain Strain (mm) Stage 1 Stage 2 Stage 6 Stage 5 Stage 4 Stage Load cycles Fig12 : Accumulative Strain Plot Acc first 25k Deformation Mag Average Resilient Modulus after 25k Slope after Stage 25K Fig13 : Results Table 6

7 WA commissioned Dr Arnold to analyse results of the testing through the finite element model he has developed, these results are measured in terms of, million equivalent Standard Axles (ESAs) to 10mm rut within aggregate, making it easier to interpret the results. Operator Sensistivity During the course of the testing programme WA had two operators (fig14) conducting the RLT testing. Fig14 : Geoff and Adrian At the outset it was identified that there was no formal procedure for conducting RLT testing. This could have led to inconsistency between the two operators and to avoid this Geoff Moore (WA) wrote a procedure for carrying out the RLT testing that was adopted for the duration of the testing programme. The procedure was strictly followed for every sample tested and this consistency of approach effectively minimized the influence on final results of different operators. Repeatability RLT repeatability was looked at over a 3 month period, incorporating 5 independent production runs of TNZ M/4 produced from a North Island greywacke. 7

8 Although the results show some variability between samples, at this stage in the evaluation of the test the emphasis is not so much on the actual result but more on the ability for the test to predict the consistency with which the basecourse will meet the performance required of it. One way of achieving this is to create performance bands in which to rank suitability of materials for certain applications. The figure which has been suggested for the upper band is > 10million ESAs and therefore on the results analysed in this paper the test would appear to be able to measure a material relatively consistently for this purpose. Sample No Date of 23/02/07 13/03/07 3/04/07 10/04/07 21/05/07 Test Volume in Run (m3) Million ESAs Source Rock: Greywacke Basecourse specification: TNZ M/4 Repeatability 14 Million ESAs to 10mm rut within Aggregate Sample No Careful selection of source rock and a high level of production control through the manufacturing process is reflected in the consistency of the material produced. This consistency is seen through both the standard TNZ M/4 source and production tests, and 8

9 it would appear that the RLT test is comparable, in terms of predicting consistency of material. There are many factors that could have influenced the individual results and further work will be undertaken to investigate the differences between the 5 samples, as well as increasing the sample population with further results that have been tested but not yet analysed. This paper does not consider repeatability between different Laboratories, but proposes that regular inter-lab testing should be conducted to establish the degree of variability between laboratories. Compaction Material was sampled from the stockpile and split to enable 3 RLT samples to be compacted as shown in table below. Sample Number % of Compaction Million ESAs 1 Under Compaction -15% of Target Target Compaction Target = 95% of MDD Over Compaction + 15% of Target

10 Influence of Compaction Million ESAs to 10mm rut within Aggregate Under Target Over Degree of Compaction As can be seen from the graph, the under compacted sample had a huge reduction in predicted life. This changed the performance of a high quality TNZ M/4 and shows the importance of achieving full target compaction. The relationship between compaction and performance is critical and it is vital that attention is paid to the compaction of the RLT samples in the lab for valid results to be achieved. This needs to be investigated further, especially when looking at test repeatability between Laboratories as the setups and equipment for vibrating compaction carried out to NZS 4402 Test 4.1.3, NZ Vibrating hammer test, do vary as highlighted by Frobel & Moulding (2006). The work to date shows that almost regardless of the high quality and consistency of the basecourse produced at the quarry, under compaction on site will result in a hugely reduced performance. The over compacted sample showed improved performance, however not to the same magnitude above target and there was significant extra compactive effort required to achieve the +15%. Water Content Previous work has highlighted the influence of water as having a significant effect on basecourse performance. 10

11 Work conducted by Arnold (2003) has shown that increasing the water content reduces the predicted performance of the basecourse and has led to the proposal that some basecourses may only be suitable for use in dry conditions. The preliminary findings support the previous work and water above OMC does have a negative influence on performance as expected and this is shown in the graph. However, RLT testing conducted on a non plastic TNZ M/4 for the degree of saturation tested to have less influence than was thought on the basecourse performance. This supports the use of performance testing to categorise individual materials on their merits rather than the empirical approach adopted by specifications such as TNZ M/4. Influence of Water 14 Million ESAs to 10mm rut within Aggregate Dry Wet Degree of Saturation Further tests, covering more source rock types and including Gap40 products have been conducted and analysis of these results will add more understanding to the effect and sensitivity of water on individual basecourse performance. 11

12 Conclusions Further research is required on the topics discussed in this paper to establish the validity of these preliminary findings and relate them to a wider range of basecourse materials. The findings presented show that the RLT test is capable, in conjunction with other testing, of characterising basecourses and appears to have real potential in enabling predicted performance and therefore classification of basecourse materials. RLT testing has allowed WA to further develop knowledge of the performance characteristics of the existing TNZ M/4 aggregates and Gap products. It has also given the opportunity to investigate new materials as alternative basecourses and work is continuing on evaluating and developing these products. The RLT test could be specified in projects to allow aggregate producers to manufacture the most suitable product that will better meet the performance requirements. Winstone focus is primarily on understanding and improving the performance of our basecourse aggregates to ensure higher quality, better performing roads. 12

13 References Arnold, G. (2004). Rutting of Granular Pavements. PhD. University of Nottingham, England, UK. Arnold, G. & Werkmeister, S. (2006). Performance tests for selecting aggregate for roads report on progress. Butkus, F. (2004, December). Reid highway basecourse test sections construction details and performance to November (Volume 1), Pavements Engineering report No. 2004/17 m, Main Roads, Western Australia. Butkus, F. and Lee (1997). Pavement moduli project, a review of repeated load triaxial test result.s Materials & pavement technology engineering report. No 97/4M, Main Roads, Western Australia. Frobel, T. and Moulding, S. (2006) Errors in vibrating hammer compaction test. Inaugural Civil Engineering Laboratories Conference Moore, G. (2007) Repeat Load Triaxial setup and implementation. Winstone Transit NZ (2006) Specification for basecourse aggregate (TNZ M/4). Transit New Zealand, Wellington, New Zealand 13