Recently, there has been an issue in regards to aggregate size which did not fit into the envelope and failed to comply with the standard specification. The issue have serious implication as it affect the wearing course of the road and this can run as high as tens of million of Ringgit. Non-compliance mean the contractor will have to scrap the existing asphaltic course, delaying their work and borne the additional cost to reconstruct another layer of wearing course in order to complete the work satisfactorily.

2. Here, an engineer have to understand the two elements of supervision, which are; quality control (QC) and quality assurance (QA). QA is the a field of supervision where all issues are thoroughly reviewed to avoid failure or at least reduce the failure frequency to expected minimum. QC on the other hand is a scope of work to check work-in-progress or completed work based on samples in allowed sampling frame. QA will come into the picture once more when QC indicates there are continuous failures which need to be controlled and will cease from reoccur.

Therefore, I was tasked to probe into this issue as the contractor could not trace the crux to this problem. During initial discussion, I specifically reckon that the issue is due to human error that may occur at the laboratory. Their QAQC manager however opined that it was not a big issue due to sampling frame and sample size. A non-compliance is a non-compliance. Open up of works and investigation have to be conducted in order to conclude and resolve the root of cause.

3. Based on limits of allowance for particle size distribution non-compliance is relative low (via IRI test requirement) which is 0.5%, unlike concrete which allows 5% of the population to be defective or remain under the characteristic strength. Here, we have a highly stringent allowance to abide by. The magnitude of this issue is significantly high due to cost or weightage of the project and the high standard requirement(s).

4. After checking the hot bin and the batching processes, I found out that there is no problem with the system as all processes are automated. All systems are good and fully calibrated and no sign of disintegration after putting it to test. All gauges are functional, proper computation and no latency to the system.

Marshall Stability Test indicates acceptable result and complied to the design and trial mix. Therefore, viscosity is not the issue here and such test is redundant and futile. I do not regard the Bitumen Extraction Test to be indicative as hydrocarbon such as asphaltic concrete will lost its bitumen content (due to disintegration) when exposed to heat. It will be relatively lower by percentage when compare to the approved design mix.

5. I have proven that the issue of hot bin and batching to be non-existent. I came back to my initial thought and the only thing that may affect the sampling which is human error which is related to the duty of care and/or degree of care in executing PSD test. Although PSD test is considered as one of the most fundamental tests in geotechnical and road laboratory, there are high chances that sampling and processes may not be as adequately conducted as per BS 1377 Part 1. This means and reflect on the additional processes taken place as impartial and missing on elements which were not specifically mentioned or reflected in the standard specification.

6. Sample of particles was made available and I started to divide samples into two categories. QA sample conducted by the consultant and QC conducted by in-house laboratory with one exception; I will do the quartering instead of others. Samples are prepared into a tray and then poured into Riffle Sample Divider. This acts as a traceable action and as controlled samples versus samples (that were done traditionally by the in-house laboratory). Here, the crux of the issue shall be determined when PSD tests were completed and concluded.

7. After completing sieving for both QA samples, samples remain at the mean of the envelope perfectly which concludes that the sample I have taken randomly and the design mix is acceptable. QC samples complied with the requirement but leaning at the edge of the envelope for coarser grains and then normalized after 5mm aperture sieve.

8. From this particular exercise, the outcome is the exact prediction which I have projected initially before conducting investigation and traceability tests. It is human error and deterministic by nature. The failure is due to the way sample quartering were performed previously. It is not reckless but the lack of attention taken when handling dust or mineral fillers as well as the method of sampling. It took an engineer with statistics knowledge into call out such small predicament in sample handling and to refine the proper method when engaging in normal distribution sampling.

9. This exercise renders the importance of engineer to be very articulate when it comes to QAQC and additional knowledge would substantially help in troubleshooting issues. Lesson learned!