Los Angeles Abrasion Result Outside Tolerance
A technician runs a Los Angeles abrasion test (AASHTO T 96, Grading B) on a crushed gravel source. The sample is prepared from material passing the 25 mm and retained on the 1.9 mm sieves, and the mass of the test sample before testing is 5,005 g. After 500 revolutions with the standard charge of 11 spheres, the retained mass on the 1.7 mm sieve is 4,180 g. The computed loss is 16.5%, which passes the 40% maximum. A second technician repeats the test on a split of the same sample and gets 21.2% loss. The two results differ by 4.7 percentage points, and the specification requires reporting a single value. The project engineer asks which result to use and whether the test needs to be rerun.
1. Verify the percent loss calculation from the first test and assess whether the two results are within expected single-operator variability.
Read the worked answer
Percent loss = (initial mass − retained mass)/initial mass × 100 = (5,005 − 4,180)/5,005 × 100 = 825/5,005 × 100 = 16.5%. The arithmetic checks. The two results, 16.5% and 21.2%, differ by 4.7 percentage points. For AASHTO T 96, single-operator precision is typically on the order of about 2 percentage points (multi-laboratory somewhat larger); a 4.7-point spread between two runs on a split of the same sample exceeds expected variability. That points to a preparation or procedural difference rather than true material variation. The technician should not simply average the two; the cause of the discrepancy must be found, because averaging hides an error.
2. What procedural factors most likely explain the discrepancy, and what should be done before reporting a value?
Read the worked answer
Likely causes include: (1) the charge - wrong number, mass, or type of spheres, or worn spheres; (2) drum speed or revolution count not at the specified 30–33 rpm for exactly 500 revolutions; (3) sample grading - the mass retained on each sieve in Grading B must match the specified proportions, and a poorly split sample changes the coarse/fine balance and thus the loss; (4) initial mass not at the specified 5,000 ± 10 g; (5) material not washed and dried to constant mass before testing, so adhering fines inflate or deflate the loss. The correct action is to inspect the machine charge and counter, verify the grading of the split, and rerun on a properly prepared split. The reported value should come from a valid run, not an average of a valid and an invalid one. If both reruns agree within tolerance, report that value; if not, investigate the apparatus.
3. If the true loss were near the 40% limit, why does the choice between 16.5% and 21.2% matter less than the choice between a valid and invalid run?
Read the worked answer
Both 16.5% and 21.2% pass a 40% maximum, so for this specification the acceptance decision is the same either way. The real risk is that an out-of-control test process could produce a result on the wrong side of the limit on a different sample. The discrepancy signals that the lab's T 96 process is not repeatable, so future results near the limit cannot be trusted. The correct judgement is to fix the process (charge, speed, grading, preparation) and document the investigation, rather than to pick the more favorable number. Acceptance should rest on a valid, repeatable test, and near-limit results should trigger confirmation testing.