Q: We are supplying limestone to a project for which there is a requirement to provide testing prior to supply and also to test during the course of the job. We submitted the initial testing, but we’re afraid that if the next tests results come in lower than the initial, they might use that as grounds to reject the material. Is there a precedent for this?
A: What you are describing is a common omission from the specifications of projects with an ongoing quality control program of testing stone during supply. It is specified to perform the production testing, yet there is no clarification regarding the acceptance criteria of the production tests or the course of action to be taken if those test results fall short of the acceptance criteria.
A relatively new standard, ASTM C1799, Standard Guide to Dimension Stone Test Specimen Sampling and Preparation, suggests that production testing should be considered for any project that requires more than 10,000 ft² (900 m²) of stone. Some stone deposits are so massive and uniform that I probably wouldn’t spend the money for production testing, regardless of the size of the project. Some stone deposits are so variable that I would consider it even for a much smaller project. The 10,000 ft² benchmark in ASTM C1799 is merely a suggestion saying that for that volume of stone, there likely is considerable distance within the quarry from where the first block was harvested to where the last block was harvested, and the initial test data may not necessarily be representative of that stone which is supplied at the end of the job.
Where the problem occurs is that while many specifications require production testing, very few of them define what level of deviation from the initial tests is acceptable. Obviously, some variation between the initial test results and the production test results has to be acceptable, since we’re dealing with a natural product. A document that I reference very frequently, ASTM C1242, Standard Guide for Selection, Design, and Installation of Dimension Stone Attachment Systems, lists a table of recommended factors of safety, and also suggests that if the coefficient of variation (standard deviation divided by the mean) exceeds 20%, the factors of safety should be increased. That’s a bit confusing, since the table of suggested factors of safety is already tailored so that a higher factor of safety is recommended in those stone types would typically have higher COVs. Limiting all of them to a COV of 20% seem unfair to a stone type like travertine, for which ASTM C1242 has already recommended a high factor of safety of 8 due to typically large COVs in test results of those stones. In a 1988 paper by Bortz and Wonneberger, published in the Marble Institute’s Through the Ages, the authors included a table of recommended factors of safety based on COV in which the table went up to a maximum COV of 20%. So there seems to be some agreement between the ASTM document and the Bortz and Wonneberger document that 20% is perhaps the high end of acceptability for COV. Yet both of these documents are talking about variability within the results of a single set of test specimens or intra-lot variability. In the case of comparing production test results to initial test results, the inter-lot variability is the concern.
Factors of Safety are in place to cover the unknowns. Variability of stone strength, both intra-lot and inter-lot, are unknowns. By implementing a production testing program, the inter-lot variability is now known, so it seems reasonable that a lower factor of safety should now be acceptable since we have one less unknown value to guard against. Perhaps 20% is a reasonable allowable amount, in that a design which had a factor of safety of 5 using the initial test data, now has a factor of safety of 4 using the production data, which is more representative of the supply. If the project’s specification doesn’t prescribe what deviation is allowable, and what corrective action is required if the deviation is greater than the allowable, it will likely just become an argument as differing interpretations of an ambiguous spec are debated.