The time of tests
Led by Europe, more tests are being devised all the time to try to evaluate stone. In this column Barry Hunt explains the tests and discusses what the results show… and what they don’t. This time he discusses how many samples of stone need to be tested to provide a true picture.
Over the past year I have taken you through the main tests required by the various European product standards that are applicable to natural stone.
Testing is important but the way it is typically conducted renders much of it completely meaningless. I can hear the howls of derision, but the problem underscoring all testing of natural stone is the fact that it is rare for the samples to represent properly a resource.
Many stones exhibit a range of colours that reflect different weathering grades. Then there are features such as veining, cleavages, large fossils or inclusions, alteration effects, mineralization… the list goes on.
It might be argued that a separate test is necessary for each feature or change in stone quality. An alternative view would be to determine which is the lowest quality of the material that is acceptable and only test that.
All too often material has been tested and the results upon which decisions have been made prove to be erroneous.
The most common problem is the over-estimation of strength and with it the potential durability.
The introduction of the lower expected value calculation might lead to considerable under-estimation of the strength in some instances as a few samples with high variation can result in values approaching zero. Testing the different qualities separately results in smaller variation and thus better lower expected values even for the lower quality materials.
There are many ways to spin test figures, depending on how the testing is carried out and which materials are selected. So, wherever possible, keep a photographic record of the samples that have been tested.
Ideally, a geologist, and more specifically one who understands the issues surrounding natural stone, ought to select the samples. A sampling plan needs to be devised and this should include an attempt to predict potential future changes in quality.
The test standards mention the need for representative sampling, but none of them actually includes a method for selecting representative samples. This relies on individual interpretation, which cannot be avoided as every situation in the natural world is different.
With no standard way to determine just how samples should be selected and tested we are left with statistical analyses to confound us all.
Many specifications are written where sampling is carried out, say, every fifth block of stone quarried.
While this might generate a lot of potentially useful data, it might also completely miss important features that could result in failure.
Again, having a geologist look at the material carefully will provide far more information than scattergun testing. It will also cost much less and potentially identify potential issues the testing could easily miss.
I have long despised the scattergun approach but for some reason it often seems to give box-tickers the most comfort.
It is worth remembering that if you get the testing right, there should be no reason for you to do project specific testing, which in many instances is a truly colossal waste of resources.
When the European Construction Products Regulation kicks in next year, testing will be unavoidable. So, if you have to do it, make sure you get proper value for your money.