Sampling is one step of the overall measurement process. Its variability depends on the means you use. What it is important is the value (in the sense of money) you attached to your measurement and what is the financial impact on the overall measurement variability. Generally, this financial impact increases when the overall measurement variance increases. Conversely, the investment and operating costs of the measurement increases when the overall measurement variance decreases. So the money you have to spend for your measurement process (including sampling and analysis) has to be covered by the gain in the financial impact.

The choice of the sampling system and analytical devices has to be done to reach the desired variance. The best is to have the same magnitude of variability for analysis and sampling.

Perhaps you may find this article of interest. http://is.gd/AyhJeZ

It addresses the financial benefits of correct and representative sampling. This could answer your question regarding the value in investing in high quality sampling devices.

I don't think that the expenditure needs to correlate exactly. However, the amount of time and funds spent on system design, operation, and maintenance needs to be proportionate to understanding and addressing the challenges in the plant much as it is in the lab. After all, if the samples are not representative the money spent in the lab is wasted.

Thanks very much for the reference. My (semi-rhetorical) question was spurred by thoughts like last statement: "After all, if the samples are not representative the money spent in the lab is wasted" or another by Francis Pitard: "Your decisions are only as good as your samples".

I think the key is to ensure that we spend enough to sample:

Correctly
To reflect the certainty we wish to have in our results

The first is the hurdle that seems to be most common - like buying a dish washer and we "can't afford" to pay a plumber to connect the water line?

The second hurdle is more difficult to pre-empt in the mining game. It is easy to drill more holes in the resource but plant sampling may mean a completely new sample station to achieve the acceptable certainty.

The metallurgical´s sampling equipment are expensive if your intention is to process with the very best design of plant and the very best people.

If your ores are exotic minerals and your feeds of plant have lot variability, then the metallurgical sampling will be your best friend and your equipments important in your cost of the operation.

And if your intention is have not equipment of sampling, your intention of to drive the plant will be aleatory and undirected, this is very expensive.

Yes, the sampling equipment and to prepare the samples for wet chemistry and as well for XRF is definite not cheap, but real question for all of us within the mining, cement and metal industry is, if we make mistake at this stage we have to pay for it within the cost of operation. And that will be quite expensive.

One issue in this context is to make a cost study, where you compare the costs for a sampling system versus the cost for the "human sampling" when taking in account the possible production losses and the "unevenness/evenness" of the product quality as referred to above. The analyzing system (XRF, etc.) can then serve several sample flows by optimizing the analyzing sequences referring to process control requirements of each process step in the mining operations. This means also that the analyzing results can be fed automatically into the process control system. Savings in chemical consumptions, etc. will also be achieved.