[Stoves] Replication of test results, until you get what you want

[Crispin Pemberton-Pigott]

Dear Friends

There is an interesting post and brief comments at the Alliance for Green Heat here

https://www.facebook.com/AllianceforGreenHeat/photos/a.128835288983.103745.120546153983/10155593812988984/?type=3&theater

The question addressed is why so many stoves are tested and get a PM emission rate of 4.4 grams per hour, just under the 4.5 g/hr EPA limit.

[cid:image001.png at 01D2F056.2EC67F20]

One of the answers is that with performance so variable, if one gets some results above the 4.5 limit, just keep re-testing until some of the results are below the limit, and accept those.

This raises an interesting question about what to do with the variability of test results, caused either by an inherent variability of the stove performance or the inability of the equipment to measure performance. There are different approaches taken by different countries on this point.

The issue is uncertainty. As has been discussed in detail by Fabio Riva in a couple of papers so far, the uncertainty of test results limits one's ability to make performance claims. Well, you can make a claim, but how uncertain is that claim?

A point made in the comments is that once manufacturers have a result that 'meets the 4.5 g limit' they are not interested in testing more to obtain a lower result. If the target were lowered to 4.2 additional testing would eventually yield a set of 'acceptable' results with an average of 4.1.

What is the variability in results? If you use % of value, Norbert Senf says in the comments it is ±35%. Interestingly, as the stoves get cleaner, the variability is larger and larger. This has implications for 'tiers of performance' because placing a stove on a tier requires one to claim that it is on one tier and not one or two others. The 'tighter' the performance requirement, the less clear it is that the target has been met.

Regards
Crispin

-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 73631 bytes
Desc: not available
URL: <http://lists.bioenergylists.org/pipermail/stoves_lists.bioenergylists.org/attachments/20170628/b2d3c3ac/attachment.bin>