barry wrote:the flat temp target was one of the 'givens' for the project. iirc, the test itself is suitable for other profiles as well, with a modified analysis. iow, run the same test protocol, and then determine deviation between successive data curves.
as far as the insanely quick terminal portion of the test, which is often criticized as "unrealistic", well... yeah. it was done that way intentionally, as a means of finding out what the machines were capable of. the analogy we thought of was taking a car for a test drive... anything will hit 55mph and run there all day long, but what reviewer with a test track and a helmet wouldn't "open 'er up" and find out "what she's got"?
--barry "100+F plus in here today"
The WBC protocol has very simplified analysis built into it that are reasonable for the first crack at looking at some machine's performance. But once you get some data you might wanna change the way you look at it. For example, we know now that lots of machines exhibit humps in the shot temperature profile and it would be very interesting to look at the ability of machines to reproduce the humps. In doing so, you might want to examine the brew temperature measured at a specific time during the shot and compare it to the average test temperatures taken at that elapsed time during the 14 test points. this would give you reproducibility information independent of the temperature profile. The problem is that the analysis isn't real obvious to folks who don't do this stuff a lot, and so people will screw it up, or not be able to do it, or draw more incorrect conclusions than are already drawn.
Even as written, the WBC protocol is pretty useful. It gives a systematic test method that demonstrates baseline machine performance in the absence of machine-specific group flushing rituals. And that includes LM rituals, too. This is an entirely reasonable test if one assumes no knowledge of how a specific machine is gonna work. Once the behavior is know, then it is a pretty simple matter of modifying the flush routine in order to fix the problems that are exposed.
Now I'm gonna pick on Mike McCoffee because he's given us a good example. The first few data sets show us a machine that idles at 10 degrees F higher than is remotely reasonable. That should be very interesting to Mike because it shows that this machine's thermosyphon loop is maintaining intermittent duty temperature at a very high value - a problem if he is actually running the thing at only 1 bar of boiler pressure. The fact that the machine still operates at high temperature in constant duty conditions of the later data sets is both good and bad. The good thing is that the machine can actually keep up. The bad thing is that the temperature is still way too hot. Now that Mike knows how hot the machine runs without flushing, he has to develop a flushing ritual that reduces temperature to a reasonable value time after time. I'd suggest he try a flushing ritual, modify the WBC procedure by incorporating the flush method into the test, and repeat the series. He'll soon know what is required for group flushing, and he'll know whether it works over a variety of duty cycles.
If I was the manufacturer of the machine, I'd be pretty unhappy and I'd be thinkin about how to get that brew temperature down while holding boiler pressure at a reasonable pressure for steaming. As it is, the WBC protocol shows Mike's bric to be a pretty crappy performer. Ken's Cimbali does much better and the Simonelli Aurelia shows that sub-1 degree temperature reproducibility is entirely possible with a hx machine. And you might say "well what does that matter in the cup?" My response is that if you question the result, then try it for yourself. Tune your machine to produce 214 degree water and make some coffee. Now tune it for 200. But don't use that argument to throw up a smokescreen.
So you can diss the WBC test for what it doesn't do. But don't lose sight of the usefulness of the test. Even if it ain't perfect there's still a lot of useful information to be gained from it.
-Greg
