I noticed that earlier in the thread there was a reference to an experiment I did a few years ago and thought I'd provide a little more context on that.

First, the main concern that I had in doing this was production consistency in a small scale commercial roasting environment, so while a fairly large range was tested it's not too far out of the bounds of what I see as typical (if a bit careless) production variation. A lot of the people who will say that yes, absolutely there is a difference are suggesting far more extreme variations to test than you would ever have in a normal production environment. Under those circumstances it would be expected that there are differences, but I think it's a mistake take the difference between a 2 minute drying time (which I've never been able to achieve on any roaster I've worked on aside from that time I was using a sample roaster I wasn't used to and failed to turn on the external fan, 2 minute French Roast, not so good) and a 24 hour drying time and use that to suggest that there should be differences in more realistic circumstances. In both extremes you're probably going to have significant (but different) roast defects forming and the importance of water noted up thread is spot on, but in normal conditions the differences do not seem to be significant to the end result. It would be interesting to nail down where the thresholds at which this is no longer true are, but that's a matter of personal curiosity rather than a legitimate practical concern. Comparing years of production data before and after turning on profile translation during the roast, production is measurably more consistent when pretending that drying time is the same among different batches of the same coffee. It's possible that's just me getting better at what I do, but I don't think that can account for all of the difference and it would be quite the coincidence.

I wasn't the only roaster involved in this experiment. Other roasters have also done this both at the time and since, and I think the more interesting finding is the value of profile translation as a technique for comparing roasts and the usefulness of having that as a software feature to use during roasting. Some roasters attempts at replicating this experiment do show significant differences among test batches, but in every such case that I've been able to examine the data, profile translation reveals differences later in the roast in line with what you would expect to produce the observed cup differences. Even very experienced roasters tend to be both confident that they can roast with that early variation and keep the rest of the roast the same, and very bad at successfully executing that without technological assistance. Software isn't strictly needed. You could plot the roasts out on transparencies to shift things around. If you're going to say that there is an observable difference in the taste of the coffee, you really need to double check that the roasting data supports that, otherwise I think there's a good chance that the cause is being attributed to intended differences in the roast when they're really caused by unintended differences.

As to the implications for any scheme involving a percent of total roasting time, I do not know of any process in roasting that progresses in relation to total roasting time so on the face of it there is no reason to expect that to be true. I've been talking with some other roasters who have been doing some experiments along these lines and time spent in different temperature ranges seems to be much more useful than percent of total time.

Due to the difficulty of successfully matching roasts post-drying, I do, however, think that best practice should be to try to be as consistent as possible for the entire duration of the roast if you want production consistency.

N3Roaster wrote:I noticed that earlier in the thread there was a reference to an experiment I did a few years ago and thought I'd provide a little more context on that.

First, the main concern that I had in doing this was production consistency in a small scale commercial roasting environment, so while a fairly large range was tested it's not too far out of the bounds of what I see as typical (if a bit careless) production variation ...

Thanks for this post.

I was one of the people saying that drying time does affect flavor; but this was in reference to drying times that reduce the moisture in the bean to below a critical level (around 5% to 7% I think) I doubt any commercial production or sample roaster is prone to this, but early models of home drum roasters, underpowerd and overloaded, so they produced 25 minute plus roasting times were.

The more interesting question is this ... Green coffee can have between 10% and 15% content. The olive-yellow lower moisture beans typically get to the bright yellow, 310F temperature point of the roast a minute or two faster than very green, high moisture beans. Given these variations in original moisture content, slightly different drying times are perfectly reasonable. But do all beans have the same moisture content when they hit the bright yellow, 310F point, or is that still a variable?

That is an interesting question. Someone should measure it.