Today in the UK we have unusually high atmospheric pressure. In my region presently it is 1049 mb
I have just completed a roast to exactly the same settings as a roast from a couple of weeks ago when air pressure was much lower. Same beans batch; same everything else.
Screenshot from 2020-01-20 11-41-06.png (23.99 KiB) Viewed 4226 times
The roast finished about 8 seconds earlier than previously but the cool-down was 30 seconds earlier. The cool-down gives a spectacular insight into the difference air density makes for air to bean temperature transfer without any system control in play.
First crack was 4C later today at 214C and the roast finished 8 seconds earlier. Both these factors throw development-time out with the bath water.
How do I make the Nano 7 compensate? Or am I confined to only roast on bog-standard 1013 mb days?
Quite a lot of factors here, but it is interesting to see the thinner air (lower pressure) showing the roast 'getting away' from the algorithm at around 180 deg, and the heavier air (higher pressure) sticking better. This is a small effect, not enough to invalidate your DTR calculations I would say. It could be caused by random effects, there being a certain amount of chaotic behaviour in the circulating bean mass. More data would give more confidence. But we must expect heavier air to be less responsive to exothermic heat, thereby giving the control system more time to react and correct. We could theoretically compensate by increasing the Kp and Kd gains for lower pressure air. With a bit of simple physics the ratios should be easy enough to estimate, although it might not work out exactly as expected due to there being many other factors in play. This is definitely an aspect of R&D once we have the expansion module that includes a barometric air pressure sensor, and just having this conversation helps reinforce for me the need to ensure that a barometric air sensor is part of that module.
Looking at the difference in roast end times, I suspect that a small 'flick' in the second roast caused an early end. Perhaps for your profile an increase in 'roast end by time ratio' would help mitigate against this, maybe increase to .66?
Differences in total cool down time are not so relevant - it's the first few seconds of cooling that count the most and the cooling rates are almost identical regardless of air pressure.
Thinking about the difference of 4 deg C in FC temperature.
One thing to think about is that detection of first crack is a statistical exercise. If each bean has its own moment of cracking, then first crack is that period of time during which we notice that beans are cracking. If we assume (dangerously, perhaps) that the cracks are going to form a normal distribution in time, and we count FC from the first 3 cracks out of about 600 beans in the batch, then we are at roughly the third standard deviation from the mean when we call FC start. So we might expect some statistical variation?
However, we know that probe temperature is affected by air temperature such that the probe sits maybe 5 to 15 deg above true bean surface temperature. If you think about the air entering the bean mass and giving up some, but not all of its heat to the beans, then the air temperature as measured by the probe depends on how much heat has been given up. Denser air at the same temperature in contact with the same beans will give up the same amount of heat (perhaps this is another assumption that can be challenged), but as it is denser it starts off with more heat. Therefore the denser air stays warmer and the probe stays further above the bean surface temperature than thinner air. So if first crack happens when the bean is at a certain temperature, the probe will read higher if the air is denser - according to you by 4 deg C for the pressure difference you were experiencing.
This means that at higher air pressures the roaster is essentially following a slightly different profile. You could adjust the profile in Studio with the appropriate transformation (say multiply the temperature by 1.02) but you would need to be confident in your figure of +4 deg C for FC. In this way you could make a high pressure and a low pressure profile.
Thanks Chris, interesting analysis. I pointed out cooling simply to illustrate how obvious it is air density affects heat transfer.
The first crack, for me, is the moment when the fight within the bean to burst against the constraint of the external air pressure is won. Higher air pressure means the external forces on the bean are higher. The bean needs more energy to bust, thus it bursts later at a higher temperature.
I think what you are saying is the dense air has a higher specific heat capacity per given volume (not mass) affecting probe and bean. We've been here before and we can agree on that.
The temperature kick at roast end didn't worry me too much as it gave a good roast result that I was hoping to repeat. But I'll try adding to the 'roast end by time ratio'. Thanks.
Did you enjoy your holiday?? This forum was dead for two weeks!
Thanks, yes enjoyed the holiday very much. The West Coast of NZ is always a special place to visit.
I hadn't thought of the air pressure difference changing the crack temperature of the bean. It must do so, but I think that air density differences would dominate. Plenty of scope here for research though.
Well, as far as coffee roasting goes, air pressure and air density are really much the same thing; one is just an analogue of the other if other factors remain constant. Air density affects heat transfer and air pressure affects bean cracking points. But air pressure (and temperature and relative humidity) contribute to air density.
It would be helpful to know what to do with a profile if 'air pressure' is higher or lower than average by a significant amount on a roast day. Professional drum roasters must make compensations - so what would we need to do?
Perhaps, instead of waiting for hardware, moving towards a manual input method of an air pressure value into the machine for software to consider, might be the way to go before you go to the barometer fix? It could be a menu item that calls up 1013mb and the + and - buttons could be used to adjust to the day's value? My phone has a weather app that gives me local barometric readings.
Really my quest is for accurate repeatability of a roast.
Ah, the Wet Coast! 7 metres of annual rainfall. Miles of black sand and no-one about. Is is still there? I thought it all got washed away in the floods.
