I've recently noticed my roast cool down times are quite large compared to earlier times. My cool down time on the Classic profile took just under 7 minutes the other day. So thinking around the issue I wondered if air humidity was having an effect? Today my house humidity is 57%; in the Winter its sits mid thirties with heating on.
I cannot seem to get roast-quality repeatability on my now very limited stock of Yirgacheffe. My Log #81 had an intense floral aroma straight after roasting but that has so far eluded me as a repeat.
So further, in my wondering, does our most Esteemed Designer (Ed!) have anything to say on the relationship between charge weight in the Nano and environmental humidity? I always dump 120g in but with higher humidity the air density is much lower so all fluid-bed calibration, surely, flies out the window?
(I seem to be hogging the board and don't really wish to - apologies!!)
Humidity
-
kaffelogic
- Site Admin
- Posts: 508
- Joined: Thu 18 Oct, 2018 9:38 am
- x 74
- Contact:
Re: Humidity
I'm not sure whether it's humidity or ambient temperature that is the greatest factor here, but you are definitely experiencing what we southerners call tropical conditions when cool down gets up to 7 mins. I'm not really sure that the Kaffelogic properly corrects for these differences. It attempts to estimate ambient air temperature from the probe temperature (this assumes the machine has been off for long enough to come to thermal equilibrium, and if you do several roasts in succession it keeps using the same estimate it made when you turned it on). It allows for differences in air density caused by temperature, but takes no account of either barometric air pressure or humidity. The result is that the initial heat applied at the very start of the roast can vary away from the ideal and the roast will not follow the curve until about 130 deg C. After that it locks on to the curve.
The theory is that all differences in humidity, ambient temperature, barometric air pressure, and bean moisture level all count for nothing provided the roast follows the curve. So once the roast is under way and the control software has got its number so to speak, the energy being provided is the right amount to follow the curve today, and therefore the right amount of energy to compensate for all of those variables as they are today. In short, we do not need to measure all of those variables in order to compensate for them.
That is all very well, but the part of the roast before about 130 deg C does not always follow the curve perfectly. This is due to the imperfections in the preheat calculation. In effect the roaster is playing catch-up over the first couple of minutes, and the more catching up it has to do the more effect there will be on flavour. It is easy to dismiss the very early part of the roast has having no effect because there are no chemical reactions under way yet, but the energy that drives the later part of the roast and the temperature gradient within each bean is being set up here. Improvements to the preheat calculation are in the pipeline (they are in the backlog
). A short term solution might be to tweak the preheat power setting, effectively creating a winter profile and a summer profile for the same roast.
It would be really great if you could post your log #81 side by side with a log that doesn't taste as good. I would really like to be able to compare the two.
(I'm more than happy for you to 'hog' the board. I'm enjoying the conversation and you are driving a number of improvements both with your questions and your suggestions.)
The theory is that all differences in humidity, ambient temperature, barometric air pressure, and bean moisture level all count for nothing provided the roast follows the curve. So once the roast is under way and the control software has got its number so to speak, the energy being provided is the right amount to follow the curve today, and therefore the right amount of energy to compensate for all of those variables as they are today. In short, we do not need to measure all of those variables in order to compensate for them.
That is all very well, but the part of the roast before about 130 deg C does not always follow the curve perfectly. This is due to the imperfections in the preheat calculation. In effect the roaster is playing catch-up over the first couple of minutes, and the more catching up it has to do the more effect there will be on flavour. It is easy to dismiss the very early part of the roast has having no effect because there are no chemical reactions under way yet, but the energy that drives the later part of the roast and the temperature gradient within each bean is being set up here. Improvements to the preheat calculation are in the pipeline (they are in the backlog
). A short term solution might be to tweak the preheat power setting, effectively creating a winter profile and a summer profile for the same roast.It would be really great if you could post your log #81 side by side with a log that doesn't taste as good. I would really like to be able to compare the two.
(I'm more than happy for you to 'hog' the board. I'm enjoying the conversation and you are driving a number of improvements both with your questions and your suggestions.)
Re: Humidity
Received wisdom is a funny thing. All that may be true for industrial drum roasters. Stefan Schenker in his Doctorate Thesis Investigations on the hot air roasting of coffee beans suggests otherwise for hot air roasting,The theory is that all differences in humidity, ambient temperature, barometric air pressure, and bean moisture level all count for nothing provided the roast follows the curve.
The humidity of hot air presents another important parameter influencing the roasting process. Figure 13 demonstrates significant differences during isothermal laboratory roasting in dry and humid hot air at the same temperature. RL and ORL progress slightly faster at elevated air humidity. Even dehydration runs slightly faster after an initial lag phase. It can be assumed that the heat transfer in humid air is more efficient due to its greater specific heat capacity. The effect of a smaller vapor pressure gradient between the beans and the humid air may be compensated by a faster temperature increase and therefore cause a faster progress of roasting. (Pdf p73 Doc p60). RL=Roast Loss and ORL=Organic Roast loss - dry matter loss, as percentages
To back that up I've had weight loss of 16.8% typically for Yirgacheffe up to 17.3 when the humidity was high. On the other hand the beans may have absorbed moisture. But they are in polybags and I squash air out on resealing.
But I was thinking more simply. Elsewhere you've said the airflow holds the beans on the fluid boundary just before chaos. I was imagining the lower moist air density would affect the boundary conditions so maybe lower mass in the chamber would help. As for humidity not affecting air density too much. Taking a PPL changed my mind on that one! Atmospheric moist air can rise to 10,000m! It may start off being warm but air's adiabatic lapse rate of ~ 2︒C per 300m soon makes it pretty cool; but it keeps on rising! (Take 5 molecules from Air; 4 Nitrogen, 1 Oxygen, atomic mass 144; take five molecules of water vapour, atomic mass 90. Whereas air density affected by temperature on the day 26︒C design temp 20 ︒C a ratio of 293/299. Let me not get carried away I must remember this is a home roaster!)
The logs don't show anything obvious to me. See what you think.
- Screenshot from 2019-08-12 13-05-33.png (85.78 KiB) Viewed 4837 times
- Screenshot from 2019-08-12 13-25-40.png (86.29 KiB) Viewed 4835 times
Log #81 was the good one with aroma after roasting; pleasant tea-like florals changing to cherry and hint of mint-chocolate and then nothing but then a heavy body slight chocolate as it aged. It responded best in a French Press initially.
Log #91 no floral aroma, slight chocolate but is is very young yet dated 10 August. Humidity 57%
Log#92 was a 90g charge weight. Same result as #91.
- Attachments
-
log0092.klog- (91.85 KiB) Downloaded 170 times
-
- log0091.klog
- (101.88 KiB) Downloaded 170 times
-
- log0081.klog
- (95.97 KiB) Downloaded 184 times
-
kaffelogic
- Site Admin
- Posts: 508
- Joined: Thu 18 Oct, 2018 9:38 am
- x 74
- Contact:
Re: Humidity
First comparing your #91 to #92 you can see how the smaller load causes faster heating for the first 30 secs or so, then the control system pulls them both back onto the line.
Apart from that difference all of your three roasts show remarkable consistency in curve following. So when you experience a significant difference in flavour there must be some other factor at play.
The first thing I would think about is freshness of the green beans. I know that the most fussy of the roasters I know consider that beans should be roasted within three months of harvest. I'm not sure how long there was between your #81 and #91 roast, but this is a factor that needs to be considered.
You have done some deep thinking about humidity. The paper you refer to talks about isothermic comparisons. However, with a roast control system the systems will not be isothermic. The roast control system should adjust the air temperature to give the desired bean surface temperature. Hence eliminating the effects of differing humidity. Of course the Kaffelogic does not measure pure bean surface temperature at its probe, there is a degree to which it is affected by air temperature, which is also dependent on air speed. This means that differing humidity could cause some differing bean surface temperature, but I contend that variations in humidity inside a habitable space are usually small enough to be safely ignored as a source of difference in bean surface temperature in a temperature controlled roast environment.
Bean moisture content is itself another factor that must be considered. This will have a significant effect on bean inner temperature whilst the roast system is controlling the bean surface temperature. This is one of the reasons that professional roasters will seek humidity controlled storage for green beans. However, even at a professional level it is often sufficient to manipulate temperature and ventilation, active humidification is not normally required. If the beans have been in your house for at least 4 days they should have come to equilibrium with your household conditions, provided you keep your house at a comfortable temperature (this isn't the case for us here in Dunedin where we only heat our kitchen in the evenings).
So I'm going to suggest that age of the beans and moisture content of the beans are the things to control for here. You can add to your data by measuring bean density prior to roasting. Just weighing a measured volume of beans gives you suitable data for comparisons and ratios.
There is also a wildcard here: the ineffability of coffee. Sometimes a batch will do something that cannot be replicated or explained. From a scientific point of view you will strive to replicate, but it is pretty difficult to do when the next batch you buy may have been grown in a different plantation and harvested at a different time.
Apart from that difference all of your three roasts show remarkable consistency in curve following. So when you experience a significant difference in flavour there must be some other factor at play.
The first thing I would think about is freshness of the green beans. I know that the most fussy of the roasters I know consider that beans should be roasted within three months of harvest. I'm not sure how long there was between your #81 and #91 roast, but this is a factor that needs to be considered.
You have done some deep thinking about humidity. The paper you refer to talks about isothermic comparisons. However, with a roast control system the systems will not be isothermic. The roast control system should adjust the air temperature to give the desired bean surface temperature. Hence eliminating the effects of differing humidity. Of course the Kaffelogic does not measure pure bean surface temperature at its probe, there is a degree to which it is affected by air temperature, which is also dependent on air speed. This means that differing humidity could cause some differing bean surface temperature, but I contend that variations in humidity inside a habitable space are usually small enough to be safely ignored as a source of difference in bean surface temperature in a temperature controlled roast environment.
Bean moisture content is itself another factor that must be considered. This will have a significant effect on bean inner temperature whilst the roast system is controlling the bean surface temperature. This is one of the reasons that professional roasters will seek humidity controlled storage for green beans. However, even at a professional level it is often sufficient to manipulate temperature and ventilation, active humidification is not normally required. If the beans have been in your house for at least 4 days they should have come to equilibrium with your household conditions, provided you keep your house at a comfortable temperature (this isn't the case for us here in Dunedin where we only heat our kitchen in the evenings).
So I'm going to suggest that age of the beans and moisture content of the beans are the things to control for here. You can add to your data by measuring bean density prior to roasting. Just weighing a measured volume of beans gives you suitable data for comparisons and ratios.
There is also a wildcard here: the ineffability of coffee. Sometimes a batch will do something that cannot be replicated or explained. From a scientific point of view you will strive to replicate, but it is pretty difficult to do when the next batch you buy may have been grown in a different plantation and harvested at a different time.
Re: Humidity
I bought two 1kg bags of those beans at the same time and yes log #91, ten days later, was from the second bag but I'm unconvinced that's the reason. A further 5Kg is coming!
I'm finding the less-than-stellar Yirgacheffe blend nicely with some Colombian. I actually achieved a coffee taste reminiscent of coffee flavour used in chocolates!
Single origin is fine but...
I'm finding the less-than-stellar Yirgacheffe blend nicely with some Colombian. I actually achieved a coffee taste reminiscent of coffee flavour used in chocolates!
Single origin is fine but...
Re: Humidity
I find similar, I enjoy Single Origin much with my own roast even from local roaster, but combining 2 or 3 kind in different ratio gives some new sensation. My past fav was Brazil or Mexican + Ethiopian, some case I also add Colombian to. I do mix in post roast just before grinding,
Las time i roasted Mexican from same bag it was about week apart and fist one (not first time from bag) lost way match weight compare to one week after. Same profile and similar DTR to stop. Log show not much difference. Taste wise quite different. I haven't conclude or find what caused this but must be on that particular run, Different beans on same day turns out good result. I'm still in trial and error stage...
Ben
Re: Humidity
error. I seem unable to delete this.
Last edited by Angela on Sat 17 Aug, 2019 10:01 pm, edited 2 times in total.
Re: Humidity
But that is the rub. Exactly that. The bean surface temperature is an unknown. It has some relationship to the air temperature the probe is reading, but that is all. You can't say when the air temperature is X then the beans will be at X-Y, because it seems to vary from bean variety to bean variety and from day to day- as evidenced by First Crack being an unknown in the Nano 7.kaffelogic wrote: ↑Thu 15 Aug, 2019 10:49 am The paper you refer to talks about isothermic comparisons. However, with a roast control system the systems will not be isothermic. The roast control system should adjust the air temperature to give the desired bean surface temperature.
In the paper referred, the author talks of the bean being a miniature chemical reactor cell. The conditions for the reaction seem to me be set up by; the bean outer temperature; the bean core temperature; the bean moisture content at that instant; the air pressure acting on the bean; humidity; the bean density and the bean chemical composition. I think the paper's author listed those.
What the paper suggested was that rate heat is given to beans by moist air is more than for dryer air.
The more humid air is just a better conductor, not any hotter. If more energy is being given to the bean at the air temperature X and the bean surface is X-y both surface and the bean core will result in being hotter than it would be with lesser rate of heat transfer. (little y is less than big Y used above.) All the nano 7 is doing is taking the the probe tip temperature along a profile curve pretty accurately. We have no real knowledge of the bean's temperature; we can only surmise.
In more humid air the temperature probe will be reading air/bean mix temperature X accurately, as always, but the surface bean temperature will likely be hotter because of better thermal transfer. It will be closer to X. The core bean temperature will also be higher. How much higher? I haven't a clue. Is a degree to two realistic? If so that is a level change.
You can argue because the probe is in contact with the beans it reads bean temperature. But it is also in contact with the air. That there is, on my system, a whopping 17 degrees difference between 196 (accepted FC temp) and my bean's FC probably indicates the probe reads closer to true air than true surface-bean temperature. I also suggest the relationship between air temperature and bean temperature is not a fixed linear relationship but a relationship that varies as the thermal transfer rate to the bean varies. In other words humid more conducting air changes profile curve following from a linear to a variable relationship with probe temperature.
It seems to me to get repeatable results, in so far as the beans allow, one would need dry air and a means of setting the pressure in the roast chamber to a constant. And that is not feasible for a home roaster.
For a home roaster we've probably got the best we are going to get though with the Nano 7. Although there may come a day when the Nano's 107'th iteration checks humidity and air pressure and adjusts the profile to suit.
In the meantime it may be best to roast to a slightly lower level on high humidity days if you want closer repeatability of a roast? I can see atmospheric pressure will have a bearing too and that is something else to untangle - any one else want to have a go - my brain hurts?
This thought exercise has been useful in that I think i have worked out that adjusting the level down a fraction may get me to my Yirgacheffe's best flavour point. Now to experiment... Again!
-
kaffelogic
- Site Admin
- Posts: 508
- Joined: Thu 18 Oct, 2018 9:38 am
- x 74
- Contact:
Re: Humidity
If we consider the four factors of ambient air temperature, pressure, humidity, plus bean moisture level, there is no roasting system that I am aware of that either controls or automatically adjusts for them. The difference between home roasting and professional roasting is that the professional may put more effort into managing the environment in which the roaster operates so as to reduce variation. Most often this is a matter of managing air temperature, especially for storage of beans. Once the setup is has become familiar the professional will make adjustments for these factors, much as you are doing Angela experimenting with adjusting the level down to get back the qualities you are looking for.
We do have a barometric air pressure, temperature, and humidity sensor planned as a future add-on for the Nano 7. Part of developing that will be the necessary research to understand how to automatically adjust the profile using that data. However, without that facility the Kaffelogic Nano 7 is in use as a professional sample roaster at a number of sites in New Zealand. These users are consistently praising the replicability of roasts, and using the Nano 7 to make commercial decisions on the basis of those samples.
That said, you have noticed a significant difference between roasts only 10 days apart. If you are able to make adjustments to bring back the original roast qualities, and can understand the factors behind those adjustments, I may be able to tune the Kaffelogic control system to go some way towards automatically adjusting. I'm looking forward to hearing your findings...
We do have a barometric air pressure, temperature, and humidity sensor planned as a future add-on for the Nano 7. Part of developing that will be the necessary research to understand how to automatically adjust the profile using that data. However, without that facility the Kaffelogic Nano 7 is in use as a professional sample roaster at a number of sites in New Zealand. These users are consistently praising the replicability of roasts, and using the Nano 7 to make commercial decisions on the basis of those samples.
That said, you have noticed a significant difference between roasts only 10 days apart. If you are able to make adjustments to bring back the original roast qualities, and can understand the factors behind those adjustments, I may be able to tune the Kaffelogic control system to go some way towards automatically adjusting. I'm looking forward to hearing your findings...
Re: Humidity
What rubbish you lot let me get away with! That makes little sense to me today!Angela wrote: ↑Sat 17 Aug, 2019 9:40 pm
In the meantime it may be best to roast to a slightly lower level on high humidity days if you want closer repeatability of a roast? I can see atmospheric pressure will have a bearing too and that is something else to untangle - any one else want to have a go - my brain hurts?
This thought exercise has been useful in that I think i have worked out that adjusting the level down a fraction may get me to my Yirgacheffe's best flavour point. Now to experiment... Again!
The feedback loop controls the temperature of the probe. At high humidity/ lower pressure it will just pump more Watts into the system.
What the feedback controller cannot do is control what happens with the reactions in the bean at differing external pressures though. And it is that, perhaps, causing my issues. NZ is in the cold Southern ocean with many high pressure, low temperature days. Whereas the UK is at the end of the Atlantic conveyor delivering one low-pressure system after another - or so it seems.