Any interest in a different roasting automated workflow?

[eoprede]

As I understand it, right now the roasting automation is effectively copying the previous roast and trying to repeat it.

I was looking for a slightly different workflow, where I can provide sort of set of desired steps for the roasting and the software executes it. Here's the sample profile (my BT thermometer is a bit off, so ignore the numbers):

Phase	Time	Heat	Fan	Advances on
Preheat	hold at 158°C	PID hold 158°C	manual	CHARGE
Charge	until TP	fixed 70% burner	30%	TP
Ramp 125°C	2 min	PID ramp to 125°C	45%	temp ≥ 125 or time up
Ramp 140°C	1 min	PID ramp to 140°C	55%	temp ≥ 140 or time up
Ramp 160°C	1:30	PID ramp to 160°C	65%	temp ≥ 160 or time up
Ramp 172°C	1 min	PID ramp to 172°C	75%	temp ≥ 172 or time up
Hold	until FC (5 min cap)	fixed 25% burner	75%	FC or timeout
FC → Drop	until DROP	fixed 20% burner	80%	DROP → workflow ends

The software then tries to reach the desired ramp temperature in the desired time (the build in PID is quite accurate, seems to hit it within a few seconds).

I have burned some tokens and have something that seems to work reasonably well for my needs.
What I am wondering - is this workflow something that may be of interest to others?
I am happy to clean up my work and make a PR, but I want to make sure there's an interest and developers are not against this approach to the workflow before burning tokens on this.

[MAKOMO]

This looks very much like how I roast using ramping replay of SV events. For this you need to use one of the custom sliders to model the SV events with an action to actually set the PIDs SV. The beauty of this is that you can modify the ramps while roasting by dragging them with the mouse. If you choose follow by time/BT and your PID is controlling the inlet temperature of your machine than you have what is known as ET/BT profiling. Works extremely well for me (max +-0.3 deviation)

https://www.youtube.com/watch?v=iUOO6tkkCbc

[eoprede]

Thank you, it looks like I have done something very similar indeed, the main difference is the way input is received (clicking in UI vs a JSON file). And I don't change the SV, just modulate power - but those are small details.
If you are interested, you can see what I did here: https://github.com/eoprede/artisan/tree/feature/workflow-automation
But it sounds like it's not substantially different from existing tools to warrant adding it upstream.

[wbj828]

It seems to me that we could improve the workflow by enhancing the roast designer. Right now, the PID can follow either the ET line or the BT lines. If we choose a particular ET profile, we don't know what that will do to the BT. If we choose a specific BT profile, we have no idea what ET we need to produce that line.

A big improvement would be to use the ET line to calculate the BT line. Newton's cooling law tells us how to do that.

$$\frac{dT}{dt} = -k(BT - ET)$$ or $$RoR = \frac{1}{\tau}(ET-BT) $$

Of course, the issue is that $$\tau$$ varies with time as well as roaster design and batch size. In practice, if you figure out what the tau curve looks like for a particular batch size on your roaster, you can reliably calculate the BT from the ET. In my app, I change the tau values based on BT at each 10F increment. Artisan can also do this calculation directly using RoR, BT and ET from an existing profile. (That makes for a very interesting and insightful curve).

This means that you can design a roast before you even warm up your roaster and have a pretty good idea of what the roast will actually do.

In my implementation, I have sliders for the Dry End, First Crack targets, Roast Length. The Dry End and First Crack are set at fixed temperatures but can slide in time. To design a roast, you decide on the duration, then slide the DE and FC markers to your desired "gates". Then you adjust the ET handles (which can move in time and temperature) to adjust the shape of the ET line. The BT line is calculated from the ET line and a set of tau values using the Newton equation. Also, by starting the BT curve at the current ambient temperature we also do away with the troublesome turning point issues.

Right now, my workflow is to design a profile in my app, export a CSV file in the artisan format, import that file into artisan and then use either the artisan PID, or my roaster's onboard PID to control the burner using the ET curve. The BT curve has no choice but to follow the physics based path.

I really like the ability to replay a roast that worked well. (that is what we have today). But even better is to be able to experiment with other roasting styles (ET isothermal for example) and see what happens to the BT.

Cheers,
bill