Maillard Profiles Phase 1

[aaronbeach]

Hi newbie here, got my KL about 2 weeks ago. Thanks so much for the profiles, in particular the Raost profile from which I started making my own.

I set out to create a set of profiles I could systematically use to explore certain roasting variables for new beans, in particular: roast level and the Dry-to-Maillard ratio, while holding other variables constant (e.g., DTR, Dev. Time). The idea is to be able to vary Roast Level (Dev Time, Temp and DTR) independent of the Maillard to Dry phase times and ratio.

I created 5 profiles that:
1. Hit FC at 6:00 (Dry + Maillard always take 6:00)
2. At roast level 3.0 DTR=20% and Dev Time=1:30
3. Support roast levels 1.0 - 5.0 with similar DTR/Dev Time at each roast level
4. I vary the Maillard % from 25%, 30%, 35%, 40%, 45% (Dry=55%, 50%, 45%, 40%, 35%)

Here is a screenshot of the 5 profiles compared:

Maillard levels.png (575.17 KiB) Viewed 3947 times

As you can see the profiles converge at FC and at Roast Level 3 (219.5C)
Roast level 1 is usually about 1:00 after FC and roast level 5 is usually about 2:30 after FC

The idea is that this gives me a 5x5 optimization space for new beans
I plan to test a 6 step process for finding the "best" roast for a new bean

M1 = 25% Maillard, M2=30%, M3=35%, M4=40%, M5=45%
R1 = Roast Level 1, R2=2, R3=3, R4=4, R5=5

Screen Shot 2023-05-24 at 2.42.37 PM.png (59.8 KiB) Viewed 3947 times

Sorry for the cryptic labeling, here's what it means:
1. Start with 5 roasts (A, B, C, D, E) and rate them on a scale of 1-10 (based on any criteria I choose)
2. Based on their ordering (best to worst) I roast 1 more for comparison*
* This logic hypothesizes that the search space is convex and does not have local maxima

Example: If A was the best and C,E were tied for second (A CE), I roast M3 at R2 and compare to A

My hypothesis is that by Day 7 (roast 7) I will be confident in the "optimal" roast level and Maillard/Dry ratio for a given bean. At that point I could tinker with roast total length, fan speeds, etc... or just enjoy it until I run out.

I started with Roast A last night and will report back the results here in a week. I wanted to post this to keep myself honest.

I attached the profiles. I wanted to keep them all as simple as "Raost" but had to make some tweaks to shape the high and low maillard roasts. Maybe I can simplify in the future. Also, they all have fan speeds adjusted to Dry Time, Preheat=66% of suggested and they all end with a -3 boost, KD=3, KP=1.73 starting just before FC because I've found it helps with jitter (I suspect these settings may be specific to my machine and environment). I wanted to keep these "adjustments" to a minimum for the test but will inevitably do some tweaking after the test. I'll post bean description with the results

Maillard 1.kpro

(1.69 KiB) Downloaded 154 times

Maillard 2.kpro

(1.69 KiB) Downloaded 166 times

Maillard 3.kpro

(1.69 KiB) Downloaded 170 times

Maillard 4.kpro

(1.68 KiB) Downloaded 149 times

Maillard 5.kpro

(1.69 KiB) Downloaded 181 times

[aaronbeach]

Also wanted to add, I roast ~82G batches (because this usually results in the amount I need for one pot of coffee, I am using the Breville Precision Brewer each day for these tests, not that it really matters as it assumes ratings are decided independently of the test). I will simply be rating them based on how much I (or my wife) likes them, I'll probably track notes for fruit, chocolate, sweetness, etc...

[rosemarycole6]

Crikey, you've certainly jumped down the rabbit hole Alice. The only problem is you can't control what the bean chooses to do

[aaronbeach]

So 4 roasts in, wanted to share some findings:

1. I understand this varies by coffee, but the long drying phase roasts seem to hit FC up to 30 seconds early (and about 6C lower) than expected while the long maillard (short dry) roasts seem to hit FC late and at a higher temperature. This could be important because I can only get the RoR steadily declining on the low Maillard roast up to roast level 3, but it seems it's actually closer to a roast level 5

[RupertPupkin]

Interesting approach indeed!
May I ask why you chose FC at 6 minutes?

[aaronbeach]

4-5 min FC profiles put a lot of pressure on the high Maillard curves since they have to start fast to quickly get through the Dry Phase (not impossible, but I wanted flexibility). And longer FC 7-8 usually hadn't tasted as good in the initial tests I did.

My goal is to incorporate roast length into the variables once I get comfortable with Maillard/Dry and Light/Dark roast levels

[aaronbeach]

Alright, the process seemed to work and brought up more questions than answers
I used this coffee: https://www.sweetmarias.com/ethiopia-dr ... -7369.html

I rated the coffees on a scale oof 1.0 - 10.0 (1.0-4.0 bad, 5.0-6.0 ok, 7.0-8.0 good, 9.0-10.0 great)
I want to spend most of my time drinking 7.0+, this coffee in these settings never got there

First I did the 5 tests (A,B,C,D,E)

Screen Shot 2023-06-01 at 3.34.02 PM.png (130.71 KiB) Viewed 3803 times

By the logic in the table the "best" was expected to be M5, R3 (Maillard 5, Roast 3) due to the rank B > A > E
I also tested the levels around it M4,R3 (A>B=E>DC) and M4,R4 (B>A>E=DC)

Screen Shot 2023-06-01 at 3.35.03 PM.png (130.98 KiB) Viewed 3803 times

Turns out the prediction was the best. This is definitely not what I would have predicted for this dry process Ethiopian (35% Dry, 45% Maillard, 20% Dev=1:30/219.4C)

log0027.png (83.29 KiB) Viewed 3803 times

It's basically a flat power profile. It's interesting that the strongest correlation with my preference was the Maillard % and the roast level far smaller correlation. Basically Maillard at 45% was better from roast levl 1.0 too 5.0 compared to any roast level with Maillard < 40%

Next I will be testing
a. Different beans (Columbian vs. Ethiopian Natural)
b. Length of Roast (FC=4:00 to 8:00)
c. If FC and Color Change occur at different temps for different Maillard profiles

I think my current goal is to match the Maillard and optimal roast time ranges to different types of beans (based on origin/altitude) so that I can quickly search (3-5 roasts) for the "optimal" when I get a new bean.

[fnq]

Hi aaronbeach

As previously mentioned ... you certainly have eagerly gone down the rabbit hole.

If I may ...... I think committing to a FC at 6 minutes irregardless of roast level or style is a bit of a false pillar to base your testing on.

Some of my good roasts ( ie taste good to me - brewed at my preference /style) have Fc near 6 minutes but plenty of my other good roasts for different beans do not. I would rather have a FC time slightly out of your ballpark time if it held a nice even power line rather than have a power line sharply powering on and then off etc

Not trying to rain on your parade, but i would also recommend

finding your preferred style / taste - eg one of mine is med to medium dark milk based espresso

finding a bean and a KL profile that has been recommended around your style ( honestly KL classic was a good starting point for me and the preference stated) and hopefully get a GOOD coffee quite quickly.

Roasting feverishly, cross testing etc for weeks and only scoring a max 7 /10 will get tiring real quick.....

[aaronbeach]

I like this idea - posting the next phase of my experiments which will vary FC time with Maillard/Dry ratio

[aaronbeach]

Added roast profiles with FC times from 4 min to 8 min here:
https://kaffelogic.com/community/viewto ... ?f=2&t=482