BBQ Logging and Brisket Analytics

toddcc1

Well-known member
Joined
Feb 28, 2015
Location
Monument...
I think it's great to see all the "1st Brisket" cook questions. The reply information is even better and fairly consistent. Rather than reply to each one, I thought I'd begin a thread discussing the entire process of logging your cooks. Searching "log" turns up quite a variety, so I'll go with "logging".

I'd like to share the importance of logging your cooks. I too learned via the threads, but also by a tremendous amount of trial and error. Every time I do a brisket, I share that it's about a $1,500 dollar piece of meat, as that's how much brisket was ruined trying to get it right. I also logged each and every cook. I'm a data scientist, so the data was of interest to me. Every hour I logged the cooker temp, probe at grate temp and middle of flat temp. I also logged if I spritzed, added wood and what grate the meat was on. After 15 brisket specific cooks, I finally was able to repeat with extreme accuracy the duration, times and taste that was acceptable to me and more importantly to others. Now, I'm no judge, but I think I know when I've ruined a brisket.

What I've learned:

The quality of the meat matters more than I originally thought. After 4 cooks, I sought out and established a relationship with a local distributor that carried Creekstone Farms briskets. Not that I advocate them over any other, but you can see the coloring difference in CF briskets vs. the CAB briskets I had been using. At this point I made a choice to only use CF briskets. It helped that I was getting them at about $4/lb. direct from the distributor. Even within the CF brisket boxes they had, I began to find the ones with the thickest fat cap on the flat and would test them by bending them from end to end, checking for pliability. If there was any strain in bending them, I wouldn't select it. I also became sensitive to the package date. If it was 50 days, I would age it another 8 in my own fridge, prior to the cook.

Rubs:
Like many, I've bought so many, mixed them up, added some from others, modified to the point that I've come full circle and now only use salt, pepper and seasoning salt. Nothing more, nothing less. I don't slather with mustard, pickle juice or anything else. I do score the fat cap, but make sure the rub doesn't "stuff" in there. After the rub, I wrap tightly and let rest in the fridge for about 10 hours.

Spritz:
I do spritz. Dale's Marinade. I've used mixtures of all recommendations and one day saw it and decided to try it. It's not too salty, but adds a great beef flavor to the brisket without overcomplicating the taste. I dilute it down and it's all I use now. I spritz every hour, on the hour.

Temps:
I've ruined more briskets due to poor temp control and eventually became comfortable with my cooks and smoker to deal with maintaining a constant temp. I start at 190 degrees and hold between that and 200, until the brisket reaches 160 degrees, I then ramp the smoker up to 230 degrees and cook until it's finished. I do not wrap.

Rest:
I lay the brisket out on the counter, put my temp pen in the middle of the flat and let it sit until it reaches 160 degrees.

What the data tells me about being done:

For the 15 briskets:

Mean: 200 degrees
Standard Deviation: 2 degrees

Though, I would be a poor data scientist if I didn't share that 30 samples is considered adequate. But, with modern statistical programming, we're able to create a data sample on 100 cooks, using the data from the 15.

What is it saying?
Well, based upon 100 cooks, with 95% certainty, a brisket with a mean weight of 12.6 lbs, will be finished cooking between 201 and 202 degrees. So, when someone says "it's done when it's done", that's actually not accurate. Though, there will be 5% of the cooks that will be outside of that temperature. If we want to be super accurate, 2.5% will be finished above 202 degrees and 2.5% will be finished below 201 degrees.


y3m7pfy9eTHFtqwazN4N7FhE0JZR1bd8vsEhN0vpGbmUciQjs5YXe6ztcATjuD8LqBFGOjnv2o5fH7iXnu5Hl2NPl9pjctC-jebXuivuotDsFsmGTM_rJI5ZjejVCXrtFQD4Vk6E8YzdfxRrTjbjbD5zcf_MISOTuvNE4iieMNjmes


That has been my experience and hopefully some newer aspiring BBQ cooks find it helpful.
 
I've seen advice on keeping logs. I do it- but only with a new cooker when I'm trying to see how it performs- being they are all a little different. Lot of folks "log their cooks" - I'm guessing after a while they figure out what they are doing -stop with the data collection and just roll with it, enjoying the process instead.

I sure hope so anyway.

be well.
 
Nice post and I do appreciate it!!!!

Soon as I make my decision for my new smoker purchase, I'll be trying out some of this and others I've read.

Again, Thank you!!
 
If I was in the business of furnishing grub to customers, I'd be mighty grateful for that information.

I don't smoke briskets but 2 or 3 times a year. I do a whole lot more butts and yardbirds.

It's been decades since I took statistics (back when the smallest calculators were the size of a large cash register); so please bear with me. I also understand that you select for size before starting out, but I'd be curious to know if your data can be extrapolated to larger briskets and, if so, how?

Your sample size information is news to me. Back when I was doing data collection and analysis we went to great extremes to randomize sample selection and to get as many as we could. A sample of 30 if properly random is statistically valid, but my profs wanted more data - probably just sweatin' us a bit though.
 
As primarily a competition cook these day, keeping good logs is essential for reproducibility and consistent results.

I really can't argue with any of your data, but will take issue with your conculsion...

Well, based upon 100 cooks, with 95% certainty, a brisket with a mean weight of 12.6 lbs, will be finished cooking between 201 and 202 degrees. So, when someone says "it's done when it's done", that's actually not accurate. Though, there will be 5% of the cooks that will be outside of that temperature. If we want to be super accurate, 2.5% will be finished above 202 degrees and 2.5% will be finished below 201 degrees.

I have no doubt that this is true for your cooker (you didn't mention what you are cooking on) at the temperatures that you are cooking at, with your methods and the briskets that you prefer. BUT... That is not my experience. I've also cooked over 100 briskets of varying types (choice, prime and Wagyu) at a variety of temperatures over the years, and I alway go by "it's done when it is done". Different grade briskets will finish at different temps. Also, the cook temperature will influence when the brisket is done. The hotter the cook temp, the hotter the finish temp.

So, data and logging is good, but you have to avoid making generalizations based on a limited set of inputs. I've had as many briskets finish below 200 as I have had above 200 and will always go by feel.
 
I feel keeping a log is very helpful to new cooks or when you are learning to use a new cooker.After learn to cook on that one cooker the log is not as important for the cooking part.But keeping good records of how a rub or sauce affected a certain kind of meat or what ever is very important to me.Even keeping records on little changes in a rub or sauce is important.And if you happen to be like me,you don't get to cook nearly as often as you would like.So when it's a good deal of time between cooks you need to check your log to see how you did that last cook that was so good.And when you get on up there in age as I am ,your memory just ain't as good as it use to be.
 
I was ready to type up a rebuttal to your conclusion, and then saw ronl beat me to it. Briskets, based on many factors besides size, such age of the beast and the level of marbling can finish at temps from the high 180s to 210+. In fact, I stop looking at internal temps at 190 and go exclusively by feel. Final temps are taken once it feels right only to determine if and how long I may need to vent the brisket before completing or resting it. I've cooked 100s of briskets. And many time briskets of identical weights will finish at 2 different durations with final temps 10 degrees apart.
 
I like reading the data too. I will say I really agree the quality of the meat matters greatly. I'm not sure if it's how the meat was processed shipped froze whatever but they are all not equal. I try to buy brands that give me the greatest success. I do notice a difference in the end product.

Doesn't Aaron Franklin use creek stone?
 
As primarily a competition cook these day, keeping good logs is essential for reproducibility and consistent results.

I really can't argue with any of your data, but will take issue with your conculsion...



I have no doubt that this is true for your cooker (you didn't mention what you are cooking on) at the temperatures that you are cooking at, with your methods and the briskets that you prefer. BUT... That is not my experience. I've also cooked over 100 briskets of varying types (choice, prime and Wagyu) at a variety of temperatures over the years, and I alway go by "it's done when it is done". Different grade briskets will finish at different temps. Also, the cook temperature will influence when the brisket is done. The hotter the cook temp, the hotter the finish temp.

So, data and logging is good, but you have to avoid making generalizations based on a limited set of inputs. I've had as many briskets finish below 200 as I have had above 200 and will always go by feel.

Yes higher temps lead to higher finishing temps, I concur. Also poor temp control shouldn't really matter for brisket. I've started hot and finished lower and vice versa. It's all about knowing how to read the signs of brisket. I'm a big advocate of using temp as a guide, but still knowing the look and feel is important.
 
I always take detailed notes with new cookers and once I get it down I still take notes but not as detailed. I haven't counted briskets but it has to be at least a hundred. I cook all kinds except high end ones. In my experience I can cook 8 select briskets and they will be done at different temps. Also seem to have 2 that take an extra hour or 2. Maybe cooking all cf in the same cooker at the same temp you have hit the sweet spot for your cooker and your cooks. Nice job on getting your system down.
 
If I was in the business of furnishing grub to customers, I'd be mighty grateful for that information.

I don't smoke briskets but 2 or 3 times a year. I do a whole lot more butts and yardbirds.

It's been decades since I took statistics (back when the smallest calculators were the size of a large cash register); so please bear with me. I also understand that you select for size before starting out, but I'd be curious to know if your data can be extrapolated to larger briskets and, if so, how?

It absolutely is able to. I'm assuming when you say "larger briskets", you mean weight? If so...

Per the image capture that I posted, this is from a program called R. R is the standard for statistical analysis and conducting data science research. There are others, but R is open source and is considered the de facto program.

>fin.temp and all the outputs after it are a sample of 100. The "fin.temp" sample is based on two things, the mean or average weight of my 15 cooks and the standard deviation of those 15 cooks, which was 1.79 degrees. We also utilize what's referred to as standard error. Because it's not feasible to run the test, literally, 100 times, 1,000 or how many we choose, we use what's called standard error (SE). Distilled down, it allows us to make statistical calculations based on random samples because it is the mean of that sample population we're testing.

What we would see with an increase in weights is the same curve or normal distribution. The weight of the briskets may increase, but there will be a normal distribution in the weights; which would equate to a normal distribution in finish temps. If we plotted it and did a "connect the dots" curve, we would see what many refer to as a "bell curve", which is also referred to as a normal distribution.


The only variable in all of this is the brisket brand, which I controlled by only selecting Creekstone Farms briskets. It wasn't their "gold" or comp waygu briskets, which I suspect would alter the temp finish times.

Your sample size information is news to me. Back when I was doing data collection and analysis we went to great extremes to randomize sample selection and to get as many as we could. A sample of 30 if properly random is statistically valid, but my profs wanted more data - probably just sweatin' us a bit though.

Depending on when you were doing stats and data collection, there is a strong likelihood that you did have to do everything by hand, such as, calculating the mean, mode, range, standard deviation, sampling error, alpha, confidence intervals and such. Possibly you used SPSS, maybe. With the creation of R, it makes all of that much easier, though understanding them is vital. In R we're able to generate a random sample based upon the curve we select, ie., normal, bimodal, uniform, which is what I did above.
 
Two things contribute to a tough cut like brisket getting tender. Temperature of the meat, and time at that temperature. With Sous vide, 48 hours at 155 will make a super tender piece of brisket that hasn't come close to 200 +/- 2 degrees.

A brisket smoked at 350 will be done sooner than at 220 but it will have gotten hotter in the process. The higher cook temp will heat it faster and the internal meat will spend less time going from 140 to 200, so when it gets to 200, it may still need more time in the cooking range to soften up. If the smoker stays at 350, the meat will keep warming and you will find a higher "done temp". If you started hot and immediately dropped them smoker temp to 200 when the IT reaches 200, you would find the brisket is always done at 200. If you start hot and drop the smoker to 180 when the IT reaches 180, it will take a bit longer but your "done temp" will be 180.
 
It absolutely is able to. I'm assuming when you say "larger briskets", you mean weight? If so...

Per the image capture that I posted, this is from a program called R. R is the standard for statistical analysis and conducting data science research. There are others, but R is open source and is considered the de facto program.

>fin.temp and all the outputs after it are a sample of 100. The "fin.temp" sample is based on two things, the mean or average weight of my 15 cooks and the standard deviation of those 15 cooks, which was 1.79 degrees. We also utilize what's referred to as standard error. Because it's not feasible to run the test, literally, 100 times, 1,000 or how many we choose, we use what's called standard error (SE). Distilled down, it allows us to make statistical calculations based on random samples because it is the mean of that sample population we're testing.

What we would see with an increase in weights is the same curve or normal distribution. The weight of the briskets may increase, but there will be a normal distribution in the weights; which would equate to a normal distribution in finish temps. If we plotted it and did a "connect the dots" curve, we would see what many refer to as a "bell curve", which is also referred to as a normal distribution.


The only variable in all of this is the brisket brand, which I controlled by only selecting Creekstone Farms briskets. It wasn't their "gold" or comp waygu briskets, which I suspect would alter the temp finish times.



Depending on when you were doing stats and data collection, there is a strong likelihood that you did have to do everything by hand, such as, calculating the mean, mode, range, standard deviation, sampling error, alpha, confidence intervals and such. Possibly you used SPSS, maybe. With the creation of R, it makes all of that much easier, though understanding them is vital. In R we're able to generate a random sample based upon the curve we select, ie., normal, bimodal, uniform, which is what I did above.

Garbage in garbage out. You are not capturing the proper variables in your analysis so your results are flawed. Done time and IT are dependent on cooking temperature, heat transmission mechanism (conduction convection radiant) thickness volume and surface area, and the specific qualities of the meat.

Your data is based on an extremely small sample that is not representative of the full range of brisket sizes and shapes. You don't seem to have any variation in cook temperatures or heat transmission etc. You assume a normal distribution (or several) but have no evidence that shows your population can be reasonably assumed to be normally distributed.

The software is spitting out results, but you need to understand the assumptions behind the formulas and realize that where you are violating those assumptions, the resulting R squared, T stat and standard error will be incorrect, potentially grossly incorrect.
 
I will also add that you absolutely cannot extrapolate outside the data set unless you know the relationships are linear within the data range and remain so out beyond where you are extrapolation. If you test 1000 14-16lb briskets, you may have great data on a 14.75 pound brisket, and you can probably make a good guess at how a 13.5 or 16.5 lb brisket will cook, but you could and likely would have tremendous error trying to estimate a 7 lb or 25 lb brisket cook based on that data.
 
As primarily a competition cook these day, keeping good logs is essential for reproducibility and consistent results.

I really can't argue with any of your data, but will take issue with your conculsion...



I have no doubt that this is true for your cooker (you didn't mention what you are cooking on) at the temperatures that you are cooking at, with your methods and the briskets that you prefer. BUT... That is not my experience. I've also cooked over 100 briskets of varying types (choice, prime and Wagyu) at a variety of temperatures over the years, and I alway go by "it's done when it is done". Different grade briskets will finish at different temps. Also, the cook temperature will influence when the brisket is done. The hotter the cook temp, the hotter the finish temp.

So, data and logging is good, but you have to avoid making generalizations based on a limited set of inputs. I've had as many briskets finish below 200 as I have had above 200 and will always go by feel.


I was cooking on a Good One Marshall, it didn't vary much in temp, provided I maintained the charcoal every 2 hours. I start at 190 and will cook between that and 200, until I hit an IT of 160. Then I ramp it up to 230.

Yeah, I'm not disputing different grades of brisket and their finish temps or even cooking hot and fast. I'm controlling the temp of my cooker, selecting the same brisket grades from the same processor; which is vital to having a normal distribution curve, which is what is being represented.

Temp as a guideline can be used to know when a brisket is done, as can a temp probe going in like room temp butter or by watching how it jiggles and shimmies under a slight push. All are very valid reasons and equate to "it's done when it is done" . Though, as has been pointed out, it takes 100's of briskets, if not more to know when a brisket is completed by those methods.
"Start probing at 195 and pull when the probe feels like it's going into room temp butter", a little unsettling if you dropped $50-$80 on packer.

As with anything that's new to us, it is absolutely intimidating and we approach it with a certain level of apprehension. Actual analytics applied to finish temps would be far more accurate and instill confidence in new and upcoming BBQ hobbyists. Which may lead to more people participating in KCBS, more great BBQ trailers popping up across the country and an even healthier ecosystem surrounding BBQ.


I was ready to type up a rebuttal to your conclusion, and then saw ronl beat me to it. Briskets, based on many factors besides size, such age of the beast and the level of marbling can finish at temps from the high 180s to 210+. In fact, I stop looking at internal temps at 190 and go exclusively by feel. Final temps are taken once it feels right only to determine if and how long I may need to vent the brisket before completing or resting it. I've cooked 100s of briskets. And many time briskets of identical weights will finish at 2 different durations with final temps 10 degrees apart.


As with Ron_L, I have no intention in disputing long standing, experienced pit guys or women. But, if with having cooked 100's of briskets, I would still say there would be the same tendencies in the cook. Another way to approach it would be to lower the Confidence Interval, which would be saying 90% confident or changing it to 85% confident. As we're dealing with percentages, we can do 1- confident interval or 100% minus 90 or 85 percent. That way, it would cover the larger swing in temps.

Again, all valid points, but if you used the same processor and brisket grade, you would see a normal distribution occur with finish temps. I don't know if I'd include Sam's Club/Wal-Mart or other larger markets like that, as I have no idea if they source their meat from the same processor week in week out. Either way, the same would hold true if you ran the same test with Snake River Farms or even CAB's from where ever. If massive swings were consistent from the same processor, that would indicate exactly what you're pointing out, large fluctuations in marbling/fat content vs. lean; which would be a sign of inconsistencies in the cow. From a market perspective, those inconsistencies are bred out, "proteined out", "grained out", farmed out, with a good degree of success; which equates to a much more consistent product for purchase.
 
"Start probing at 195 and pull when the probe feels like it's going into room temp butter", a little unsettling if you dropped $50-$80 on packer.

I feel exactly the opposite :-D Dropping a $200 wagyu into the smoker and saying "it will be done between 201 and 202 degrees scares the crap out of me :twitch:
 
Back
Top