A D V E R T I S E M E N T


Author Topic: Sourdough starter quantity predictive model  (Read 121993 times)

0 Members and 1 Guest are viewing this topic.

Offline dciolek

  • Registered User
  • Posts: 21
  • Location: Ohio
  • I Love Pizza!
Re: Sourdough starter quantity predictive model
« Reply #260 on: October 28, 2017, 10:12:55 AM »
If it was a best fit curve approach -- it should be reasonably simple to add those constants to the spreadsheet formula as variables.  In the event that anyone needed to adjust the predicted completion curve to a different set of fermentation conditions.

Offline TXCraig1

  • Supporting Member
  • *
  • Posts: 23224
  • Location: Houston, TX
  • Pizza is not bread.
    • Craig's Neapolitan Garage
Re: Sourdough starter quantity predictive model
« Reply #261 on: October 28, 2017, 09:18:50 PM »
I'm not exactly sure what you're suggesting, but to be sure I was clear in my earlier comment, the curve fitting is not the model. It was simply a convenient way to port the model to that spreadsheet.
"We make great pizza, with sourdough when we can, baker's yeast when we must, but always great pizza."  
Craig's Neapolitan Garage

Offline dciolek

  • Registered User
  • Posts: 21
  • Location: Ohio
  • I Love Pizza!
Re: Sourdough starter quantity predictive model
« Reply #262 on: October 29, 2017, 05:12:03 AM »
Sorry -- upon further reflection, I think my suggestion is a bit out of context.  The question from mikeoz about the 4th order polynomials got me to thinking about "the spreadsheet" and the math.  To which I of course agree -- the curve fit and the polynomial expression is not the model.

My suggestion was in reference to the curve fitting polynomial formula that is used in your google docs spreadsheet at https://docs.google.com/spreadsheet/ccc?key=0AuvMQbzk5INUdGZScWx6U2lYSEtZVkJuVGJiR19NaXc#gid=0 which translates your model's experimental data into a formula that calculates the percentage completion of fermentation for a series of time and temp steps.  That spreadsheet has inputs for up to 10 time/temp steps and then uses the "best fit curve" math for each step to sum up to a total percentage completion and solves for how much starter you need to get there.

I used that same math to make a similar spreadsheet that sums up the percentage completion of time/temp steps from the temperature output of a real time dough temp monitoring probe, so you could track the progress percent to completion and then calculate the remaining predicted time to completion at a given temp based on actual progress since starting the ferment.  Data from first trial run here:  https://www.pizzamaking.com/forum/index.php?topic=49858.msg502822#msg502822

In this monitoring spreadsheet -- I took one of the constants from the math in the google docs spreadsheet that is used to calculate the percent of starter quantity required and turned it into a variable, calling it "Target Completion Factor" (and using the constant as a default setting).  That Target Completion Factor seems to be determining how many times the percent added starter must double in order to grow to that level of fermentation completion.  For my first trial run, I used a different starter culture than was used in building the original model -- and noticed that the pH bottomed out at 95% completion and stayed there for the next three hours until it reached 100% completion. 

So if on the next trial dough run, I wanted to experiment and adjust my time to completion to hit the point where my pH starts to bottom out (maybe signaling that LAB function is declining -- and maybe the flavor contribution to the dough), I could adjust down that Target Completion Factor to 95% of the default factor and basically reset my completion target in the spreadsheet.

Likewise, my suggestion about being able to change constants in the curve fit polynomials from the monitoring spreadsheet into variables is along those same lines.  In the event that someone else has a different set of experimental data, using a different set of starting conditions (like the starter type, or flour type that is malted, or less/more salt, etc) -- they could adjust the variables for the polynomial to match the best fit curve from their own experimental data and the monitoring spreadsheet would automatically adapt its calculation of percent completion to match (without changing the spreadsheet formulas).

I am not suggesting anything to change the model, your spreadsheet or question the underlying math.  My experiment in real time monitoring is simply adapting your math and your work to provide feedback from the actual progress and dough temps over time during the fermentation.
« Last Edit: October 29, 2017, 05:16:30 AM by dciolek »

Offline bifi85

  • Registered User
  • Posts: 27
  • Location: Germany
Re: Sourdough starter quantity predictive model
« Reply #263 on: November 22, 2017, 05:26:47 AM »
Do I understand this topic correct?
When I want to make a pizza with 20% starter (aka poolish) in it and my room temperature where I put the starter is 64F. So I look at TXCraig1's model go to the point where 20% and 64F meet (18h). Now I know my starter for the pizza is ready in 18h. That means in 18h I put the starter into my pizza dough and from here I do my pizza dough thing (rest overnight in fridge etc.).

Offline TXCraig1

  • Supporting Member
  • *
  • Posts: 23224
  • Location: Houston, TX
  • Pizza is not bread.
    • Craig's Neapolitan Garage
Re: Sourdough starter quantity predictive model
« Reply #264 on: November 22, 2017, 08:45:55 AM »
Do I understand this topic correct?
When I want to make a pizza with 20% starter (aka poolish) in it and my room temperature where I put the starter is 64F. So I look at TXCraig1's model go to the point where 20% and 64F meet (18h). Now I know my starter for the pizza is ready in 18h. That means in 18h I put the starter into my pizza dough and from here I do my pizza dough thing (rest overnight in fridge etc.).

No. You interpret the same set of numbers like this:  if you want your dough to be ready in 18 hours, you would use 20% fully active starter in your dough and ferment the dough for 18 hours at 64F.  Keep in mind that the model is just a prediction and that there are literally dozens of variables that can affect the timing, so some testing and tweaking should be expected.
"We make great pizza, with sourdough when we can, baker's yeast when we must, but always great pizza."  
Craig's Neapolitan Garage

A D V E R T I S E M E N T


Offline ccgus

  • Supporting Member
  • *
  • Posts: 201
  • Location: Mukilteo, WA
    • Maybe Pizza?
Re: Sourdough starter quantity predictive model
« Reply #265 on: November 22, 2017, 03:16:01 PM »
No. You interpret the same set of numbers like this:  if you want your dough to be ready in 18 hours, you would use 20% fully active starter in your dough and ferment the dough for 18 hours at 64F.  Keep in mind that the model is just a prediction and that there are literally dozens of variables that can affect the timing, so some testing and tweaking should be expected.

I'll just drop this little anecdote to reinforce this idea.

The starter I'm currently using, which is at about 1 1/2 years old now, behaves way different from when I first started using it. For a 24 hour ferment, I would originally use about 10% starter. I found over time the starter became way more active as I trained it. These days, I'm down to using about 4% of starter. Same aromas, same starter, but different behavior as the starter matured.

Each starter is different in subtle ways, even the same starter a year later is going to act different.

Offline TXCraig1

  • Supporting Member
  • *
  • Posts: 23224
  • Location: Houston, TX
  • Pizza is not bread.
    • Craig's Neapolitan Garage
Re: Sourdough starter quantity predictive model
« Reply #266 on: November 22, 2017, 06:31:00 PM »
I'll just drop this little anecdote to reinforce this idea.

The starter I'm currently using, which is at about 1 1/2 years old now, behaves way different from when I first started using it. For a 24 hour ferment, I would originally use about 10% starter. I found over time the starter became way more active as I trained it. These days, I'm down to using about 4% of starter. Same aromas, same starter, but different behavior as the starter matured.

Each starter is different in subtle ways, even the same starter a year later is going to act different.

That's a great point. Mine has slowed over time as well. Where I originally used 1.7%, I'm, now using 1.9% - 2.1%.

As Yoda would say, one with your starter you must be.
"We make great pizza, with sourdough when we can, baker's yeast when we must, but always great pizza."  
Craig's Neapolitan Garage

Offline bifi85

  • Registered User
  • Posts: 27
  • Location: Germany
Re: Sourdough starter quantity predictive model
« Reply #267 on: November 23, 2017, 05:42:56 AM »
No. You interpret the same set of numbers like this:  if you want your dough to be ready in 18 hours, you would use 20% fully active starter in your dough and ferment the dough for 18 hours at 64F.  Keep in mind that the model is just a prediction and that there are literally dozens of variables that can affect the timing, so some testing and tweaking should be expected.
Thank you for your answer.
  • I make a poolish and let it rest for 18h at room temperature for example
  • After 18h I make pizza dough with 20% poolish in it
  • My fridge has 45F. So I have to put the final pizza dough 113h into the fridge?!?! (Holy %$#)

Offline TXCraig1

  • Supporting Member
  • *
  • Posts: 23224
  • Location: Houston, TX
  • Pizza is not bread.
    • Craig's Neapolitan Garage
Re: Sourdough starter quantity predictive model
« Reply #268 on: November 23, 2017, 08:58:29 AM »
Thank you for your answer.
  • I make a poolish and let it rest for 18h at room temperature for example
  • After 18h I make pizza dough with 20% poolish in it
  • My fridge has 45F. So I have to put the final pizza dough 113h into the fridge?!?! (Holy %$#)

No, you're still reading the chart wrong.  Here are the basic steps given the data you identified:

1. Make a fully active starter (when you say "poolish" you mean sourdough, correct? This table will not work with a baker's yeast poolish).
2. Make your dough which incorporates 20% starter (for example, a dough with 1000g flour would use 200g of fully active starter)
3. Mix and ferment at 64F  for 18 hours total (bulk and ball time combined).

Again, note that this is just an estimate based on how a bunch of different cultures perform. Every culture is different as are formulas, workflows, and a bunch of other variables. Some testing and tweaking of %, temp, and/or time may be needed to get things how you want them.
"We make great pizza, with sourdough when we can, baker's yeast when we must, but always great pizza."  
Craig's Neapolitan Garage

Offline corkd

  • Lifetime Member
  • *
  • Posts: 499
  • Location: syracuse, ny
  • I Love Pizza!
Re: Sourdough starter quantity predictive model
« Reply #269 on: November 24, 2017, 10:09:50 AM »
I have found this model to be very useful, rather than expecting it to be completely accurate. I use it all the time, and usually just go a little heavier on the starter %, & just slow things down with the fridge, garage, or basement as needed (depending on the season); or the 100degree bread proofing setting on my oven if I need to go the other way.

A D V E R T I S E M E N T


Offline bifi85

  • Registered User
  • Posts: 27
  • Location: Germany
Re: Sourdough starter quantity predictive model
« Reply #270 on: January 13, 2018, 11:07:25 AM »
TXCraig1, can you post 48 h chart for Sourdough starter, too? That would be awesome.  :-[

Offline TXCraig1

  • Supporting Member
  • *
  • Posts: 23224
  • Location: Houston, TX
  • Pizza is not bread.
    • Craig's Neapolitan Garage
Re: Sourdough starter quantity predictive model
« Reply #271 on: January 13, 2018, 01:26:57 PM »
TXCraig1, can you post 48 h chart for Sourdough starter, too? That would be awesome.  :-[

This spreadsheet will give precise time/temp predictions (and multi-stage predictions) based on the sourdough model: https://docs.google.com/spreadsheets/d/1UEj7IApIauBjj_9d3jdWrbjOjnWu5FICcgAFZY6p8Oc/edit#gid=0

To get predicted starter quantity for any temp, just make the top time cell 48 and the top temperature cell whatever temp you want. As long as all the other time cells are zero, the prediction will be 48 hours at the temperature you put in at the top temp cell.
"We make great pizza, with sourdough when we can, baker's yeast when we must, but always great pizza."  
Craig's Neapolitan Garage

Offline bifi85

  • Registered User
  • Posts: 27
  • Location: Germany
Re: Sourdough starter quantity predictive model
« Reply #272 on: January 13, 2018, 05:49:45 PM »
Do I use the google docs wrong?

Offline TXCraig1

  • Supporting Member
  • *
  • Posts: 23224
  • Location: Houston, TX
  • Pizza is not bread.
    • Craig's Neapolitan Garage
Re: Sourdough starter quantity predictive model
« Reply #273 on: January 13, 2018, 06:03:55 PM »
I have no idea how you did what you did. Somehow you edited two cells that were locked for editing and changed the number format. I fixed it, and it seems to work fine for me. Try just changing the temperature in the top cell and nothing else and see if it works.
"We make great pizza, with sourdough when we can, baker's yeast when we must, but always great pizza."  
Craig's Neapolitan Garage

Offline bifi85

  • Registered User
  • Posts: 27
  • Location: Germany
Re: Sourdough starter quantity predictive model
« Reply #274 on: January 13, 2018, 06:17:18 PM »
Thank you very much. I did not know I changed other stuff than hours and F. I am sorry, didn't know that a google docs hacker sleeps in me.

A D V E R T I S E M E N T


Offline Jr07

  • Registered User
  • Posts: 94
  • I Love Pizza!
Re: Sourdough starter quantity predictive model
« Reply #275 on: January 16, 2018, 06:30:50 AM »
just to be clear, the hours in the predictive model are from when to when? from mixing to ready to bake?

Offline TXCraig1

  • Supporting Member
  • *
  • Posts: 23224
  • Location: Houston, TX
  • Pizza is not bread.
    • Craig's Neapolitan Garage
Re: Sourdough starter quantity predictive model
« Reply #276 on: January 16, 2018, 09:01:07 AM »
just to be clear, the hours in the predictive model are from when to when? from mixing to ready to bake?

Assuming you follow relatively normal mixing processes* the hours would be from finished mixing to ready to use.

*non-standard processes might include things like multiple long rests, extended autolyse, etc. 
"We make great pizza, with sourdough when we can, baker's yeast when we must, but always great pizza."  
Craig's Neapolitan Garage

Offline yarbrough462

  • Registered User
  • Posts: 211
  • Location: San Quirino, Italy
  • An American in Italy
Re: Sourdough starter quantity predictive model
« Reply #277 on: January 18, 2018, 07:09:12 AM »
Just wanted to pop in and say thanks for this model.  I did 29 pizzas yesterday for some of my younger folks at work.  We have been working them in to the ground in preparation for an upcoming event so this provided a nice break.  I used your garage process and did a 48 hr ferment based on your model.  They fermentation was dead on and the young folks said that this is the best pizza they have ever had.  Keep in mind that these are all young Americans that live in Italy so they have experience in awesome pizza.  You made a lot of young Airmen happy yesterday! 

Offline TXCraig1

  • Supporting Member
  • *
  • Posts: 23224
  • Location: Houston, TX
  • Pizza is not bread.
    • Craig's Neapolitan Garage
Re: Sourdough starter quantity predictive model
« Reply #278 on: January 18, 2018, 08:53:46 AM »
Just wanted to pop in and say thanks for this model.  I did 29 pizzas yesterday for some of my younger folks at work.  We have been working them in to the ground in preparation for an upcoming event so this provided a nice break.  I used your garage process and did a 48 hr ferment based on your model.  They fermentation was dead on and the young folks said that this is the best pizza they have ever had.  Keep in mind that these are all young Americans that live in Italy so they have experience in awesome pizza.  You made a lot of young Airmen happy yesterday!

Cheers! Thanks for letting me know.  ;D
"We make great pizza, with sourdough when we can, baker's yeast when we must, but always great pizza."  
Craig's Neapolitan Garage

Offline yarbrough462

  • Registered User
  • Posts: 211
  • Location: San Quirino, Italy
  • An American in Italy
Re: Sourdough starter quantity predictive model
« Reply #279 on: January 18, 2018, 01:31:17 PM »
Couple pics.  Didn't get many since I was cooking...

« Last Edit: January 18, 2018, 02:07:43 PM by yarbrough462 »

A D V E R T I S E M E N T


 

wordpress