Sourdough starter quantity predictive model

[TXCraig1]

Yes, I wondering the same. But do you think that bacterial fermentation is a factored into your chart? Or is it a non-issue?

John

No. It doesn't take LAB fermentation into account. The yeast and LAB activity in response to temperature is nearly identical below 80F. I wouldn't expect a model that included LAB fermentation to deviate much until you got north of 85F or so which really wasn't in the true scope of the project.

CL

[DannyG]

I think you're basically reading it right. Every data point on the chart represents the theoretical hours needed for a particular starter%/temperature combination. 40%/40F/169h is one possible combination. I color coded the chart for exactly the reason you noted. I don't think the points in red are viable and the ones in yellow are questionable (based on my experience - not science per se). The examples you gave are all solidly in the red.

Remember, this chart assumes 100% of the fermentation takes place at the given temperature. It does not take into account the time it takes the dough to get to that temperature nor does it contemplate fermentation at multiple temperatures. I'm not an advocate of cold fermentation- particularly with SD. I only included those temperatures in the chart for discussion purposes such as this.

All that being said, I'm still surprised to hear that 3 days in the fridge + 4 hours at RT is enough. Is your starter sourdough or is it made from commercial yeast?

Wow, where is my head at? I use the refrigerator only for IDY. When I use my Ischia sourdough starter I put it in a small wine cooler refrigerator which is kept at 54 degrees. According to the chart this knocks the time down to 104 hrs/4.34 days. Figuring that I bench rest the dough for about an hour before cooling, and I take it out for a few hours before baking, I come much closer to the times on your chart.

[TXCraig1]

The chart below shows the changes in predicted fermentation hours between the original chart I posted and Rev1 (I replaced in the chart in the original post). I was pleased to see that the changes inside my original green zone (the original color zones are shown below) were for the most part negligible. The revised predictions fit the 65F test data better than the original. They also fit much better with other data I have both significantly cooler and warmer than the test data (not illustrated).

To be clear - the chart below only shows the difference between the original chart and Revision1. The chart at the beginning of the thread is the revised chart.

[Jet_deck]

Why are the numbers negative or is it just a dash?

[enter8]

Why are the numbers negative or is it just a dash?

The negative numbers are just showing the difference between the original and the revised figures. The corrected figures are on the first post of this thread.

[TXCraig1]

The negative numbers are just showing the difference between the original and the revised figures. The corrected figures are on the first post of this thread.

Correct.

[kiwipete]

Great work, Craig, thank you.

Like you, my workflow has been established over a long time with lots of trial and error and is currently reasonably well established.

But if you are starting out with SD, or want to try something completely different, this is a great tool.

To prevent this thread from dropping down over time, I'd like to suggest to post the chart in the tools section or make the thread sticky or something along those lines.

Cheers

Peter

[DannyG]

Craig,
A couple of questions...
1. How do you determine what a fully fermented dough is?
2. Because we are talking a sourdough starter, is there really any taste or dough quality difference between a high percentage starter/short time vs. a low percentage starter/long time? If not, one can use your chart by determining how much time one has for fermentation and use the appropriate percentage of starter. This is a great tool.

[TXCraig1]

Craig,
A couple of questions...
1. How do you determine what a fully fermented dough is?
2. Because we are talking a sourdough starter, is there really any taste or dough quality difference between a high percentage starter/short time vs. a low percentage starter/long time? If not, one can use your chart by determining how much time one has for fermentation and use the appropriate percentage of starter. This is a great tool.

1) It's loosely defined as the dough is ready to be baked - typically this would be determined visually by the amount of rise.
2) I don't know. My gut feeling is there are some differences, but there are so many trade-offs and ways to compensate for one factor with another. Take gluten for example, it can be developed mechanically or simply with time. As for flavor, that is probably even more effected by the specific culture than fermentation time.

[deb415611]

Thank you Craig, this is great.  I often shy away from starter because I can't find a recipe that works for what I want my timing to be,  this will make it much easier to create my own.

I agree with others that it would be good to put in the tools section or stickied.

[bfguilford]

Thanks for doing the work on this, Craig. Your commitment to pizzamaking and this forum is really appreciated. Lately, I've been using my Ischia only for bread baking, and this is inspiring me to try NY style using starter again.

Barry

[TXCraig1]

Mitch (mitchjg) asked me a question about using the data in the model to calculate the effects of multiple time/temp combinations on the same dough. Looking at the model, it occured to me that you can easily do just that with the model exactly as it is. I added a few more columns (starter%) to the chart so there is more data to work with (less extrapolation needed). I also stripped off the red zones. The data itself and model are unchanged. You could use this same technique with the chart on the first page of this thread. The charts below are optimized to make it even easier.


Example 1
Say you made a batch of dough w/1.5% starter that normally takes 48 hours at 64F to be ready. Being that it is winter and cooler in the house that normal, your fermentation set up (http://www.pizzamaking.com/forum/index.php/topic,18509.msg179991.html#msg179991) is running cooler than normal - 60F rather than 64F. How could you adjust your workflow to have the dough ready on time? Perhaps you could move it to the oven with the light on which is holding at about 80F, but for how long? This is the problem I faced over the past couple days.

We know from the model that 1.5% starter @ 60F should take ~69hours (1 below). I fermented the bulk for 24 hours at 60F. I then balled it and put it back into the box at 60F for another 12 hours – a total of 36 hours at 60F. According to the model, I should need another 33 at 60F (2), but the model tells us a whole lot more than that. Every data point up and down that column represents a time/temperature combination that will finish the dough. All you have to do is slide down to 80F, and you see that dough will be ready in 9 hours at 80F.

I hadn’t thought about doing this when I made my dough Thursday night – nor on Saturday morning when I moved it to the oven with the light on. It turns out however, that my results were spot-on what the model would have predicted.

IMPORTANT: Keep in mind that dough doesn’t instantly go from 60F to 80F. Actually, my oven with the light on is closer to 85F, but the average dough temp over the 9 hours was probably around 80F – just a gut feel adjustment. You might need to make adjustments recognizing this sort of thing. The shorter the window of time at a temperature, the more important the adjustment becomes as the dough temperature transition time becomes a much larger portion of the total time.

[TXCraig1]

Example 2
How about two temperature changes? Let’s say you made a 1.5% starter dough and wanted to do a 2 hour bulk ferment at your room temperature of 78F followed by 24 hours in balls at 65F, and then bring the dough back to room temperature to bake. How long is the dough going to need at 78F at the end to get ready? (Ignoring any adjustments discussed above for simplicity of illustration.)

Start at 1.5% and 78F (1) then move over two hours to represent the 2-hour bulk ferment. Like before, every data point in this column (2) represents a time/temperature combination that will finish the dough. Go up to 65F, and you see that you need 39 hours to finish the dough at this temperature (3). But we don’t want to finish it at this temperature – we just want 24 hours here, so slide over to the right 24 hours. This column (4) tells us what we need at any given time/temp to finish. If we slide down to 78F, we see that we’ll need about 6 hours.

The dough coming out of our mixer was probably neat 78F, so we probably don’t need to adjust that, and the second leg is 24 hours, so maybe that is OK too as the dough temperature will adjust long before 24 hours. However, going from 64 to 78 in 6 hours might not be representative of what the model is calculating. We might want to adjust the 78F down a couple degrees to account for the fact that it will take a meaningful amount of the total time at 78F to get there. We can look back up the column (4) a couple degrees and see that we should probably be prepared for it to take perhaps 7 hours or maybe a little more.

Looking up and down the column, above and below your predicted time, will give you an idea of how sensitive the prediction is. In this case, you can go up 3F and still only have a 1 hour change. There are other places on the chart where a 1F change could result in a time change of several hours or more. Those are the predictions you need to keep a close eye on when trying.

[TXCraig1]

Example 3
Rather than predict when a given dough will be ready, this time we will predict how much starter to use in the first place. This can be done by working backwards through the model.

For example, if you wanted 21 hours at 62F followed by 12 hours @ 70F, how much starter do you use? To figure it out, you start where you want to finish. Go across the 70F row until you find 12 hours (1). Now go up that column (2) until you hit the 62F row. Take the number you find there – in this case 23 hours - and add 21 hours to it for the bulk ferment to get 44 hours. Now go left across the row (3) until you find 44 (4). From there shoot up to the top and you will find the predicted amount of starter needed - 4% (5).

You can easily add additional fermentation stages to the prediction. If you wanted 5 hours at 77F before the 21 hours at 62F, from (4) instead of going up to find the starter % as described above, you would drop down to the 77F row, add 5 to the number you found there (15+5=20), slide left until you found that number, then shoot up to the top to find the starter % - 1.5% (red dotted line)

[TXCraig1]

Example 4
Interpolation

If you wanted to do 2 hours bulk at 78F followed by 36 hours in balls at 65F, how much culture should you use? This time, the numbers you are looking for are not going to be expressly on the table. You will need to interpolate to find them.

Again, start where you want to finish – in this case, 36h/65F – here we go down the 65F row and we see we need to interpolate between 34 and 37 hours (1).   Next, slide down that column (2) until you get to your bulk temperature row of 78F (3) and interpolate the result between 15 and 16 hours. Since 36 is about 2/3 of the way between 34 and 37F, I estimated the number to be about 15.7 (2/3 of the way between 15 and 16). To that number, add the desired amount of bulk time (2 hours) giving you 17.7 hours and slide to the left until you reach it – you’ll have to interpolate again (4), and then shoot up to the top of the table to estimate the predicted starter quantity needed – 2.2% (5).

I would guess that in most, if not all, cases, simply splitting the difference between numbers would plenty accurate.

[TXCraig1]

Here is a blank chart you can work from:

[kiwipete]

That is an absolutely brilliant (an unexpected) use of the model.

Many thanks!

Peter

[TXCraig1]

Mitch came up with a great methodology for using the model data to compute starter% given multiple fermentation stages. This way you don't have to backtrack through the chart; you just plug in your numbers. The chart itself is still a great tool for making changes on the fly as I described above.

I set the model up so that up to 10 stages could be entered, not that anyone would do a 10 stage ferment per-se but rather so there would be the flexibility to more smoothly model the transitions between temperatures if desired. You don't have to jump straight from room temp to 64F if you don't want to, you could make an educated guess and stick in a couple intermediate temps and times if you like. However, this isn't necessary. Unused stages have no impact.

In the yellow cells, simply enter the time and temperature for each stage. For an unused stage, just make the time 0. It doesn't matter what the temperature is if the time is 0. The model will update the predicted starter quantity as you make changes. The other calculation the model shows is how much of the fermentation is occurring at each stage. This is the percent of flora activity not the percent of time. Enter a low temperature and a long time as one of the stages as an example, and you will see how it contributes a relatively small amount of the total fermentation.

https://docs.google.com/spreadsheet/ccc?key=0AuvMQbzk5INUdGZScWx6U2lYSEtZVkJuVGJiR19NaXc#gid=0

[Serpentelli]

Here's another wrinkle I was actually thinking about last night: What about "arresting" the process somewhere in between either bulk + ball, or post-ball but pre-completion of fermentation, via freezing

It would seem that as long as your final (post freeze)fermentation temperature was high enough (say 80 degrees in may case) and you took into account the time it takes for the ball to come up to the 80 degree temperature, the model should be greatly helpful for a guy like me who wants to have some greater flexibilty & spontaneity for a midweek pizza night.

I would just take the balls out of the freezer in the morning, bring up to RT in a warm water bath (in airtight vacuum sealed bags), and then leave in an 80 degree environment until ready to use that evening.

What am I missing?

Thanks,

John

[PizzaJerk]

Craig, I thank you for creating this most helpful chart. I thought I was making a fairly predictable dough before and now I can just about pencil in the same flavor, rise and fermentation schedule every time. It can be adjusted on the fly based on demand as well. What a great tool, especially for those in a commercial environment.

Anthony