Fraction of a pack's cells still alive on a given day.
Calls yeast.ts's viabilityFromManufactureDate: 100 % on the date
printed on the pack, then exponential decay at 0.79 per 30 days for liquid
and 0.96 per 30 days for dry. A White Labs pack, recognised by its name,
decays on White Labs' own published curve instead, 0.946 per 30 days (see
PUREPITCH_RETENTION_PER_30_DAYS).
Why not the linear "20 % a month" that's usually quoted
Most calculators (Mr Malty's chart, and the usual summaries of the Yeast
book and White Labs' guidance) use a straight line losing about 20–21 % of
the original count per month, often floored at 10 % so it never reaches
zero. That's a convention, not a measurement. Real decay depends on the
strain, the cell density in the pack and, more than anything, how warm the
pack got in transit.
We model the same ~21 %/month as a continuous exponential, for the reason
given on VIABILITY_RETENTION_PER_30_DAYS: the curve stays positive without
an arbitrary floor and the monthly loss compounds. Over the ages that
matter, the two differ by less than the convention's own uncertainty:
age
linear 20 %/mo
this engine (0.79^(d/30))
1 month
80 %
79 %
3 months
40 %
49 %
6 months
10 % (floor)
24 %
The gap only gets large at ages where nobody would pitch the pack without
stepping it up anyway. Two different models would be worse: the same pack
would show 49 % in the pitch check and 40 % in the starter calculator. So
the starter calculator uses the same curve as pitchPlan.
Dry yeast
Dry yeast is made for shelf stability (vacuum-packed, low moisture, sold
with a two-year best-before), so date decay matters much less than the cells
per gram the manufacturer packed, which is where availableCells gets
its number. The 0.96/30 days is a small correction. A sachet inside its
best-before date is within a rounding error of its printed count.
Fraction of a pack's cells still alive on a given day.
Calls
yeast.ts's viabilityFromManufactureDate: 100 % on the date printed on the pack, then exponential decay at 0.79 per 30 days for liquid and 0.96 per 30 days for dry. A White Labs pack, recognised by its name, decays on White Labs' own published curve instead, 0.946 per 30 days (seePUREPITCH_RETENTION_PER_30_DAYS).Why not the linear "20 % a month" that's usually quoted
Most calculators (Mr Malty's chart, and the usual summaries of the Yeast book and White Labs' guidance) use a straight line losing about 20–21 % of the original count per month, often floored at 10 % so it never reaches zero. That's a convention, not a measurement. Real decay depends on the strain, the cell density in the pack and, more than anything, how warm the pack got in transit.
We model the same ~21 %/month as a continuous exponential, for the reason given on
VIABILITY_RETENTION_PER_30_DAYS: the curve stays positive without an arbitrary floor and the monthly loss compounds. Over the ages that matter, the two differ by less than the convention's own uncertainty:The gap only gets large at ages where nobody would pitch the pack without stepping it up anyway. Two different models would be worse: the same pack would show 49 % in the pitch check and 40 % in the starter calculator. So the starter calculator uses the same curve as pitchPlan.
Dry yeast
Dry yeast is made for shelf stability (vacuum-packed, low moisture, sold with a two-year best-before), so date decay matters much less than the cells per gram the manufacturer packed, which is where availableCells gets its number. The 0.96/30 days is a small correction. A sachet inside its best-before date is within a rounding error of its printed count.