Stalled box fermentation in San Martín: 34% slaty beans and a buyer waiting
A 42-member cooperative in San Martín, Peru, delivers 600 kg of wet Trinitario beans to a central fermentation shed. The pods were cut at peak ripeness and the pulp is sweet, but night temperatures have dropped to 23 °C. The 600 kg was split between two wooden boxes measuring 2.0 × 1.2 m, each loaded to a depth of 25 cm, and turned every 24 hours from the first day. At 48 hours the core temperature has reached only 39 °C and is falling; the box smells sharply of vinegar but none of the deeper fruity notes the buyer expects. A 100-bean cut test on day 4 gives 34% slaty, 48% purple and 18% brown; cotyledon pH is 6.1. Fermentation is scheduled to stop tomorrow and the beans will go to raised drying beds. The buyer has contracted 8 tonnes of this lot for a single-origin 70% bar and wants a fermentation report before the next container leaves in five weeks.
1. Diagnose why the fermentation stalled and set out a corrected protocol, with timings and target numbers, for the next delivery.
Read the worked answer
Diagnosis (physical, not biological). The pulp was fine and the yeast–LAB sequence started (the vinegar smell confirms acetic acid production), but the mass never retained metabolic heat. Three faults compound each other:
- Mass and depth too small. 300 kg at 25 cm in a 2.0 × 1.2 m box gives a high surface-to-volume ratio; the correct target is 40–60 cm depth, roughly 600–1,000 kg per box.
- Turned too early. The first turn at 24 h removes the small heat gain before the exothermic acetic-acid phase peaks. First turn should be at 48 h.
- No insulation. Uncovered boxes in a 23 °C shed lose heat overnight; banana leaves or jute sacking plus a lid are essential.
Corrected protocol. Load 600 kg into one 1.2 × 1.2 × 0.6 m box (≈45 cm depth); cover; do not turn before 48 h; then turn every 24 h, using core temperature rather than the clock (turn when the core exceeds 48–50 °C). Target a core of 45–50 °C by 48–72 h, cotyledon pH 5.0–5.5 by day 4–5, total 5–6 days. Verify with a 100-bean cut test each day: ≥85–90% brown and no more than 5% slaty at the endpoint.
2. The day-4 beans are still in the box. What do you do with this lot, and what do you tell the buyer?
Read the worked answer
Do not stop tomorrow. Give the lot two more days with a turn now and a cover, and re-run the cut test and pH after 48 h. Beans that are purple will often still convert; fully slaty beans improve only marginally because the cotyledon cells are already dead and their polyphenols oxidised poorly. Expect a realistic gain from 18% to perhaps 30–40% brown, not to spec.
Decision rule. Re-test on day 6. If slaty remains above ~15–20%, this is not fine-flavour single-origin material: it will taste astringent, potato/woody and lack fruit. Options, in order of preference:
- Sell as a non-single-origin blending component at up to 20–25% in a workhorse bar, where the astringency is masked by sugar and other origins.
- Sell as commodity grade at commodity price.
- Never present it as the contracted single-origin lot.
Tell the buyer first, with data. Send the temperature curve, pH series and cut-test counts, explain the cause (under-loading and premature turning), and offer either a deferred shipment from the next well-fermented delivery or a renegotiated price. Buyers forgive a documented failure; they do not forgive discovering it in the roaster. Ask whether they will accept a 20% blend trial so the 8 t contract is not lost entirely.
3. What records should the cooperative keep so that this failure is diagnosable and the buyer's fermentation report can be produced?
Read the worked answer
The report must let a third party reconstruct what the microbes experienced. Minimum dataset per batch:
- Identity: batch code, date and time of pod breaking, farm/collector, wet weight delivered, bean variety and plot altitude.
- Geometry: box dimensions, depth of load, kg per box, cover material.
- Turning log: time of every turn and the core temperature immediately before and after.
- Temperature: core (probe at 20–30 cm) and ambient, twice daily, plotted rather than tabulated alone.
- Chemistry: cotyledon pH at day 0, 2, 4 and at the endpoint; moisture at the end of drying (target 7–7.5%).
- Cut test: 100-bean counts of slaty/purple/brown at the same intervals, with photos.
- Sensory note: two or three descriptors recorded at the end of fermentation, not at the end of drying.
Add a one-line corrective-action field, because the value of the record is proving that the next batch was run differently. Present the plot of core temperature against time; a flat line at 39 °C is the single most persuasive piece of evidence a buyer can read, and it turns an argument about quality into a shared diagnosis of loading depth.