Wet Processing: Using Starter Cultures

One of the steps in processing fully-washed coffees is to ferment the pulped seeds to help remove the sticky mucilage that surrounds the seed.  Most of the time this fermentation process is a wild fermentation started anew each day and that means from day-to-day, different strains of bacteria and yeast may become dominant, likely altering the coffee’s character (fermentation can affect a coffee’s flavor, body, and other aspects).

By using a starter culture taken from the previous day’s batch, a process called ‘back slopping,’ we can accomplish two things;

  1. Achieve a more standardized flavor profile
  2. Decrease soak-time while achieving the same level of cleanliness

During the fermentation process, first lactic acid bacteria (LAB) and then yeast colonies begin to proliferate in the sugary water.  Early in the process, the LAB create an acidic environment, which helps to prevent the growth of other, unwanted bacteria not adapted to an acidic environment.   It is believed that LAB aren’t directly involved in mucilage removal but instead, create an acidic environment which helps degrade the mucilage.  Yeast join the party later in the fermentation process and some create pectinase enzymes, which degrade pectin — a major component of the mucilage.

It would be ideal to have starter cultures of yeasts that produce specific pectinases known to effectively degrade the mucilage while preserving cup quality, but in the absence of those specific yeasts the previous day’s fermentation water will at least carry over the same yeast and LAB colonies from day-to-day, thus achieving our two objectives from above.

My first experiment with this process wasn’t as accurate as I would normally like so I plan to recreate it in July.  But the anecdotal evidence indicated a quicker fermentation when using starter culture.  Normal fermentation time at this mill, with this water is around twelve to sixteen hours.  Fermentation using starter culture was nine hours.  Unfortunately, I didn’t accurately measure the ratio of starter culture to clean water.

The water used on the day of the experiment had a pH of 9.1 (surprisingly alkaline, though Bali soil is rich in limestone) and the pH of the starter culture was 4.1.  After 30 minutes of fermentation, the pH of the test bucket was 5.1.

Anecdotally, using a starter culture reduces the pH quicker (increasing acidity), which results in a shorter fermentation time.  This is desirable because we want as short a fermentation time as possible because soaking the beans leaches flavonoids from the beans into the water, thus reducing the coffee’s flavor complexity and quality. Also, the sooner we free the fermentation tanks, the sooner we can start the next batch.

I plan on revisiting this experiment in July when I can do a better job of documenting everything.

Stay tuned.

Update: I have done a more proper experiment and found that using carryover, or “yesterday’s water”, as I like to call it, reduced soak time by 38% when using 10% starter culture by volume. Experiment details below the video:

For the experiment I split a 56 kilo bag of cherries into two buckets, each containing 28 kilos of cherries. I pulped the cherries with the same pulper and thoroughly mixed the pulped seeds before separating into two buckets.

  • The experiment was performed at 1111MASL.
  • Clean water pH: 7.0 (note the large difference from 9.1 pH observed in the previous experiment!)
  • Starter culture pH: 4.2
  • Bucket #1 received one standardized scoop of starter culture and nine scoops of clean water.
  • Bucket #2 received no starter culture and 10 scoops of clean water.
  • Separate stirring sticks were used for each bucket to prevent cross contamination.
  • Before each pH reading, both buckets received five bottom-up stirs and the meter was rinsed with clean water before every reading.

Below are the individual readings:

Time of reading Bucket 1 pH Bucket 2 pH
7/22/15, 16:13 5.1 6.3
17:27 5.1 5.8
21:42 4.8 5.4
7/23/15, 06:59 4.3 5.2
10:03 4.2 4.8
11:30 4.1 4.7
13:03 4.0 4.6
15:25 4.0 4.4

At 7a on 7/23, the beans in bucket one (with starter) passed the tactile test of feeling sufficiently fermented and ready for mechanical washing, i.e. they felt gritty and rough. Bucket two (no starter) still felt slimy and needed more fermentation time.  By 15:25 bucket number two passed the tactile test and was ready for mechanical washing.

A few things to note:

  • Only keep the water one day. Don't use water from two days ago.
  • Using more starter culture further reduces soak time.
  • Frequent stirring encourages microorganism growth and proliferation and eliminates fermentation hot-spots (bonus: stirring also frees more floaters from the bean-mass).

Using a starter culture in your fermentation process will help standardize the dominate yeast and bacteria strains over multiple batches, will help reduce soak time per batch, and will possibly help normalize the flavor profile (cupping experiments to follow). When access to commercial-grade bacteria and yeast is not an option, back slopping can still improve the process.

Update June 16, 2022: One thrust area from this experiment is to dig deeper into the widely varying pH levels of the water for the two experiments. The water is sourced from a nearby river and from rain water when available. It is stored in a cement water tank that has been lined with paint-on water proofing. Bali has a lot of limestone and that may contribute to a higher alkalinity. But also worth looking into is the variance itself. If mills face such hugely different pH levels from day-to-day or month-to-month, what impact does that have on their product?

 This article is associated with a project: Ulian Project: Improve Processing Technique
Michael C. Wright

Michael is a licensed Q Grader, licensed Q Processor Pro, an Authorized SCA Trainer (AST), and most recently, a graduate with a degree in horticulture and a concentration in horticultural business management. He has over ten years experience in the coffee industry operating on both the supply and demand sides of the value chain.