How to Reduce Unripe Beans: Harvest and Sorting in Two Stages

When unripe beans get in, cup quality does drop. So how far you can bring down the unripe fraction is the quality of the harvest itself. But you cannot get it to zero at the picking stage alone. At LuLaLao Farm we reduce it in two stages, harvest and sorting.

The two-stage approach to unripe fruit on the farm

"The first stage is picking. We pick only the ripe and leave the still-green fruit. Even so, some unripe fruit gets in. So in the second stage we float the picked fruit in water. Light fruit, that is unripe, insect-damaged or internally damaged fruit, floats, and we remove it. The soaking step is not only about unripe fruit, but that story belongs elsewhere."

Research data: one pass cannot take it all; protect in layers

Unripe fruit cannot be brought to zero at picking. Even mechanical harvesting aimed at ripe fruit dropped 8% of unripe fruit along with the ripe, with a bean damage rate of 5.23% (Yu et al., 2023). So sorting after picking is needed. There is support for it too. In a study predicting flavour from the physical properties of green beans, density and moisture were the most predictive (Gonzalez-Sanchez et al., 2024). Floating fruit in water to remove the light ones is not about appearance; it is a step that separates quality by density.

But sorting is not finished in one pass either. Even with machines there are limits. Near-infrared spectral imaging identified insect-damaged beans correctly with 95% accuracy (Chen et al., 2020). Still not all-powerful. Sorters that separate by colour alone are weak on look-alike unripe and minor defects. Even with spectroscopy, only about 68% of the defective beans in a mix were removed in one pass, and about 84% after two (Berardi et al., 2026). Foreign matter like stones, on the other hand, machines removed 100%. But look-alike defects can only be taken out through several nets: density, colour, spectroscopy, and finally the human eye. Hand sorting tires the eyes in about 20 minutes and accuracy drops (Chen et al., 2020). So we reduce unripe fruit upstream, at harvest and flotation, and then layer the stages.

"Unripe cannot be zero. So we net it again and again. Pick, float, pick out."

Taking it back to the farm

Back to the farm. For each lot we record the estimated unripe fraction and the share removed by flotation (float removal rate). Kept as numbers, they show where to strengthen next year.

"This season, the estimated unripe fraction after sorting was 6%. Flotation removed 8%."

For judging ripeness see Ripeness comes before colour; for why ripeness splits within a tree, Why fruit of different timing shares one tree; for the whole picture, the overview of harvest.

References

• Yu et al. (2023). Agriculture. 10.3390/agriculture13030700
• Gonzalez-Sanchez et al. (2024). Processes. 10.3390/pr12010018
• Chen et al. (2020). Remote Sensing. 10.3390/rs12152348
• Berardi et al. (2026). Exploration of Foods and Foodomics. 10.37349/eff.2026.1010148