Ben-Gurion University study tests sensors for tracking solids in biofloc aquaculture

Ben-Gurion University researchers reported on September 30 that a six-month sensor trial in a biofloc aquaculture reactor could help operators track changes in suspended solids between slower physical measurements. The study, published in Aquacultural Engineering, found that estimates could guide alerts, but its authors say they need reactor-specific calibration and should not alone control solids removal or carbon dosing.

The trial took place in one solids-rich side-stream biofloc reactor. It did not test the system across multiple commercial farms. That distinction matters for anyone considering whether the reported accuracy would carry over to a different reactor, with different water conditions or operating practices.

Why suspended solids matter in biofloc systems

Biofloc systems use microbial communities as part of aquaculture production. A separate 2024 scientific review describes how those communities can help recycle nitrogen compounds and provide natural food to cultured animals. Their solids, however, also require management: the review says high concentrations can harm fish, and reported levels vary across species and production conditions.

Total suspended solids are a core indicator of microbial biomass in the biofloc-based recirculating system studied by the Ben-Gurion team. The conventional gravimetric method requires a physical sample and is too slow to give a continuous, real-time reading. The sensor system was designed to estimate the indicator while operators wait for those direct measurements.

What the six-month sensor trial measured

Robel Kahsu Adane, Orel Rahamim, Elyasaf Freiman, Amit Gross and Elad Levintal evaluated an open-source, low-cost Internet of Things monitoring system. For six months it measured pH, oxidation-reduction potential, dissolved oxygen, turbidity, electrical conductivity and temperature in the reactor. The published abstract gives no checked purchase price for the equipment.

After confirming sensor stability, the team conducted suspended-solids sampling campaigns and built two regression models. One estimated solids from turbidity alone. The other combined turbidity with electrical conductivity and temperature. Both approaches used sensor readings to predict a measure that operators would otherwise obtain through physical sampling.

During cross-validation in model development, the three-variable model achieved an R² of 0.77, compared with 0.65 for the turbidity-only model. Its mean absolute error fell from 599 to 514 milligrams per litre; root mean square error fell from 796 to 645 milligrams per litre. Those figures describe the model-development comparison, rather than a demonstrated accuracy for other farms.

Why field accuracy still needs checking

A separate field validation complicated the comparison. According to the study abstract, performance changed with the validation window, and the simpler turbidity-only model had lower mean absolute error under narrower local operating conditions. Adding sensor inputs therefore improved the reported cross-validation scores without making the more complex model consistently better in the separate field check.

The authors propose using estimated solids for trend monitoring, threshold guidance and early warning of solids accumulation or biomass loss. They explicitly caution against relying on the prediction as the sole automatic control parameter for solids removal or carbon dosing. Their recommended safeguards are calibration for the particular reactor and periodic confirmation by gravimetric measurement.

That advice fits the range of conditions described in a separate 2023 meta-analysis of 33 studies published between 2012 and 2020. It examined biofloc effects on water-quality measures, including suspended solids, and reported differing findings. It noted that high concentrations can stress shrimp respiration, while very low concentrations can also complicate management in some systems.

The meta-analysis also says biofloc performance can depend on stocking density, feeding regime and management practices. Those findings provide context for the new study’s call for local calibration. They do not establish that this sensor model works outside the reactor in which it was tested, or identify a universal solids threshold for operators to use.

What operators can take from the result

For a biofloc operator, the practical prospect is a continuous estimate that may flag a change before the next physical solids measurement. The study supports that use as a monitoring aid, subject to checks. Its variable field results leave the accuracy of a particular reading dependent on local conditions and calibration.

The accessible study record contains the abstract and publication details. The publisher’s full paper could not be retrieved for this report, so detailed sampling counts, sensor prices and the complete field-validation tables could not be checked. The available evidence also does not establish a commercial deployment, cost saving, animal-health benefit or environmental improvement from this system. Those outcomes remain separate questions from whether its readings can help track solids trends.