SimOps
Simulation and Optimisation Suite for Bio-Manufacturing Systems
synopsis
SimOps is a general-purpose simulation and optimisation suite for bio-manufacturing and cyber-physical systems. Sensor observations are normalised via the W3C SOSA/SSN ontology — decoupling the intelligence layer from any specific hardware — and routed to a four-agent computational pipeline: cascade (KPI computation), predictor (kinetic model forecasting), optimizer (parameter recommendation), and bioreactor modeler (model fitting and validation).
The suite was originally developed for photobioreactor management (Chlorella algae production, caloric ROI, LCC optimisation) and has been extended to SCOBY fermentation via the Ambu bioreactor. Because the SOSA abstraction is hardware-agnostic, SimOps can be deployed against any manufacturing context that produces sensor observations — replacing bespoke instrumentation logic with a unified agent interface.
agent pipeline
| simops_cascade | Computes KPI cascade from live or synthetic SOSA observations: NER, SEC, LCC, harvest energy intensity, fermentation efficiency, organic acid yield. |
| simops_predictor | Fits a logistic / exponential kinetic model to historical observations and forecasts the remaining session trajectory, harvest window, and yield with confidence intervals. |
| simops_optimizer | Compares current operating point against the Pareto-optimal frontier and recommends parameter adjustments (setpoints, duty cycles, flow rates) to improve the target objective. |
| bioreactor_modeler | Fits Monod-type kinetic models (μ_max, K_s, K_i) to session data, stores ODE parameters as JSONB, reports R² and RMSE, and flags model drift between sessions. |
architecture
Physical sensors / synthetic generator
│
▼ SOSA normalisation
sosa_observation (TimescaleDB hypertable)
│
├─▶ simops_cascade → KPI strip
├─▶ simops_predictor → harvest forecast + confidence
├─▶ simops_optimizer → parameter delta recommendations
└─▶ bioreactor_modeler→ kinetic fit · R² · RMSE
│
▼
sosa_observation (results stored as
typed observations for full audit trail)
All agent outputs are written back as SOSA observations, giving the system a complete, queryable history of both sensor readings and computational results in a single table.
reference deployments
| Deployment | Context | Primary sensor | Status |
|---|---|---|---|
| Chlorella PBR | Algae biomass production, caloric ROI optimisation | OD600, PAR, pH, temp | Synthetic · Digital Twin |
| Ambu SCOBY | SCOBY fermentation, kombucha batch tracking | Brix, temp | Synthetic · MCU pending |
related research
tags
Digital Twin — Interactive Tool
Two reference deployments (Chlorella PBR + Ambu SCOBY) with synthetic SOSA observation streams, D3 time-series charts, a time scrubber, and live SimOps agent intelligence panels. Designed to become a live dashboard when the Ambu M5Stack MCU is commissioned.