IndPenSim benchmark dataset
🧪 IndPenSim dataset
100 fed-batch runs with 2,238 measured variables — a benchmark dataset originally developed to study modern monitoring and control challenges in biopharmaceutical manufacturing. In the STAMM demo, we focus on the 39 process variables; the Raman spectra aren't used in the streaming pipeline.
What's inside
Nonlinear behaviors, time delays, and noise consistent with real industrial fermentation.
2,199 spectral points per sample — kept aside in the streaming pipeline for now.
Batches operated under different control strategies and fault scenarios — useful for drift testing.
Process variables
The IndPenSim benchmark exposes 39 process variables that combine manual, automatic, online, and offline measurements. Table 1 below summarizes the main process variables, their role, and their physical units. These variables are directly available in the STAMM Demo as part of the virtual data stream.
| Variable name from raw data | Column type | Units | Variable name in InfluxDB |
|---|---|---|---|
| Time (h) | Time | Hour | experiment_time |
| Aeration rate (Fg:L/h) | Actuator | Fg·L⁻¹·h⁻¹ | aeration_rate |
| Agitator RPM (RPM:RPM) | Actuator | RPM | agitator |
| Sugar feed rate (Fs:L/h) | Actuator | L/h | sugar_feed_rate |
| Acid flow rate (Fa:L/h) | Actuator | L/h | acid_flow_rate |
| Base flow rate (Fb:L/h) | Actuator | L/h | base_flow_rate |
| Heating/cooling water flow rate (Fc:L/h) | Actuator | L/h | heating_cooling_water_flow_rate |
| Heating water flow rate (Fh:L/h) | Actuator | L/h | heating_water_flow_rate |
| Water for injection/dilution (Fw:L/h) | Actuator | L/h | water_for_injection/dilution |
| Air head pressure (pressure:bar) | Sensor | bar | air_head_pressure |
| Dumped broth flow (Fremoved:L/h) | Actuator | L/h | dumped_broth_flow |
| Substrate concentration (S:g/L) | Actuator | g/L | substrate_concentration |
| Dissolved oxygen concentration (DO2:mg/L) | Sensor | mg/L | dissolved_oxygen_concentration |
| Penicillin concentration (P:g/L) | Soft sensor | g/L | penicillin_concentration |
| Vessel Volume (V:L) | Computed variable | L | vessel_volume |
| Vessel Weight (Wt:Kg) | Computed variable | kg | vessel_weight |
| pH (pH:pH) | Sensor | pH | pH |
| Temperature (T:K) | Sensor | Kelvin | temperature |
| Generated heat (Q:kJ) | Sensor | kJ | generated_heat |
| Carbon dioxide percent in off-gas (CO2outgas:%) | Sensor | % | CO2_percent_in_off_gas |
| PAA flow (Fpaa:PAA flow (L/h)) | Actuator | L/h | PAA_flow |
| PAA concentration offline (PAA_offline:PAA (g L^-1)) | Offline measurement | g·L⁻¹ | PAA_concentration |
| Oil flow (Foil:L/h) | Actuator | L/h | oil_flow |
| NH₃ concentration off-line (NH3_offline:NH3 (g L^-1)) | Offline measurement | g·L⁻¹ | NH3_concentration |
| Oxygen Uptake Rate (OUR:(g min^-1)) | Offline measurement | g·min⁻¹ | oxygen_uptake_rate |
| Oxygen in percent in off-gas (O2:O2 (%)) | Sensor | % | oxygen_in_percent_in_off_gas |
| Offline Penicillin concentration (P_offline:P (g L^-1)) | Offline measurement | g·L⁻¹ | offline_penicillin_concentration |
| Offline Biomass concentration (X_offline:X (g L^-1)) | Offline measurement | g·L⁻¹ | offline_biomass_concentration |
| Carbon evolution rate (CER:(g h^-1)) | Sensor | g/h | carbon_evolution_rate |
| Ammonia shots (NH3_shots:kg) | Actuator | kg | ammonia_shots |
| Viscosity offline (Viscosity_offline:centPoise) | Offline measurement | centPoise | viscosity |
| Batch ID | Experiment | – | experiment_ID |