What must a fed-batch process simulation calculate?

A decision-useful model calculates time-dependent liquid volume, biomass or viable cells, substrate and feed components, product, metabolites, dissolved gases, heat, additions, harvest and equipment constraints. It also reports numerical residuals, parameter sources, uncertainty and the conditions under which the model is valid.

How should a fed-batch model be validated?

Fit parameters to identified development batches, reserve independent batches or operating conditions for validation, and predefine acceptance criteria for trajectories, final titer or product, material closure, peak oxygen and heat demand, event timing and downstream load.

Fed-batch process simulation

Build a time-resolved fed-batch model that explains the plant load.

Connect feed policy, growth, productivity, metabolites, liquid volume, oxygen and heat limits, harvest timing, downstream demand, equipment occupancy, cost and validation evidence in one simulation basis.

Inspect the CHO benchmark Evaluate your fed-batch process

Dynamic reactor states must resolve into material demand, equipment duty and downstream load.

Calculation basis

A feed profile is only useful when its states, events and physical boundaries are explicit.

Freeze the organism or cell line, medium and feed composition, inoculum, initial and maximum working volume, temperature, pressure, pH strategy, gas train, agitation policy, batch events, harvest criterion and intended decision before fitting parameters.

Biological states

Represent biomass or viable and total cells, limiting substrates, nitrogen source, product, metabolites, dissolved oxygen and carbon dioxide, volume and relevant quality proxies. Separate growth, maintenance, production, death, degradation and transfer terms so the model cannot hide two effects inside one fitted yield.

Operating policy

Define each bolus and continuous feed by composition, flow, start condition and actuator limit. Include base, acid, antifoam, sampling, evaporation, gas composition, backpressure, agitation, cooling and induction or temperature-shift events. Event logic must use the same timestamps as the imported batch data.

Whole-process consequences

Translate the simulated harvest volume, titer, biomass, impurities and timing into centrifuge or filter load, chromatography cycles, buffer and consumable demand, hold times, schedule occupancy, utility peaks, waste, COGS and LCA inventory. This is where a reactor optimization becomes a factory decision.

Fed-batch state balanced(VC_i)/dt = F_inC_in,i − F_outC_i + Vr_i

Use the corresponding volume balance dV/dt = F_in − F_out − F_evap. Concentration equations must retain dilution, reaction, gas transfer, sampling and bleed terms. Numerical integration should report solver settings, event handling, residuals and failed physical checks.

ConstraintCalculate over timeRequired evidence

Oxygen and gasOUR, OTR, kLa margin, gas rate, enrichment and exhaust loadOUR or off-gas data, kLa method, vessel and gas-train limits

Mixing and feed zoneBlend time, local substrate exposure, osmolality and pH riskTracer data, feed-port location, scale-down or qualified CFD

Heat and volumeMetabolic heat, cooling duty, working-volume and foam marginCalorimetry or heat reconciliation, UA, coolant and operating records

Downstream interfaceHarvest mass, solids, impurity burden, cycles and hold-time riskAssays, recovery data, equipment limits and schedule rules

Calibration packageIdentified batches · parameter priors and intervals · residual plots · covariance or identifiability · validity domain · versioned solver configuration

Independent validationReserved batches or conditions · predefined trajectory and endpoint tolerances · mass closure · event timing · equipment duty · documented deviations

Decision outputLow/base/high scenarios · limiting interval · active constraint · sensitivity ranking · data gaps · screening, calibrated, validated or not decision-ready status