Continuous biogas plants#

Simulate mixed organic feeds, a continuously stirred tank reactor (CSTR), a gas holder and combined heat and power (CHP). This model carries substrate and gas inventories between intervals. It complements the independent, steady-interval simulate_digestion and simulate_profile methods.

The three supplied configurations are synthetic technical examples. They represent common agroindustrial and livestock feed combinations, not measured plants. The implementation is a reduced-order research model; detailed biological and equipment validation remains to be done and errors may remain.

Run a plant#

python -m pip install -e '.[plots]'
python use_cases/biogas_plant.py --plant agroindustrial_chp --days 30 --plot
python use_cases/biogas_plant.py --plant dairy_manure_chp --days 30 --plot
python use_cases/biogas_plant.py --plant swine_slurry_chp --days 30 --plot

Outputs under results/biogas_plant/<plant>/ include a PNG profile, CSV intervals and a JSON record containing the entire configuration, schedule, initial state, final states and summary. Omit --plot to run without Matplotlib. There is no random sampling. The worked gallery example walks through the model and compares storage against the same plant without storage.

Configuration

Wet feed

Gross CHP rating

Equivalent digester

Gas holder

agroindustrial_chp

100 t/day: food residues, slurry, sludge

1,000 kW

3,500 m³

2,500 Nm³

dairy_manure_chp

120 t/day: dairy manure and food residues

500 kW

4,000 m³

1,500 Nm³

swine_slurry_chp

85 t/day: swine slurry and food residues

250 kW

2,500 m³

800 Nm³

All feed properties, decay rates, vessel sizes, gas-holder sizes, efficiencies and loads in these configurations are assumptions to vary or calibrate. Nameplate rating is a limit, not a prediction of continuous generation. Their differing feed composition and throughput prevent a direct performance ranking.

The example uses hourly steps. A 12-hour CHP outage begins at hour 120, and all feeds stop from hour 240 to 288. These events only occur when the requested run is long enough. Site demand and weather inputs are synthetic constants. Digestion continues during the outage; heat and electrical demands remain accounted for.

Inputs and state#

Use DigesterFeedstock, ContinuousDigester, BiogasPlant and PlantInterval from x2power.methods.biogas_plant.

from x2power.methods.biogas_plant import PlantInterval, simulate_biogas_plant
from x2power.utilities.biogas_plant import plant_from_config, summarize_plant
from x2power.utilities.io import load_scenario

config = load_scenario("use_cases/plants/agroindustrial_chp.json")
plant = plant_from_config(config)
results = simulate_biogas_plant(plant, [PlantInterval() for _ in range(24)])
print(summarize_plant(results)["net_electricity_kwh"])

# Continue with exactly the same plant definition and feedstock order.
next_day = simulate_biogas_plant(
    plant, [PlantInterval() for _ in range(24)], initial_state=results[-1].state
)

Plant JSON uses model: continuous_biogas_plant and schema_version: 1; it is a separate configuration from the existing x2power run route scenarios. Pass it through plant_from_config, or use the example runner above. Feed rates are kg wet/day, unlike the interval totals accepted by the older route models. feed_multipliers follows the order of plant.feedstocks; an empty tuple uses all base rates. A zero multiplier stops a feed without removing its reactor inventory.

By default, substrate pools start at steady state for the first interval’s feeds, and the gas holder starts empty. Use initial_plant_state(..., steady_state=False) for zero substrate inventory in an already full, warm digester. This does not simulate microbial inoculation, startup heating or vessel filling. Initial gas must be declared explicitly and cannot exceed holder capacity.

Reactor equations and boundaries#

For feed i, let wet flow be W_i (kg/day), density rho_i (kg/m³), wet-basis solids fraction TS_i, volatile fraction of solids VS_i, biodegradable fraction of VS b_i and first-order conversion rate k_i (1/day). Let V be working volume.

\[Q = \sum_i W_i/\rho_i,\qquad D = Q/V,\qquad \mathrm{HRT} = V/Q,\qquad \mathrm{OLR} = \frac{\sum_i W_i TS_i VS_i}{V}.\]

At zero flow, HRT is undefined and reported as None / JSON null. The model assumes constant liquid volume, equal inlet/outlet flow and perfect mixing. It neglects gas-volume effects on liquid hydraulics. There is no solids separation, recycle or independent solids retention time. Parallel tanks can be represented as one equivalent mixed volume only when that approximation is justified; tanks in series and post-digesters need separate models.

Each feed has a degradable VS pool M_i and an inert VS pool I_i, in kg:

\[\frac{dM_i}{dt} = W_i TS_i VS_i b_i - (k_i+D)M_i,\qquad \frac{dI_i}{dt} = W_i TS_i VS_i(1-b_i)-DI_i.\]

Inputs are constant within an interval. The code integrates each linear equation analytically, including its integrated reaction and effluent flows. Thus:

\[M_{i,1}=M_{i,0}e^{-(k_i+D)\Delta t} +\frac{W_i TS_i VS_i b_i}{k_i+D}(1-e^{-(k_i+D)\Delta t}).\]

The zero-rate limit is handled explicitly. Destroyed VS is the integral of k_i M_i; methane equals destroyed VS times the declared yield. This yield is Nm³ CH4/kg VS destroyed, not biogas per kg wet feed, per kg VS fed, or a batch BMP measurement. Do not insert a BMP value without reconciling the denominator and assumed biodegradability. A feed-level upper-potential check rejects methane energy greater than the dry feed’s declared LHV. This is a plausibility check, not a complete reactor energy balance.

The VS ledger is initial inventory + feed = destroyed + effluent + final inventory. Effluent VS is reported, but total digestate wet mass, nutrient content, pathogen removal and fertilizer properties are not predicted. The first-order equations are an explicit project approximation, not an implementation of ADM1. They do not resolve pH, microbial growth, ammonia/VFA inhibition or temperature kinetics. Configured HRT/OLR limits generate diagnostic flags without changing the rates.

Gas, CHP, heat and electricity#

All gas volumes use a dry normal reference of 0 °C and 101325 Pa. Holder capacity is an equivalent normal gas volume, not a geometric membrane capacity. A constant methane fraction converts methane to biogas; capture losses are reported separately as methane not captured. Cleaning electricity is included, but contaminant removal and gas composition changes are not calculated.

Each interval uses its average captured gas production for dispatch. CHP demand is limited by rating × setpoint fraction × availability × elapsed time. Available fuel comprises initial stored gas and captured production. The engine uses up to its demand; remaining gas fills the holder, and excess goes to an unlimited ideal flare. The ledger is initial storage + captured = CHP + flare + final storage. There are no holder flow limits, pressure dynamics, storage leaks, minimum engine load or startup costs. Flared volume is a destination flow, not methane emitted. Fractional availability is an equivalent derating; resolve actual outages with separate 0/1 intervals. Check sensitivity to shorter time steps when feeds change rapidly or the holder empties/fills. Within-interval gas fluctuations are not resolved.

The engine has constant LHV-based electrical and recovered-heat efficiencies. Recovered heat supplies the digester first. Heating demand is:

\[H_{feed}=\frac{m_{wet}c_p\max(T_d-T_{feed},0)}{3600},\qquad H_{loss}=UA\max(T_d-T_{ambient},0)\Delta t_h.\]

c_p is kJ/(kg K), UA is kW/K and each heat term is kWh. Fixed digester temperature is maintained by an unlimited electric backup heater when recovered heat is insufficient. Its electricity is deducted from output, including during outages. There is no thermal storage, cooling load, boiler or pasteurization step. Any upstream sanitization demand must be modelled separately. Remaining recovered heat can serve declared site heat demand; unmet site heat and rejected heat are explicit.

Plant net electricity deducts mixing, fixed auxiliaries, feed preparation, gas treatment and backup heating. Site electricity demand is then subtracted to split positive grid exports from grid imports. Export is unconstrained, without a grid connection limit or demand tariff. Negative plant net output remains visible. The CHP follows the declared setpoint; it is not an optimizing dispatch controller.

Technical sources and validation#

These references establish the process context and measurement definitions; they do not validate the synthetic configurations or supply their numerical parameters:

  • U.S. EPA AgSTAR, Project Development Handbook, third edition, April 2020, EPA 430-B-20-001: chapters 3–5 on digestion, feedstocks and gas/CHP equipment. Official publication.

  • U.S. EPA AgSTAR, Anaerobic Digester/Biogas System Operator Guidebook, November 2020, EPA 430-B-20-003: sections 3.1–3.5 on retention, loading, temperature and methane potential; section 7.4 on gas utilization and CHP heat recovery. Official PDF.

  • Bernhard Drosg, Process monitoring in biogas plants, IEA Bioenergy Task 37, 2013: biological stability and the need for plant-specific monitoring. Official report page.

Tests compare steady-state and transient results against independent analytical solutions and check VS, gas, electricity and heat ledgers, storage limits, outages, state continuation, small steps and invalid inputs. Before treating results as plant predictions, calibrate feed TS/VS, yields, rate constants, measured gas composition, heat losses, engine curves and auxiliary loads. Validate against independent measured intervals, including disturbances. See validation guidance.