Verification and calibration#

What the automated tests establish#

  • Agreement with independent arithmetic examples for generation, usable heat, fuel capture, thermal conversion, VS destruction and COD conversion.

  • Generator-boundary conservation across capacity limits and availability values.

  • Correct limiting behaviour for zero fuel, zero capacity, complete outages and net-negative operation.

  • Rejection of invalid units-as-strings, non-finite numbers, impossible fractions, incompatible energy yields and unknown input fields.

  • Consistent interval aggregation, deterministic seeded sampling and immutable source scenarios during experiments.

  • Executable example scripts, CLI JSON/CSV output and package entry points.

  • Plot data preserve signed net electricity, heater input versus delivered heat and unequal interval durations. PNG/SVG/PDF files render, the worked example reproduces its numerical outputs, and core simulation runs without the plotting dependency.

These are software verification checks. They do not establish empirical accuracy. All seven included scenarios use synthetic parameters and must be calibrated before their outputs can be interpreted as a prediction of specific equipment.

Calibrating a scenario#

  1. Define the interval, feedstock sampling method and measurement conditions.

  2. Establish wet mass, moisture basis, dry-feed LHV, ash basis, VS/COD or dry normal gas properties, as appropriate to the route.

  3. Obtain conversion yields with their exact denominator and standard conditions. Convert VS-fed yields to VS-destroyed yields when required.

  4. Reconcile gross and net efficiencies, LHV versus HHV, power rating and operating availability with the generator boundary used here.

  5. Account explicitly for pumps, preparation, gas cleaning, compression and other electrical consumers. Set heat demand and temperature-compatible heat recovery.

  6. Compare measured fuel, gross electricity, recovered heat, auxiliary use and net electricity separately. Matching only net electricity can hide compensating errors.

  7. Use independent observations for validation. Record parameter sources and their uncertainty; examine sensitivity and correlated inputs before interpreting Monte Carlo quantiles.

Continuous biogas plant verification#

The continuous plant model has independent analytical checks for steady and transient first-order substrate pools, VS/gas conservation, stored-gas release after outages, heating loads and grid import/export accounting. Restarted simulations preserve state, and zero-feed cases retain residual gas production. These checks establish numerical consistency, not empirical plant performance.

BMP and financial screening verification#

The banana-residue analysis tests the VS-fed BMP conversion, separate capture/availability deductions, annual heat/electricity accounting and outage imports against independent arithmetic. Financial tests cover year-zero costs, discounting, replacement/closure timing, payback, undefined ratios and break-even NPV closure. The analysis runner is checked from a different working directory, and its committed reference results are compared with a fresh run.

These checks do not validate transferring a laboratory BMP from Gran Canaria to Tenerife, a continuous mono-digester, equipment prices or local project economics.

Extension boundaries#

Detailed reactor kinetics, chemical equilibrium, tar/contaminant handling, part-load efficiency curves, microbial startup, full product mass balances and emissions require additional models. The continuous plant module adds simple gas-holder dispatch and first-order substrate transients with explicit limitations. Add each as an explicit module with its own boundary and reference cases rather than embedding undocumented corrections in the existing recovery factors.