PropulsionLab · Methodology
What the model does, and what it does not claim.
PropulsionLab is an educational reduced-order cycle simulator. It captures the thermodynamics and the right trends, station by station, so you can reason about how an engine behaves. It is not a high-fidelity design tool, and this page is explicit about where the line is.
The cycle model
Steady, one-dimensional, station-based. Each engine is a chain of stations (inlet, fan, compressor, combustor, turbine, nozzle) carrying stagnation temperature and pressure. There is no spatial flow field, no blade-row geometry, no swirl.
Perfect gas with constant specific heats by region. The working fluid is air, then combustion products, each with a fixed cp and gamma (a cold-side and a hot-side value). An optional Cantera-backed real-gas correction recomputes the whole cycle — compressor, turbine and nozzle — with temperature-dependent (variable-cp) properties and reports it side by side with the constant-cp deck; the default, and the core station table, stay constant-cp.
Isentropic processes with component efficiencies. Compression and expansion are reversible-adiabatic, scaled by an isentropic efficiency you set; the combustor adds heat with an efficiency and a pressure loss. These are the standard preliminary-design idealisations.
Off-design by constant-ratio matching. With a choked turbine and nozzle, the turbine temperature and pressure ratios are constant, so the operating point is found by a spool work balance rather than a full transient. Compressor/turbine maps are offered as a higher-fidelity path, but they are synthetic, illustrative characteristics sized to the design point, not measured manufacturer data.
Per-engine notes
Turbojet is the most fully featured model: dry and afterburning, choked/unchoked convergent nozzle, full station table, off-design matching, single-spool transient spool dynamics, variable-area nozzle scheduling, and an afterburner flame-stability loop. The optimisation, sensitivity and transient analysis tools all run on it.
Turbofan is separate-flow, two-spool, with a bypass stream and an optional third stream; off-design and map matching hold the bypass ratio at its design value. The optimisation, sensitivity and transient analysis tools also run on it; turbofan transient is two-spool (light HP core leads, heavy LP fan lags, thrust follows the fan).
Turboprop adds a free power turbine and a propeller model; reported power is shaft/equivalent-shaft power, and propeller efficiency is a simple model, not a blade-element one.
Ramjet and scramjet are high-speed, compressor-less cycles: the ramjet uses MIL-spec inlet recovery and subsonic combustion; the scramjet burns supersonically with an equivalence-ratio input and dissociation warnings. Both only produce useful thrust above their characteristic Mach numbers, and the model says so.
Things the analysis tools assume
Transient spool dynamics is the bare rotor-inertia response; a real engine adds a fuel-control acceleration schedule that slows spool-up further. Variable geometry and the afterburner stability loop are turbojet-only and are reduced-order estimates, not combustor chemistry. Optimisation, sensitivity and transient run on the turbojet and turbofan decks (the turbofan transient is two-spool).
Off-design matching reports its iteration count and a work-balance residual on every solve, so the convergence quality is visible rather than hidden.
Bleed and HPT cooling are modelled as bulk mass-flow fractions taken at the HPC exit and re-introduced at the turbine inlet — not as channel-resolved cooling-network geometry. The fractions reduce effective Tt4 by mixing; they do not resolve film cooling.
Combustion is by default a constant-pressure energy balance. Two opt-in higher-fidelity paths are available: an equilibrium combustor (Cantera, all major species), and a separate reduced reactor-network emissions module that returns NOx and CO emission indices and an ICAO LTO aggregate. Both degrade to the constant-cp model when Cantera is not installed.
What is not claimed
No manufacturer-level validation. The model has not been matched to flight-test or manufacturer data for a named engine. Presets are public-estimate starting points, clearly labelled "-like", not the real engine's deck.
Not certification-grade. Everything is reduced-order and educational. Treat every number as trend-correct, not a design figure: useful for understanding why thrust rises with turbine temperature or how bypass trades against specific thrust, not for sizing hardware.
No CFD, no real geometry, no 3-D effects. There are no blade rows, no channel-resolved cooling networks (only the bulk fractions noted above), and no spatial flow field. The maps and the afterburner stability loop are synthetic, sized to the design point, not measured component data.
PropulsionLab is an educational project by Ainesh Das. The models, assumptions and limits described here are stated in good faith; nothing on this page or in the tool is a certified engineering result.