Getting started

SITL mission workflow

The SITL page explains how to connect an exported powertrain to an autopilot for simulated mission testing. Integration details opens the firmware setup, vehicle configuration, and recorded examples. Export requires Pro and a completed single-point, single-pack simulation. Compare controller behavior and battery use within the exported operating envelope; the vehicle geometry and mass properties also affect flight predictions.

Validation evidence overview

The validation summary compares measurements with predictions from the named release and engine build. PROM 5.3 uses coefficient mean absolute error (MAE), rather than propeller percentage error, as its primary statistic. All captured zero and negative coefficients remain included. These are component regression benchmarks, not a fresh blind study or an aircraft flight-time guarantee.

In Compare the actual performance, select a propeller, static or forward-flight condition, and an exact measured RPM. Forward-flight plots show Cₜ and Cₚ versus advance ratio, which describes forward speed relative to propeller rotation. Static plots show thrust in newtons and shaft power in watts versus RPM. Dots are measurements; the line joins predictions at those same conditions. It is not a fitted curve or an extrapolation. Use the Measurement slider, arrow keys, or plotted dots to inspect one test point. Newton/watt readouts are converted from coefficients using that point's RPM, diameter and the capture's reference air density of 1.225 kg/m³.

The Lowest error, Median example and Highest error buttons rank propellers by the average of their Cₜ MAE and Cₚ MAE over all of their captured points. Ties are alphabetical. The initial example is the median-ranked propeller; its initial forward-flight sweep has the most measured points, with ascending RPM breaking ties. Every propeller is available in the selector.

Under Agreement across the dataset, family and operating-condition filters update the sample counts, MAE, 95th-percentile absolute error, parity plots and empirical cumulative error curves. Parity axes use identical scales, and the diagonal means perfect agreement. Families use different colors and shapes. Cumulative curves include every sample and show the fraction at or below each absolute error. The 50th and 95th percentiles use the nearest-rank rule. All statistics weight operating points equally, so propellers with more rows contribute more. No average or percentile is a guaranteed error bound.

The technical report adds signed-error plots. Choose Advance ratio J or RPM as the horizontal axis, and use the family and operating-condition filters to isolate patterns. Error is prediction minus measurement: positive means over-prediction and negative means under-prediction. Full tables retain the original test/development groups separately. Historical reports retain their original metrics and scope. Motor current and speed continue to use percentage errors; their datasets remain separate.

Pricing billing interval

On Pricing, Monthly and Annual update the Pro price and checkout destination together. Tab to the control and use arrow keys to change the interval; Home selects Monthly and End selects Annual. Read the billing cadence and availability note before continuing. The interval does not change plan capabilities.

ThrustLab predicts how an electric drone powertrain will perform before you build it. You describe the parts — a propeller, a motor, an electronic speed controller (ESC), and a battery — and ThrustLab simulates how much thrust they make, how much power they draw, and how hot they get. This page explains how a simulation works and how to read what comes back.

If you have never run one before, start here: pick a propeller, motor, ESC, and battery, press run, and read the thrust and power numbers. Everything else on this page fills in the detail behind those numbers.

How a simulation works

A simulation answers one question: at a given throttle, in a given airspeed, what does this powertrain do? You give ThrustLab four components and an operating point, and it solves the physics that couples them together.

• Propeller — converts shaft power into thrust. Its diameter and pitch set how much air it moves.
• Motor — turns electrical power into shaft power. Its velocity constant (Kv) sets how fast it spins per volt.
• ESC — feeds the motor the right current. It sets switching behavior and adds a small loss.
• Battery — supplies the voltage and current, and sags under load.

ThrustLab balances all four at once: the propeller demands torque, the motor and ESC supply it from the battery, and the battery voltage drops as current rises. The result is a single self-consistent operating point — the thrust, current, power, and efficiency you would actually measure on a bench.

Reading your results

Every result reports the same core numbers so you can compare designs directly. Thrust is the force the propeller produces, usually in grams or newtons — higher is more lift. Power is the electrical power drawn from the battery in watts — lower is better for the same thrust. Efficiency ties them together as thrust per watt; a higher value means a longer flight on the same battery.

Example simulations

The homepage links to five complete public examples: a freestyle quad sweep, a tricopter transition, a fixed-wing cruise point, a heavy-lift octocopter hover point, and an endurance hexacopter sweep. These are saved runs produced by the solver, not illustrative marketing numbers; the engineering-use disclaimer still applies.

Homepage account actions

Sign up and Start free open account registration. When these buttons scroll above the page, they appear in the header. Signed-in users see Go to dashboard instead.

Homepage motion

The homepage uses Blender-rendered videos with manufacturer geometry for the motor, propeller, and airframe. The videos illustrate the hardware; flight and spin-up are visual presentations, not simulated operating results. The propeller's outer blade follows APC station data and standard airfoil profiles; its hub uses published dimensions, with a modeled root transition. Battery and ESC shapes are reconstructed from product photographs and carry no manufacturer branding. Use the Pause animations button beneath the cards to stop both videos and Play animations to resume. Reduced-motion preferences show stills automatically. Videos pause when offscreen or when the tab is hidden. No WebGL is required.

The drone sequence shows arming and idle before takeoff, flight out of frame, return and landing while the rotors remain powered, then disarming and spin-down. Red port-side and green starboard-side lights identify the aircraft's sides. The close-up powertrain sits on a modeled bench test rig, with a spinner nut and the propeller seated against the adapter. These are presentation scenes, not experimental footage.

Homepage simulation example

The output showcase contains screenshots of the current public result-page UI, captured September 7, 2026, displaying the five saved examples. These runs are not recomputed when visiting the homepage. Single point, Sweep, and Dynamic select a simulation type. The five small selectors choose individual examples, and Open example opens the full result. Screenshots advance every six seconds while visible. Hover, keyboard focus, manual selection, or the pause button stops the slideshow; reduced-motion preferences disable automatic cycling. On small screens the screenshot can scroll horizontally so its text remains readable.

Homepage validation link

Validation report opens the latest published report with its measured data, methodology, and scope.

Propeller model card

Hover, focus, or tap Propeller to reveal its model summary. Blade-element momentum analysis uses blade geometry and section aerodynamics, accounting for local angle of attack, Reynolds number, and Mach. Its outputs include thrust, torque, power, and aerodynamic efficiency.

The card lists the complete component output inventory, grouped into performance, blade-station results, oblique/transient flow and diagnostics. Each card's panel scrolls independently. For keyboard access, open the card, press Tab to focus its output panel, then use arrow or Page Down keys to scroll. The lists include model-level quantities; saved result tables vary by analysis mode. Aliases for the same quantity are grouped together.

Motor model card

Hover, focus, or tap Motor to reveal its model summary. The d–q electromagnetic model includes back-EMF, torque, winding resistance, and inductance, with copper, iron, friction, and PWM-related losses. Thermal behavior includes winding and magnet temperatures.

Its scrollable output list covers shaft/electrical performance, detailed d–q currents and voltages, back-EMF, resistance, the returned loss terms, and thermal/transient quantities. Model-level electrical readouts are labelled separately from saved-run quantities.

Battery model card

Hover, focus, or tap Battery to reveal its model summary. The cell-based Thévenin model uses state of charge, open-circuit voltage, internal resistance, and RC polarization, with series/parallel pack composition and core/case temperatures.

The output list includes pack and cell electrical quantities, capacity and estimated endurance, per-cell Joule and reversible entropic heat, core/case temperatures, polarization histories, and per-pack current/power when multiple packs are enabled.

ESC model card

Hover, focus, or tap ESC to reveal its model summary. The averaged inverter model includes six-step and field-oriented modulation, PWM frequency, timing, synchronous rectification, and conduction, switching, and dead-time losses. It does not claim to be a transistor-level time-domain circuit simulator.

The output list separates bus/terminal performance, the returned inverter/PCB/ripple/freewheel loss breakdown, and timing/phase-angle/torque-per-amp quantities. These groups distinguish the component readout from quantities available in saved results.

A reasonable hover point for a 5-inch quad rotor sits near 1:1 thrust-to-weight at part throttle, drawing tens of watts per motor. If a number looks far outside what you expect, check the component specs you entered before trusting the prediction.

Accuracy & fidelity

A simulation is a prediction, not a measurement — it is best used to compare designs and find the right ballpark, then to confirm the final build on a bench. ThrustLab couples four physics models — the propeller aerodynamics validated against wind-tunnel data — but the result can only be as accurate as the parts data you feed in.

Where the numbers come from:

• Propeller — ThrustLab uses a vortex-BEM solver: a Drela vortex-based blade-element model that closes the local inflow at each blade station and corrects for tip Mach and compressibility. It is validated against the UIUC propeller database over 17 two-bladed APC Thin-Electric propellers, 12–21 in diameter, pitch/diameter 0.44–0.83, at advance ratios up to J ≈ 0.95 and tip Mach up to 0.43 — propellers far outside that envelope are extrapolations, not validated predictions. Because the model works the blade station-by-station, it flags any stations that did not converge or that are running past drag divergence.
• Motor & ESC — an electromagnetic + thermal model predicts current, RPM, losses, and winding/magnet temperatures. Its accuracy depends on the motor's Kv, resistance, and pole count being correct — published specs vary, so a motor with good data predicts better.
• Battery — voltage sag and drain are modeled from the pack's capacity, resistance (or C-rating), and chemistry. A pack with an accurate internal resistance sags realistically under load.

Two practical rules of thumb. First, garbage in, garbage out: if a motor's published resistance or a prop's pitch is wrong, the prediction inherits that error — sanity-check the specs you entered before trusting an answer that surprises you. Second, ThrustLab flags when it is operating outside its comfort zone (a convergence warning, blade tips past drag divergence, or a battery running hot), so treat those warnings as a signal to look closer rather than ignore.

Next steps

Once you can read a single run, the rest of the User Guide goes deeper: the Glossary defines every term, New simulation explains each input field, Results explains each output, and Warnings explained covers what to do when a run flags a problem.

Units (Metric or Imperial)

The Units toggle on your Account page chooses whether inputs and results are shown in Metric or Imperial. It is an account-wide preference: changing it converts every unit-bearing field and result across the whole app at once — thrust, temperature, airspeed, torque, weight, and so on — so you never have to pick a unit per field. New accounts start on Imperial.

Switching units never changes the underlying numbers. ThrustLab always stores your data in metric and converts only for display, so toggling back and forth is lossless — the same simulation reads "1000 gf" in metric and "2.20 lb" in imperial, but the stored thrust is unchanged.

When to change it. Set it once to whichever system you think in. Find it on the Account page under your plan; the change takes effect across every page after a quick refresh.

Request a password reset

Select Forgot password? on the sign-in page, enter your account email, and select Send reset link. ThrustLab always shows the same confirmation message, even if the address is not attached to an account. This protects account privacy. If the account exists and is active, the email contains a link that expires in one hour.

If you make several requests in a short period, ThrustLab may ask you to wait before trying again. Check spam or junk folders before requesting another link.

Choose a new password

Open the link from the reset email and enter the new password twice. Passwords must be 8–128 characters. A reset link works only once: after the password changes, opening the same link again shows an invalid-or-expired message. Use Request a new reset link if the link is expired, malformed, or already used, then sign in explicitly with the new password.