Move clearly from Study to Explore to Compare.

Choose a Study model

Studio starts with Half Car · 4 DOF, the flagship model for coupled heave and pitch, front and rear suspension, and explicit axle roads. Quarter Car · 2 DOF remains a simpler foundational model. Both support Transient and Frequency Study and Compare, Transient Explore, and completed-evidence exports. Transient Study and Compare also support Motion; Study Motion can export local video. Follow the Half-Car Study guide.

Three focused workspace modes

Study is the default and pairs compact configuration controls with the dominant result area. Explore replaces that workspace with range controls and sweep evidence. Compare replaces it with Baseline/Variant A setup and paired outputs. Switching modes does not create new browser routes or persistence.

Run provenance and Model & assumptions are keyboard-accessible disclosures near the workspace. Common physical and road settings remain visible; initial state, duration, RK4 step, and detailed range guidance remain available under Advanced settings.

Within Study, choose Transient for prescribed road events and RK4 time histories, or Frequency for normalized steady-state harmonic response. The same physical parameters feed both analyses; road shape, duration, initial state, and RK4 step remain Transient-only settings. Compare uses the same secondary Transient/Frequency selector. Explore remains Transient-only and does not advertise a disabled Frequency mode.

Your engineering journey

The persistent Journey map groups available capabilities under Study, Explore, Compare, and Report. It distinguishes the current workspace from explored history; an explored capability can still have edited inputs that require a rerun. A capability counts as explored only after a successful authoritative Worker result, or after a PDF report is successfully created.

The discovery store contains only versioned capability IDs. Resetting it does not reset inputs, completed records, project metadata, or exports. It is independent of project import/export and unavailable or malformed LocalStorage fails safely.

Running a Frequency Study

1. Choose Study, then the keyboard-operable Frequency tab.
2. Review shared physical parameters and the 0.2–30 Hz, 240-point logarithmic default.
3. Run the definition in the browser Worker and inspect the model's magnitude plots and stored point values.
4. Export deterministic magnitude/phase CSV, individual 1600 × 1000 PNG figures, or the PDF Engineering Report.
5. After editing, the prior completed record and its exports remain unchanged until a full rerun replaces it.

Frequency projects remain vehiclelab.project.v1: they save the selected analysis and reproducible frequency definition only. Import clears old results and requires a fresh authoritative run. Older projects without an analysis field still open as Transient Study definitions.

Working with VehicleLab projects

Open Data & project to add an optional project name, engineering objective, and plain-text notes. Export projectdownloads a deterministic, human-readable .vehiclelab.json file for the active Study, Explore, or Compare definition. A project file records what should be run; it does not contain time-history arrays, completed metrics, plot images, or transient browser state.

1. Configure Study, Explore, or Compare and optionally add project details.
2. Choose Export project to save the definition locally.
3. Use New project to return to the generic Study default.
4. Choose Import project. VehicleLab parses and validates the complete file locally before replacing the workspace.
5. Rerun the restored definition to regenerate authoritative numerical evidence. Compare and Explore imports never display old result state from the previous project.

Only vehiclelab.project.v1 is supported in this release. Malformed files, unknown schemas, invalid values, and incompatible model or solver identities are rejected without modifying the current workspace. The Zhao benchmark context and its evidence link are restored when present explicitly in a supported exported benchmark project.

Export provenance downloads a compact deterministic JSON manifest with the project schema, product/model/solver identities, execution environment, Float64 precision, benchmark evidence identity where applicable, and the saved definitions. There is no project library, engineering autosave, IndexedDB, account, or cloud save. Journey LocalStorage contains only capability-discovery IDs for the Journey map, never engineering values.

Quarter-Car vehicle parameters

VehicleLab uses SI units throughout. This reference shows the Quarter-Car fields; the Half-Car Study guide covers its body, front, rear, and axle parameters. Runtime validation remains authoritative in the typed study schemas. Non-positive masses, stiffnesses, and damping are invalid for the intended passive models.

Road inputs

Opening a published benchmark

The stable URL /studio/?benchmark=zhao-2015-passive-bumploads the source-audited Zhao et al. passive bump definition without running it. Choose Run study to calculate its response. A visible benchmark-context notice links to the detailed evidence page.

• Reset to benchmark restores the audited parameters, road, zero state, duration, and step.
• Return to generic study removes benchmark context and restores the generic Quarter-Car definition without running it.
• Compare remains available, so Variant A can explore one physical-parameter change under the benchmark road and solver scenario.
• Plots and optional motion playback work from the completed benchmark result exactly as they do for generic studies.
• Unknown benchmark query values safely fall back to the generic default without a benchmark claim.
• Exporting and importing the benchmark project preserves its explicit context, exact audited definition, reset behavior, and evidence link; evidence data remains built into VehicleLab rather than embedded in every project file.

Solver settings

Simulation duration

0.01 to 30 s

Requested fixed RK4 step

0.0001 to 0.02 s

Maximum workload

100,001 samples

Sweep workload

3–25 cases; 500,000 total samples

Repeatability

Deterministic for identical valid input

A smaller requested step increases both sample count and computation. If duration is not exactly divisible by the requested step, VehicleLab uses one shorter final step so the final sample lands exactly on the requested duration.

Running a study

1. Studio opens on the flagship Half-Car Transient Study without running it.
2. Choose the selected model's Run Study action after editing valid inputs.
3. The run status reports preparation, completion, or a validation/runtime error.
4. The dedicated browser Web Worker executes the fixed-step study.
5. Choose the selected model's reset action to restore its generic definition.

Exploring one physical parameter

Choose Explore from the mode navigation or Study actions. Quarter Car supports suspension damping and stiffness; Half Car supports twelve physical parameters, including front and rear suspension and tire properties. Enter a finite minimum, maximum, and integer case count; the minimum must be lower than the maximum and both must remain inside the same physical bounds shown for ordinary inputs. Invalid values are rejected rather than clamped.

VehicleLab includes both requested endpoints exactly and generates every other candidate in deterministic ascending order. All non-swept physical parameters, road inputs, initial state, duration, requested step, model, and solver come from the current Baseline and are identical for every candidate. Progress identifies the active candidate. Cancel stops work by terminating and recreating the local Worker; no simulation data is uploaded.

1. Run the sweep and read the canonical table: four metrics for Quarter Car or eight for Half Car.
2. Inspect the model's parameter-response and trade-off plots. Markers are actual simulations; VehicleLab does not interpolate or optimize.
3. Select a candidate from a table radio control or keyboard-focusable plot marker.
4. Promote the selected candidate to Compare. Its physical configuration becomes Variant A while the Baseline and shared scenario remain unchanged.
5. Run the configured comparison explicitly.

Export sweep CSV downloads a deterministic UTF-8 summary containing the swept parameter and unit, the selected model's metric values and units, schema and release identity, model, solver, execution environment, and precision. It contains no arbitrary browser state. After any input or sweep edit, “Edited — rerun required” appears while the prior completed record remains unchanged; only a fully completed rerun replaces it.

Comparing Baseline and Variant A

Choose Compare to copy the current physical parameters into Variant A. The road inputs, initial state, duration, RK4 step, model identity, and solver identity remain one shared scenario so both configurations receive the same excitation and numerical treatment.

1. Edit either physical-parameter panel.
2. Run the configured comparison explicitly.
3. Review changed parameters, neutral metric deltas, and original-channel overlays.
4. If an input changes afterward, the Studio reports Edited — rerun requiredwhile retaining the last complete comparison snapshot.
5. Select Study to leave Compare and return to the single-study workspace with the current baseline configuration.

Choose the Frequency analysis to compare both physical configurations on one exact logarithmic grid with the direct frequency solver. The completed record stores aligned Baseline/Variant magnitude and phase responses, changed physical parameters, sampled interior-peak shifts, and native-point deltas. See the Frequency Compare guide for interpretation and export details.

Inspecting frequency samples

In Frequency Study and Frequency Compare, expand Inspect a computed frequencyto choose a stored sample and read the model's response quantities. The inspector starts collapsed so the peak summary and response plots remain the main evidence surface.

Motion playback

Completed Transient results expose Response | Motion beside the result heading. Response is selected by default. Choose Motion, then Model | Vehicle: Model shows the selected model's schematic; Vehicle shows physics-driven contextual 3D geometry. Play/Pause sits above the Model schematic or over the Vehicle canvas. The road-input trace sits directly below the canvas, before the legend and model-boundary disclosure. Restart, the time slider, and 0.5×, 1×, or 2× speed remain below; all controls are shared. Switching renderers preserves playback; returning to Response pauses it while retaining the selected time.

A completed comparison adds a Baseline/Variant A selector. Both use the same simulation time, road input, and visual scale. Movement is amplified and not drawn to geometric scale. The playback linearly interpolates between stored samples for display only; plots, metrics, and V&V evidence continue to use stored outputs. Editable inputs do not affect playback until a new completed result replaces the snapshot.

Vehicle offers static 3/4 and Side cameras and generic springs and dampers that follow the displayed body/wheel positions. A compact road-input trace shows the completed samples, a shared-time cursor and the current displacement in millimetres. Draft road edits do not rewrite it. The trace remains available without WebGL; it shows time, not distance, vehicle speed or a travelling road profile. Camera selection survives renderer and Compare-source switches and is not saved in project files.

Quarter-Car Motion shows vertical body and wheel response; Half-Car Motion also shows coupled heave and pitch with front and rear axle response. Wheels are mirrored for display; neither model calculates roll or independent corners. The displayed scale is common to Model, Vehicle, Baseline and Variant A; readouts retain actual engineering values. Vehicle loads only when selected. If WebGL is unavailable, Response, Model, simulation and reports remain usable. Frequency Motion is intentionally absent. Renderer and scene state are not saved in engineering project JSON.

Exporting engineering evidence

The persistent Engineering Report action is always visible. It is unavailable until the active workspace has completed evidence, marked current when the record matches current inputs, and marked stale after an edit while remaining available for the last completed evidence. It never substitutes edited draft inputs into retained evidence. Secondary CSV and project actions live under Data & project, while each completed plot keeps its direct PNG action. Filenames use the sanitized project name, or vehiclelab when the project is untitled. CSV uses deterministic columns, candidate/sample order, UTF-8 text, CRLF line endings, and JavaScript’s round-trip-safe number representation without display rounding.

• Transient Study CSV contains the exact completed time and response channels for the selected model, including road inputs, body and unsprung response, suspension travel, and tire deflection.
• Comparison CSV aligns Baseline and Variant A channel pairs on the authoritative shared time base. Summary CSV contains changed physical parameters and the stored metric deltas, units, relative deltas, and direction.
• Sweep CSV preserves canonical ascending candidate order and the existing metric/unit contract.
• Export PNG on each Study, Compare, or Explore plot directly renders the authoritative arrays to a 1600 × 1000 image with curves or candidate markers, title, legend, axes, grid, and units. It does not copy uPlot, screenshot the page, resample, smooth, normalize, or change plot semantics.
• Engineering Report generates a selectable-text A4 PDF locally from the active completed immutable Study, Frequency Study, transient/frequency Compare, or Sweep record. The report includes configuration, authoritative metrics or deltas, the same professional plot renderer, provenance, assumptions, and limitations. Zhao benchmark reports also include the numerical-reproduction evidence context.

Project JSON defines what to reproduce; CSV preserves raw numerical output; PNG is a professional engineering figure; PDF communicates the completed engineering evidence. DOCX and customizable report builders are not included.

Data and privacy

In completed Transient Study Motion, select Export Video for a local wide or vertical Model or Vehicle artifact. The video uses completed evidence and shows physical readouts, road context and amplification disclosure. See Engineering Video Exportfor timing, browser support and cancellation details.

Simulation parameters, imported project files, and generated JSON, CSV, PNG, and PDF artifacts remain in the browser. VehicleLab has no simulation/report backend, account system, or saved-project database. Cloudflare Web Analytics applies only to ordinary production page traffic; no custom simulation parameters or result channels are intentionally sent to analytics.

Half-Car Frequency Compare

Study → Half Car → Frequency → Compare a design change. Baseline and Variant A use the same harmonic experiment, with explicit physical changes, complex evidence and completed CSV/PNG/PDF exports. Follow the Frequency Compare guide.