PEDESTRIAN PROTECTION EVALUATOR

Runs pedestrian protection (PedPro) LS-DYNA simulations by combining a vehicle baseline model (or pre-positioned inputs) with one or more impactor types (headform, lower leg, upper leg) at defined strike positions. Submits a batch of HPC jobs for each position and collects the resulting study and simulation records for downstream post-processing.

When to use

Tagged: pedestrian_impact, batch_simulation, headform, hpc, impactor, lower_leg, ls-dyna, pedestrian_protection.

Inputs

Label ID Type Default Required Description
Study Name study_name text PedPro Simulations Human-readable label assigned to the PedPro simulation study; defaults to ‘PedPro Simulations’ and is used to group all submitted jobs.
Impactor Type impactor_type select headform   Selects the pedestrian impactor type to use — ‘headform’, ‘lower_leg’, or ‘upper_leg’ — which determines the impactor FE model and applicable injury criteria extracted during post-processing.
Vehicle Input Files main_input_files file Vehicle LS-DYNA input deck(s): upload multiple flat files or a single ZIP archive; if zipped, the main include file must contain the word ‘main’ in its filename.
Input file type input_file_type select baseline   Specifies how impactor positions are encoded in the uploaded files: ‘baseline’ (single model with embedded positions), ‘positioned_folders’ (ZIP with one sub-folder per position), or ‘positioned_files’ (individual pre-positioned decks).
Impactor Positions impactor_positions dataset   Optional dataset supplying impactor strike positions (translations, direction cosines, rotation centre, angle) when ‘baseline’ input_file_type is selected; leave empty if positions are baked into the file structure.
Impactor Position Mapper impactor_positions_col_map mapper   Column mapper that aligns dataset fields from impactor_positions to the required position schema (X/Y/Z-Translation, X/Y/Z-Cosine, X/Y/Z-Rotation-Center, Angle); must be configured whenever impactor_positions is supplied.
Part and Sensors Map for Post processing part_and_sensors_map keyvalue (complex) Key-value map of LS-DYNA part IDs and sensor/load-cell IDs used during post-processing to extract contact, tracking-node, femur, and tibia responses; defaults cover a standard single-impactor model setup.
Impactor Node Set Id impactor_node_set_id text 1   Impactor node set id
Impactor Transform Id impactor_transform_id text 1   Impactor transform id
Template ID template_id remote_lookup    
Project ID project_id remote_lookup    
Study ID study_id remote_lookup    
Saved HPC Config saved_hpc_config remote_lookup    
Run Baseline Only baseline_only select yes Option to run baseline or all positions
Impactor Position file pattern position_file_pattern text PedShot   Impactor Position File Name pattern. This file will be replaced
Main input file pattern main_input_file_pattern text Main   Main input file pattern. This file will be used to replace the positioned files
Termination Time term_time text 0   Optional termination time. Only supported for positioned files
Include Files pattern include_files_pattern text INCLUDE   Include file pattern. Only files matching this pattern will be included in the simulation basides the main and the pedshot file
Update main input file name with Impactor filename update_main_file select no Helps to distinguish when the filename is identical in all runs
Analysis analysis select simulation   Simulation runs every point. Eigenvalue does a localized modal run per point. ML-Assisted simulates a chosen subset and predicts HIC for the rest. Select-only returns the picked-vs-skipped points for review, with no simulations.
Saved ML model saved_model remote_lookup   ML-Assisted only. Supply a saved model to skip simulation and predict HIC for all points directly.
Subset sizing sizing_mode select fixed   ML-Assisted only. How the simulated subset is sized.
Subset size (K or fraction) num_subset_points text 0.2   How many points to select: an integer count, or a fraction 0<f<1 of all points (0.2 = 20%).
Include eigen features include_eigen_features select no   ML-Assisted only. When Yes, run eigenvalue at all points to add eigen frequency/mass/stiffness as features.
Accuracy target accuracy_target text 0.9   ML-Assisted only. Target prediction accuracy (R^2) for adaptive / fixed-plus-one sizing.
Feature source ml_feature_source select compute   Where features come from. Compute: measure per-point metrics from the model. Use supplied rows: take the input positions as-is (they already carry the metric columns, e.g. a clearance spreadsheet).
Selection method ml_selection_method select kmeans   How the simulated subset is chosen (all deterministic). K-means: uniform cluster representatives. Latin-hypercube: even coverage per feature. Farthest-point: the extremes. Combined: union of those three, trimmed to K by consensus (combined-all keeps every unique point). Zone lines: per line keep both ends, the center, and N interior points, along constant-Px (zone-x), constant-Py (zone-y), or both (zone-line).
Feature columns ml_feature_columns select (complex)   Metric columns the selection spreads over (not used by zone lines). Empty = use all numeric metrics present, including any new ones (HIC/id auto-excluded). Or pick a subset, or type custom names for dataset features (e.g. MIN, CON1).
Zone-line interior points (N) zone_line_interior_points text 3   Zone lines only. Interior points kept per line, plus its two ends and center.
Zone-line grouping tolerance zone_line_px_tolerance text 10   Zone lines only. Points farther apart than this (along the grouping axis) start a new line. Raise if one line is split, lower if lines merge.
Analysis scope analysis_scope select entire_vehicle   Entire vehicle: analyse the model as supplied. Hood parts only: extract a self-contained submodel of the hood part ids and their connections and analyse that (requires Hood part ids). Eigenvalue only.
Keep connections keep_connections select yes   Hood parts only. Keep the submodel’s connections (spotwelds, rivets, tied contacts, BCs) among included parts. Eigenvalue only.
Connection levels attachment_levels text 0   Hood parts only. Connection levels beyond the hood to include: 0 = hood + its internal connections, 1 = also directly-connected parts, 2 = one level further, etc. Eigenvalue only.
Submodel delivery submodel_delivery select copy   Copy: place the model/submodel in every simulation. Store once: stage it in one host simulation and point the rest at it via *INCLUDE_PATH (needs a shared file system). Eigenvalue only.
Constraint type constraint_type select free_free   SPC for the localized run. Free-free: none (full modal). All-outside: fix all nodes outside the impact box. Boundary: fix only the box-edge ring. All-outside-except-connected: as all-outside but leave spotweld/rivet/tied nodes free. Box shell: fix only the shell between an outer box and the impact box (fast; for hood-only submodels). Eigenvalue only.
Headform radius (mm) headform_radius text 82.5   Pedestrian headform radius (165 mm dia = 82.5 mm). In-plane box half-extent is 1.5x this (82.5 -> 123.75 mm), with a tall z. Eigenvalue only.
Impact-box half-extent override (mm) head_impactor_diameter text   Optional in-plane box half-extent override. Blank = size from the headform radius (1.5x). Eigenvalue only.
Box half-extents override (mm) box_half_extents text   Optional explicit dx,dy,dz half-extents that override the head-impactor-diameter box. Leave blank to size the box from the head impactor diameter. Eigenvalue analysis only.
Number of eigen modes num_eigen_modes text 50   Number of modes to compute (neig). Eigenvalue analysis only.
ID offset for generated cards id_offset text 9990000   Eigenvalue analysis only. Added to the generated box and node-set ids so they do not clash with the baseline (e.g. offset 9990000 gives id 9990001). Increase if your baseline already uses ids in this range.
Deck mode deck_mode select per_point   Per-point creates one simulation for each impact point; single_case creates one simulation with a case per point. Eigenvalue analysis only.
Remove previous study simulations remove_previous_study_sims select no   When a study id is provided, delete that study’s prior simulations before creating new ones.
Fetch results after jobs finish fetch_results select no   When jobs are finished, aggregate eigout (eigenvalue) or HIC/responses (simulation) into a result dataset.

Outputs

Label ID Type Description
Study Id study_id integer Integer identifier of the d3VIEW study created to group all submitted pedestrian-protection simulations.
Simulations simulations dataset Dataset listing each individual simulation record (one row per impactor position) with metadata linking position parameters to their respective jobs.
Hpcjobs hpcjobs dataset Dataset of HPC job records corresponding to each submitted simulation, carrying queue status and job IDs for monitoring and downstream result extraction.
Results results dataset After jobs finish (Fetch results = Yes), a dataset of results: for Eigenvalue analysis, one row per mode per simulation (eigenvalue, radians, frequency in Hz, period); for Simulation analysis, the extracted responses.
Metrics metrics dataset ML-Assisted only. The per-point feature rows the point selection ran over (computed metrics, or the supplied dataset features).
Selected Points selected_points dataset ML-Assisted (select points only). Every candidate point tagged selected (yes/no) along with its metric columns, so the picked-vs-skipped sweep can be reviewed before simulating.
Selected Count selected_count scalar ML-Assisted (select points only). Number of points the sweep selected.

Disciplines

  • cae.preprocessing.scenario
  • cae.solver
  • engineering.crash.dynamics
  • engineering.crash.occupant_safety
  • platform.job_submission

Auto-generated from platform schema. Worker id: pedpro_evaluator. Schema hash: fc4409e955a1. Hand-curated docs in workerexamples/ override this page when present.