pfc-rutting-test

Design, implement, and audit PFC3D 5.0 asphalt wheel-tracking and intersection-rutting simulations using RVE-to-strip-to-full-size scaling, PFC5-compatible vector reaction control, Burger-calibrated c

SKILL.md

GitHub

PFC5 Wheel-Tracking and Rutting Test

Use this skill after specimen generation and Burger calibration. Prefer PFC3D because wheel width, lateral confinement, contact patch and rut geometry are intrinsically three-dimensional.

When to use

  • plan a wheel-tracking/rutting model in PFC 5.0;
  • choose a moving, rolling-faceted or fixed equivalent cyclic load;
  • servo a wall/head to a target reaction rather than assign a read-only contact force;
  • extract rut-depth histories and an edition-specific dynamic-stability metric;
  • scale an RVE to a strip/pilot and then a full-size slab.
  • study braking/acceleration with vertical-horizontal coupled loading, asymmetric deformation, particle migration or shear-stress depth profiles;
  • audit an equivalent fixed patch or accelerated-time route against an unaccelerated moving-load pilot.

Do not use this skill for Marshall or for specimen generation.

Required inputs

  • exact PFC3D 5.0 build, units and runtime budget;
  • governing standard edition, method ID, clauses and machine/specimen coefficients;
  • slab, mold, wheel and travel geometry;
  • conditioning/test temperature;
  • wheel motion definition and whether “count” means one-way pass or complete cycle;
  • physical load waveform/frequency/rest history and every step used to replace it with a moving, cyclic or constant equivalent route;
  • prescribed pressure/load and the physical/equivalent contact-area definition;
  • vertical and horizontal load-control implementation, coordinate/sign convention, target vector-reaction ratio and tolerances;
  • source and meaning of any F_h/F_v or friction-demand factor, kept separate from wheel-contact and material-contact friction coefficients;
  • lateral/front/back boundary conditions and the planned free-versus-confined sensitivity comparison;
  • load-ready specimen plus calibrated contact parameters;
  • rut-depth measurement region/reference and stop condition;
  • laboratory rut curve/dynamic-stability targets.

The current reviewed route is JTG 3410-2025 T 0719-2025, effective 2025-10-01. JTG E20-2011 T0719 is a legacy route and must not supply current default dimensions, time windows or coefficients.

Modeling hierarchy

  1. RVE/material check — calibrate creep/recovery and timestep sensitivity.
  2. Strip/pilot model — verify boundaries, contact patch, load servo, wheel motion, rut measurement and computational cost.
  3. Full-size model — use only after the first two gates pass.

The exact dimensions and test values come from the selected standard edition or laboratory protocol, not from a remembered default.

Loading routes

Route Meaning Acceptance gate
moving faceted wheel/head translation over slab reaction and footprint match target throughout travel
rolling faceted wheel translation plus angular velocity kinematic sign, slip and facet-resolution sensitivity
fixed wall with facet conveyor fictitious surface conveyance without moving facets label as conveyor surrogate; validate against moving/rolling footprint and reaction
fixed equivalent cyclic patch accelerated surrogate correlation to moving-wheel pilot; never label as normative motion
fixed equivalent constant patch creep-oriented accelerated surrogate preserve declared impulse/dwell or constitutive invariants and pass an unaccelerated comparison
vertical-horizontal coupled head intersection braking/acceleration surrogate both reaction components, slip/work, direction reversal and boundary sensitivity pass

In PFC5, contact force is normally an observed reaction. Apply load by a verified servo or velocity/position control and demonstrate that the reaction meets the target. Pressure-to-force conversion requires a declared reference area; the evolving DEM contact area is not automatically that area.

Workflow

  1. Freeze standard edition, physical parameters and numerical equivalents separately.
  2. Restore the load-ready specimen, audit boundaries/contact groups and equilibrate.
  3. Build and motion-test the wheel/head with no specimen. For a rolling wall, PFC5 translational velocity and angular spin are different attributes; set the center of rotation on the wheel axis and verify |v| = |omega| R plus the sign at the contact point. The default rotation center is the global origin, so relying on it is unsafe for an offset wheel.
  4. Run a small contact/load-servo probe; quantify reaction error and oscillation.
  5. Run the strip pilot for a short declared number of cycles/passes.
  6. Establish a vertical-only baseline. For an intersection route, add positive and negative horizontal cases and verify both reaction components before interpreting asymmetric deformation.
  7. Record wheel position, commanded motion, vertical/horizontal reactions, reference surface, rut depth, the seven fixed-position measurements or an approved equivalent, timestep, equilibrium/inertia indicators, one-way pass count and round-trip count.
  8. If time is compressed, compare lambda=1 and every candidate scale on a small model; audit rut, recovery, contact state and energy, not only the final displacement.
  9. For coupled loading, export forward/reverse-face deformation, lateral heave, phase-separated particle displacement and a tensor-defined shear-depth profile.
  10. Confirm no wall penetration, particle escape, axis/sign error or callback duplication.
  11. Run the production duration with restart saves.
  12. Compute the standard metric only from the selected edition's formula, intervals and coefficients; otherwise report the raw rut history with the metric pending.
  13. Compare against laboratory curves and run seed/resolution/rate/load-route, contact-mixture, thickness and boundary sensitivity.

Working rules

  • Keep PFC5 command/FISH syntax version-pure.
  • Do not set wall.force as if it were a prescribed actuator without proof from the exact PFC5 API; use a verified motion/servo controller.
  • Do not assume a rectangular patch, accelerated cycle or fixed load is equivalent to a moving rubber wheel. Calibrate and label the surrogate.
  • Do not call translation-only motion “rolling.” A rolling wall needs verified spin, axis/center of rotation and no-slip sign; a facet conveyor is another distinct surrogate because it changes contact relative velocity without moving the facets.
  • Do not call a fixed constant patch a dynamic load merely because its duration came from a wheel-pass calculation. State which physical quantities the mapping preserves and which it discards.
  • F_h = mu F_v may define an external braking/acceleration demand, but that mu is not automatically a DEM contact friction coefficient. Verify commanded and measured vector reactions, slip and work independently.
  • A free horizontal outlet can exaggerate material migration; a fully fixed boundary can suppress it. Report at least one free/confined sensitivity pair before field interpretation.
  • A measure-region “shear stress” is incomplete without tensor component, axes, sign, radius/volume weighting, overlap, sample time and empty-region handling.
  • Published strip dimensions, load levels, contact fractions, friction-demand factors and lambda values are case evidence, not package defaults.
  • For T 0719-2025, compute DS from the current method's pass-count interval and width coefficient. Do not use the legacy 45 min/60 min shorthand or an unverified extra correction coefficient.
  • Preserve RVE, strip, pre-load, pilot and production milestone saves.
  • Full-size output without pilot load-control evidence is not validated.

Output contract

  • PFC build/units and standard edition/method/clause;
  • physical versus DEM-equivalent parameter table;
  • specimen/wheel/boundary/contact assignment;
  • actuator/servo pseudocode or PFC5-verified code plus reaction-error history;
  • physical-to-DEM load/time equivalence ledger and accelerated-versus-unaccelerated comparison;
  • rut-depth measurement definition and raw time/pass/cycle history;
  • vertical/horizontal command and reaction histories, face-asymmetry/lateral-heave metrics, particle displacement fields and tensor-defined shear-depth profiles;
  • edition-correct metric calculation with units/coefficients, or pending status;
  • RVE/strip/full-size stage evidence and sensitivity results;
  • laboratory comparison and runtime_validated status.

Local contents

  • references/overview.md — current T 0719-2025 measurement, count and PFC3D mapping.
  • references/jtg-t0719.md — current-version gate, user-input template and symbolic metric; no standard data tables.
  • references/modeling-strategy.md — PFC5-safe hierarchy and controller pseudocode.
  • ../pfc5-asphalt-workflow/references/intersection-rutting-research-evidence.md — external-paper evidence and non-default boundaries for mixed contacts, time compression and vertical-horizontal loading.
  • ../pfc5-asphalt-workflow/references/standards-policy.md — standards policy.
  • scripts/rutting_contact_pilot.p3dat — executable curved-wheel PFC3D 5.0 contact/translation pilot; not a normative wheel-tracking model.
  • dependencies.json — package-level sibling-skill assets required for the complete workflow.
  • agents/openai.yaml — Agent metadata.