Multiphysics & geotechnical analysis
Rebuild coupled or staged geotechnical models from a spec and compare cases.
20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.
- Typical applications
- COMSOL · PLAXIS · GeoStudio
- Class GUI-need label
G3· GUI-only
Task proposals · 20
Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.
R07.02.01.02-T01 COMSOL: Joule-heating busbar maximum temperature
Original instruction
Rebuild the busbar in input/busbar_spec.pdf in COMSOL 6.2 (copper, 400 A DC through the end faces, convective h = 10 W/m²K at 293.15 K, titanium bolts), with Electric Currents + Heat Transfer coupled by Electromagnetic Heating. Solve stationary and report max temperature to output/busbar.json.
- Platforms
win· Windowsmac· macOSlin· Linux- Applications
- COMSOL Multiphysics
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsgeom· Precise geometry- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Grader opens the .mph via COMSOL server (read-only): physics interfaces and multiphysics coupling present, materials per domain, BCs; recomputed max T within 1% of reference. Agent has no LiveLink/Java API or Method editor (disabled); model history must come from the GUI.
Proposed partial credit
25% physics and coupling; 25% materials/BCs; 50% max temperature.
R07.02.01.02-T02 COMSOL: bimetal strip tip deflection under a 100 K rise
Original instruction
Model the 50 × 5 × (0.5 + 0.5) mm steel/brass bimetal strip from input/bimetal.pdf in COMSOL with Solid Mechanics and Thermal Expansion (ΔT = 100 K, one end fixed). Report the tip deflection in mm.
- Platforms
win· Windowsmac· macOSlin· Linux- Applications
- COMSOL Multiphysics
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Model has thermal expansion coupling with reference temperature and per-layer materials; deflection within 2% of reference (Timoshenko bimetal formula cross-check). COMSOL's Method Editor is disabled, LiveLink for MATLAB is not licensed and comsolbatch/mphserver are absent from the agent's PATH, so the model can only be built in the Model Builder.
Proposed partial credit
30% geometry/materials; 30% thermal expansion setup; 40% deflection.
R07.02.01.02-T03 COMSOL: capacitance of a two-electrode geometry
Original instruction
Import input/electrodes.step into COMSOL, set up Electrostatics with one electrode at 1 V terminal and the other grounded in an air box 10× the device size, compute the capacitance via the terminal and report it in pF.
- Platforms
win· Windowsmac· macOSlin· Linux- Applications
- COMSOL Multiphysics
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 2 / 5
Proposed verifier
Terminal and ground features on correct boundaries; air domain extent; capacitance within 2% of reference converged value. COMSOL's Method Editor is disabled, LiveLink for MATLAB is not licensed and comsolbatch/mphserver are absent from the agent's PATH, so the model can only be built in the Model Builder.
Proposed partial credit
40% electrostatics setup; 60% capacitance value.
R07.02.01.02-T04 PLAXIS 2D: diaphragm-wall excavation with three strut-stiffness cases
Original instruction
Build the staged excavation in input/excavation_spec.pdf in PLAXIS 2D (HS-small soil layers, 0.8 m diaphragm wall, two strut levels, 5 excavation phases with dewatering). Run three cases of strut EA (0.5×, 1×, 2×) and report max wall deflection and max surface settlement per case to output/cases.csv.
- Platforms
win· Windows- Applications
- PLAXIS 2D
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsgeom· Precise geometry- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro applong· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Grader reads the .p2dx via PLAXIS output (read-only): layers, materials, phases and activation states; 6 values within max(0.3 mm, 5%) of hidden reference. Remote-scripting server is disabled for the agent.
Proposed partial credit
20% geometry/materials; 30% phase sequence; 50% six results (≈8 each).
R07.02.01.02-T05 PLAXIS: effect of adding a strut row at −3 m on settlement
Original instruction
Open input/pit_base.p2dx (solved). Add a strut row at −3.0 m (EA from input/strut.json) activated in the phase before excavation reaches −4 m, rerun, and report the change in final-phase maximum surface settlement (mm) and wall moment.
- Platforms
win· Windows- Applications
- PLAXIS 2D
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
AP· API-sufficient- Interaction primitives
click· Point & clicktype· Text & formsgeom· Precise geometry- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Model diff: new anchor/strut at −3.0 m with correct EA and activation phase; recomputed deltas within max(0.3 mm, 5%) of reference. The PLAXIS remote-scripting server and bundled Python are disabled for the agent, so the model can only be built in PLAXIS Input.
Proposed partial credit
40% strut defined and activated correctly; 30% settlement delta; 30% moment delta.
R07.02.01.02-T06 PLAXIS: slope safety factor by phi-c reduction
Original instruction
For the embankment slope in input/slope.p2dx, add a Safety (phi-c reduction) phase after the final construction phase, run it with 100 steps, and report the safety factor Msf and a screenshot of the incremental shear strain mechanism.
- Platforms
win· Windows- Applications
- PLAXIS 2D
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsvis· Visual judgment- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Phase type Safety with correct parent phase; Msf within ±0.02 of reference; screenshot shows shear band (grader compares slip-surface location from output with reference, VLM not used). The PLAXIS remote-scripting server and bundled Python are disabled for the agent, so the model can only be built in PLAXIS Input.
Proposed partial credit
40% safety phase set up; 50% Msf; 10% mechanism output.
R07.02.01.02-T07 GeoStudio SLOPE/W: Morgenstern-Price factor of safety with a water table
Original instruction
In GeoStudio SLOPE/W, build the embankment in input/embankment.pdf (two soils, piezometric line given), analyse with Morgenstern-Price, entry-exit slip search, and report the minimum FoS and slip-surface centre.
- Platforms
win· Windows- Applications
- GeoStudio SLOPE/W
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickgeom· Precise geometrytype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Grader opens the .gsz (zipped XML) read-only: method M-P, soils, piezometric line coordinates ±0.05 m; recomputed FoS within ±0.02 of reference; reported FoS matches file results. GeoStudio's scripting API and command-line solver are not available to the agent, and a file-system audit log flags a .gsz not written by GeoStudio (hand-built XML fails).
Proposed partial credit
30% geometry/soils; 20% piezometric line; 50% FoS.
R07.02.01.02-T08 GeoStudio: rapid drawdown with SEEP/W transient feeding SLOPE/W
Original instruction
For the earth dam in input/dam.pdf, set up a SEEP/W steady-state then transient drawdown (10 m over 5 days) and a SLOPE/W analysis using the transient pore pressures at each saved step. Report the minimum FoS over time and when it occurs.
- Platforms
win· Windows- Applications
- GeoStudio SEEP/W · GeoStudio SLOPE/W
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicktype· Text & formsgeom· Precise geometry- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro applong· Long-horizon- Proposed difficulty
- 4 / 5
Proposed verifier
Analysis tree: SLOPE/W parent is transient SEEP/W, drawdown boundary function correct; FoS-time series within ±0.03 of reference; minimum and time step exact. GeoStudio's scripting API is not reachable from the agent VM, which has no Python, so the analysis must be built in the GeoStudio GUI.
Proposed partial credit
30% seepage setup; 20% drawdown function; 50% FoS minimum and timing.
R07.02.01.02-T09 GeoStudio SIGMA/W: consolidation settlement after one year with drains
Original instruction
Model the embankment over soft clay in input/soft_clay.pdf in SIGMA/W coupled consolidation, with prefabricated vertical drains modelled as equivalent permeability per the spec, and report centreline settlement at 365 days and degree of consolidation.
- Platforms
win· Windows- Applications
- GeoStudio SIGMA/W
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicktype· Text & formsgeom· Precise geometry- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Coupled stress-PWP analysis type; staged fill loading; settlement at 365 d within 5% of reference; U within ±3 points. GeoStudio's scripting API is not reachable from the agent VM, which has no Python, so the analysis must be built in the GeoStudio GUI.
Proposed partial credit
30% coupled analysis; 20% PVD representation; 50% results.
R07.02.01.02-T10 COMSOL: thermal warpage of an electronics package
Original instruction
Model the package stack in input/package.pdf (die, die-attach, substrate; temperature-dependent CTEs) in COMSOL, cool from 175 °C to 25 °C stress-free at 175 °C, and report max warpage (µm) and its sign (smile/cry).
- Platforms
win· Windowsmac· macOSlin· Linux- Applications
- COMSOL Multiphysics
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Thermal expansion reference temperature 175 °C, interpolation functions for CTE; warpage within 5% and sign correct. COMSOL's Method Editor is disabled, LiveLink for MATLAB is not licensed and comsolbatch/mphserver are absent from the agent's PATH, so the model can only be built in the Model Builder.
Proposed partial credit
30% materials with T-dependent CTE; 20% thermal load; 50% warpage.
R07.02.01.02-T11 COMSOL: piezoelectric cantilever harvester resonance and output
Original instruction
Build the bimorph harvester in input/harvester.pdf (PZT-5A layers, brass shim, tip mass) with Piezoelectricity + Electrical Circuit (load 100 kΩ), find the first resonance by eigenfrequency, then frequency-domain response at 0.5 g base excitation; report resonance (Hz) and RMS power (µW).
- Platforms
win· Windowsmac· macOSlin· Linux- Applications
- COMSOL Multiphysics
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro applong· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Physics/coupling and circuit present; poling direction per layer; resonance within 2%, power within 8% of reference. COMSOL's Method Editor is disabled, LiveLink for MATLAB is not licensed and comsolbatch/mphserver are absent from the agent's PATH, so the model can only be built in the Model Builder.
Proposed partial credit
30% coupled physics; 20% circuit; 25% resonance; 25% power.
R07.02.01.02-T12 COMSOL: microfluidic T-mixer mixing index
Original instruction
Model the T-mixer in input/tmixer.pdf with Laminar Flow + Transport of Diluted Species (D = 1e-9 m²/s, 1 mol/m³ vs 0 inlets, 0.5 mm/s each), solve stationary, and report the mixing index at the outlet (1 − σ/σ0).
- Platforms
win· Windowsmac· macOSlin· Linux- Applications
- COMSOL Multiphysics
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Flow and species interfaces coupled; mesh adequate (Peclet-related refinement); mixing index within 0.02 absolute of reference. COMSOL's Method Editor is disabled, LiveLink for MATLAB is not licensed and comsolbatch/mphserver are absent from the agent's PATH, so the model can only be built in the Model Builder.
Proposed partial credit
30% physics; 20% mesh; 50% mixing index.
R07.02.01.02-T13 COMSOL: induction heating of a steel billet to a time-at-temperature
Original instruction
Model input/induction.pdf (2D axisymmetric coil and billet, 10 kHz, nonlinear B-H and Curie-dependent properties) with Induction Heating, and find the coil current needed so the billet core reaches 900 °C at t = 60 s ±5 s. Report current and surface temperature at 60 s.
- Platforms
win· Windowsmac· macOSlin· Linux- Applications
- COMSOL Multiphysics
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsvis· Visual judgment- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro applong· Long-horizon- Proposed difficulty
- 5 / 5
Proposed verifier
Frequency-transient study; material functions; grader re-solves with the reported current: core T at 60 s in 895–905 °C; surface T within 3%. COMSOL's Method Editor is disabled, LiveLink for MATLAB is not licensed and comsolbatch/mphserver are absent from the agent's PATH, so the model can only be built in the Model Builder.
Proposed partial credit
30% physics/study; 20% materials; 50% current meets target.
R07.02.01.02-T14 Elmer GUI: steady heat conduction in a heat sink
Original instruction
In ElmerGUI, mesh input/heatsink.step, set aluminium conductivity, a 15 W heat flux on the base footprint and h = 25 W/m²K on fins, solve Heat Equation and report max base temperature (ambient 25 °C) and export VTU for ParaView.
- Platforms
lin· Linuxwin· Windowsmac· macOS- Applications
- ElmerGUI · ParaView
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Generated .sif inspected for BC assignments to correct boundary IDs; grader reruns ElmerSolver: max T within 2% of reference; VTU present.
Proposed partial credit
40% BCs/material; 40% max T; 20% VTU export.
R07.02.01.02-T15 PLAXIS 3D: pile group settlement and load distribution
Original instruction
Model the 3 × 3 bored pile group from input/pilegroup.pdf in PLAXIS 3D (embedded beams, cap plate, layered soil), apply 9 MN vertical load in stages, and report cap settlement and the corner/centre pile head loads.
- Platforms
win· Windows- Applications
- PLAXIS 3D
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsgeom· Precise geometry- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro applong· Long-horizon- Proposed difficulty
- 4 / 5
Proposed verifier
Model structure inspected (9 embedded beams, interface/skin resistance per spec); settlement within 5%, pile loads within 7% of reference. The PLAXIS remote-scripting server and bundled Python are disabled for the agent, so the model can only be built in PLAXIS 3D Input.
Proposed partial credit
30% model build; 20% staging; 50% results.
R07.02.01.02-T16 Rocscience RS2: staged tunnel excavation with liner moments
Original instruction
In RS2, model the circular 8 m tunnel in input/tunnel.pdf (Hoek-Brown rock, K0=1.5), relax core by 40% in stage 2 before installing a 300 mm shotcrete liner in stage 3, and report maximum liner bending moment and crown displacement.
- Platforms
win· Windows- Applications
- Rocscience RS2
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsgeom· Precise geometry- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Stages and load-split (internal pressure reduction) read from model; liner properties; results within 5%. Rocscience Python scripting is disabled for the agent's license and its VM has no Python, so the model must be built in the RS2 modeler.
Proposed partial credit
30% stages/relaxation; 20% liner; 50% results.
R07.02.01.02-T17 Rocscience Slide2: probabilistic slope with variable cohesion
Original instruction
In Slide2, open input/cut_slope.slmd, make cohesion and friction of layer 2 normally distributed per input/stats.json, run 1,000 Latin-Hypercube samples (seed 42) with Spencer, and report mean FoS and probability of failure.
- Platforms
win· Windows- Applications
- Rocscience Slide2
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Statistics settings read from file; mean FoS ±0.01 and PF ±0.5 percentage points vs reference with same seed. Rocscience Python scripting is disabled for the agent's license and its VM has no Python, so the model must be built in the Slide2 modeler.
Proposed partial credit
40% probabilistic setup; 30% mean FoS; 30% PF.
R07.02.01.02-T18 COMSOL: mesh refinement effect on a solved thermal model
Original instruction
Open input/heater_solved.mph, change the mesh from 'Normal' to 'Finer', re-solve, and report the change in max temperature (K) and in the integrated heat flux through the top boundary.
- Platforms
win· Windowsmac· macOSlin· Linux- Applications
- COMSOL Multiphysics
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 2 / 5
Proposed verifier
Model mesh size setting 'Finer'; grader re-solves and compares reported deltas (±2%). COMSOL's Method Editor is disabled, LiveLink for MATLAB is not licensed and comsolbatch/mphserver are absent from the agent's PATH, so the model can only be built in the Model Builder.
Proposed partial credit
30% mesh changed; 70% two deltas.
R07.02.01.02-T19 PLAXIS SoilTest: calibrate HS-small parameters to triaxial data
Original instruction
Using PLAXIS SoilTest, calibrate E50ref, Eoedref, G0ref and γ0.7 of an HS-small material so simulated drained triaxial curves match input/triaxial_*.csv (3 confining pressures); save the material set and report RMSE of deviatoric stress.
- Platforms
win· Windows- Applications
- PLAXIS SoilTest
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicktype· Text & formsvis· Visual judgment- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Grader re-simulates the three tests with the saved parameters; RMSE ≤ 8% of peak deviatoric stress for full credit (scaled to 20%). The PLAXIS remote-scripting server and bundled Python are disabled for the agent, so parameters can only be fitted in SoilTest.
Proposed partial credit
20% material set saved; 80% fit quality (scaled RMSE).
R07.02.01.02-T20 GeoStudio: rerun a slope model with the water table raised 2 m
Original instruction
Open input/levee.gsz (solved, SLOPE/W). Raise the piezometric line by 2.0 m everywhere, re-solve with the same method and search, and report the new minimum FoS and the change from the original.
- Platforms
win· Windows- Applications
- GeoStudio SLOPE/W
- Interfaces
desk· Desktop GUI- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicktype· Text & forms- Work product
WP07· Simulation result- Conditions
licensed· Licensed pro app- Proposed difficulty
- 2 / 5
Proposed verifier
Piezometric points each raised by 2.00 ±0.01 m; method/search unchanged; FoS within ±0.02 of reference. GeoStudio's scripting API is not reachable from the agent VM, which has no Python, so the analysis must be built in the GeoStudio GUI.
Proposed partial credit
40% water table edited; 60% FoS and delta.
Mapped research references
Existing benchmark-to-class links from the research catalog. They are not record-level annotations, executions or product endorsements.