Laboratory tests¶
numgeo-ACT calibrates constitutive models against element tests — idealized laboratory experiments in which the stress and strain state is (assumed) homogeneous across the specimen. For every candidate parameter set, each selected test is reproduced by a numgeo element-test simulation and compared against your measured data.
This section describes the supported test types, what they measure, and how each one is simulated. The exact layout your data must follow in the Excel database is given on the Excel sheet reference page; here the focus is on the physics and the role each test plays in a calibration.
Supported tests¶
| Test | Loading | Drainage | Excel sheet | Typical role in calibration |
|---|---|---|---|---|
| Oedometric compression | monotonic | drained | OED-# |
stiffness under confined compression; controls compression parameters |
| Isotropic compression | monotonic | drained | IC-# |
volumetric stiffness; limiting void-ratio / compression law |
| Drained triaxial (CD) | monotonic | drained | CD-# |
shear strength & dilatancy at constant cell pressure |
| Undrained triaxial (CU) | monotonic | undrained | CU-# |
effective stress path, undrained strength, pore-pressure build-up |
| Undrained cyclic triaxial | cyclic | undrained | CUCYC-# |
cyclic mobility / liquefaction, pore-pressure accumulation |
| HCA — drained cyclic triaxial | many cycles | drained | CDCYC_HCA-# |
strain accumulation under high-cyclic loading (HCA model) |
| HCA — undrained cyclic triaxial | many cycles | undrained | CUCYC_HCA-# |
pore-pressure / strain accumulation under high-cyclic loading |
| Undrained cyclic simple shear | cyclic | undrained | USScyc-# |
cyclic shear response, liquefaction in simple-shear conditions |
You choose which tests to honour
A calibration uses whichever tests you pass to
globals.setup. You can
combine any subset — for example oedometer + drained + undrained triaxial for
a monotonic calibration, or add cyclic triaxial tests to capture the cyclic
response as well. Each test type is its own keyword argument
(oedometer, triaxCD, triaxCU, triaxCUcyc, triaxCDcycHCA,
triaxCUcycHCA, USScyc, iso_comp).
Monotonic vs. cyclic¶
The tests fall into two broad groups, and numgeo-ACT lets you balance them against each other through weighting:
- Monotonic tests (oedometer, isotropic compression, drained/undrained triaxial) constrain the backbone response: stiffness, strength and dilatancy.
- Cyclic tests (cyclic triaxial, cyclic simple shear, HCA) constrain the cyclic response: pore-pressure accumulation, cyclic mobility, liquefaction and — for the HCA model — long-term strain accumulation over many cycles.
By default, monotonic and cyclic groups each receive half of the total weight; this can be adjusted to emphasize the behaviour most relevant to your application.
How a test becomes a simulation¶
For every test in your database, numgeo-ACT:
- reads the initial state (void ratio, consolidation stresses, and for cyclic tests the loading amplitude and number of cycles) from the sheet;
- builds the corresponding numgeo element-test input;
- runs the simulation with the current parameter set (subject to a per-test timeout);
- extracts the same quantities you measured (e.g. stress–strain, stress path, pore pressure, accumulated strain) and compares simulation against experiment using the configured similarity measure.
The per-test scores are combined into the single objective value the optimizer minimizes.