PLAXIS is powerful, but it is also unforgiving: small setup decisions can dominate the response. The goal is not a “pretty plot” — it’s a model you can explain, check, and use to make engineering decisions.

This guide focuses on the steps that most often separate defensible models from fragile ones.

---

1) Define the question and the critical stage

Write down the primary output you need:

• deformations (e.g., wall top deflection, settlement trough)
• internal forces in structural elements
• excess pore pressures / stability margins
• construction stage constraints

If the question is “Is this slope stable?” start with a slope-focused workflow first: slope stability analysis.

2) Geometry, boundaries, and mesh: avoid artificial stiffness

Common failure modes:

• boundaries too close to the zone of influence
• mesh too coarse around interfaces and structural elements
• overly stiff boundary conditions causing non-physical restraint

Rules of thumb:

• keep boundaries far enough that displacement contours decay naturally
• refine around excavations, walls, anchors/props, and stratigraphic changes
• check mesh sensitivity for the output you care about (not for aesthetics)

3) Groundwater is a first-order driver

Model groundwater explicitly and test:

• baseline piezometric surface
• perched / artesian conditions where relevant
• dewatering assumptions (drawdown time, permeability uncertainty)

In many excavations, the “stability issue” is actually a pore pressure assumption issue.

4) Constitutive model selection: match the mechanism

Start conservative and only add complexity when it changes a decision:

• Mohr-Coulomb: transparent, robust, good for bounding
• Hardening Soil: stress-dependent stiffness and more realistic deformation
• HSsmall: small-strain stiffness for serviceability-sensitive problems

Always record which parameters are measured vs. inferred, and where correlations were used.

5) Staged construction: treat sequencing as part of the model

For excavations and retaining systems, the stage definition is the physics:

• activate/deactivate soil and structural elements explicitly
• apply loads when they occur (not “at the end”)
• model struts/anchors with realistic activation timing and pre-stress where relevant

6) QA checklist before you trust results

At minimum:

• equilibrium/convergence across stages
• interface behaviour (no unrealistic tension transfer)
• stress and pore pressure plausibility checks
• bounding comparison with simplified hand checks / LEM where applicable

If the model is influencing design decisions, a short peer review can save significant downstream risk. See services.

7) Communicate results like an engineer

Prefer:

• a single governing deformation/force metric per stage
• a sensitivity bracket (key parameters and groundwater)
• clear plots with consistent scales and stage labels

If you want examples of concise output packages, see projects or get in touch via contact.