Troubleshooting, Model QA & Report Generation
Learning Objectives
- Classify solver errors, warnings and informational messages by engineering consequence.
- Diagnose instability, disconnected structures, duplicate joints, missing properties and excessive releases.
- Use deformed shape, reactions, load totals, local axes and result magnitudes as diagnostic evidence.
- Distinguish numerical symptoms from the underlying physical/modeling cause.
- Apply a repeatable model-QA checklist before design or issue.
- Assemble a concise report containing assumptions, model basis, verification evidence, governing results and design conclusions.
Never fix a warning by hiding it
The correct troubleshooting question is not “How do I make the message disappear?” but “What model condition produced this message, and does that condition represent the intended physical structure?” A warning can be benign, serious, or a clue to a deeper modeling error; it requires disposition, not suppression.
Analysis Troubleshooting
Inject simplified modeling defects, diagnose the physical cause, and compare the kind of message an analysis program may report.
Illustrative analysis diagnostic
Exact STAAD output wording varies by release, solver context, and model. Treat these messages as diagnostic categories, not verbatim `.ANL` text.
Next step: verify equilibrium, displacement, load path, warnings, and governing results before accepting the model.
The teaching model has sufficient idealized restraints to solve, but solver completion still requires engineering verification.
A Structured Diagnostic Method
From symptom to root cause
- Reproduce the issue in one controlled load case/model revision.
- Read the complete solver/output message and note member/joint/direction references.
- Classify the symptom: syntax/input, connectivity, property, restraint/stability, load, nonlinear convergence, or design-code issue.
- Inspect the smallest relevant model region in both graphics and text/model tables.
- Check units and local axes before changing structural properties.
- Make one physically justified change at a time.
- Re-run and compare: confirm the warning/error changed for the expected reason.
- Re-run verification checks because fixing one issue can change forces, reactions and governing design results.
Errors, Warnings and Notes
Fatal/input/solver errors
Errors that prevent analysis completion may arise from invalid input, missing data, singular/unstable systems, unsupported command combinations, or failed iterative solution. Results from an incomplete run must not be treated as valid final analysis output.
Warnings
Warnings indicate conditions requiring review: unusual stiffness, instability in a degree of freedom, large displacement, multiple structures, design parameter concerns, nonconvergence or other suspicious states. Some warnings can be understood and accepted with documented justification; others invalidate the model. The engineer must distinguish them.
Informational messages
Informational notes describe solver choices, generated data or other processing details. They still deserve review when they reveal an assumption different from the intended model.
Instability and Singular Stiffness
Structural mechanism / instability
A model has an unconstrained or insufficiently stiff degree of freedom that allows rigid-body or mechanism-like motion under the analytical idealization. The assembled/restrained stiffness system may become singular or numerically ill-conditioned, so a unique displacement solution cannot be obtained reliably.
Common causes
- Missing support restraint in a global direction.
- Releases that remove required frame stiffness.
- A joint/load disconnected from the intended structure.
- A physical connection modeled as two coincident-but-unmerged joints.
- Members/elements missing section/material stiffness.
- A substructure that has been disconnected during geometry edits.
Exact singular-system condition
A unique linear displacement solution requires the restrained structural system to have adequate independent stiffness in the active degrees of freedom.
Variables
| Symbol | Description | Unit |
|---|---|---|
| Restrained/reduced stiffness matrix for the active system | - |
Do not use determinant magnitude as a practical conditioning meter
The expression above is an exact mathematical singularity statement. In floating-point structural analysis, a model can be severely ill-conditioned without an exactly zero computed determinant, and the numerical size of a determinant depends strongly on matrix scaling and dimension. Diagnose poor conditioning using the solver's stability/pivot/conditioning diagnostics together with the physical model, restraint system, stiffness contrast, deformed shape and affected degrees of freedom—not a universal “small determinant” threshold.
Do not add arbitrary restraints just to make the model run
Artificially fixing a degree of freedom can suppress a solver instability while also creating false reaction paths and unrealistically high stiffness. Add a restraint only when the physical structure/foundation provides it or when a documented analytical idealization justifies it.
STAAD Diagnostic Simulator
Illustrative diagnostic categories; exact messages and menu paths vary by installed release.
The base rotational degree of freedom was released, so the idealized cantilever becomes a rigid-body mechanism under lateral load.
Required action: Remove the unintended base moment release or otherwise restore the physical rotational restraint required by the structural idealization. Re-check all releases before re-analysis.
Multiple Structures / Disconnected Connectivity
What to inspect
If one building is reported as several analytical structures, search for nearly coincident joints, members that cross visually without sharing a joint, deleted bridging members, separated plate meshes, or imported geometry that was not merged. Display joint numbers and coordinates rather than relying on line rendering alone.
Abnormal Results as Diagnostics
Suspicious-result triggers
- A displacement is orders of magnitude larger than a hand estimate.
- A nominally symmetric model twists unexpectedly.
- One support reaction is zero when load should reach it—or enormous without physical reason.
- Total reactions do not match the applied static load.
- A member force diagram has an impossible jump without a corresponding load/release/joint.
- A mode shape is dominated by local motion that should be restrained.
- Concrete/steel design fails almost every member after a small model change.
Use the deformed shape as an X-ray
Magnify the deformed shape and animate it if helpful. Mechanisms, disconnected joints, wrong support directions, accidental weak-axis orientation, missing diaphragm/bracing behavior and incorrect releases often become immediately visible through the displacement pattern.
Nonlinear / P-Delta Convergence
Nonconvergence is not automatically collapse—and not automatically harmless
An iterative analysis can fail because the structural state truly becomes unstable, because load steps/tolerances/iterations are inappropriate, or because the model contains poor connectivity, extreme stiffness contrast or another numerical problem. Review the physical trend, model assumptions, solver settings and intermediate response before interpreting the failure.
Model QA Gate Before Design
STAAD analysis acceptance checklist
- Model revision/source is identified.
- Units and global-axis convention are documented.
- Duplicate/disconnected/zero-length geometry checks are complete.
- Materials, sections, beta angles, offsets and releases are reviewed.
- Supports represent the physical restraint system.
- Load cases, signs, mass source and combinations are checked.
- No unresolved fatal errors or unexplained warnings remain.
- Applied-load totals are independently checked.
- Reactions/equilibrium are reviewed.
- Deformed shapes and representative force diagrams are reasonable.
- Dynamic modes/participation or nonlinear convergence are reviewed where applicable.
- Governing load cases/combinations are identified.
Professional Calculation / Model Report
A report should enable review—not overwhelm the reviewer
Raw analysis output can contain thousands of lines. A useful structural report explains the model and demonstrates why the engineer accepted it. Include enough source/input data for traceability, but prioritize assumptions, design basis, verification checks and governing results.
Recommended report structure
- Project/model identification: model revision, software/version, date, engineer/checker.
- Design basis: governing standards/editions, materials, units, loading criteria and major assumptions.
- Model description: geometry, structural system, supports, releases, stiffness assumptions, mass model and important idealizations.
- Loading: primary cases, generated loads, mass source, combination/envelope basis.
- Verification: load totals, reaction/equilibrium checks, deformed shape, benchmark hand checks, mesh/dynamic/nonlinear verification as relevant.
- Governing analysis results: displacements, reactions, member forces, plate/surface results and governing cases.
- Design summary: critical utilization/reinforcement results and failed-member disposition.
- Concrete detailing handoff: RCDC/Advanced Concrete model revision and drawing/schedule status where used.
- Warnings/exceptions: disposition of solver warnings and known limitations.
- Appendices: selected input/output tables and graphics needed for auditability—not indiscriminate raw dumps.
Revision discipline
If geometry, support conditions, stiffness, load definitions or analysis settings change after design/detailing, record the new model revision, re-run analysis verification, then refresh downstream design/RCDC deliverables. Do not mix results from different analytical revisions in one report.
- Troubleshooting should isolate the root physical/model cause, not merely silence messages.
- Instability warnings often originate in restraints, releases, missing stiffness or disconnected topology.
- Deformed shape, reaction balance and model text/tables are high-value diagnostic tools.
- Nonlinear nonconvergence requires interpretation; it should not be labeled collapse without evidence.
- A professional report demonstrates assumptions, verification and governing results rather than printing every available output line.
- Model revision control must extend into downstream concrete design/detailing deliverables.