ECSS Structures — Test-Analysis Correlation DRD (space-systems/ecss/drd-test-analysis-correlation)
Use when the task is producing the test-analysis correlation report required by ECSS-E-ST-32C Annex O — comparing pre-test analysis predictions against test measurements for each measured quantity, applying correlation acceptance criteria, identifying prediction-measurement pairs that require model updating, and confirming the TAC document contains every DRD-mandated section.
Domain quick reference
- ECSS-E-ST-32C Annex O and the companion section 32-11 define the content requirements (DRD) for the test-analysis correlation report. The report must address every measured quantity, document the pre-test prediction alongside the test result, and state a pass/fail assessment against the programme-agreed acceptance threshold.
- Measured quantities are categorized into five types: frequency (natural frequency of a structural mode), mode_shape (spatial deformation pattern of a mode, assessed via the Modal Assurance Criterion), static_deflection (displacement under a static load case), stress (surface strain-gauge derived stress), and damping (modal damping ratio). Each quantity type carries a separate default acceptance threshold; programme-specific criteria take precedence over the defaults.
- Correlation is quantified by a relative delta for scalar quantities: delta = |predicted − measured| / |measured|. A delta at or below the threshold is a pass; above is a fail and triggers a model update requirement. Mode-shape correlation uses the Modal Assurance Criterion (MAC), a dimensionless number from 0 (no correlation) to 1 (perfect correlation); a MAC value at or above the threshold (default 0.90) is a pass.
- A model update is required whenever any prediction-measurement pair fails its acceptance criterion. The updated model must be re-submitted for re-correlation before the TAC report can be accepted. When all pairs pass, the model is considered validated for the test load environment.
Workflow
- Inventory all pre-test analysis predictions and their corresponding test measurement results. Each entry must identify the quantity type, the predicted value, the measured value, and the applicable acceptance threshold. An entry with an unrecognized quantity type must be rejected before it enters the correlation assessment — an uncategorized quantity is a documentation gap.
- For each frequency, static-deflection, stress, or damping entry, compute the relative delta: |predicted − measured| / |measured|. A zero measured value is undefined and must be flagged as an input error — do not substitute a near-zero value or silently skip the entry.
- For each mode-shape entry, retrieve the MAC value computed from the test and analysis mode vectors. Confirm the MAC value is in [0, 1]; a value outside this range indicates an error in the MAC computation and must be rejected.
- Apply the acceptance threshold to each entry. Compare delta ≤ threshold for scalar quantities, MAC ≥ threshold for mode shapes. Record a pass or fail status for every prediction-measurement pair.
- Aggregate all entries into the correlation summary table. Any entry with a fail status requires a model update justification. If no entries fail, the model is validated for the tested load environment.
- Determine whether a model update is required: the TAC report requires a model-update justification section whenever at least one entry has failed, even if the failure is subsequently resolved by a re-analysis.
- Verify the TAC report document contains all DRD-mandated sections: scope, applicable documents, test configuration, analysis model description, pre-test predictions, test results, correlation assessment, model update justification, and conclusions. A missing section is a document compliance gap independent of the numeric correlation outcomes.
Pitfalls
- Omitting the mode-shape correlation (MAC check) and relying solely on frequency correlation — two modes can match in frequency while being completely different shapes; a MAC check is required to confirm that the compared predictions and measurements refer to the same mode.
- Using the absolute difference rather than the relative delta for frequency correlation — a 5 Hz miss on a 10 Hz mode (50% error) looks small in absolute terms but is far outside any reasonable threshold; the relative delta normalises for mode frequency.
- Treating a measured value of zero as a valid denominator — a zero measurement makes the relative delta undefined; the entry must be flagged as an input error, not defaulted to 100% or silently passed.
- Accepting a MAC value above threshold on a single-mode pair and concluding global model validity — MAC is evaluated per mode pair; a high MAC on one mode does not imply agreement on other modes in the same frequency band.
- Skipping the model-update justification section when all entries pass — the DRD requires this section regardless of outcome; an empty or "not required" statement is the correct content, not a missing section.
Behavior contract (gate 3)
The quantity categorization, relative-delta computation, MAC assessment, per-pair correlation status, model-update determination, and DRD completeness logic is exercised by the gate 3 contract test: scripts/test_drd_test_analysis_correlation.py against scripts/drd_test_analysis_correlation_logic.py (stdlib unittest, offline). Run: python3 scripts/test_drd_test_analysis_correlation.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
