ECSS System Engineering — Analysis Tools and Models (space-systems/ecss/e10-analysis-tools-models)
Use when the task is to establish that the analysis methods, tools and
models of ECSS-E-ST-10C clause 5.3.4 are qualified before their
results are relied on -- deciding per tool whether heritage covers the
present use, and correlating the tool against independent reference
data when it does not.
Domain quick reference
- A tool's qualification basis is decided from two independent facts,
not one: whether the tool is heritage, and whether its domain of
applicability (physics regime, parameter range) has been confirmed
for this analysis. Only the pair (heritage, domain confirmed)
earns heritage reuse. A heritage tool pushed outside its confirmed
domain is treated exactly like a new tool -- it must be correlated.
- Correlation error is the relative difference between the tool's
prediction and an independently derived reference value (test data
or a validated benchmark method), expressed as a percent of the
reference. A zero reference value makes percent error undefined and
is an input error, not a finding.
- The acceptance tolerance is set by the criticality of the analysis,
not by the tool: a high-criticality analysis admits a narrow
correlation error, a low-criticality one a wide band. The same tool
and the same correlation error therefore pass in one analysis and
fail in another.
- A tool on a correlation basis is qualified only when both the
correlation error and the tolerance are on record. Missing
correlation data is an input error -- a tool cannot be quietly
passed for want of evidence.
- An analysis is valid only when every tool backing it is qualified.
Each tool is reviewed exactly once; a repeated tool identifier is an
input error, and an analysis with no reviewed tool has nothing
qualifying it at all.
Workflow
- Collect every analysis tool and model whose output feeds the
engineering decision, each with a unique identifier.
- For each tool, determine the qualification basis from its heritage
flag and the confirmation of its domain of applicability for this
analysis.
- For a tool on a heritage-reuse basis, record it as qualified and
move on -- no correlation is required.
- For a tool requiring correlation, take the acceptance tolerance
from the analysis criticality, compute the correlation error
against the reference value, and compare the two.
- Aggregate the per-tool verdicts; the analysis is valid only when
the issue list is empty, and each exceeded tolerance is reported
against the tool that caused it.
Pitfalls
- Reading "heritage tool" as sufficient on its own -- heritage earns
reuse only inside the domain the heritage record actually covers, so
a flight-proven tool applied to a new regime still needs correlation.
- Applying one fixed correlation tolerance across the programme -- the
tolerance follows the criticality of the analysis being supported,
so the acceptance band must be re-derived per analysis.
- Passing a correlation-basis tool whose error or tolerance was never
captured. Absent evidence is not weak evidence; the review must stop
rather than assume the tool is fine.
- Computing percent correlation error against a zero reference and
reporting the result as a large-but-finite error -- the quantity is
undefined and must be rejected at the input.
- Averaging correlation errors across the tools behind an analysis. A
single tool outside its tolerance invalidates the analysis; good
agreement elsewhere does not compensate for it.
Behavior contract (gate 3)
The qualification-basis, tolerance, correlation-error, per-tool verdict
and aggregation logic is exercised by the gate 3 contract test:
scripts/test_e10_analysis_tools_models.py against
scripts/e10_analysis_tools_models_logic.py (stdlib unittest, offline).
Run:
python3 scripts/test_e10_analysis_tools_models.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and
paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
1---2name: e10-analysis-tools-models3description: "Use when validate the analysis methods, tools and models behind a space system analysis before its results are used, under ECSS-E-ST-10C clause 5.3.4: decide whether a tool qualifies by heritage reuse or must be correlated, compute its correlation error against independently derived reference data, compare that error against the acceptance tolerance set by the analysis criticality, and aggregate every tool review into a single verdict on whether the analysis is valid. Trigger: ecss, e-st-10-system-scope, analysis-tools, model-qualification, tool-correlation, heritage-reuse, domain-of-applicability, analysis-criticality."4license: Apache-2.05compliance: STANDARDS-REF6standards:7 - id: ecss8 reference-only: true9gated: false10domain: space-systems11pack: space-systems12compatibility: "agentskills.io SKILL.md; any SKILL.md host (Claude Code, Hermes, OpenClaw)"13metadata:14 source: https://skillmd.com/skills/ashfordeou/e10-analysis-tools-models15 domain: space-systems16 subdomain: ecss17 tags: [ecss, e-st-10-system-scope, analysis-tools, model-qualification, tool-correlation, heritage-reuse, analysis-criticality]18 version: 0.1.019 author: Aero Agent Skills20---2122# ECSS System Engineering — Analysis Tools and Models (space-systems/ecss/e10-analysis-tools-models)2324Use when the task is to establish that the analysis methods, tools and25models of ECSS-E-ST-10C clause 5.3.4 are qualified before their26results are relied on -- deciding per tool whether heritage covers the27present use, and correlating the tool against independent reference28data when it does not.2930## Domain quick reference3132- A tool's qualification basis is decided from two independent facts,33 not one: whether the tool is heritage, and whether its domain of34 applicability (physics regime, parameter range) has been confirmed35 for *this* analysis. Only the pair (heritage, domain confirmed)36 earns heritage reuse. A heritage tool pushed outside its confirmed37 domain is treated exactly like a new tool -- it must be correlated.38- Correlation error is the relative difference between the tool's39 prediction and an independently derived reference value (test data40 or a validated benchmark method), expressed as a percent of the41 reference. A zero reference value makes percent error undefined and42 is an input error, not a finding.43- The acceptance tolerance is set by the criticality of the analysis,44 not by the tool: a high-criticality analysis admits a narrow45 correlation error, a low-criticality one a wide band. The same tool46 and the same correlation error therefore pass in one analysis and47 fail in another.48- A tool on a correlation basis is qualified only when both the49 correlation error and the tolerance are on record. Missing50 correlation data is an input error -- a tool cannot be quietly51 passed for want of evidence.52- An analysis is valid only when every tool backing it is qualified.53 Each tool is reviewed exactly once; a repeated tool identifier is an54 input error, and an analysis with no reviewed tool has nothing55 qualifying it at all.5657## Workflow58591. Collect every analysis tool and model whose output feeds the60 engineering decision, each with a unique identifier.612. For each tool, determine the qualification basis from its heritage62 flag and the confirmation of its domain of applicability for this63 analysis.643. For a tool on a heritage-reuse basis, record it as qualified and65 move on -- no correlation is required.664. For a tool requiring correlation, take the acceptance tolerance67 from the analysis criticality, compute the correlation error68 against the reference value, and compare the two.695. Aggregate the per-tool verdicts; the analysis is valid only when70 the issue list is empty, and each exceeded tolerance is reported71 against the tool that caused it.7273## Pitfalls7475- Reading "heritage tool" as sufficient on its own -- heritage earns76 reuse only inside the domain the heritage record actually covers, so77 a flight-proven tool applied to a new regime still needs correlation.78- Applying one fixed correlation tolerance across the programme -- the79 tolerance follows the criticality of the analysis being supported,80 so the acceptance band must be re-derived per analysis.81- Passing a correlation-basis tool whose error or tolerance was never82 captured. Absent evidence is not weak evidence; the review must stop83 rather than assume the tool is fine.84- Computing percent correlation error against a zero reference and85 reporting the result as a large-but-finite error -- the quantity is86 undefined and must be rejected at the input.87- Averaging correlation errors across the tools behind an analysis. A88 single tool outside its tolerance invalidates the analysis; good89 agreement elsewhere does not compensate for it.9091## Behavior contract (gate 3)9293The qualification-basis, tolerance, correlation-error, per-tool verdict94and aggregation logic is exercised by the gate 3 contract test:95scripts/test_e10_analysis_tools_models.py against96scripts/e10_analysis_tools_models_logic.py (stdlib unittest, offline).97Run:98python3 scripts/test_e10_analysis_tools_models.py99100## Compliance101102- ECSS standards are freely downloadable (ESA); cite the source and103 paraphrase per standards-map.yaml.104- compliance: STANDARDS-REF, gated: false.