{"schema_version":1,"research_id":"eoa_inverse_innovation_exp04_external_evaluation_20260802","source_assessment_id":"layer_decay_and_expiration_management__aviation_aeronautics:RETRIEVAL_FIRST:v0","cell_id":"layer_decay_and_expiration_management__aviation_aeronautics","search_queries":["site:faa.gov AC 20-179 certification data retention PDF records readable retrievable","site:docs.aws.amazon.com computational fluid dynamics AWS hot warm cold storage lifecycle CFD","PRUNE preserving run environment provenance graph data eviction regeneration PDF","Airbus Helicopters simulation process data management certification analysis traceability","Siemens Teamcenter simulation process data management archive delete CAE traceability fact sheet PDF","PROV-IO+ cross platform provenance framework HPC paper DOI file lineage environment provenance","NASA aircraft certification by analysis guide simulation data traceability reproducibility PDF","NIST archive restore testing backups contingency plan restore test official guidance","site:aws.amazon.com/s3/pricing S3 pricing Glacier Deep Archive per GB month US East 2026","site:bls.gov Occupational Employment and Wage Statistics aerospace engineers May 2025 wage","site:bls.gov Occupational Outlook Handbook software developers median pay 2025","eCFR 14 CFR 21.49 availability type design records FAA NTSB"],"sources":[{"source_id":"S1","title":"AC 20-179 — Certification Data Retention Agreements and Government Records","publisher":"Federal Aviation Administration","url":"https://www.faa.gov/documentlibrary/media/advisory_circular/ac_20-179.pdf","source_class":"OFFICIAL_GUIDANCE","publication_date":"2013-06-10","accessed_at":"2026-08-02","claims_supported":["Type-design and substantiation data covered by a retention agreement are permanent records that must not be destroyed.","Electronic custodians must control storage, maintenance, and backup; preserve accessibility and readability; and retrieve legacy data or migrate it to retrievable media.","Test and analysis reports are among the data categories identified for retention.","Agreements must specify availability intervals, repository inventories, access controls, and control over removal."]},{"source_id":"S2","title":"Optimizing HPC Components — Computational Fluid Dynamics on AWS","publisher":"Amazon Web Services","url":"https://docs.aws.amazon.com/whitepapers/latest/computational-fluid-dynamics-on-aws/optimizing-hpc-components.html","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"undated","accessed_at":"2026-08-02","claims_supported":["CFD case data include CAD, meshes, input files, and output figures.","AWS documents hot, warm, and cold states for CFD data and lifecycle-rule transitions from active storage to object and archival storage.","Completed runs can be moved from compute storage to S3 or Glacier while remaining available for future retrieval."]},{"source_id":"S3","title":"Airbus Helicopters: Simulation Process and Data Management","publisher":"Dassault Systèmes","url":"https://www.3ds.com/insights/customer-stories/airbus-helicopters-simulation-process-data-management","source_class":"COMMERCIAL_FIRST_PARTY","publication_date":"2026","accessed_at":"2026-08-02","claims_supported":["An Airbus Helicopters methods-and-tools engineer expressly describes enormous simulation-data volume and a need to manage, access, retrieve, reuse, and trace it.","Airbus Helicopters adopted an SPDM backbone connecting simulation inputs to results, including information used during certification.","The adopter associates SPDM with certification objectives, productivity, data-search reduction, reproducibility, and digital continuity."]},{"source_id":"S4","title":"PRUNE: A Preserving Run Environment for Reproducible Scientific Computing","publisher":"IEEE eScience / University of Notre Dame Cooperative Computing Lab","url":"https://par.nsf.gov/servlets/purl/10047189","source_class":"PRIMARY_RESEARCH","publication_date":"2016","accessed_at":"2026-08-02","claims_supported":["PRUNE represents tasks, files, execution environments, lineage, and progeny in a workflow graph.","It treats derived files as disposable and regenerable while protecting non-regenerable roots and retaining task, result, resource, and checksum metadata.","Its proposed eviction priorities already include age, size, graph position, likelihood of reuse, and reproduction cost.","Regenerated outputs can be checked against retained checksums, substantially colliding with the candidate's graph-aware ranking and reconstruction logic."]},{"source_id":"S5","title":"Simulation Process and Data Management","publisher":"Siemens Digital Industries Software","url":"https://static.sw.cdn.siemens.com/siemens-disw-assets/public/2E75uZ8EUCGXAX766Kv3EK/en-US/siemens-simulation-process-data-management-factsheet.pdf","source_class":"COMMERCIAL_FIRST_PARTY","publication_date":"2024-03","accessed_at":"2026-08-02","claims_supported":["Siemens describes rapidly growing simulation volume, obsolete inputs, search burden, duplication, and data loss as operational problems.","Teamcenter Simulation manages typed simulation objects and automatically maintains relationships from product revisions to models, analyses, and results.","The product already supports independent lifecycle actions including revision, release, archive, and deletion while retaining contextual relationships.","It detects analyses made stale by parent revisions and supports role-based access and regulatory-compliance workflows."]},{"source_id":"S6","title":"NIST SP 1339 — OT Backup Quick Start Guide","publisher":"National Institute of Standards and Technology","url":"https://csrc.nist.gov/pubs/sp/1339/final","source_class":"OFFICIAL_GUIDANCE","publication_date":"2026-06-17","accessed_at":"2026-08-02","claims_supported":["Effective backup management includes regularly testing backups and reviewing them during recovery exercises.","Restore testing is therefore established safety practice rather than a novel mechanism, although the source is not specific to CFD certification bundles."]},{"source_id":"S7","title":"Amazon S3 Pricing","publisher":"Amazon Web Services","url":"https://aws.amazon.com/s3/pricing/","source_class":"COMMERCIAL_FIRST_PARTY","publication_date":"dynamic pricing accessed 2026-08-02","accessed_at":"2026-08-02","claims_supported":["Archive economics depend on storage class, object count, ingest and transition requests, retrievals, temporary restored copies, and data transfer.","Glacier classes impose minimum storage durations and metadata overhead, so data volume, object granularity, and restore frequency materially affect recurring cost.","AWS provides a directly available archive and restore path suitable for a bounded technical pilot."]},{"source_id":"S8","title":"National Employment and Wage Data by Occupation, May 2025","publisher":"U.S. Bureau of Labor Statistics","url":"https://www.bls.gov/news.release/ocwage.t01.htm","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2026-05-15","accessed_at":"2026-08-02","claims_supported":["May 2025 mean wages were $142,060 annually for aerospace engineers and $148,100 for software developers.","These wage benchmarks support labor-equivalent cost bands, subject to employer overhead, specialized certification expertise, and commercial-license costs."]}],"problem_evidence":{"support":"STRONG","rationale":"The problem is visible across an aerospace adopter, a major SPDM vendor, CFD-specific storage guidance, and primary workflow research. Airbus expressly reports enormous multidisciplinary simulation-data volume and difficulty managing it; Siemens describes rapidly growing simulation volume, obsolete inputs, duplication, search burden, and loss risk; AWS treats CFD lifecycle management as vital; and PRUNE identifies derived workflow files as the dominant storage burden. FAA guidance makes erroneous disposition consequential because protected substantiation and analysis records must remain readable, accessible, and retrievable. Public evidence does not quantify the candidate pilot's actual hot-storage volume or document a specific prior reconstruction failure, so local prevalence and magnitude remain unmeasured.","source_ids":["S1","S2","S3","S4","S5"]},"stakeholder_evidence":{"support":"MODERATE","rationale":"Airbus Helicopters is an identifiable adopter that has publicly expressed a need to manage enormous simulation-data volumes, make data retrievable and reusable, improve reproducibility, and preserve certification traceability. FAA aircraft-certification offices and design-approval holders are identifiable authorizers or custodians for protected certification records. However, no source expresses demand for this exact ranking-plus-gate-plus-quarantine-plus-restore-test package or commits funding for a pilot.","source_ids":["S1","S3"]},"prior_art":{"proximity":"SUBSTANTIAL_COLLISION","closest_analogues":[{"name":"PRUNE preserving run environment","similarity":"Already combines workflow dependency graphs, environment capture, protection of roots, deletion of regenerable derived files, retained metadata and checksums, regeneration, and multi-factor eviction priorities including age, size, graph position, reuse probability, and reproduction cost.","remaining_difference":"It is a general research prototype and does not document certification-record overrides, reversible quarantine, recurring stratified archive-restore drills, or the candidate's 40%/95% joint performance target.","source_ids":["S4"]},{"name":"Teamcenter Simulation SPDM","similarity":"Already manages typed CAE objects, automatically preserves product-model-analysis-result relationships, detects stale analyses, controls access, and supports archive and delete lifecycle states independently of the product lifecycle.","remaining_difference":"The opened documentation does not specify a hard inbound-reference disposition block, impact-sized quarantine, or recurring end-to-end reruns of archived certification-cited CFD bundles.","source_ids":["S5"]},{"name":"AWS CFD hot/warm/cold lifecycle","similarity":"Directly applies hot, warm, cold, and archival tiers to CAD, meshes, solver inputs, and outputs from completed CFD runs using lifecycle rules.","remaining_difference":"It does not add certification-aware dependency adjudication, reversible deletion, or proof that restored cited results remain reproducible.","source_ids":["S2","S7"]},{"name":"FAA retention controls plus tested-backup practice","similarity":"FAA guidance supplies permanent-record, inventory, access-control, readability, backup, and retrievability obligations, while NIST establishes recurring backup testing and recovery exercises as standard safety practice.","remaining_difference":"These controls do not prescribe value-ranked cleanup of non-record CFD artifacts or evaluate the proposed storage-reproducibility trade-off.","source_ids":["S1","S6"]}],"distinctive_claim_remaining":"In a defined certification-linked CFD collection, adding a hard inbound-dependency gate, reversible quarantine, and recurring end-to-end result reproduction to existing SPDM and hot/warm/cold tiering will reduce hot-storage residency by at least 40% while at least 95% of a predeclared, stratified sample of cited runs reproduces within the agreed restore-time limit; neither age/access-only tiering nor graph-aware regeneration without that closed loop will meet both thresholds.","confidence":"HIGH"},"implementation_evidence":{"support":"MODERATE","rationale":"The constituent technologies are credible: commercial SPDM maintains simulation relationships and lifecycle states; PRUNE demonstrates graph-aware preservation, eviction, and regeneration; AWS documents CFD tier transitions and archive restores; and NIST recommends tested recovery. Workflow feasibility is narrower because certification custodians must classify protected records and approve removal controls. Data feasibility depends on resolving proprietary links across CAD, mesh, solver, scripts, environments, reports, and external references. Legal and safety feasibility requires permanent exclusion of FAA-retained records from deletion, readable formats, controlled access, auditable approvals, and halt/rollback on restore or classification failures. No source demonstrates the integrated candidate on an aerospace certification dataset, and proprietary solver licenses, obsolete environments, nondeterministic comparisons, and hidden dependencies remain unresolved.","source_ids":["S1","S2","S4","S5","S6","S7"]},"scores":{"meaningful_impact":{"score":4,"rationale":"The intervention addresses an explicitly reported aerospace data-management burden while protecting evidence relevant to certification and later safety review; local storage magnitude and failure incidence remain unmeasured.","source_ids":["S1","S3","S5"]},"stakeholder_pull":{"score":4,"rationale":"Airbus has adopted SPDM and publicly tied its need to simulation-data volume, retrieval, reproducibility, and certification, but has not requested this exact lifecycle extension.","source_ids":["S3"]},"incremental_advantage":{"score":3,"rationale":"The combined dependency block, quarantine, and recurring reproduction test could provide a safer storage-reduction loop than age/access tiering alone, but its joint advantage has not been measured and most components already exist.","source_ids":["S2","S4","S5","S6"]},"distinctiveness_plausibility":{"score":2,"rationale":"PRUNE, Teamcenter, AWS lifecycle tiering, FAA retention controls, and tested-recovery guidance cover nearly all mechanisms separately or in substantial combinations; only the closed-loop certification deployment and quantitative trade-off remain.","source_ids":["S1","S2","S4","S5","S6"]},"technical_implementability":{"score":4,"rationale":"Provenance graphs, lifecycle states, object archives, restore procedures, checksums, and workflow controls are demonstrated capabilities, although legacy solver environments and semantic result comparison may be difficult.","source_ids":["S2","S4","S5","S6"]},"adoption_authority_feasibility":{"score":3,"rationale":"A design-approval holder can govern its non-record simulation estate, while FAA agreements and certification-record custodians define protected data and access obligations. Joint approval is feasible but organization-specific authority and hold rules must be verified.","source_ids":["S1","S3"]},"evidence_readiness":{"score":3,"rationale":"A non-destructive retrospective study is well bounded, but decisive evidence requires proprietary bundle inventories, citation links, storage telemetry, licenses, environments, and live restore execution.","source_ids":["S1","S3","S4","S5"]},"safety_net_benefit":{"score":4,"rationale":"Hard dependency blocks, protected-record overrides, reversible quarantine, audited approvals, and tested restores directly reduce irreversible-loss risk; incomplete dependency discovery and unrepresentative restore samples remain hazards.","source_ids":["S1","S4","S6"]},"scalability":{"score":3,"rationale":"Commercial SPDM and object-storage lifecycle systems can scale, but resolving cross-tool dependencies, maintaining old execution environments, reviewing holds, and running representative restores may become labor-intensive across programs.","source_ids":["S2","S5","S7"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"A non-destructive retrospective pilot over approximately three closed projects and 100 provenance-linked bundles, including inventory extraction, custodian classification, dry-run scoring, archive copies, and 30 stratified end-to-end restores.","confidence":"MODERATE","assumptions":["Approximately 1,000-3,000 combined hours from software/data engineering, aerospace analysis, records custody, and security staff.","Existing SPDM, solver, and storage access can be reused without a new enterprise license.","Cloud or internal archive and restore charges are small relative to specialist labor at pilot scale.","Loaded labor is higher than the BLS cash-wage benchmarks."],"source_ids":["S7","S8"]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Production-quality integration for one CFD group: bundle schema, provenance connectors, identity and access controls, hold workflow, audit trail, archive path, quarantine state, restore automation, monitoring, and validation documentation.","confidence":"LOW","assumptions":["Four to eight full-time-equivalent staff-years are not required; effort is limited to roughly two to six FTE-equivalent months plus specialist review.","An existing SPDM or PLM platform is available.","Commercial connector, security, or solver-license costs do not exceed several hundred thousand dollars.","No bulk permanent migration is included."],"source_ids":["S5","S7","S8"]},"operational_launch":{"band_2026_usd":"1M_TO_5M","scope":"Controlled launch across one aircraft program or business unit, including historical ingestion, dependency remediation, security and records validation, operator training, archive capacity, restore infrastructure, governance, and staged reversible transitions.","confidence":"LOW","assumptions":["Launch spans multiple simulation tools and repositories and requires roughly 5-15 loaded FTE-equivalents over a year.","Historical metadata quality requires manual remediation.","Existing enterprise PLM and HPC contracts remain usable, but integration and validation are material.","Scope excludes fleet-wide or enterprise-wide rollout and excludes permanent destruction."],"source_ids":["S1","S5","S7","S8"]},"annual_recurring":{"band_2026_usd":"250K_TO_1M","scope":"Ongoing policy administration, custodian reviews, exception revalidation, provenance maintenance, archive storage, retrievals, quarterly stratified restore drills, incident handling, audit reporting, and tool support for one program.","confidence":"LOW","assumptions":["One to four loaded FTE-equivalents provide engineering, records, platform, and audit support.","Archive volume is tens of terabytes to low petabytes rather than multi-petabyte annual growth.","Restore drills are sampled rather than exhaustive.","Commercial maintenance and solver-license costs remain within the band; storage price alone is not the dominant expense."],"source_ids":["S1","S6","S7","S8"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"A named aerospace adopter reports enormous simulation-data volume and retrieval/reproducibility needs; Siemens, AWS, and PRUNE independently document simulation-data overload, lifecycle management, and preservation-versus-storage tension.","source_ids":["S2","S3","S4","S5"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"Airbus Helicopters is an identifiable SPDM adopter with expressed certification and reproducibility needs; FAA certification offices and design-approval holders are identifiable authorities or custodians for retained certification data.","source_ids":["S1","S3"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The residual claim specifies a defined workflow, explicit comparator practices, a 40% hot-storage threshold, a 95% cited-run reproduction threshold, and a restore-time criterion, while acknowledging substantial prior-art overlap.","source_ids":["S1","S2","S4","S5","S6"]},"bounded_next_evidence_step":{"status":"YES","reason":"A copy-based retrospective pilot can inventory a finite collection, apply the controls without changing live states, and compare storage and restoration outcomes against predeclared thresholds.","source_ids":["S2","S4","S5","S6"]},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"For the first evidence step only, permanent deletion and live-state changes are excluded; FAA-retained records remain protected, custodians classify holds, archive tests operate on copies, and failures halt progression. A later live pilot still requires local records, security, engineering-owner, and custodian approval.","source_ids":["S1","S6"]},"credible_cost_scope_and_range":{"status":"YES","reason":"The ranges are deliberately broad, scoped by stage, and anchored to current aerospace/software wage data and first-party object-storage cost drivers. Confidence remains low beyond the first pilot because proprietary licenses, metadata quality, data volume, and integration complexity are unknown.","source_ids":["S7","S8"]}},"next_evidence_step":"With an aircraft-program data owner and certification-record custodian, select three closed CFD projects containing at least 100 bundles and predeclare 30 cited runs stratified by age, solver, artifact class, and environment. On read-only metadata and duplicate archive copies, compare: (A) current retain-hot/ad-hoc cleanup; (B) standard age/access-only hot-to-cold lifecycle; and (C) the proposed value ranking plus hard inbound-dependency block, preservation overrides, reversible quarantine simulation, and end-to-end restore/re-execution. Before testing, define result tolerances appropriate to deterministic and nondeterministic solvers and a restore-time limit. Measure eligible hot bytes, reviewer time, unresolved-reference rate, false disposition blocks, unexpected access requests, restore latency, environment reconstruction success, and reproduced-result rate. Falsify advancement if projected hot-storage reduction is below 40%, fewer than 95% of sampled cited runs meet the reproduction and timing criteria, any protected record is classified as eligible, unresolved dependencies exceed 5% after remediation, or the eligible population is too small to materially change storage. Do not alter live storage or permanently destroy data.","blocking_evidence":["No public source quantifies hot-storage volume, growth, or avoidable cost for the proposed pilot population.","No proprietary bundle inventory or certification-citation graph has been examined, so dependency completeness and eligible volume are unknown.","No end-to-end restore has shown that historical CAD, meshes, solver versions, libraries, credentials, licenses, and execution environments remain usable.","The 40% storage and 95% reproduction thresholds are proposed criteria, not externally validated requirements.","No adopter has committed to this exact intervention, supplied a data custodian, or authorized even a reversible live transition.","Commercial SPDM, solver, archive, integration, and cybersecurity costs are unavailable, limiting deployment-cost confidence.","The search did not establish whether an existing aerospace deployment already implements the exact four-part closed loop.","World novelty, patentability, freedom to operate, market size, and realized impact remain unmeasured."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"Bounded web research establishes substantial collision, not worldwide novelty. Graph-aware ranked eviction and regeneration, SPDM traceability and lifecycle states, CFD storage tiering, certification-record preservation, and tested recovery are already documented. The search did not find one direct source documenting their exact certification-specific closed loop with the proposed 40%/95% performance thresholds. Patentability, freedom to operate, exhaustive worldwide prior art, market size, and realized impact were not assessed.","arm":"RETRIEVAL_FIRST","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":true,"material_progress_observed":true,"progress_targets":["Secure one aircraft-program data owner and certification-record custodian as named pilot partners.","Measure the closed-project hot-storage baseline, annual growth, access distribution, and current storage cost from proprietary telemetry.","Resolve bundle provenance and certification references and demonstrate at least 95% dependency-map completeness with manual spot checks.","Predeclare deterministic and nondeterministic result-comparison tolerances, a restore-time service level, and the cited-run sampling plan.","Execute copy-based end-to-end restores across solver, age, artifact, and environment strata and report failures, timing, and remediation effort.","Compare retain-hot/ad-hoc cleanup, age/access-only tiering, and the full dependency-gated loop using the same bundle population.","Obtain records, legal, cybersecurity, engineering-owner, and storage-operator approval before any reversible live transition.","Replace broad deployment cost bands with vendor quotes, measured data volumes, expected retrieval frequency, and staff-effort logs.","Conduct a targeted additional search of aerospace deployment manuals, conference proceedings, and patents for the exact four-part closed loop before making any distinctiveness claim."],"reason":"Web evidence verifies a consequential problem, a credible adopter and authorizer, technical building blocks, substantial prior-art overlap, and a bounded safe study. The remaining 40%/95% incremental claim cannot be resolved by further public web search because it requires proprietary inventories, authority decisions, dependency validation, and live archive/reproduction testing."}}