{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp11_mechanism_context_external20_20260804","research_id":"eoa_inverse_innovation_exp11_external_scrutiny_20260804","cell_id":"deadweight_loss_reduction__systems_cybernetics","opaque_id":"deadweight_loss_reduction__systems_cybernetics__C","search_lanes":{"direct_problem":{"queries":["industrial PID loops poorly tuned energy valve wear alarm operator workload study","\"PID tuning\" \"management of change\" plant","controller tuning management of change plant PID retuning approval latency poor performance loops"],"source_ids":["SRC1","SRC2"],"no_result_note":"The retained sources establish prevalent poor loop performance, its consequences, and change-review requirements, but no direct study was found that quantifies controller-retuning approval latency or attributes degraded performance specifically to a uniform approval queue."},"closest_prior_art":{"queries":["industrial PID controller retuning change approval management of change low risk pre-approved parameter bounds automatic rollback","industrial controller adaptive tuning approved operating regions validated model parameter limits","risk-based management of change low risk pre-approved change industrial plant standard"],"source_ids":["SRC3","SRC4","SRC5","SRC8"],"no_result_note":null},"historical_terminology":{"queries":["1980s industrial self-tuning regulator adaptive PID process control field trial","\"self-tuning regulator\" process industry PID historical","Åström Hägglund automatic tuning PID 1984 relay feedback industrial controller PDF"],"source_ids":["SRC6"],"no_result_note":null},"products_practices_standards":{"queries":["site:emerson.com DeltaV adaptive control automatically retunes PID controller online limits","site:osha.gov process safety management management of change replacement in kind control system parameters","risk-based management of change low risk pre-approved change industrial plant standard"],"source_ids":["SRC2","SRC3","SRC4"],"no_result_note":null},"non_english_regional":{"queries":["taratura automatica PID adattativa impianto industriale limiti parametri sicurezza approvazione modifica","PID Selbstoptimierung industrielle Anlage Parametergrenzen Änderungsmanagement Freigabe"],"source_ids":["SRC7"],"no_result_note":"Non-English first-party documentation confirmed adaptive and self-tuning terminology and practice, but no retained non-English source addressed approval-queue latency specifically."},"composition_subproblems":{"queries":["control loop performance monitoring retuning management of change risk classification rollback baseline","adaptive PID tuning industrial controller validated bounds automatic retuning safety envelope","industrial controller adaptive tuning approved operating regions validated model parameter limits"],"source_ids":["SRC1","SRC3","SRC4","SRC5","SRC8"],"no_result_note":"No single retained source combined plant-specific approval-delay diagnosis, risk-tier governance, bounded automatic retuning, matched evaluation, and automatic rollback; the components appear across several sources."}},"sources":[{"source_id":"SRC1","title":"Fossil Plant Instrumentation and Control Guidelines, Volume 3: Monitoring and Improving Control Loop Performance","url":"https://restservice.epri.com/publicdownload/000000000001019684/0/Product","publisher":"Electric Power Research Institute","date_or_year":"December 2010","source_type":"SECONDARY_RESEARCH","language":"English","claims_supported":["Roughly one-third of industrial control loops perform poorly.","Poor tuning can cause oscillation, increased heat rate, equipment wear, emissions, excursions, and trips.","Historian-derived metrics can monitor error, oscillation, responsiveness, valve travel, limits, and alarms.","Changing process characteristics can make previously selected tuning unsuitable."]},{"source_id":"SRC2","title":"1926.64 Appendix C—Compliance Guidelines and Recommendations for Process Safety Management","url":"https://www.osha.gov/laws-regs/regulations/standardnumber/1926/1926.64AppC","publisher":"U.S. Occupational Safety and Health Administration","date_or_year":"1992; current OSHA guidance page","source_type":"OFFICIAL_GUIDANCE","language":"English","claims_supported":["Changes beyond replacement in kind require identification and review before implementation in covered processes.","Operators may retain flexibility within established operating parameters, while operation outside them requires written review and approval.","Temporary changes require time limits, monitoring, documentation, and restoration to designed conditions.","The employer is responsible for the management-of-change system and approvals."]},{"source_id":"SRC3","title":"Guidance on meeting expectations of EI Process Safety Management Framework Element 12: Management of Change and Project Management","url":"https://www.energyinst.org/industry/publications/topics/process-safety/guidance-on-meeting-expectations-of-ei-process-safety-management-framework-element-12-management-of-change-and-project-management","publisher":"Energy Institute","date_or_year":"September 2015","source_type":"OFFICIAL_GUIDANCE","language":"English","claims_supported":["Industry guidance explicitly calls for approved criteria classifying changes as high or low risk.","Change classification should be reviewed by competent nominated personnel with delegated authority.","Both low- and high-risk changes require proportionate risk assessment and defined controls.","The framework supplies performance measures for overdue assessments, controls, and temporary changes."]},{"source_id":"SRC4","title":"DeltaV Adapt: Continuous Closed Loop Adaptive Control","url":"https://www.emerson.com/is/content/emerson/en/systems-and-software/deltav-distributed-control-system-dcs/product-data-sheets/documents/deltav-adapt-continuous-closed-loop-adaptive-control.pdf","publisher":"Emerson","date_or_year":"October 2024","source_type":"FIRST_PARTY_PRODUCT","language":"English","claims_supported":["A commercial industrial product continuously adjusts PID tuning as process conditions change.","Engineers can define operating regions, approve models, and constrain identification and tuning parameters.","The product offers full-adaptive, approved-region partial-adaptive, and learn-only modes.","Models are quality-validated before use, and parameter changes are constrained per trigger event."]},{"source_id":"SRC5","title":"Stability-Preserving Automatic Tuning of PID Control with Reinforcement Learning","url":"https://arxiv.org/abs/2112.15187","publisher":"arXiv / paper authors","date_or_year":"2021; revised February 2022","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["The proposed tuner starts from a conservative stable baseline controller.","A supervisor monitors running tracking error and restores the baseline controller when a benchmark is exceeded.","Simulation on a second-order-plus-dead-time system maintained stability during parameter exploration.","The work demonstrates the technical pattern of monitored tuning with automatic fallback, though not plant MOC governance."]},{"source_id":"SRC6","title":"Experiences with Self-Tuning Control in the Process Industry","url":"https://www.sciencedirect.com/science/article/pii/S1474667017613120","publisher":"International Federation of Automatic Control / Elsevier","date_or_year":"July 1984","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["Adaptive regulators were already commercially available by the early 1980s.","ASEA NOVATUNE entered the market in 1982.","Industrial applications included a chemical reactor and a skin-pass mill.","Older relevant terminology includes adaptive regulator and self-tuning regulator."]},{"source_id":"SRC7","title":"ABB Quaderno Tecnico HVAC—Controllo adattativo e PID auto-ottimizzante","url":"https://cataloghi.it.abb.com/view/133633271/59/","publisher":"ABB","date_or_year":"October 2025","source_type":"FIRST_PARTY_PRODUCT","language":"Italian","claims_supported":["Italian first-party guidance describes self-optimizing PID control that continually adjusts gains as plant dynamics change.","It identifies self-tuning, auto-learning, and gain scheduling as established implementation forms.","It states that fixed tuning can degrade as loads, aging, faults, or seasonal conditions change.","It presents reduced manual tuning effort and improved efficiency as motivations for adaptive control."]},{"source_id":"SRC8","title":"Efficient Safe Learning for Controller Tuning with Experimental Validation","url":"https://arxiv.org/abs/2310.17431","publisher":"Engineering Applications of Artificial Intelligence / paper authors","date_or_year":"March 2025","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["Safe controller-parameter optimization can impose explicit performance constraints while exploring candidate settings.","The method was experimentally validated on an industrial-grade precision-motion system.","It achieved approximately 30% better tracking than a compared safe-learning method and improved on default autotuning.","The validation supports bounded experimentation but does not establish safety for coupled chemical-process loops."]}],"problem_evidence":{"status":"PARTLY_SUPPORTED","finding":"Poor or stale controller tuning is an externally documented industrial problem with measurable effects on control error, oscillation, energy use, wear, alarms, and trips. Industrial MOC guidance also confirms that parameter or operating changes may require controlled authorization and that risk-tier classification is legitimate. However, the proposal's decisive causal premise—that a uniform approval path makes procedural queue time dominate retuning latency and thereby causes the observed degradation—was not directly demonstrated by the retained sources.","source_ids":["SRC1","SRC2","SRC3","SRC7"],"uncertainty":"Poor performance has many competing causes, including valve stiction, sensor faults, loop coupling, incorrect mode, unsuitable controller structure, and substantive hazard analysis. Plant logs are needed to distinguish approval friction from technically necessary review and other failure modes."},"adopter_evidence":{"status":"SUPPORTED","finding":"The adopting organization is an industrial plant operator or employer with configuration and process-safety responsibilities. The authorizing roles are identifiable as the control-system owner and competent personnel holding delegated MOC/process-safety authority; OSHA places responsibility on the employer, and Energy Institute guidance explicitly requires competent nominated approvers.","source_ids":["SRC2","SRC3"],"uncertainty":"The sources identify role classes rather than the actual named personnel or delegation instruments at the candidate plant. Local regulatory, insurance, quality, cybersecurity, and union requirements could add approvers."},"implementation_evidence":{"status":"SUPPORTED","finding":"The technical components are feasible: commercial systems already perform continuous or approved-region PID adaptation with model validation and tuning limits; research demonstrates safe-set optimization and monitored restoration of a stable baseline; standard loop-performance metrics can supply predetermined monitoring. What remains untested is the complete plant-governance package and its effect on approval latency.","source_ids":["SRC1","SRC4","SRC5","SRC8"],"uncertainty":"Evidence ranges from vendor documentation and simulation to industrial-grade motion hardware, not a redundant low-criticality loop in the proposed plant. Interactions among coupled loops and rare process-safety events remain material limits."},"prior_art":{"disposition":"ADJACENT_PRIOR_ART","closest_analogues":[{"name":"Emerson DeltaV Adapt approved-region adaptive PID control","source_ids":["SRC4"],"same_problem":true,"same_causal_lever":true,"overlap":"Continuously or conditionally retunes PID loops as dynamics change; permits engineer-approved operating regions, validated models, constrained tuning parameters, and a learn-only mode before online changes.","remaining_difference":"It does not establish that a uniform human approval queue is binding, prescribe risk-tier MOC governance, use the proposal's matched expiring pilot, or specify restoration on the proposed safety and performance triggers."},{"name":"Energy Institute high-/low-risk management-of-change classification","source_ids":["SRC3"],"same_problem":false,"same_causal_lever":true,"overlap":"Explicitly classifies changes by risk and assigns proportionate assessment and approval to competent delegated personnel.","remaining_difference":"It is generic process-safety governance, not controller-retuning preauthorization inside validated parameter envelopes, automated monitoring, or rollback."},{"name":"Stability-preserving automatic PID tuning with baseline restoration","source_ids":["SRC5"],"same_problem":false,"same_causal_lever":true,"overlap":"Explores PID settings under a supervisor and replaces a degrading controller with a stable baseline when a monitored benchmark is exceeded.","remaining_difference":"It addresses algorithmic exploration safety in simulation rather than plant approval latency, formal authority, interlocks, cybersecurity, or a matched operational pilot."},{"name":"Industrial self-tuning regulators and safe-learning controller tuning","source_ids":["SRC6","SRC8"],"same_problem":true,"same_causal_lever":true,"overlap":"Automatic adaptation to changing dynamics has industrial roots dating to at least 1982, and modern constrained learning has demonstrated bounded performance improvement on industrial-grade hardware.","remaining_difference":"These sources do not integrate risk-tier change authorization, independent low-risk classification, plant-specific queue diagnosis, and governance-triggered rollback."}],"contrastive_claim_remaining":"At a plant where retrospective logs first show that procedural approval materially dominates low-risk retuning latency, an independently defined class of redundant low-criticality loops can be preauthorized for changes within prevalidated bounds, reducing median approval-to-deployment latency and improving control-band time, cycling, energy per service unit, alarms, or interventions versus a matched baseline without worsening trips, envelope breaches, overrides, or maintenance burden. This is a plant-specific causal and governance claim, not a claim that adaptive tuning or risk-tier MOC is new.","contrastive_claim_falsifier":"The remaining claim is falsified if procedural approval contributes less than 5% of latency; queued requests do not predict degradation; the low-risk class cannot be defined before observing outcomes; or a controlled pilot produces no material latency or performance improvement, breaches an envelope, causes an attributable trip, or worsens any protected safety metric. It would also collapse if the candidate plant already routinely operates substantially the same approved-region adaptive-tuning and risk-tier MOC package.","confidence":"MODERATE","search_limitations":"The search retained exactly eight opened public sources across seven publishers and six lanes. It did not access proprietary plant MOC procedures, incident databases, paywalled standards in full, vendor field-performance datasets, or the candidate plant's logs. Search results cannot establish worldwide novelty, routine prevalence across all industries, or comparative safety for rare events."},"researchability_gates":{"externally_supported_problem":{"status":"PASS","rationale":"Industrial guidance documents prevalent poor loop performance, stale tuning as process characteristics change, and consequential variability, energy, wear, alarms, and trips. The narrower approval-delay mechanism remains a testable uncertainty rather than an unsupported premise.","source_ids":["SRC1","SRC2","SRC7"]},"identifiable_adopter_or_authorizer":{"status":"PASS","rationale":"The plant employer or control-system owner is the adopter, while competent personnel with delegated MOC/process-safety authority are identifiable authorizers.","source_ids":["SRC2","SRC3"]},"distinct_testable_incremental_claim":{"status":"PASS","rationale":"Although adaptive tuning, approved operating regions, parameter limits, risk-tier MOC, and baseline fallback all have prior art, the plant-specific claim that uniform procedural review is binding and that the integrated governance package improves measured outcomes remains distinct and falsifiable.","source_ids":["SRC1","SRC3","SRC4","SRC5"]},"bounded_next_evidence_step":{"status":"PASS","rationale":"A retrospective log audit can first test the stated 5% latency falsifier without touching the plant. Conditional on that result, four weeks of shadow operation on one redundant low-criticality loop is technically bounded and measurable using established loop-performance metrics.","source_ids":["SRC1","SRC4","SRC5","SRC8"]},"no_unresolved_safety_or_authority_stop":{"status":"PASS","rationale":"The proposed boundary excludes interlocks, trips, actuator limits, cybersecurity controls, high-criticality loops, and out-of-envelope changes; it retains competent authorization, human override, monitoring, expiry, and restoration. External guidance supports risk classification but still requires local competent approval before any live pilot.","source_ids":["SRC2","SRC3","SRC4","SRC5"]},"adequate_search_evidence":{"status":"PASS","rationale":"All six required lanes were searched adversarially. Eight opened direct sources span official guidance, first-party products, primary research, older terminology, and Italian regional terminology, with substantially more than two independent publishers and more than two primary, official, or first-party sources.","source_ids":["SRC1","SRC2","SRC3","SRC4","SRC5","SRC6","SRC7","SRC8"]}},"strict_success":true,"screen_survival":true,"remaining_research_value":"MODERATE","recommended_next_step":"Before any live change, extract 12 months of tuning-request, approval, historian, alarm, trip, override, maintenance, and energy records for candidate low-criticality loops. Pre-register the low-risk classification and test whether procedural approval exceeds 5% of total latency and whether delay predicts degradation after controlling for load and known faults. If both conditions hold, obtain delegated control-owner and process-safety approval, run four weeks in learn-only shadow mode, then conduct the proposed four-week matched pilot with immutable bounds, interlocks, automatic restoration, and the stated halt rules.","world_novelty_boundary":"This bounded public search found close and long-standing prior art for self-tuning controllers, approved-region constrained adaptation, safe parameter exploration with baseline fallback, and risk-tier management of change. It did not find one opened source demonstrating the complete plant-specific problem–intervention–evaluation package. That residual contrast is researchable but is not evidence of world novelty, patentability, freedom to operate, market size, routine adoption, realized impact, or safety outside the defined pilot boundary."}