{"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 controller tuning change management approval parameter changes safety envelope adaptive tuning product","site:isa.org controller tuning poor performance process plants PID loops percentage","site:osha.gov process safety management management of change control system setpoint parameters"],"source_ids":["SRC1","SRC2","SRC3"],"no_result_note":"The search established poor or stale controller tuning and formal change-control obligations, but found no direct public evidence that a plant uniformly sends every low-risk retuning through a slow human queue."},"closest_prior_art":{"queries":["site:emerson.com DeltaV adaptive tuning user limits automatic tuning controller","safe automatic PID tuning industrial process constraints rollback controller gains research","industrial adaptive PID tuning parameter bounds validated models automatic rollback","risk based management of change process control parameter changes tiered approval"],"source_ids":["SRC5","SRC6","SRC7","SRC8"],"no_result_note":null},"historical_terminology":{"queries":["1970 self-tuning regulator process control automatic controller parameters industrial historical paper","1980 adaptive control self tuning PID industrial process automatic tuning historical","\"self-tuning regulator\" industrial process controller parameter adjustment historical"],"source_ids":["SRC8"],"no_result_note":null},"products_practices_standards":{"queries":["site:iec.ch industrial automation change management controller parameters IEC 62443","site:nist.gov industrial control systems configuration change control parameter changes","site:osha.gov laws regs regulations standardnumber 1910 1910.119 management of change employer authorize","site:emerson.com DeltaV adaptive tuning user limits automatic tuning controller"],"source_ids":["SRC3","SRC4","SRC5","SRC6"],"no_result_note":null},"non_english_regional":{"queries":["Regelkreis Optimierung automatische PID Anpassung Parameter Grenzen Industrie Anlage Genehmigung","ajuste automático PID industrial límites parámetros seguridad control adaptativo español"],"source_ids":["SRC6"],"no_result_note":"Spanish Siemens documentation supplied a regional first-party analogue with bounded optimization and automatic cancellation; it did not document approval latency or tiered authorization."},"composition_subproblems":{"queries":["safe automatic PID tuning industrial process constraints rollback controller gains research","risk based management of change process control parameter changes tiered approval","industrial PID controller tuning change management approval parameter changes safety envelope adaptive tuning product"],"source_ids":["SRC3","SRC4","SRC5","SRC6","SRC7"],"no_result_note":null}},"sources":[{"source_id":"SRC1","title":"Automation Basics: Closed-loop control troubleshooting","url":"https://www.isa.org/intech-home/2018/march-april/departments/closed-loop-control-troubleshooting","publisher":"International Society of Automation","date_or_year":"2018","source_type":"TRADE_PROFESSIONAL","language":"English","claims_supported":["Process plants commonly have poorly performing control loops.","Incorrect tuning can cause oscillation, while sluggish tuning can permit large deviations.","Retuning is not always the remedy because valves, noise, disturbances, and interacting loops can be root causes."]},{"source_id":"SRC2","title":"Analytics is changing our understanding of regulatory control","url":"https://www.isa.org/intech-home/2020/may-june/features/analytics-is-changing-our-understanding-of-regulat","publisher":"International Society of Automation","date_or_year":"2020","source_type":"TRADE_PROFESSIONAL","language":"English","claims_supported":["An assessment using data from 116 production facilities classified 20 percent of controllers as needing tuning and another 25 percent as fair.","Control-loop performance monitoring products can identify changing dynamics and recommend tuning adjustments from routine operating data.","The article links poor control to excess energy and operational losses but warns that older industry-wide estimates were anecdotal or generalized."]},{"source_id":"SRC3","title":"29 CFR 1910.119 — Process safety management of highly hazardous chemicals","url":"https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.119","publisher":"U.S. Occupational Safety and Health Administration","date_or_year":"1992; amended through 2013 on the cited page","source_type":"OFFICIAL_STANDARD","language":"English","claims_supported":["For covered hazardous-chemical processes, the employer must maintain written procedures for changes other than replacement in kind.","Before a covered change, the technical basis, safety and health impact, operating-procedure effects, duration, and authorization requirements must be addressed.","Controls, sensors, alarms, and interlocks are within the standard's process-equipment scope, supporting the need to preserve safety authority and hard protections."]},{"source_id":"SRC4","title":"NIST SP 800-171 Revision 3: Protecting Controlled Unclassified Information in Nonfederal Systems and Organizations","url":"https://nvlpubs.nist.gov/nistpubs/SpecialPublications/800-171r3/NIST.SP.800-171r3.html","publisher":"National Institute of Standards and Technology","date_or_year":"2024","source_type":"OFFICIAL_GUIDANCE","language":"English","claims_supported":["Configuration-change control defines which change types are controlled and requires review, approval or disapproval, implementation records, and monitoring.","The guidance identifies authorizing officials, system owners, security engineers, and risk-management roles as relevant actors.","Its scope is CUI confidentiality rather than general process safety, so it is an adjacent cybersecurity-governance constraint rather than direct evidence about PID approval queues."]},{"source_id":"SRC5","title":"DeltaV Adapt: Continuous Closed Loop Adaptive Control","url":"https://www.emerson.com/is/content/emerson/es/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":"2024","source_type":"FIRST_PARTY_PRODUCT","language":"English","claims_supported":["A commercial industrial-control product continuously adjusts PID tuning as operating conditions change.","Control engineers can define operating regions, approve regional models, and constrain model-identification and tuning parameters.","The product validates models, offers full-adaptive, approved-region partial-adaptive, and advisory-only modes, and starts identification from the last good model with per-event change limits."]},{"source_id":"SRC6","title":"Límites del valor real — PID_Temp (S7-1200, S7-1500, S7-1200 G2)","url":"https://docs.tia.siemens.cloud/r/es-es/v21/regulacion-pid/regulacion-pid-s7-1200-s7-1500-s7-1200-g2/uso-de-pid_temp-s7-1200-s7-1500-s7-1200-g2/configurar-pid_temp-s7-1200-s7-1500-s7-1200-g2/ajustes-del-valor-real-s7-1200-s7-1500-s7-1200-g2/limites-del-valor-real-s7-1200-s7-1500-s7-1200-g2?contentId=DNR7tSVe0sjW~g5v2H2Ssg","publisher":"Siemens","date_or_year":"2025","source_type":"FIRST_PARTY_PRODUCT","language":"Spanish","claims_supported":["Siemens PID_Temp permits absolute upper and lower process-value limits to be configured.","Controller optimization is automatically cancelled if those limits are crossed.","The documented safeguard is a concrete regional product analogue for bounded tuning, although it does not establish risk-tiered organizational approval."]},{"source_id":"SRC7","title":"Safety-Aware Cascade Controller Tuning Using Constrained Bayesian Optimization","url":"https://arxiv.org/abs/2010.15211","publisher":"arXiv; authors from ETH Zurich and Inspire AG","date_or_year":"2020 preprint; published in IEEE Transactions on Industrial Electronics in 2023","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["The researchers experimentally implemented automated performance-based controller tuning on a CNC grinding-machine axis.","They restricted optimization using experimentally determined unsafe gains and a pre-decided maximum-overshoot bound.","The experiment demonstrates technical feasibility of bounded automated tuning but does not test approval latency, plant change governance, or process-safety event rates."]},{"source_id":"SRC8","title":"Self-Tuning Regulators","url":"https://lucris.lub.lu.se/ws/files/55837187/Self_tuning_regulators_MIT_workshop_1975.pdf","publisher":"Lund Institute of Technology, Lund University","date_or_year":"1975","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["Automatic parameter adaptation under the older terms self-tuning regulator and adaptive regulator was already developed and tested in industrial applications by the mid-1970s.","Reported applications included paper machines, digesters, ore crushers, an enthalpy exchanger, and a supertanker autopilot.","The paper also describes stability and convergence difficulties and reports misapplications, showing that automatic adaptation did not eliminate the need for safeguards and competent system design."]}],"problem_evidence":{"status":"PARTLY_SUPPORTED","finding":"The generic technical problem exists: industrial loops can be poorly tuned, changing operating conditions can make fixed tuning inappropriate, and poor control can increase variability, energy use, and intervention. Formal safety and cybersecurity change controls also exist. The proposal's central procedural assertion—that one plant applies the same slow human approval path to every retuning and that this queue materially strands safe improvements—was not directly established.","source_ids":["SRC1","SRC2","SRC3","SRC4","SRC5"],"uncertainty":"No public plant-level approval logs linked procedural latency to tuning age or downstream performance. Poor control can instead arise from valves, sensors, disturbances, nonlinearities, or loop interaction, and covered hazardous processes may require substantive analysis rather than merely procedural routing."},"adopter_evidence":{"status":"SUPPORTED","finding":"An organizational adopter and authorizer are identifiable. In OSHA-covered processes the employer owns the management-of-change procedure and must define authorization requirements; NIST identifies system owners and authorizing or security roles; Emerson assigns operating-region model approval and tuning limits to control engineers. The exact accountable title and delegation rules remain plant-specific.","source_ids":["SRC3","SRC4","SRC5"],"uncertainty":"The public sources do not verify that a designated process-safety engineer at any particular plant can legally delegate repeated PID changes through standing preauthorization."},"implementation_evidence":{"status":"SUPPORTED","finding":"Bounded automated or semi-automated retuning is technically implementable. DeltaV Adapt supplies validated-model, constrained, full-adaptive and approved-region modes; Siemens cancels optimization on configured process-limit violations; and published CNC experiments used constrained gain ranges and overshoot bounds. These analogues support a shadow test and a low-criticality pilot, not unrestricted deployment.","source_ids":["SRC5","SRC6","SRC7","SRC8"],"uncertainty":"None of the retained sources demonstrates the complete proposed package—risk classification, standing organizational preauthorization, automatic rollback, cybersecurity controls, and matched process-safety outcomes—in the specified industrial-plant setting."},"prior_art":{"disposition":"ADJACENT_PRIOR_ART","closest_analogues":[{"name":"Emerson DeltaV Adapt approved-region and full-adaptive tuning","source_ids":["SRC5"],"same_problem":false,"same_causal_lever":true,"overlap":"It continuously changes PID gains as dynamics change, validates models, lets an engineer approve operating-region models, constrains parameters, and retains last-good regional tuning.","remaining_difference":"It addresses nonlinear or changing process dynamics rather than proving that a uniform approval queue is binding; the source does not describe risk-tiered change governance, automatic rollback to the last formally approved gains, or a matched evaluation of approval latency and protected outcomes."},{"name":"Siemens PID_Temp bounded optimization","source_ids":["SRC6"],"same_problem":false,"same_causal_lever":true,"overlap":"It combines controller optimization with configured process-value bounds and automatic cancellation when a bound is exceeded.","remaining_difference":"It does not supply the proposed organizational risk tiers, standing authorization, approval-queue redesign, or plant-level incidence and distributional review."},{"name":"Safety-aware constrained Bayesian controller tuning","source_ids":["SRC7"],"same_problem":false,"same_causal_lever":true,"overlap":"It automates gain optimization within experimentally derived safety and stability limits and evaluates performance on physical industrial equipment.","remaining_difference":"It is a controller-tuning experiment, not a study of change-approval latency, formal plant authority, automatic restoration of formally approved gains, or long-duration process-safety outcomes."},{"name":"Historical self-tuning regulators","source_ids":["SRC8"],"same_problem":false,"same_causal_lever":true,"overlap":"Real-time identification and automatic controller-parameter adjustment were tested in multiple industrial applications by 1975.","remaining_difference":"The historical work lacks the proposal's explicit risk-tiered authorization, preserved interlocks and cybersecurity controls, predetermined halt rules, and matched governance evaluation."}],"contrastive_claim_remaining":"At a plant that currently requires case-by-case human authorization for all PID gain changes, an ex-ante, independently defined low-risk class with validated parameter bounds, retained interlocks, logged changes, operator override, and automatic restoration of last-approved gains will reduce median approval-to-deployment latency and improve control-band time, cycling, energy per service unit, or intervention load without worsening trips, alarms, maintenance burden, cybersecurity findings, or service quality relative to the existing workflow and advisory-only tuning.","contrastive_claim_falsifier":"The claim is falsified if retrospective logs show procedural review contributes less than 5 percent of tuning latency or delayed requests do not predict degraded performance; or, after a binding gate is shown, the bounded pilot fails to materially reduce latency or improve any prespecified performance measure, breaches an envelope, increases any protected safety or maintenance measure, or merely reproduces benefits already obtained by an enabled adaptive-tuning product.","confidence":"MODERATE","search_limitations":"This was a bounded public-web search using exactly eight retained direct sources. Plant procedures, controller audit trails, vendor customer deployments, incident data, paid standards, and internal hazard analyses were unavailable. Product availability does not prove routine use, and the search cannot distinguish all local interpretations of management-of-change requirements."},"researchability_gates":{"externally_supported_problem":{"status":"INDETERMINATE","rationale":"Poor and stale tuning and the existence of change-control obligations are externally supported, but the proposed causal bottleneck—uniform slow approval of every low-risk retune—was not directly evidenced. Plant logs are required to determine whether the asserted problem exists in the target setting.","source_ids":["SRC1","SRC2","SRC3","SRC4","SRC5"]},"identifiable_adopter_or_authorizer":{"status":"PASS","rationale":"The employer or plant owner, system owner or authorizing role, and control engineer are identifiable organizational actors capable of owning change control, model approval, and tuning limits, subject to local delegation rules.","source_ids":["SRC3","SRC4","SRC5"]},"distinct_testable_incremental_claim":{"status":"PASS","rationale":"Despite close adaptive-tuning prior art, a distinct governance claim remains: risk-tiered standing preauthorization can reduce demonstrated procedural latency and improve matched outcomes beyond case-by-case or advisory-only workflows without degrading protected metrics.","source_ids":["SRC3","SRC4","SRC5","SRC6","SRC7","SRC8"]},"bounded_next_evidence_step":{"status":"PASS","rationale":"One redundant, low-criticality loop can first be evaluated retrospectively and in shadow mode, then in a time-limited matched pilot with fixed eligibility, parameter bounds, outcome metrics, halt triggers, and restoration rules.","source_ids":["SRC5","SRC6","SRC7"]},"no_unresolved_safety_or_authority_stop":{"status":"PASS","rationale":"No source establishes a categorical prohibition on a properly authorized bounded pilot. Safety and cybersecurity require a stop unless the employer's change procedure approves the class, interlocks and access controls remain unchanged, the envelope is independently validated, and operators retain override authority; those are feasible preconditions rather than unresolved categorical barriers.","source_ids":["SRC3","SRC4","SRC5","SRC6","SRC7"]},"adequate_search_evidence":{"status":"PASS","rationale":"All six required lanes were searched adversarially, including historical self-tuning terminology, official standards and guidance, commercial products, Spanish terminology, and combinations of governance, constraints, rollback, and safe-tuning subproblems. Eight direct sources from seven publisher organizations were retained and opened or fetch-attempted, including official, first-party, and primary sources.","source_ids":["SRC1","SRC2","SRC3","SRC4","SRC5","SRC6","SRC7","SRC8"]}},"strict_success":false,"screen_survival":false,"remaining_research_value":"MODERATE","recommended_next_step":"Before any live retuning, select one candidate redundant low-criticality loop and audit at least 6–12 months of tuning requests and controller histories. Pre-register a low-risk classifier independent of outcomes; decompose total latency into substantive hazard analysis versus routing or queue delay; test whether delay predicts control error, cycling, energy, alarms, and interventions; and replay candidate bounded changes in four weeks of shadow mode. Proceed to the four-week matched live pilot only if procedural delay is material, the envelope is independently validated across load, disturbance, sensor-error, saturation, and loop-interaction cases, and the plant's formal safety and cybersecurity authorities approve the standing authorization and rollback procedure.","world_novelty_boundary":"This bounded search identifies close historical, research, product, and governance analogues and leaves a plant-specific incremental governance claim. It does not establish world novelty, patentability, freedom to operate, market size, routine adoption, or realized impact."}