{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp09_archetype_breadth150_20260804","cell_id":"load_balancing__chemistry_materials","arm":"BREADTH_PROBE_ONE_SHOT","candidate_id":"load_balancing__chemistry_materials__P1","proposal_index":1,"version":0,"title":"Capacity-Aware Routing of Exothermic Polymer Batches Across Parallel Reactors","problem":"A polymer pilot plant has several validated stirred-tank reactors capable of running overlapping recipe families, but new batches are assigned by fixed reactor ownership or simple first-available rules. Because reactors differ moment to moment in cooling headroom, cleaning state, compatible prior contents, queue length, and equipment health, one reactor can accumulate thermally demanding batches while another eligible reactor remains idle or lightly loaded.","actors":["Process supervisor responsible for batch assignment","Polymer chemists submitting batch recipes","Reactor operators executing approved batch records","Process-safety engineer maintaining operating envelopes and interlocks","Quality personnel governing equipment qualification, traceability, and change control"],"observable_state":"At each batch-release decision, timestamped records show the recipe's vessel-eligibility constraints and thermal demand alongside each reactor's committed queue, estimated cooling headroom, cleaning and contamination status, qualified operating range, sensor and agitator health, and expected release time. The problem state is present when an eligible reactor is projected to exceed its approved queue or thermal-headroom threshold while equivalent qualified capacity is idle elsewhere.","consequence":"Concentrated assignments can delay batch starts, increase the chance of temperature-profile deviations during closely spaced exothermic runs, leave qualified equipment unused, and create pressure for rushed cleaning or manual reassignment. A routing intervention must not relax reaction-safety, contamination, validation, or traceability requirements.","affected_objective":"Complete approved polymer batches within validated processing windows while preserving thermal safety, product segregation, equipment qualification, operator workload limits, and batch traceability.","intervention":"Place a decision-support dispatcher before batch release. For every incoming approved recipe, it first excludes reactors that fail hard eligibility, cleaning, maintenance, sensor-health, agitation, containment, or thermal-envelope checks. It then uses capacity-aware routing to rank the remaining reactors by forecast cooling-headroom consumption, committed processing and cleaning time, queue risk, and reassignment stability. The supervisor receives the ranked assignment with its signals and may approve or override it with a recorded reason. Active reactions are never migrated; only not-yet-started batches are routed, and a cooldown prevents repeated reassignment as estimates fluctuate.","structural_mapping":[{"archetype_element":"Incoming work or flow","domain_realization":"Approved, not-yet-started polymer batch recipes awaiting assignment to a reactor."},{"archetype_element":"Multiple viable receivers","domain_realization":"Parallel stirred-tank reactors qualified for overlapping recipe families."},{"archetype_element":"Capacity signal","domain_realization":"Forecast cooling headroom, committed reaction and cleaning time, queue length, and operator coverage for each reactor."},{"archetype_element":"Health check","domain_realization":"Maintenance state plus required temperature, pressure, flow, and agitation instrumentation availability."},{"archetype_element":"Affinity and consistency constraint","domain_realization":"Recipe qualification, construction-material compatibility, contamination sequence, scale range, and batch-record traceability restrict eligible reactors."},{"archetype_element":"Distribution policy","domain_realization":"Hard eligibility filtering followed by weighted ranking of viable reactors using thermal and temporal headroom."},{"archetype_element":"Feedback and monitoring","domain_realization":"Actual heat-removal margin, cycle duration, cleaning duration, overrides, and deviations update later capacity estimates."},{"archetype_element":"Target invariant","domain_realization":"No qualified reactor is assigned beyond its approved envelope while equivalent eligible capacity remains unused."}],"mechanism_mapping":[{"mechanism_slug":"capacity_aware_routing","role":"Routes each released batch toward a qualified reactor with sufficient forecast thermal and temporal headroom rather than treating all reactors as equal.","counterfactual_removal":"Without current capacity signals, the dispatcher reduces to a static or equal split and can reproduce hot spots despite nominally balanced batch counts."},{"mechanism_slug":"health_checked_distribution","role":"Removes or downweights reactors whose required equipment, instrumentation, cleaning state, or maintenance status makes them unavailable for ordinary assignment.","counterfactual_removal":"Without health checks, the policy can direct a batch to a nominally free but unsafe or unusable reactor."},{"mechanism_slug":"weighted_routing","role":"Accounts for differences in reactor scale, cooling capability, expected cycle time, and recipe burden when ranking eligible destinations.","counterfactual_removal":"Without weights, equal assignment can overload a lower-capacity reactor while underusing a higher-capacity one."}],"causal_chain":["Static or first-available assignment ignores changing differences among otherwise qualified reactors.","Thermally demanding or long-duration batches therefore concentrate in one reactor's queue while eligible capacity elsewhere can remain unused.","Hard eligibility and health checks define which reactors are genuinely viable for each incoming recipe.","Capacity-aware weighted routing assigns the not-yet-started batch using current thermal headroom, committed time, cleaning burden, and queue risk.","Assignments shift away from projected hot spots toward qualified capacity with more usable headroom.","Recorded outcomes and overrides update capacity estimates, while cooldown rules limit route flapping.","If the diagnosis is correct, projected overload episodes should decline without increasing safety-envelope, compatibility, or traceability violations."],"baseline":"The baseline is fixed recipe-to-reactor ownership supplemented by operator judgment and a first-available rule. It preserves familiar qualification and accountability but does not systematically compare current usable capacity across all eligible reactors at each batch-release decision.","nearest_rivals":["Production scheduling: primarily changes the order and timing of batches; the proposed move selects among multiple currently viable reactors for each admitted batch using receiver state.","Capacity expansion: adds reactors or cooling equipment; the proposed move tests whether existing qualified capacity is being stranded by assignment before adding capacity.","Rate limiting or admission control: delays or rejects batch releases when the plant is busy; the proposed move first redistributes admitted work across usable reactors.","Failover: transfers work after a reactor becomes unavailable; the proposed policy distributes ordinary batches among healthy reactors before failure.","Campaign planning by chemistry family: groups compatible recipes to reduce cleaning; it can supply eligibility and changeover costs but does not itself react to current reactor load and health."],"remaining_contrastive_claim":"The intervention's distinguishing claim is limited to assignment topology: among already approved batches and already qualified reactors, live thermal, queue, cleaning, and health signals can identify avoidable localized overload that a static ownership or first-available baseline misses. It neither asserts that total plant capacity is sufficient nor replaces scheduling, campaign design, admission control, or safety interlocks.","authority_safety":{"decision_authority":"The designated process supervisor retains batch-to-reactor assignment authority, subject to existing process-safety, quality, equipment-qualification, and change-control approvals.","authorized_first_step":"Run the dispatcher in retrospective and then shadow recommendation mode using existing records; show rankings and reasons without changing any batch assignment or equipment state.","excluded_actions":["Autonomously starting, stopping, transferring, or rescheduling a reaction","Moving an active reaction between vessels","Changing validated recipes, charge sequences, operating envelopes, relief assumptions, or safety interlocks","Assigning a recipe to an unqualified, incompatible, uncleared, or unhealthy reactor","Suppressing alarms, required cleaning, quality review, or batch traceability","Using inferred capacity to override an operator or process-safety hold"],"halt_rollback":"Stop the shadow test if required input signals are missing or stale, if any recommendation violates the independently maintained eligibility matrix, or if recommendations oscillate beyond the preset reassignment limit. Rollback consists of discarding the recommendations and retaining the existing supervisor-controlled assignment process; no plant-control configuration is altered."},"negative_tests":{"strongest_counterevidence":"The apparent spare reactor capacity disappears after accounting for a shared cooling-water header, common operator constraint, incompatible recipe qualification, or mandatory cleaning sequence; the observed congestion is then aggregate or shared-bottleneck undercapacity rather than uneven assignment.","problem_falsifier":"Across the bounded sample, after applying all qualification, compatibility, health, cleaning, operator, and shared-utility constraints, there are no intervals in which one reactor is beyond the defined queue or headroom threshold while another equivalent eligible reactor has unused capacity.","intervention_falsifier":"In preregistered replay, the dispatcher does not reduce projected hot-spot intervals relative to the recorded baseline, or any apparent reduction requires more eligibility violations, safety-margin breaches, unstable reassignments, delayed critical batches, or unmodeled shared-utility overload.","risks":["Stale or optimistic cooling-headroom estimates could route a demanding batch to a marginal reactor.","A hidden shared cooling, staffing, feed, or analytical bottleneck could make vessel-level balancing ineffective.","Incorrect compatibility or qualification data could recommend an invalid assignment.","Frequent signal changes could cause assignment thrashing and disrupt material preparation or operator plans.","Optimizing utilization could conflict with contamination control, campaign efficiency, priority batches, or equitable operator burden.","The dispatcher could become an opaque coordination bottleneck if recommendations and overrides are not explainable and auditable.","Historical replay may underrepresent unusual reaction behavior or maintenance conditions."]},"next_evidence_step":"Using at most 30 consecutive historical batch-release decisions from one reactor train, reconstruct the information available at each decision, freeze an independently approved reactor-eligibility matrix, and compare the recorded assignment with shadow capacity-aware rankings. Count projected hot-spot intervals, idle eligible reactor-hours, eligibility errors, recommendation changes, and shared-utility threshold breaches. Do not operate equipment or change assignments; proceed beyond shadow analysis only if every recommendation passes independent safety and qualification review.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"No comparison with prior proposals was performed under runtime isolation. This candidate was derived solely from the supplied load-balancing structure and the chemistry-and-materials domain card.","revision_record":{"parent_version":null,"progress_targets_addressed":[],"conceptual_changes":[],"operational_changes":[],"evidence_changes":[],"claim_changes":[]}}