Answers are grounded in this paper and cite their sources.
Mature waterflood reservoirs contain long injection–production histories, yet routine surveillance remains challenged by delayed producer responses and interference from multiple injectors. Existing diagnostic plots, CRM-type analyses, and data-driven models are useful, but they often require preselected patterns, aggregate long-period behavior, or provide limited event-level traceability. This study presents an event-window evidence-fusion workflow (EWEF) for screening candidate injection–production responses without treating the screened objects as causal proof. EWEF uses material injection-rate perturbations as observational triggers and constructs auditable event-window evidence for candidate injector–producer objects. The workflow includes dynamic-record checking, injection-event extraction, candidate-pair construction, pre/post response-window quantification, lag scanning of oil-rate and water-cut responses, static-prior scoring from geometry and transmissibility proxies, relative multi-injector attribution, and rule-model disagreement routing. A transparent rule layer assigns engineering review classes, and a random-forest model trained on rule-derived weak labels serves only as a consistency-review gate. The workflow was applied to three anonymized field waterflood samples comprising 54 dynamic wells, 153 injection perturbation events, 297 candidate injector–producer evidence objects, and 30 final review-list objects. In the weak-label consistency model, lagged oil-rate correlation, water-cut shift, oil-rate shift, static prior, and water-cut lag correlation were recurrent influential input variables. In one representative event, the six-month post-event window showed that oil rate increased by 5.49 m3/d, whereas water cut decreased by 0.52 percentage points relative to the pre-event baseline. Lag scanning yielded modest correlations, with maximum absolute values of approximately 0.34; these correlations are treated as descriptive screening evidence rather than as statistically confirmed connectivity indicators. EWEF therefore provides a traceable workflow for screening and prioritizing field-review candidates. It does not establish causal interwell connectivity. Confirmation requires independent tracer, pressure, intervention, logging, or simulation evidence, and field-specific calibration remains necessary.
No takes yet. Share an insight, caveat, or question.
Li et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: