Answers are grounded in this paper and cite their sources.
This study integrated selective data reconstruction and mass-load analysis to evaluate the long-term performance of a centralized industrial wastewater treatment plant before and after capacity expansion. Monitoring records from 2018–2024 contained 59–60% missing influent data and 61–63% missing effluent data, including prolonged gaps. Influent gaps of up to three days were reconstructed by linear interpolation, while Random Forest (RF), selected from four models, reconstructed eligible effluent gaps of up to 14 days. Longer gaps remained unfilled. Discharge-limit assessment used observed effluent measurements exclusively, whereas load-based removal analysis included observed and selectively reconstructed data. The largest phase difference occurred in total suspended solids (TSS) removal, with median efficiency increasing from 79.70% to 91.98% (p < 0.001; rrb = 0.701). This positive difference persisted in measured-only analyses and sensitivity tests excluding operational phase as an RF predictor or varying hypothetical influent–effluent time lags. Temporal validation and missingness-pattern assessments further highlighted limitations affecting model reliability and interpretation. However, extensive missingness, unequal monitoring coverage, and unavailable process measurements prevent causal attribution of the improvement to capacity expansion. The proposed framework combines traceable reconstruction, observed-data-based compliance assessment, and sensitivity analyses to support cautious interpretation of historical treatment performance while acknowledging uncertainty arising from incomplete monitoring.
No takes yet. Share an insight, caveat, or question.
Cheraghi et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: