Response surface methodology optimization of ozonation for COD removal from real industrial wastewater
摘要
The escalating discharge of recalcitrant industrial effluents necessitates robust tertiary treatment technologies capable of meeting stringent chemical oxygen demand (COD) discharge limits. Ozonation is a promising advanced oxidation process for degrading refractory organics; however, its efficacy is governed by multiple interdependent operational parameters that resist empirical optimization. This study employed response surface methodology (RSM) to systematically investigate and optimize the ozonation of real petrochemical wastewater within a three-factor five-level full factorial design (5³ = 125 runs). The individual and combined effects of initial pH (3–11), reaction time (10–50 min), and ozone dosage (5–25 mg/min) on COD removal efficiency were evaluated. A quadratic polynomial model was developed and validated through sequential ANOVA, residual diagnostics, and predicted-versus-actual verification. The model exhibited strong predictive fidelity (R² = 0.8806, adjusted R² = 0.9869, predicted R² = 0.9683, RMSE = 3.58%) and satisfied all statistical assumptions of normality (Shapiro–Wilk p = 0.4565), homoscedasticity, and independence. ANOVA revealed a clear factor hierarchy: the quadratic pH term was statistically dominant (F = 244.98, p < 0.0001), followed by linear reaction time (F = 588.61, p < 0.0001) and linear ozone dosage (F = 74.14, p = 0.0003), whereas all two-factor interactions were statistically negligible (p > 0.05). Response surface topology identified a broad, robust optimum plateau centered at pH ≈ 7, 40–50 min, and 15–20 mg/min ozone dosage, where COD removal consistently exceeded 85% with minimal run-to-run variability. Under strongly acidic (pH 3) or alkaline (pH 11) conditions, efficiency deteriorated markedly due to suppressed hydroxyl radical generation and enhanced scavenging, respectively. These findings demonstrate that near-neutral pH is the primary lever for maximizing COD removal, while reaction time and ozone dosage serve as secondary, independently tunable parameters. The validated RSM model provides a defensible, data-driven foundation for the design and control of full-scale ozonation reactors treating recalcitrant industrial effluents.