<![CDATA[Ensuring the consistent quality of drinking water remains a major challenge in Indonesia, particularly due to natural variability and operational limitations in regional water companies (PDAMs). Statistical quality control methods such as the Multivariate Exponentially Weighted Moving Average (MEWMA) chart, are widely applied for monitoring; however, their assumption of independent and identically distributed observations reduces their effectiveness when applied to autocorrelated time-series data. This study proposes a Vector Autoregressive (VAR)-based MEWMA control chart for monitoring water quality parameters, turbidity, and residual chlorine at PDAM Tirta. Daily observations from 2023 (n = 365) were analyzed. The VAR(3) model was selected using the Akaike Information Criterion (AIC), and residuals were validated to be free from autocorrelation. These residuals were then incorporated into the MEWMA framework with a smoothing parameter λ = 0.03. A comparative analysis was conducted between the standard MEWMA and the VAR-based MEWMA through Monte Carlo simulations (5,000 replications) across three shift scenarios. Results showed that both methods achieved comparable ARL₀ values (≈3), confirming stability under in-control conditions. However, the VAR-based MEWMA consistently demonstrated lower ARL₁ values in detecting small shifts, especially in turbidity, with improvements of up to 22% compared to the standard MEWMA. These findings highlight the VAR-based MEWMA as a more sensitive and reliable monitoring tool, offering water utilities an early-warning system that enables timely corrective actions and ensures compliance with drinking water quality standards.]]>
