Randomised Field Trial for the Performance Diagnostics and Yield Optimisation of Rural Water Treatment Systems in Kenya

Rural water treatment systems in sub-Saharan Africa frequently underperform, but evidence-based, field-validated methodologies for systematic performance diagnostics and optimisation are scarce. This study aimed to develop and test a randomised field trial methodology to diagnose operational failures and quantify the potential yield improvement of small-scale water treatment facilities. We conducted a randomised controlled trial involving 120 community-managed water treatment facilities. The intervention arm received a structured diagnostic protocol and targeted engineering interventions, while the control arm continued with routine operation. Performance was measured via volumetric yield and water quality testing. The treatment effect was estimated using a linear mixed-effects model: $Y_{ij} = \beta_0 + \beta_1 T_i + \gamma X_{ij} + u_j + \epsilon_{ij}$, where $T_i$ is the treatment assignment and $u_j$ is a random intercept for region. The diagnostic protocol identified coagulation-flocculation unit under-dosing as the predominant failure mode (68% of facilities). The intervention increased mean daily treated water yield by 42% (95% CI: 34% to 50%; p<0.001) compared to the control group. A structured, field-based diagnostic approach can significantly enhance the functional performance of rural water treatment infrastructure. We recommend integrating the diagnostic protocol into routine maintenance schedules by local water service providers and for its adoption in national performance benchmarking frameworks. water treatment, randomised controlled trial, performance diagnostics, yield, rural water supply, maintenance This paper provides a novel methodological framework for the randomised field evaluation of water treatment system performance, generating a robust dataset on remediable inefficiencies.