A model-based approach to predicting BOD5 in settled sewage

Abstract

Rapid, accurate and reliable measurements of BOD are a very desirable basis for monitoring and controlling wastewater treatment works. Unfortunately, however, producing satisfactory measurements using hardware instrumentation has proved difficult. This paper addresses the issue of BOD estimation using a model-based approach. Two models are developed from historical data using neural network methods. The first model estimates the five-day BOD of the settled sewage using flow, solids, chloride and ammonia data and produces accurate predictions. The second model uses only flow and solids data as inputs yet still produces acceptable (although less accurate) predictions. It was concluded from this that satisfactory estimates of five day BOD can be produced using information that is relatively common to most works. The models are straightforward to apply on-line and offer a method of estimating five day BOD in real-time that is likely to be cheaper, more reliable and easier to maintain than hardware instrumentation.