System Description

Southern Water provides drinking and wastewater services to 2.3m customers in the Southeast of England with approximately 0.5Gl/day of production and 0.7Gl/day of recycling. The Tenterden catchment is in Ashford, Kent and has a population of approximately 15,000. The model comprises 53km of sewers, two CSOs, 16 pumping stations, and one wastewater treatment works. For the purpose of model calibration/verification, a 12-week short-term flow survey was undertaken with six rain gauges, 12 flow monitors, two depth monitors, and MCERTS (the Environment Agency’s Monitoring Certification Scheme) data at the treatment works.

Purpose

AMP8, the eighth Asset Management Period for UK water companies in England and Wales, is a five-year regulatory cycle, overseen by Ofwat (the Water Services Regulation Authority in England and Wales). It focuses on improving efficiency, long term sustainability, customer satisfaction, and keeping costs low. A key priority is reducing CSO spills, which demands accurate modelling to guide informed investment decisions. Modelling is essential for informed decision-making, but effective models are only possible with accurate data, which is often imperfect. Southern Water faced the challenge of reducing timeconsuming and iterative verification processes, allowing modellers to focus on tasks like understanding and analysing catchment hydraulics. Additionally, the utility sought to improve data governance, increase consistency, and minimize human error.

Project Scope

The stated objective of the pilot was to use Optimizer to achieve a level of verification similar to that of a hydraulic modeller with improved control over variables and consistency. There were notable obstacles with the survey data quality. In particular, one critical monitor was reporting data that did not align with MCERTS-verified data at the treatment works. With guidance from Southern Water, Optimatics created a solution within the Optimizer platform to perform verification for ICM models across dry weather, storms, and ground infiltration. It was decided that storm verification would be tested for Tenterden, utilising the contributing area change decision across monitor catchments and the curve fitting design criteria. The curve fitting was based on the Nash-Sutcliffe Efficiency (NSE) and provided a way to quantify and improve curve fits at each monitor simultaneously.

Outcomes

Verifying the model in Optimizer provided a similar and slightly improved result to the modeller and was 80% faster than completing it manually. Optimizer was able to solve the problem across all monitors and storms in a single run and produce results within appropriate thresholds. Where 0 is a perfect data match using 1-NSE curve fitting, the initial model had an average of 182. The modeller was able to achieve an average of 28.65 and Optimizer was able to achieve an average of 0.35. In this use-case, Optimizer for verification was able to address the challenges for Southern Water by providing a clear and automated audit trail (what was changed, where, and by how much) through visualisation and communication of data. In turn, this created efficiencies in modeller workflows and improved accuracy of results while delivering quantified improvements in curve fitting.