Challenge
A major UK water utility needed to plan upgrades across an ageing wastewater network — deciding which pipes to upsize, where to add storage tanks, which pumps to upgrade, and where to install sustainable drainage. Every intervention changes how water flows downstream. The planning problem is combinatorial: thousands of assets, five upgrade types, and two competing objectives — minimise flooding and minimise cost.
The test catchment alone contained 7,886 pipes, 8,040 junctions, 1,829 subcatchments, 47 pumps, and 6 offline storage locations. Under a 30-year design storm, the baseline network generated 52,000m³ of flooding. Engineers manually trialling a handful of scenarios couldn’t come close to searching a space this large.
7,886 pipes. 8,040 junctions. 52,000m³ of flooding under a 30-year storm. Too many combinations for engineers to search by hand.
Approach
The tool uses NSGA-II — a multi-objective evolutionary algorithm — to search the space of possible network upgrades. Each generation creates 50 candidate solutions. Each solution specifies a combination of pipe upsizes, storage tanks, sustainable drainage installations, and pump upgrades. Every candidate runs a full hydraulic simulation, and the results — total flooding volume and total investment cost — feed back to the algorithm to evolve the next generation.
5,425 eligible pipes reduced to 1,359 groups. Grouping pipes by network topology and engineering rules makes the search space tractable without losing hydraulic accuracy.
Before handing the problem to the genetic algorithm, a Design of Experiments stage seeds the search with 50 strategically chosen scenarios. Each sampling strategy explores a different region of the solution space — from extreme max-upgrade configurations to engineering-informed starting points.
The optimiser ran for 49 generations after the initial experiments, evaluating 2,500 total scenarios across a 3-machine workgroup over 54 hours. Each scenario creates a full network configuration, runs a hydraulic simulation under a 30-year design storm, extracts flooding and cost results, and feeds them back to the algorithm. The tool handles dropped connections, non-convergent simulations, and storage cleanup automatically — it runs unattended over a weekend.
Planners explore results through an interactive dashboard where they can filter the Pareto front by cost range or asset type, drill down into individual solutions, and push any promising scenario back into the hydraulic model for detailed engineering review. A heatmap of the most commonly upgraded assets across optimal solutions surfaces no-regret investments — upgrades that appear regardless of which trade-off point is chosen.
Result
52,000m³ of flooding reduced to 22,000m³ — a 57% reduction — at just under £100M investment. 2,500 scenarios evaluated in one weekend of compute, replacing weeks of manual planning.