Build a calibrated, cloud-running, agent-queryable EPANET model of your own distribution system — then turn it into a decision engine that forecasts demand, scores pipe risk, and generates a defensible five-year capital plan.
One system — yours — carried through all sixteen modules. Each course closes with a capstone built live on the final afternoon: not a bolt-on exercise, but the accumulated work of the two days.
Hydraulic first principles, automated model construction from GIS, cloud simulation at scale, an MCP server that exposes EPANET to an AI agent, sensor strategy, data QA/QC, and automated calibration. You finish with a working, trusted model.
A live twin fed by SCADA, weather-driven demand forecasting, optimal sensor placement, ML pipe-break risk scoring, automated design optimization, and a risk-based five-year CIP written back into your asset management system.
You will write code. Comfort with basic Python (variables, loops, functions, reading a script) is expected, along with working familiarity with EPANET or another distribution system hydraulic model. New to modeling? Start with the free Water Modeling Fundamentals course. Self-paced pre-work covering the Python environment, cloud workspace setup, and a refresher on pressure-zone hydraulics is included and must be completed before day one. No machine learning or AI background is assumed. If you lead a utility rather than model for one, the one-day Executive Program is built for you instead.
Before automating anything, be certain what the solver is doing. This module rebuilds the hydraulic fundamentals that every downstream automation depends on, then draws a hard line around where an AI agent may and may not operate.
A model you cannot rebuild by script is a model you cannot automate. Take the INP file apart, then construct one programmatically from your own GIS and billing data.
One run on a laptop is an exercise. Hundreds of runs on demand is infrastructure. Containerize the solver and drive it at scale, without a surprise bill.
The core technical skill of the program. Build a Model Context Protocol server that exposes your hydraulic model to an AI agent as a set of safe, validated tools — the bridge between natural language and a running simulation.
Calibration is only as good as the field data behind it. Design a monitoring program deliberately, and specify instruments that will still be telling the truth in three years.
Every failed calibration this instructor has reviewed failed here first. Build the automated validation pipeline that stands between raw telemetry and any model that informs a decision.
Move from weeks of manual roughness tuning to a repeatable, agent-driven calibration loop — with honest acceptance criteria and a clear statement of what the model is not fit to answer.
The final afternoon of day two. Assemble everything into one working system, built on the data you brought to class, and defend it to the room.
Connect the calibrated model to live telemetry. The step from a model you run to a twin that runs itself is architectural, not algorithmic.
Demand is a weather signal. Build forecasting that drives pumping and storage decisions before the day starts, rather than reacting after it does.
Where the next meter goes is an optimization problem, not a site-visit judgement call. Quantify what an instrument is worth before purchasing it.
Build likelihood and consequence of failure models that survive an engineer’s cross-examination — and learn the specific ways risk models quietly lie.
Let the agent generate and test thousands of design alternatives against competing objectives — then choose from the Pareto front defensibly.
The flagship module, and the seam with the Executive Program. Turn risk scores and hydraulic deficiencies into a sequenced, budgeted, defensible capital program.
A recommendation that never reaches a work order changes nothing. Close the loop between the twin and the field crew, in both directions.
The final afternoon of day two. Wire the full decision chain together on your own data, from live telemetry through to a ranked capital program.
In Advanced Module 6 you produce a risk-based capital plan. The one-day Executive Program teaches your GM, director, or CFO to interrogate and defend that same plan — so the work you do actually gets funded.
Sign up for the Drinking Water practitioner track and leave with a working agentic digital twin of your own distribution system.