Find what fails first — and what it costs.
A configurable likelihood × consequence matrix, scored on your own model — into a defensible renewals plan.
Asset risk turns the register and the hydraulic model into a prioritised capital plan. Likelihood of failure — age, material, condition, deterioration — meets consequence of failure — which mains a break strands, how much service is lost, who's affected, the dollars at stake. The result is a configurable risk matrix and a worst-first register, every score traceable to the factors that drove it. The difference: consequence is read straight from your network model — no second hydraulic licence, no export-and-reconcile.
Likelihood × consequence, kept separate.
Risk is two questions, kept apart so each stays interrogable: how likely is failure, and what does failure cost? Each is a configurable, weighted set of factors — no hard-coded formula, because utilities differ. Combine them per asset and you get a banded risk score and a matrix cell; price the consequence and you get Business Risk Exposure in dollars.
How likely it is to break.
A weighted set of factors — age, material, leak and break history, and condition (increasingly from coded inspection). Deepened over time by a deterioration model that turns condition into a failure-probability curve and a remaining-useful-life estimate. The probabilistic and differentiable variants are held to a research bar — they must beat a plain baseline first, and we make no claim to a self-correcting live twin.
What it costs when it does.
Read from the model itself: hydraulic criticality (what a break strands), customers and critical customers in the impact zone, service severity, repair cost, environmental and regulatory exposure — drawn from the data Foundation. Because the engine is native, consequence updates when the network does — no separate hydraulic product to license and reconcile.
One configurable matrix, every asset placed.
Likelihood and consequence each bin to a band; the pair lands every asset in a cell of a configurable matrix — 5×5 by default, with your axes, weights, and colours. Click a cell to see its assets, or sort the register worst-first by risk band, dollars at risk, or remaining life.
Consequence →
Illustrative. The matrix size, binning, weights, and palette are configured per utility.
Configurable, not hard-coded
Define the factors, weights, bands, and matrix per utility — the score reflects your risk policy, not ours.
Business Risk Exposure, in dollars
Where consequence is priced, risk reads as BRE = probability × cost — a $-at-risk number a board understands, summed across the portfolio.
Explainable by construction
Every score carries its per-factor contributions — what drove this asset's risk — so an engineer can defend it and an auditor can replay it.
Hydraulic criticality, on the map.
The consequence side starts with hydraulics: run the model, fail each pipe in turn, and rank by what the network loses. The pipe-criticality overlay and ranked panel are built and running on dev today — the first risk surface on the register, and the native input the full matrix scores against.
Criticality overlay
Every pipe shaded by the consequence of its failure, on the same map you model in.
Ranked panel
A worst-first list — the assets whose failure costs the network most, ready to act on.
Model-grounded
Computed from the hydraulic run, not a desktop spreadsheet — so it moves when the network does.
From a condition grade to a failure curve.
Likelihood gets sharper when condition becomes a forecast. Deterioration models turn today's grade and history into a probability-of-failure curve over time and a remaining-useful-life estimate — the input that lets the matrix and the renewals plan look forward, not just backward.
Markov condition-state
Coded condition (1–5) transitions through states over time — the established approach for CCTV-graded sewers.
Survival & ML break-rate
Break-rate and survival models, and a machine-learned likelihood from environmental covariates, for water mains.
Validated against the classical oracle
A Bayesian variant runs beside the classical model and must match or beat it before it ships — never a self-correcting twin.
A capital plan, optimised against a budget.
Set a budget by year and an objective — minimise risk, minimise whole-life cost, or hold a condition floor — and the planner prioritises repair, reline, and replacement against it, grouping work by proximity and accounting for what each intervention buys in extended life and avoided risk. Compare what-if plans side by side; see spend against risk reduced.
By dollars-at-risk avoided
Rank candidates by the Business Risk Exposure each renewal removes per dollar spent — within the budget you set.
Not just oldest-first
Replacement cost, deterioration, and intervention effects feed a whole-life view — condition-based beats age-based.
Defect → risk → plan, in place
A coded defect raises likelihood, meets criticality, and reprioritises the plan — in the platform that holds the model and the register, not three tools and a spreadsheet.
Every number stays traceable — which data, which assumption, which run — and the framing is built for ISO 31000 / AS/NZS 31000 and AWWA J100 defensibility. Risk you can take to a board and to a regulator.
Where it stands: the hydraulic-criticality overlay and ranked panel, the configurable LoF × CoF matrix with Business Risk Exposure and per-factor explainability, deterioration / remaining-life models, the budget-constrained renewals optimiser, and a portfolio risk dashboard are all built and running on dev; the probabilistic and differentiable variants are research-gated. HydroDSS is in private build ahead of a 2026 launch — everything here is coming soon, not yet generally available.