Skip to content

Comparing two models

A client asks what is different between the 2019 model and the 2024 one, and what it did. A text diff answers neither question. Model data ▸ Compare two models answers both.

Re-exporting a model nobody touched rewrites section order, decimal precision and comment lines, so diff reports thousands of changes to an unchanged network. Then a single line changing FLOW_ROUTING from KINWAVE to DYNWAVE - which changes every number downstream - appears as one unremarkable row among them.

This tool asks a different question of every difference: can it move water?

Counts cosmetic change, never lists it. Reprojecting a model moves every node. Those coordinates are cartographic only: SWMM takes conduit length and subcatchment area from explicit fields, not from geometry, so a moved node changes nothing the solver sees. Burying twelve real changes inside 1,240 coordinate rows would be worse than no diff at all.

Compares numbers numerically. 0.013 and 0.0130 are the same roughness. A writer that rounds differently is not a model change.

Puts options first. A renumbering changes 900 ids and nothing hydraulic; a routing-method change touches one line and invalidates every comparison drawn below it. Options that redefine the solution - flow routing, units, infiltration method, link offsets, force-main equation, ponding, surcharge method - are badged as such.

An id change looks like a deletion plus an addition, so an asset-register renumbering looks like the whole network being replaced. Where a removed and an added link share the same endpoints and the same geometry, and the match is unique, it is reported as a probable rename.

The structural comparison is instant and offline. Run both and compare is opt-in, because two simulations take minutes on a real network. It reports peak link flow, flooded nodes, summed peak flooding rate and continuity error, side by side with the change.

If the models are hydraulically identical and the results still differ, the tool says so in as many words. That difference is in the run, not the model: a different rainfall series, a different scenario or active-topology selection, or a different simulation window. Reading it as a model change sends someone hunting for an edit that does not exist.

Equally, where a major option changed, the tool warns that the differences cannot be attributed to the network edits without re-running one model under the other’s options.

Click any element id to fly the map to it.

  • Export CSV gives the full change list - element, status, field, before, after - for a change register.
  • Comparison report (PDF) prints the whole comparison in the order it should be read: what it means, then the options that redefine the solution, then the network, then what was set aside as presentational. Written for an appendix, and deliberately not shaped like a text diff - a table of elements and values a reviewer can check against their own records.

Requires a Pro plan.