10optimization algorithms per product
~N×faster with N CPU cores in parallel
99objectives supported in one run
5professional desktop applications
The Product Family
One optimization platform. Five specialized tools.
TRNSYS MO, EES MO, and COMSOL MO share one battle-tested optimizer core — the same ten algorithms, the same live Pareto plots, the same CSV reports — each speaking its host solver’s native language. CFD AI adds neural-network surrogates, and Datacenter Cooling is a complete standalone simulator.
TRNSYS MO
Parallel multi-objective optimization for TRNSYS decks. Your existing Type 945/946 projects run unchanged.
EES MO
Optimization for Engineering Equation Solver — nothing to install inside EES, just $Import and $Export.
COMSOL MO
Optimize COMSOL models with zero scripting — parameters and result tables read straight from the .mph.
CFD AI
Physics-informed neural network surrogates trained from COMSOL data — full CFD fields in milliseconds.
Datacenter Cooling
A purpose-built thermal simulator for data centers: air, liquid, and immersion cooling on one canvas.
Why Multi Optimization
Built for real engineering workflows
Truly parallel
Whole generations of designs evaluate simultaneously across your CPU cores — each simulation isolated in its own working copy. A study that took a week runs overnight.
Ten algorithms, one click
NSGA-II, NSGA-III, U-NSGA-III, MOPSO, SPEA2, MOEA/D, AGE-MOEA, AGE-MOEA-II, SMS-EMOA, RVEA — switch algorithms without touching your model, bounds, or reports.
Survives the night shift
Failed simulations are penalized, hung ones are killed after a time limit, and stray solver dialogs are auto-dismissed. Unattended runs finish — every time.
Instant licensing
Monthly, annual, lifetime, or free-trial plans, delivered the moment you check out. Timed licenses start at activation — buying ahead never wastes a day.
Optimization Methods
Every serious algorithm, one dropdown
No algorithm wins every problem. Switch between seventeen of them with one click — your model, bounds, and reports stay exactly as they are.
AI Training
Eight neural-network architectures for CFD surrogates
CFD AI trains physics-informed surrogates with the architecture that fits your model — from classic MLPs and Fourier-feature networks to graph networks that learn on the mesh itself. The best one is suggested automatically.
MLP
Fourier features
DeepONet
Mesh GNN
TRNSYS MO vs. GenOpt: setting a new standard
| TRNSYS MO | GenOpt | |
|---|---|---|
| Objectives | Single or multi-objective — up to 99 simultaneously, with 2D/3D Pareto visualization | Single-objective only |
| Algorithms | 10 modern methods: NSGA-II, NSGA-III, MOPSO, SPEA2, MOEA/D, SMS-EMOA, RVEA… | Classic single-objective methods; no NSGA-II / MOPSO |
| Parallel execution | Whole generations in parallel — near-N× speedup on N cores | Limited parallelism |
| Setup | GUI: point at your deck, tick variables, run — legacy decks convert automatically | Hand-written configuration and template files |
| Results | Live plots, Pareto metrics, suggested best design, CSV reports | Text output |
Video Guides
See it working
Two real TRNSYS studies, start to finish — setup, parallel run, and reading the Pareto front.
Two-objective optimizationEnergy vs. cost on a solar thermal system
Triple-objective optimizationReading a 3D Pareto front and picking a design
Peer-Reviewed
Verified in published research
The methods behind our software are applied and reviewed in peer-reviewed journals.
Try any product free — on your own model
Every product offers a free trial license. Activate it in minutes, run a real optimization on your own model, and decide afterwards. Timed plans only start counting once you activate.
