The technology
A physics engine for coolant network transients, and the AI surrogates trained on it to run in real time.
01
The engine
A transient thermal-hydraulic solver for coolant networks: pipes, pumps, CDUs, manifolds and cold plates, modeled together as a single network rather than in isolation.
Single and two-phase flow. Pump trip, valve slam, load-step and N+1 failure scenarios solved explicitly, not approximated with safety factors. Standards-based: Crane TP-410, ASME, ASHRAE.
02
AI surrogates
Neural networks trained on the physics engine to deliver real-time prediction across a full network, thousands of nodes, without re-solving the underlying transient equations at every timestep.
This approach was presented in the founder's talk, “Neural Network Surrogate Models for Pipeline Systems,” at the SLB User Meeting, Abu Dhabi, 2025.
Roadmap
Where this is going
2027
Network Analyser v1
Steady-state and transient network modeling for coolant loop design review.
2027
Transient Engine
Full pump trip, valve slam, load-step and failure-scenario solver.
2028
AI Surrogate Twin
Real-time neural surrogate for live digital twin monitoring of operating loops.
2029+
Multi-Carrier Platform
Extending the engine to hydrogen, CO2 and thermal energy carrier networks.