TRM vs SolidWorks Flow — PCB thermal, honestly.

A category mismatch

Most teams shopping “PCB thermal” land on Flow because their MEs already have it. That’s reasonable on the budget but wrong on the physics. Flow is a CFD tool — its job is to solve the Navier-Stokes equations for air (and other fluids), then derive temperature from heat transfer into and out of the simulated geometry. It is excellent at this. It does not know that a 6 mil trace is carrying 4 A through a 0.5 oz copper sheet that gets less conductive at 85 °C. To use Flow on a PCB you simplify the board, pre-compute dissipations as boundary conditions, and feed it as a heat-source.

TRM goes the other direction. It models the board as the electrical + thermal network it actually is — conductors, vias, planes, components with their R_θJC — and solves the DC IR-drop, current density, copper heating, and temperature distribution as one coupled system. It models convection with heat-transfer coefficients, not with a solved flow field. That’s the deliberate simplification that makes it fast.

What Flow does that TRM doesn’t

• Chassis air, ducts, fans. Selecting a fan, sizing a duct, modeling recirculation across multiple boards, debugging a hot spot you suspect is starved of air. CFD is the only honest answer here.
• Liquid cooling. Cold plates with internal flow, microchannels, two-phase. TRM can take a cold-plate boundary condition; Flow simulates the coolant.
• Outdoor / natural-convection at scale. Modeling a sealed industrial enclosure with sun-load and ambient wind needs CFD.

What TRM does that Flow can’t

• Current density on copper. “Where is the bottleneck on this 60 A rail?” needs an electrical solver. Flow doesn’t have one.
• Coupled DC IR-drop + thermal. Copper resistance rises with temperature; that changes IR-drop, which changes dissipation, which changes temperature. TRM iterates to consistency. Flow can’t.
• Transient board electrical events. Inrush, cold-crank, fault, brown-out — TRM solves these as time-domain electrical scenarios on the board.
• Setup time. First credible result in ~5 minutes. Flow needs a meshing pass, boundary conditions, and an analyst who knows CFD.

The sane dual-tool workflow

For most products with both a PCB and an enclosure, the right setup is:

1. EE owns TRM. Every board spin gets a TRM model: current density, copper heating, transient. Sign-off comes with the report.
2. ME owns Flow. Every chassis spin (and only those) gets a Flow study: airflow paths, fan selection, hot-spot starvation.
3. They exchange: TRM hands Flow the per-component dissipation map; Flow hands TRM the local heat-transfer coefficient or ambient.

That’s how most disciplined hardware programs run. The two solvers are complementary, not competitive.

When you can skip one

Skip Flow when convection is uninteresting: a small board in still air with a generous T_amb margin; an indoor consumer device with a heat-sink and no fan; a sealed module where the conduction path dominates.

Skip TRM when the board is electrically boring: low-current digital, single rail under 5 A, no controlled-impedance differential routing, no point-of-load fan-out. At that point Flow with a simplified heat-source is enough.

Almost no high-performance hardware sits in either of those extremes. If your product matters, you want both.

The right question isn’t “TRM or Flow?” It’s “what physics is causing the failure I’m trying to prevent?” Copper → TRM. Air → Flow. Both → both.

Try it

TRM has a 14-day trial; for a wider tool sweep see TRM vs Icepak vs HyperLynx, and if you’d rather have us run the board, that’s PhySignoff.