Why is CFD torque or shaft power wrong in a rotating machine?

Problem: rotating performance does not match expectations

Typical symptoms include torque with the wrong sign, power off by a factor close to 2π or 60, a rotor producing almost zero moment, CFD power that disagrees strongly with pressure ratio, or transient torque that never reaches a repeatable mean.

These symptoms should first be treated as measurement and conservation problems. A correct-looking blade pressure field does not prove that the moment report is defined correctly.

Check the rotation axis and moment origin

Torque is the moment of pressure and viscous forces about a specified axis. If the axis vector or origin is wrong, the reported moment can be wrong even when the local forces are correct.

• Use the same axis direction as the rotating-frame or mesh-motion definition.
• Confirm the origin lies on the physical shaft axis.
• Check whether the report returns a vector component or magnitude.
• Verify the sign convention with a simple known-force direction.
• For sector models, confirm whether the reported torque is sector torque or automatically scaled to the full annulus.

Check RPM, angular speed and shaft-power units

The conversion from revolutions per minute to angular speed is:

Using rpm directly in P = τω introduces a large unit error. Also confirm whether the software input expects rpm, rev/s or rad/s.

Make sure the moment report includes the correct surfaces

Inspect individual surface contributions. If one blade or hub surface has an unexpected sign or magnitude, the problem becomes much easier to localize.

Then cross-check the solved physics and interfaces

Once the report definition is correct, compare mechanical power with the fluid's change in energy. The balance will not be exact in every formulation or station choice, but a large unexplained discrepancy is a strong diagnostic signal.

• Check mass flow through inlet, outlet, bleed, leakage and cooling paths.
• Use consistent total pressure and total temperature stations.
• Verify rotating-frame and rotor-stator interface setup.
• Check periodicity and blade-count scaling.
• Inspect whether poor mesh or interface conservation affects blade loading.
• For compressible machinery, ensure density and energy models are appropriate.

If torque is correct but pressure ratio is not, focus on flow physics and losses. If both are wrong in the same direction, investigate the operating point, boundary conditions and mass flow.

Transient torque needs cycle convergence and time averaging

Sliding-mesh rotor-stator interaction naturally produces periodic torque. An instantaneous value should not be compared directly with a steady performance map. Average over a sufficient number of blade-passing cycles after startup transients have decayed.

1. Plot instantaneous torque and mass flow.

   Identify startup and periodic behavior.

2. Check cycle-to-cycle repeatability.

   Compare mean and amplitude across successive cycles.

3. Reduce timestep.

   Verify the mean and oscillation amplitude are temporally resolved.

4. Check mesh sensitivity.

   Blade loading and tip-gap resolution can materially affect torque.

5. Compare like with like.

   Use the same corrected speed, mass flow and efficiency convention as the test or map.