Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 1

Modeling the Cooling Air Flow in an Electric Generator

6th OpenFOAM workshop

Pirooz Moradnia, Hakan Nilsson

PennState University, USA

2011-06-14

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 2

Importance of cooling in generators

• Hydroelectric power generation: ≈ 50% of the electricity generation in Sweden

• Modifications to the existing units→ significant contributions to the total energy production

• Increased power output→ more heat generation

• The two large sources of energy losses in the generators:

- Thermal: electric resistance in the generator coils

- Ventilation: when cooling down the unit

• The stators should be cooled by air flowing through the stator air channels

• Axially cooled generators

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 3

Experimental test rig

• A small generator at Uppsala University,

Sweden

• 4 cooling channel rows

• 108 cooling channels in each row

- Some channels impossible to access

• Stator outer casing with 12 openings

- 5 openings almost completely blocked

• 12 poles

• Rotational speed: 500 rpm

• Inner radius of the stator: 0.365 m

• Outer radius of the stator: 0.437 m

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 4

Modelling in OpenFOAM

• Geometry:

- A periodic 1/12 sector

- Generated with blockMesh

- Less air blockage than the experimental rig

• Boundaries:

- No in/outlet→ No prescribed mass flow

- Extra space for the air recirculation

- Mass flow given by the solution

• Solver

- OpenFOAM − 1.5.x

- MRFSimpleFoam (frozen rotor concept)

- LaunderSharmaKE turbulence model

- Y + < 10

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 5

Complete generator model

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 6

Stator cooling channels

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 7

Study of the flow properties

• Inlet flow pattern

→ Smoke visualization vs. Steady computations

• Inlet velocity profiles

→ Traverse system, total pressure tube vs. Steady computations

• Outlet velocity profiles

→ Total pressure rake vs. Steady computations

• Volume flow

→ In/outlet measurements vs. Steady computations

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 8

Inlet flow pattern

Smoke pen Schematic flow pattern, Experiment Unit vectors of the flow, OpenFOAM

Remarks:

• Radially inward flow near the horizontal stator baffle

• Gradual growth of the axial velocity component farther from the baffle

• Purely axial velocity close to the rotor rings

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 9

Inlet velocity profiles

Location of the inlet, Experiment & OpenFOAM Comparison of the velocity Profiles

Remarks

• Height of the profiles:

- Experiment 29 mm (due to the angle of attack)

- OpenFOAM 38 mm

• Same behaviour in the experiments and OpenFOAM

• Difference in the magnitudes:

- Less air blockage in OpenFOAM

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 10

Outlet velocity profiles

Channel outlets, Experiment Available channels Location of the rake in the channels Outlet velocity profiles

Remarks

• 5 (out of 9) channels per row available for measurements

• Same behaviour in the experiments and OpenFOAM

• Difference in the magnitudes:

- Less air blockage in OpenFOAM

- Non-periodicity in the experimental rig (scattered experimental data)

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 11

Volume flow

• Experiment: ≈ 0.09m3/s

• OpenFOAM: ≈ 0.16m3/s

→≈ 43% difference

• Less air blockage in OpenFOAM

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 12

Conclusions

• Non-periodicity in experiments:

- The largest source of error

• Less air blockage in OpenFOAM:

- More qualitative comparison

• Flow behaviours predicted by OpenFOAM:

- Close to the experimental results

• Better quantitative comparison:

- Unsteady simulations

- Exact geometrical similarity

Pirooz Moradnia, Chalmers / Applied Mechanics / Fluid Dynamics 13

Thank you!

Acknowledgements

The work has been financed by SVC (www.svc.nu):

Swedish Energy Agency, ELFORSK, Svenska Kraftnat, 1

Chalmers, LTU, KTH, UU

SNIC (Swedish National Infrastructure for Computing) and C3SE (Chalmers Centre for Com-

putational Science and Engineering) have provided the computational resources.

1Companies involved: CarlBro, E.ON Vattenkraft Sverige, Fortum Generation, Jamtkraft, Jonkoping Energi, Malarenergi, Skelleftea Kraft, Sollefteaforsens,

Statoil Lubricants, Sweco VBB, Sweco Energuide, SweMin, Tekniska Verken i Linkoping, Vattenfall Research and Development, Vattenfall Vattenkraft, VG Power,

Oresundskraft, Waplans and Andritz Inepar Hydro