Numerical Investigations on the Noise Characteristics of a Radial - - PowerPoint PPT Presentation

Numerical Investigations on the Noise Characteristics of a Radial Fan with Forward Curved Blades Manoochehr Darvish Stefan Frank

darvish@htw-berlin.de stefan.frank@htw-berlin.de

Contents

• Introduction to FC Blade Fans
• Review of some previous results
• Acoustic measurements
• CFD/CAA simulations (STAR-CCM+)
• CAA simulations (ACTRAN)
• Conclusions
• Next steps

2

Radial Fan with FC Blades

 Forward Curved Blades  Operation at low speeds  Low level of noise  Used mainly in small and medium sizes ─ Relatively low efficiency ─ Scroll Housing is required Application: ─ Automotive industry ─ HVAC applications

─ Compactness vs. Efficiency

3

Cordier Diagram

4

The best combination : SST K-Omega + Polyhedral

5

Flow rate= 450 m³/h SST K-Omega Steady Unsteady Experiment (PIV)

Anechoic Room with Measuring Duct acc. DIN EN ISO 5136

6

Turbulence Screen (Sampling Tube) microphone applicable up to v= 40 m/s

STAR-CCM+ Simulation

• Simulations start from converged steady solutions
• RANS (SST K-Omega),DES (2nd order SST K-Omega ; Hybrid

Bounded Central Differencing) ,LES

• Non-reflecting Inlet/Outlet boundaries (Free-Stream)
• Segregated solver
• Compressible flow
• 2nd order Temporal Discretization
• Rotational speed :1000rpm
• Number of blades :38
• Time Step becomes gradually smaller :

1°,0.75°,0.5°,0.25°rotation of the fan-wheel

7

 BPF = 16.667 * 38 ~ 633 Hz

STAR-CCM+ CFD/CAA Simulation

• Monitoring the maximum pressure at the inlet & in the fan

discharge

• FW-H Receiver is placed at the outlet
• The whole model is assigned to FW-H Surface (Noise Source)

8

• Exporting the results (Density and Velocity) for CAA

simulations using ACTRAN

9

4.5M Cells 16.5M Cells 7 days/rev 26.5M Cells 10 days/rev 12.5M Cells 5 days/rev

• 20 cells per wavelength
• Polyhedral mesh with 4.5M cells  not suitable for CAA
• 12.5M cells model is a solution adapted mesh

1 day/rev

10

Low-Re Wall Resolution

11

Different Turbulence Models  Different Flow Features LES URANS RANS DES

12

Proudman Acoustic Power dB (steady RANS)

13

Proudman Acoustic Power dB Curle Surface Acoustic Power dB

Analyzing different designs using RANS simulations

surface based dipole sources volume based quadrupole sources

14

DES sometimes has nonmonotonic response to grid refinements

65 dB 75 dB

Pressure Monitor FW-H

15

65 dB 75 dB

FW-H Pressure Monitor

CAA simulations with ACTRAN

Lighthill Surface brings the acoustic fluctuations flowing from Rotor to the Stator

4-6 Elements/Wave Length Tetrahedral mesh (STAR-CCM+)

Modal Surface Boundary Condition makes the channel infinite

16

Transient CFD solution (.CCM)ACTRAN

17

ACTRAN results

DES_16.5M  inaccurate results DES_26.5M  underprediction in the tonal noise

Contribution of Lighthill Surface (LS) and Lighthill Volume (LV)

18

URANS becomes inaccurate in lower range of the frequencies 2 hours (ICFD)+ 1 day (ACTRAN solver) on a single core (2.3 Ghz)

ACTRAN results

Close correlation between STAR-CCM+ and ACTRAN 633 Hz 100 Hz STAR-CCM+ ACTRAN

19

Transient Surface data helps to gain insights into the characteristics that cannot easily be obtained from charts & diagrams f=633 Hz

Surface Pressure Level (SPL) obtained from DES simulation

20

Conclusions

• Good agreement between experimental and

simulation results

• URANS and DES are both suitable for the

aeroacoustic simulations using STAR-CCM+

• Close correlation between STAR-CCM+ and ACTRAN
• Using ACTRAN, DES input results provide more

accurate aeroacoustic results (comparing URANS)

• All the presented CFD (URANS,DES and LES) and

CAA methods (i.e. FW-H, Pressure Monitors and ACTRAN) are going to be used and validated for the

• ther operating points of the fan

21

What’s next:

22

• Improving current results
• Design modifications
• Noise determination at the inlet side

What’s next:

23

102M Cells 26.5M Cells LES using a finer grid

Thank you for your attention!