Open Access
Kim, Geon-Seok
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
October 04, 2007
Committee Members:
  • Gerald Clyde Lauchle, Committee Chair
  • Timothy A Brungart, Committee Chair
  • Dean Capone, Committee Member
  • Gary Hugo Koopmann, Committee Member
  • cooling fan noise
  • hot-wire
  • cross-correlation of incoming turbulence
This effort focuses on measuring and predicting the discrete-frequency and broadband noise radiated by diesel engine cooling fans. Unsteady forces developed by the interaction of the fan blade with inlet flow are the dominant source for both discrete-frequency and broadband noise of the subject propeller fan. In many cases, a primary source of discrepancy between fan noise prediction and measurement is due to incomplete description of the fan inflow. Particularly, in such engine cooling systems where space is very limited, it would be very difficult, if not, impossible to measure the fan inflow velocity field using the conventional, stationary hot-wire method. Instead, the fan inflow was measured with two-component x-type hot-film probes attached very close to the leading edge of a rotating blade. One of the advantages of the blade-mounted-probe measurement technique is that it measures velocities relative to the rotating probe, which enables the acquired data to be applied directly in many aerodynamic theories that have been developed for the airfoil fixed-coordinate system. However, the velocity time data measured by this technique contains the spatially non-uniform mean velocity field along with the temporal fluctuations. A phase-locked averaging technique was successfully employed to decompose the velocity data into time-invariant flow distortions and fluctuations due to turbulence. The angles of attack of the fan blades, obtained from inlet flow measurements, indicate that the blades are stalled. The fan's radiated noise was measured without contamination from the engine noise by driving the fan with an electric motor. The motor operated at a constant speed while a pair of speed controllable pulleys controlled the fan speed. Narrowband and 1/3-octave band sound power of the cooling fan was measured by using the comparison method with a reference sound source in a reverberant room. The spatially non-uniform mean velocity field was used in axial-flow fan noise theory to predict the discrete-frequency noise at the blade passing frequency (BPF) and harmonics. The unsteady lift was predicted by considering transverse and longitudinal velocity fluctuations. The influences due to an upstream finger guard were also investigated. The radiated sound power spectra that were measured for the fan are shown to be in excellent agreement with those predicted. The agreement between prediction and measurement is only fair at the fundamental BPF tone. Further experimental investigations revealed that the interaction noise between the fan blades and a shroud surrounding the fan might be the dominant source for the radiation at the first harmonic. The space-time correlation functions of the inflow velocity fluctuations were measured and utilized in stochastic lifting surface theory to calculate the unsteady blade lift and resulting broadband fan noise. The integral length scale of the inlet flow was found to be much smaller than the blade-to-blade separate distance of the fan. Therefore, contributions from blade-to-blade correlations of the various elements on different blades were found to be negligible and hence ignored; only the correlations between the strip elements on a given blade were considered. The cross-correlations measured between elements separated by more than the integral length scale were also found to be negligibly small. The predicted broadband sound power spectra agree well with those measured for frequencies greater than 400 Hz. There are deviations between the predictions and measurements at lower frequencies. These are likely due to fan blade stall, and possibly, anomalies in the noise measurement environment. In order to reduce the sound radiation at the blade rate tones, the baseline fan was replaced with a skewed fan. The backward skew angle of 30¡Æ was found to effectively reduce the 2nd and higher harmonics of the blade rate tone. The interaction of the shroud opening and the blade tips dominates the sound level at the fundamental tone. This tone was successfully reduced by incorporating a serrated shroud opening. Overall, a 2.8 dB sound power level reduction was achieved by applying the skewed fan and the serrated shroud opening simultaneously. Almost all blade rate tone levels were reduced without adversely affecting the flow performance of the system.