Coal, O&M

Testing Predicts Fan Erosion and Leads to Design Changes

Issue 2 and Volume 112.

By Stephen Mick, Robinson Industries Inc.

Industrial facilities of all kinds frequently must deal with particulate matter impinging the wheel and housing of industrial fans and eroding fan components. This is true even of fans manufactured with wear-resistant materials. However, a fluid flow testing method shows promise in analyzing the complex process of erosion, helping create a wheel design that minimizes wear in challenging power-generation environments.

Research indicates that computational fluid dynamics (CFD), a method of analysis where computers simulate the interaction of fluids and gases with complex surfaces (in this case, fan components), successfully estimates erosion rates and predicts erosion locations on fan components. The analysis gives design engineers accurate data to improve wear-resistance. While paint injection tests offer similar results, CFD can predict additional behaviors in the fluid streams that painting cannot. CFD also saves time because it requires no physical model. This allows designers to experiment with many preliminary designs before settling on one to test further. This often leads to better results through more valuable design information.

A recent project initiated by Robinson Industries tested CFD’s performance in predicting fan erosion. The project was done on behalf of a customer who wanted to look into the possibility of a wheel design with better wear resistant characteristics. The project involved physical model testing to verify performance, fly ash particle analysis to determine particle size distribution and CFD analysis, which was compared to paint injection testing for accuracy. The goal was to offer a new fan wheel design inside the current housing that could better withstand a highly erosive environment.

Researchers designed two models. Design 1 was a double width, double inlet model that included the housing, inlet boxes, shaft, inlet pieces and wheel. Design 2 incorporated the same housing as Design 1 but with a new, revised wheel and inlet pieces (Photo 1). The model size was 36.42 percent of full size, which satisfied the necessary Tip Speed Mach Parameter and Fan Reynolds Number set forth in the Air Movement and Control Association International (AMCA) Publication 802-02, “Industrial Process/Power Generation Fans: Establishing Performance Using Laboratory Models.”

Figure 1. Researchers designed two models to verify CFD’s ability to predict fan erosion. Design 2 incorporated the same housing as Design 1, but with revised wheel and inlet pieces. Shown are the fan wheel, shaft and inlet pieces with the top half of the housing removed in Design 2.
Click here to enlarge image

For the sample material, fly ash particle analysis helped isolate the larger particles (100 microns) for the CFD analysis, because larger particles—with their increased kinetic energy—are stronger contributors to erosion. The particle distribution was determined for the fly ash in the gas stream and the inlet ductwork was also modeled with CFD to explore particle behavior between the dust collector tubes and the fan inlets.

Air performance testing was conducted according to guidelines outlined in AMCA Standard #210-99, “Laboratory Methods of Testing Fans for Rating.” Correcting the air test results to full size confirmed that the model performed similarly to the fan’s design conditions. Testing was also conducted on the Design 2 wheel to verify its performance inside the same housing. It required a slight tip-out and minor cut-off modification to meet the performance requirements.

Improving the CFD Erosion Model

CFD erosion modeling is a recent development that is difficult to apply in rotating fan models. For this project, researchers used erosion equations developed by Wang and Shirazi1, which assumed that the fluid particles were diluted and spherical and that effects from the turbulent flow velocity fluctuations were not considered. The erosion model for this project was based on a semi-empirical relation by Ahlert2:

ER=AV 1.73 f (β) B -0.59
ER is the erosion rate, which is the ratio of removed material mass to the mass of particles in the fluid
β is the impact angle
V is the particle impingement velocity
A=1.95E-5, an empirical constant
B is the Brinell Harness number of the material

f(β) = a β 2 + b β 2 for 0≤ β<15°
X cos 2 (β) sin 2 (β) + Y sin 2 (β) + Z for 15°< β ≤ 90°.

The constants a, b, X, Y, Z have different values for wet and dry surfaces (Table 1).

Click here to enlarge image

Researchers modified the equation for this CFD analysis. The entire wheel was modeled with the same material and a base Brinell Hardness of 100. The -0.59 Brinell exponent was not used to correct for different material types. Instead, results were corrected to correspond with Robinson Industries’ erosion test results, which measured the erosion rates of various material samples. A value in the equation was also corrected and adjusted.

The erosion equation is applied after the CFD program solves the model and determines velocity, particle path and mass transfer data. This mass transfer data is particularly important—it requires a fine grid near the model walls as well as correct turbulent patterns to fully resolve the mass transfer boundary layer near solid boundaries.3

Researchers analyzed the wheels on Designs 1 and 2 at 560,000 CFM and 864,000 CFM conditions with a particle loading of 217 lb/min of 100 micron particles. Table 2 displays the predicted material loss from erosion in inches per year, corrected for the appropriate material on each component.

Click here to enlarge image

Table 3 shows the approximate time it would take to wear through each main wheel component. The approximation lists the time to wear through the thickness of the material for both the maximum amount of wear and the average amount of wear over the entire component, assuming each fan will operate for 33 percent of the time each year. Realistically, replacement should occur between the average and the maximum, as the wheel was modeled using a moving reference frame. Researchers based the correction factor applied for different materials and Brinnel hardness on Robinson Industries’ material testing.

Click here to enlarge image

Engineers should avoid using the absolute numbers in Tables 2 and 3 as guaranteed, concrete numbers for wear rate. Instead, they should use the CFD erosion numbers to compare the two wheel designs. In this study, the comparison showed that the Design 2 analysis predicts better wear on all components except for the housing, shroud and leading edge bar.

The analysis indicated high wear in the same general areas on both wheels: the leading edge bar near the web, the center of the blades, the centerline of the housing, the web and the hub donut. On the housing, CFD results showed increased erosion on Design 2 vs. Design 1, most likely because Design 2’s wheel diameter was larger. The highest area of wear on the housing appeared on the far edge of the scroll. (Because only half the fan is modeled for symmetry, this corresponds to the fan housing center.)

Design 2 showed a significant decrease in erosion on the web and the front side of the fan blades. The main location of blade erosion on the actual fan in the field was in the center of the blade, especially near the tip. Results of the CFD also confirm this, as the wheel is modeled with symmetry.

Figure 2. Paint injection testing confirmed Robinson’s CFD results. Shown is paint accumulation on the web plate around the leading edge of the blade web.
Click here to enlarge image

To test the accuracy of the CFD analysis, researchers used paint injection testing as an established baseline. With injection testing, paint sprayed into one inlet box during simulated operating conditions accumulated in areas that experienced the most wear (Photo 2). In Design 1, paint accumulated on the web donut, on the web itself, and especially on the leading edge bar and the blade, near the center plate. In Design 2, reduced paint accumulation indicated better wear on the web and front side of the blade, confirming the CFD results.

Paint testing also supported the CFD results shown for the leading edge of the blade, which showed increase wear near the center of the blade. The CFD analysis also indicated that Design 2 had a slightly higher erosion rate on the leading edge bar than Design 1.

Opportunities for Improved Design

This research project gives power plant design engineers an example of CFD’s capabilities in predicting fan erosion. The CFD erosion results, while giving a rough estimate of erosion in terms of inches of material loss per year, are also useful for comparing two different designs as well as for matching various fan components with an appropriate wear-resistant material.

Predicting erosion with CFD is a promising development that should become more accurate and powerful as better models are developed to analyze the complex erosion process. CFD provides design engineers with a powerful tool for developing fans with better wear resistance thereby leading to improved efficiencies, lower maintenance and reduced operating costs for power plants.


  1. Wang and Shirazi, “A CFD Based Correlation for Erosion Factor for Long-Radius Elbows and Bends.” Journal of Energy Resources Technology, Vol. 125, March 2003.
  2. “Effects of Particle Impingement Angle and Surface Wetting on Solid Particle Erosion on ANSI 1018 Steel,” M.S. thesis, The University of Tulsa.
  3. “Prediction of Two-Phase Erosion-Corrosion in Bends,” Second International Conference on CFD in the Minerals and Process Industries, December 1999.