Cutting through the mysteries of impeller trimming

Centrifugal pumps are a common and integral part of many industrial applications related to fluid flow of any kind. They require significant amounts of energy to operate. Unfortunately, many pumps are oversized for their applications. When pumps are not properly sized this can lead to operational difficulties that must be addressed. A pragmatic solution to this problem may be to reduce the outer diameter of the pump impeller to better meet flow requirements and avoid costly operating procedures such as throttling or expensive replacements. Changing the impeller diameter will affect the performance of the pump, so special consideration should be paid to such modifications. Even so, such changes could provide significant energy and cost savings.

To properly begin, the root problem should be addressed. Why are pumps oversized in the first place? The answer, of course, lies in how the pumps are selected and sized. With several parties involved with this task, there is always room for misunderstanding or even poor quality of analysis. To compensate, usually the vendor, the engineer(s), and sometimes even the owners will “pad” the selection with a larger pump. Doing so is thought to insure meeting the flow requirements of the pumped system in question. This could cover up any miscalculations and misunderstandings in the system’s resistances. However, there are significant drawbacks to this custom. While the practice seems obvious enough to correct, it is common knowledge that this happens and is even openly acknowledged in industry, yet it continues. While the ultimate solution is to change the paradigm of padding the selection, that information cannot change the thousands of already installed pumps in operation today.  

What does it mean when a pump is oversized? As concisely as possible, the pump can produce more head and flow than is necessary. While this may sound like a good problem to have, some applications require strict tolerances in regards to flow demand. Additionally, some systems may not be able to handle the higher pressures the pump can produce, presenting a real safety issue. In either case, the system must then be operated or modified in such a way that these are no longer concerns.

Some of the possible solutions are as follows. First, the situation could be avoided entirely with a properly sized pump. Barring that, the pump could then be replaced. This, of course, requires another selection process - removal of the existing equipment and installation of the new pump and all other associated tasks. If the costs for this avenue are too high, then a variable frequency drive could be installed on the pump motor and utilized to reduce the pump’s speed. While this article will not go into significant detail on this option, attention should be paid to whether reducing the speed will still be able to meet the requirements of the system. The next possible step could be to throttle the system; more on this below. Finally, and surely the reason you are reading this now, trimming the impeller. Each of these has their own benefits and drawbacks, so the cost as how easily each may be implemented should be determined when choosing an option.

Throttling the system may perhaps be the most common and first solution thought of by engineers when beginning to address the problems of having an oversized pump. While simple and usually readily achievable, this can potentially be very costly. Here is why. Below are two pressure profiles of similar systems with the same piping requiring 500 gallons per minute of water. One has an oversized pump and one has a properly sized pump.

   Figure 1 The pressure profile for a 1,000-foot pipeline, supplying water uphill 200 feet in elevation at 500 gpm

Figure 1 The pressure profile for a 1,000-foot pipeline, supplying water uphill 200 feet in elevation at 500 gpm

 Figure 2 The same system with an additional control valve and oversized pump. Area in red is wasted energy, removed by the control valve

Figure 2 The same system with an additional control valve and oversized pump. Area in red is wasted energy, removed by the control valve

Notice that the oversized example has a greater head rise. The performance curve of this pump has approximately twice the head available for any given flow compared to the properly sized pump. It has been throttled down to 500 gallons per minute with a control valve, which must remove close to 173 pounds per square inch of pressure to insure the appropriate flowrate is being supplied to the system.

 Figure 3 Operating characteristics of the throttling control valve

Figure 3 Operating characteristics of the throttling control valve

While the wasted energy should be apparent enough, assuming electricity costs at $0.11 per kilowatt-hour for a five-year cycle, the throttled system costs just over twice what the properly sized pump cost to operate, $353,000 compared to $171,000. This cost assumes only the pump energy and not any of the maintenance and other costs associated with the control valve in place or the purchase prices of the pumps themselves. Note that while hyperbole is effective in illustrating a concept, this is a real phenomenon that operational systems experience in industry today. Throttling is effective in obtaining desired operating conditions but does very little to protect against wasted energy and money.

This leads the discussion directly to trimming the impeller. Now, before continuing too far, it should be noted that there are multiple ways to modify the original impeller to change the performance of the pump. Ranging from reducing the thickness of the vanes to modifying a shroud, there are many ways impellers may be modified. Reducing the outer diameter of the entire impeller will be the specific modification referred to as trimming the impeller in this article.

It has already been stated that trimming an impeller changes the pump’s performance, but why? To answer this, the relationship to the casing wall and the impeller’s outer edge should be understood. When the vanes of the impeller impart momentum onto the fluid by increasing its velocity, this velocity is almost immediately converted to potential energy in the form of pressure as it is stopped against the casing wall. Understandably so, the distance from the edge of the impeller to the casing is important in regards to design. So, when this distance is increased, two things occur simultaneously. First, the local velocity at the impeller’s new edge is lower than before because this point on the impeller has a lower angular velocity. There is also a greater distance for the fluid to travel before converting the energy to pressure at the casing. This is important and covered later. In effect, the efficiency of the impeller drops because the same amount of work produces less head rise for the fluid. Even with this drop of efficiency, this solution may be the best option depending on the significance of the change. By reducing the impeller diameter, the pump now has a different performance with less capacity and head rise.

Trimming the impeller is only beneficial if the new characteristics of the pump satisfy the system requirements from before. To determine this information, the pump’s performance can be predicted in one of two ways. First would be using vendor supplied data. Often, Vendors will test pumps with varying impeller diameters and will report these curves on data sheets. New impeller size performances can be determined by interpolating between these published curves. The second option is to use the affinity laws if no vendor data is available. Although commonly used in terms of speed, the affinity laws can also be applied with a ratio of impeller diameters instead. Using AFT Fathom, both methods can be used to model the change in the pump’s performance. In fact, it is strongly encouraged that prior to modifying the impeller, these calculations be performed to help guarantee the desired outcome of the modification. Failure to do so could result in unnecessary down time and an unsolved problem.

Consider a system like those described above. A lower reservoir full of water must supply a reservoir 200 feet higher in elevation. The piping is 1,000 feet of ANSI standard steel at STD scheduling with the pump suction being only 10 feet long. The pump has been sized for a necessary head rise of 370 feet at 500 gallons per minute. However, due to padding, a larger pump utilizing an 8.5-inch impeller was installed. For the sake of time, it is known that with this pump a 7-inch impeller is most appropriate and will satisfy the flow and head requirements.

 Figure 4 System layout

Figure 4 System layout

Beginning with vendor data, assume there are curves for both the 8.5-inch and a 6-inch impeller. The curves are shown below. Again, because it is known that close to a 7-inch impeller will be best, a few iterations may be skipped to go straight to the validation. To use the first method in AFT Fathom, the supplied curves must be entered as multiple configurations in the Pump Property window. This is accessed by clicking the “Enter Curve Data…” button and clicking the “Multiple” radio button in the top left of the Pump Configuration window. These configurations must be saved after entry.

 Figure 5 Vendor data for varying diameters

Figure 5 Vendor data for varying diameters

Once saved, the multiple choices become available in the pump model tab, allowing users to access the stored configurations. With the configurations entered the Actual Impeller feature becomes available. Entering 7 for this value modifies the input curve, in this case the information from the 8.5-inch impeller. If the rest of the input information necessary is present, then the model can be run and the pump’s performance can thus be determined.

 Figure 6 Multiple configurations can be entered and stored for future use

Figure 6 Multiple configurations can be entered and stored for future use

 Figure 7 The pump model section in the Pump Property Window with modified performance curves

Figure 7 The pump model section in the Pump Property Window with modified performance curves

Running the model, both the hydraulic and monetary benefit can easily be seen in the Output Results window of AFT Fathom. Assuming a going rate of 7.5 cents per kilowatt-hour for ten years of operation, and keeping in mind the requirements of 500 gallons per minute available around 370 feet of head, the impact of the impeller reduction can be seen below. The 7-inch impeller provides 482 gpm at 367.6 feet of head and costs about half as much in energy compared to the 8.5-inch impeller.

 Figure 8 Results using the Actual Impeller feature in AFT Fathom

Figure 8 Results using the Actual Impeller feature in AFT Fathom

This leads to the second method; the affinity laws. If multiple curves are not available, then only the performance curve on hand can be entered and the “Ratio as Percent” radio button may be entered. Using the ratio of 7 to 8.5, 82.3% is used to characterize the change in how the pump operates.

 Figure 9 Using the affinity laws to predict performance

Figure 9 Using the affinity laws to predict performance

Running the model shows that the results are very close to the Actual Impeller feature results, allowing for rounding. RA Mueller has a great depiction of the Affinity Laws used both for speed and impeller diameter here: http://www.ramueller.com/pump-handbook/affinity-laws/index.html.

 Figure 10 Ratio as Percent results

Figure 10 Ratio as Percent results

Before you remove your impeller and get it ready to send to the shop, it should be noted that there is a point where too much of the impeller can be removed. Recall that the pressure is built along the casing wall when the velocity approaches zero. If the distance between the edge of the impeller and the casing wall becomes too great, recirculation increases. The fluid would, at this point, leave the vanes of the impeller but fail to contact the casing as fast, and thus fail to create as much pressure. In researching information for this article, it was very difficult to find a hard standard for this distance. However, excellent heuristics are stated on a Hydraulic Institute Tip Sheet, #7, here: https://energy.gov/sites/prod/files/2014/05/f16/trim_replace_impellers7.pdf. In the essay, it is stated that if vendor data is available, one should not reduce the diameter of the impeller smaller than the smallest size reported. In the above example, this would be the 6-inch impeller. However, if multiple impeller curves are not available and one is using the affinity laws to determine how the modification will impact the system, then the impeller reduction should be limited to no more than 75% of the original impeller diameter. Beyond this point, the validity of the affinity laws will break down in terms of diameter.

While these provide excellent guidelines, it is hard to say whether recirculation would occur at larger diameters. Note that AFT Fathom allows users to specify reductions much greater than this and will assume the affinity laws continue to apply and an actual impeller size may be specified below the smallest configuration saved. Users should exercise due diligence to insure the results will be congruent with best practices.

Also available on the mentioned tip sheet is an example highlighting the monetary benefits of a proposed impeller modification. Additional practice can be had by following this example as well.

industry is currently plagued with oversized pumps. The industrial sector is responsible for close to 22% of the energy consumption in the US (https://www.eia.gov/energyexplained/?page=us_energy_home) and of that, anywhere from 10% to 60% of consumption comes from rotating equipment energy requirements (https://www.eia.gov/todayinenergy/detail.php?id=13431), special attention should be paid to the cost of operation and pump selection. Customary practices lead towards oversized pumps and costly solutions. Rotating equipment is not limited to categories of pumps but may include the likes of compressors. The potential to reduce energy consumption by 2-12% is nothing to ignore. Understanding that trimming the impeller can help, care should also be taken and calculations performed to make sure such modifications will be beneficial. AFT Fathom makes this part of the change quick and simple.

While a change in paradigm may present a cure, one of the best treatments that could save significant amounts of energy costs would be to reduce the diameter of the pump impeller. The cost to send the impeller to your trusted machinist and having them trim it on their lathe, will likely cost less than most of the other possible solutions. Their work could save your system huge sums of money- so do me a favor and thank your machinist for me!