Case Study: Field Efficiency Testing for Pumps

My recent series titled “Case Study: Troubleshooting Seal Problems in Cooling Water Pumps,” published in Pumps & Systems (December 2016, and January and February 2017, find them here), described using automated pump field performance data acquisition systems to assess pump components’ reliability (in seals and bearings) via vibrations monitoring and through online efficiency measurements for energy optimization. The series generated significant input and questions from readers about the system’s details. This column describes the process of the automated data acquisition system dubbed “PREMS-2A” (Pumps Reliability and Efficiency/Energy Monitoring System, rev. 2A). It is important to understand pump efficiency. By integrating efficiency data into the repair and upgrade scheduling logistics process, users can generate significant energy savings. For a large pump, even a few points of efficiency degradation can translate to tens of thousands of dollars wasted if proper periodic adjustments are not implemented. Consider, for example, a typical 1,500-horsepower (hp) cooling water pump at a power plant operating non-stop at $0.10 per kilowatt-hour (kWh) cost. This translates to (1,500 x 0.746) x 24 x 365 x 0.1 = $980,244 per year, and each percent of efficiency degradation would mean $98,024 a year. After approximately 10 years in service, it is not uncommon for pumps to easily drop 5 to 10 percent in efficiency due to wear and operating away from the best efficiency point (BEP) because of process changes. Trending and monitoring pump efficiency helps users decide when to schedule efficiency repair upgrades and evaluate the return on investment against the proposed cost of such repairs/upgrades. A typical PREMS system consists of instrumentation (pressure transducers, electric current coil transformer [CT]); ultrasonic flow meter (or alternative 4-20 mA inputs from the existing meter); vibration accelerometers (with full Fast Fourier Transform [FFT] capability) and thermocouples; and data acquisition chassis hardware. The PREMS system also includes a PC to receive data locally or transmit it via modem gateway for a remote display. The display shows the actual pump performance curves (head, power, efficiency versus flow) displayed in real time and plotted against the expected performance, based on original equipment manufacturer (OEM) figures. The difference between actual and expected efficiencies is calculated and displayed numerically and trended as “dollars being burned.” A typical field setup (see Image 1) takes a few hours to install and activate to watch the data streaming live.

The larger the pumps, the greater the impact on energy savings. To illustrate the concept, this article will use a small setup of two pumps in parallel. As the test results show (see Figures 1 and 3), one of the pumps (Pump No. 2) operates more efficiently than the other (Pump No. 1). The “expected” BEP for these pumps is 2 gallons per minute (gpm), 11.2 feet, 0.11 hp and 5.1 percent efficiency. The reason for such an unusually low value of efficiency is not important in this case; the example pumps are small, low-efficiency mag-drive units. The streaming raw data from sensors is first translated into a “pump language,” with gage elevation corrections, pressures-to-head conversion, etc. Figures 1 and 3 show the data transformed by the software into pump performance curves (real vs. OEM). Figure 1 shows the results for the “stronger” No. 2 pump. The data was recorded for about an hour, sampling every two seconds and plotting every 10 seconds (these and other settings are user-adjustable). In Figure 1, notice four distinct areas of operation: Point A is just past the BEP flow; Points B and C are to the left of the BEP; and Point D is close to shutoff. The head “clouds of data” indicate that the actual Pump No. 2 head-capacity performance curve is somewhat higher than the OEM head-capacity curve. The horsepower data also displays above the OEM curve, with the resultant calculated and plotted efficiency curve shaping up below the OEM curve, perhaps a point or so drop in efficiency. Note that the “smudge” of the data cloud of head gets wider as it gets closer to shutoff, reflecting internal hydraulic instabilities resulting in head and flow fluctuations. For a larger pump, these instabilities could be very detrimental for the pumps’ reliability as far as impact on bearings and seals. The PREMS system also monitors vibrations to help shed light on reliability. Three sample vibration channels (two on the pump and one on the motor) are shown on Figure 1. Other traces (individual temperatures, vibrations FFTs, pressures, flow, power, etc.) can be similarly viewed live during pumps operation. Figure 3 shows performance of Pump No. 1 with “clouds” of data connected via curves, indicting the actual entire performance curves. Overall trend is shown, and FFT vibration spectrum details can be viewed via the reference noted at the end of the article. Comparing Figures 1 and 3, the No. 2 pump has lost more head and efficiency as compared to No. 1 pump, and it would likely be the first candidate for the repairs to restore its OEM efficiency. Clearly, for small pumps in this example, the energy “wasted” per year is very small (also dynamically computed and displayed continually at the lower left corner of the PREMS screen), but for more significant power levels, the impact on cost is clearly significant. As a parting quiz for readers, what is depicted in Figure 4? The other figures in the article provide clues. The reward for the best answer is admission to the next Pump School session.