Often when troubleshooting a system, many people only consider the “obvious choice.” Common occurrence and experience sometimes blind us to other possibilities and may lead us to spend time and resources solving the wrong problem. In the following example, we will explore a situation where the “obvious choice” for the source of pump noise did not add up.
A backup system using three centrifugal pumps had been added to a waste treatment plant in State College, PA. The primary system consisted of a pair of centrifugal pumps running from a sewage storage tank to the discharge tank, and the backup system drew from the same storage tank and discharged into the same discharge tank. An initial sketch of the pump suction piping is shown in Figure 1.
The main pumps were sized to operate at 600 gpm, and the three backup pumps were sized to operate at 1,200 gpm. With the main pumps running there were no issues, however any time a backup pump was operated they would experience noise and vibration. As a result of the excessive vibration, it was decided to investigate the operation of the backup pumps.
Jeremy Frank, president of KCF Technologies, Inc. in State College, PA, was working with the client to determine the root cause of the issues with the backup pumps. Several wireless sensors were installed close to the backup pump suction nozzle and the system was started up again to analyze what was happening in the pump suction pipeline. Figure 2 shows the vibration sensor data in terms of time history (upper graphs, MS values) and corresponding frequency spectrum (lower graph, Hz) when one of the three backup pumps was started.
Determining the Cause of “Pump Cavitation”
The genesis of this joint project started when KCF was looking for a Senior Capstone Project for Penn State students that would demonstrate how continuous monitoring technology and piping system simulation software can be used together to identify a problem in a pumping system. Initial identification of excessive vibration is an important step during diagnosis. Digging further into the root cause with system simulation allows for a total system view and efficient troubleshooting and optimization of pump performance. Together, vibration instrumentation allows for real time identification of system issues, while simulation allows for a systems approach to issue resolution. After some discussion, the client indicated they were willing and we agreed this would be an excellent project.
The members of the Penn State’s senior capstone design team, along with the members of the client’s operations staff, installed the wireless monitors on the suction side of the three pumps in the system. Once the sensors were installed, the plant staff operated the system to determine the source of the noise and vibration.
As shown in the data in figure 2, high magnitude vibrations occurred within the time history for all backup pumps. The corresponding frequency data had elevated broad band noise typically seen with cavitation, air entrainment, or recirculation. A twenty g’s peak was present, indicating there would be rapid wear of the components if left uncorrected.
Based upon the results of the vibration sensor spectral data, it was believed that culprit was cavitation likely due to inadequate Net Positive Suction Head.
During a review of the vibration data, the following questions surfaced:
- Why did the backup pump operate for ten minutes before vibration occurred?
- Since all pumps shared suction on the same sewage tank, why was it that only the backup pumps exhibited excessive noise and vibration?
- What were the suction and discharge pressure gage readings for the operating backup pumps?
- What was the calculated value of NPSH available for the operating backup pumps?
- What was the NPSH required for the backup pump when operating within the system?
Since these points had not yet been addressed, a definite determination of the cause could not be made.
NPSH calculations were subsequently performed via a system simulation of the pump suction only (Figure 1), and indicated the NPSHa at the pump suction was 39 ft and the NPSHr for the pump was only 11 ft and thus the pump should not have been cavitating. Since the calculations did not match the observations it was time to see what was occurring in the actual system.
Building the Entire Piping System Model
Since the piping system was vital to operations, it was not permissible to shut the system down to run tests to gain an understanding of what was happening. Instead, an accurate total system simulation was developed to understand the true source of the measured vibrations.
First the students developed a model of the total piping system, both suction and discharge sides, to gain insight on how the total system operated. The students created a model with the aid of PIPE-FLO® Professional from Engineered Software, Inc.
Once the Penn State capstone design team had access to the piping simulation software, they were able to easily create the model by inserting design data for all the items within the system. Figure 3 shows the connections for all the items found in the backup piping system. Note that the main pumps are not displayed on the drawing. This is because during the walk down it was determined the primary system was not interconnected with the backup system (which, as an aside, highlights the benefit of visually verifying the model and the system match).
All piping systems regardless of size or function are made of interconnected primary elements. The pump elements add all the hydraulic energy, the process elements are used to make or transport the product or provide the service, and the control elements improve and manage the quality of the product or service within the system. Without an understanding of how these three types of elements work together you cannot fully comprehend how a system operates. In looking at figure 3 you can see the pump elements consist of the three Backup Pumps. The process elements consist of the Sewage Storage Tank and interconnecting pipelines along with the Outlet tank. The control elements consist of the tank levels and on/off control of the pumps to prevent the tank from overflowing on high level and operating the pump dry with low tank level.
The piping system model contains the design parameters for each item included in the system. We will see how the equipment in each element is used in making the model.
For the process equipment, the bottom tank elevation and liquid level in the sewage storage and outlet tanks are used to define fluid energy at the boundaries of the piping system. The pipe size, length, and valve and fitting coefficients are used to describe each pipeline. The Darcy method is used to calculate the head loss in the pipelines.
The pump elements are defined by the pump curve supplied by the manufacturer, created in accordance with multiple ANSI/HI standards. Once this information is entered, the model can determine pump operation across a wide range of conditions.
The control elements in this system consist of the level switches within the tank level indicators. When the level in the sewage storage tank is low the pump stops to prevent it from running dry. With a high level in the tank the pumps are started to maintain the tank within working level.
Analyzing the Piping System Model
Once the model was created, the system was calculated with Backup Pump 101 in operation and the fluid levels in the tanks set to operational levels. The piping system model results indicated that the flow rate through the pump was in excess of 4,736 gpm which was well outside design flow rate of 1,200 gpm.
Looking at the pump’s NPSH, the simulation showed an NPSHa at the pump suction of 38.4 feet of fluid, with a pump NPSHr of 30 ft, initial indications were that the noise and vibration on the pump suction was not in fact due to cavitation. However, the simulation also calculated a system flow rate much greater that the intended design flow of the system.
With NPSHa values greater than NPSHr and a calculated flow rate over three times the expected flow rate of 1,200 gpm, clearly there is more afoot than the commonly experienced pump cavitation. But what could it be? Send thoughts to SimGuyRay@eng-software.com and tune in next month when analyses are performed to walk through troubleshooting via system simulation, then validating the result using test data.