top of page

Data-Driven Pump Maintenance

Manufacturing facilities have made great strides in collecting and archiving historical process data, but many are failing to fully utilize this information in their maintenance programs. Data-based pump reliability programs can transform traditional maintenance programs to yield game changing results.

Manufacturing facilities have come a long way in terms of measuring and archiving historical process data. However, despite robust data collection, many facilities are failing to fully utilize this information in maintenance programs. From identifying patterns and trends, to predicting equipment failures, the benefits of properly integrating data into a pump reliability program is a game changer. For this reason, it is important to understand the critical variables that contribute to pump failure and the potential pit falls that may prevent an organization from profiting from existing data-based systems.

Traditional pump maintenance methods rely mostly on mechanical techniques to identify root cause of failure and prescribe what is required for equipment overhaul. This has been implemented effectively since early into the industrial revolution. As computing and data collection has progressed, robust data systems have been integrated into manufacturing systems, and have become commonly used for optimizing production rates, ingredient additions, and machine settings. To a lesser degree, some resourceful maintenance programs have capitalized on data driven programs.

It is hard to say why modern pump maintenance procedures have been slow to catch on. Albeit, in the past decade, the industry has recognized the gap and data-based maintenance has been a hot topic. But where to start? It can be overwhelming with all the broadly defined methodologies and analytic systems. It is important to look past the acronyms and understand the physics behind pump failures.

Accurate measurement of pressure, temperature, and flow combined with the knowledge of where, when, and what are essential in understanding upset conditions and process design issues. By monitoring these parameters and identifying potential issues before they lead to equipment failure, it is possible to prevent downtime and prolong the lifespan of the equipment.

Pressure is a fundamental parameter in many industrial processes, and variations in pressure can indicate a problem. However, it is just as critical to understand where each pressure variable is measured. In a pump system, pressure measurements are typically taken at various points, such as the suction and discharge side of the pump, as well as at the inlet and outlet of the system. The pressure measurements at the suction and discharge side of the pump provide information about the performance of the pump, such as flow rate, total head, and efficiency. The pressure measurements at the inlet and outlet of the system provide information about the system as a whole.

Pump flow rate is another critical parameter in predicting, solving, and troubleshooting pump reliability issues. A great example is measuring flow rate per RPM. When the pump was initially installed, the flow rate at a given RPM is at its maximum value. As internal components wear, the flow rate per RPM decreases. Understanding the limitations of your pump and your system speed can allow for a predictive maintenance schedule.

Flow rate is also an effective tool in diagnosing centrifugal pump issues. When there is a deviation from a centrifugal pump’s performance curve it can indicate a suction issue or a permanent mechanical failure. For instance, if the discharge pressure is low and the flow is below its expected flow rate, then it can be an indication that something is wrong with the suction or is mechanically damaged in the impeller or volute. knowing that flow rate is too low for the given pressure should be immediate red flag to pull the pump and inspect it.

IMAGE 1: Damaged Centrifugal Pump Impeller Vanes

Another essential variable is temperature. Temperature is useful for both mechanical and process driven failures. Great examples of mechanical temperature indicators are:

· Temperature of a bearing will increase as it is nearing the end of its life. T

· When a mechanical seal is operated without fluid, a dramatic stuffing box temperature increase, proceeded by seal failure is an indication that more system interlocks are required to avoid future dry run conditions

· If a downstream valve is closed on a centrifugal pump, causing the pump to deadhead, there will be heat generation in the volute that will continue to increase as a result of mechanical frictions and the inability of moving fluid to take away that heat.

When it comes to mechanical temperature analysis the bottom line is that friction creates heat and friction is the main limiting factor of many mechanical systems. For this reason, it is very important to understand and measure mechanical temperatures where possible.

IMAGE 2: Progressive Cavity Pump Failure with Heat Marks from Dry Run Conditions

Temperatures can also explain issues in process systems. For instance:

· Cavitation creates heat that would cause the temperature to increase.

· Contamination of chemicals could cause an exothermic or endothermic reaction increasing or decreasing temperature respectively.

· An improperly sized valve could create a choke flow causing phase change resulting in a temperature increase.

Knowing where an issue occurred can provide insight into the root cause of the problem and help identify potential issues before they lead to equipment failure.

· Where: Location that data is being measured dictates how the data needs to be interpreted. For instance, if a pressure sensor is located upstream of a modulating valve, closest to the discharge of a centrifugal pump, then the pressure value would be inversely proportional to flow rate (higher pressure = lower flow). If the pressure measurement is downstream of a modulating valve and all other variables are held constant, then the pressure will be become directly proportional (more flow at higher pressures). The point is that the location of the same type of variable can display opposite or noncorrelatable results depending on where the information is measured.

· When: Time correlation is a prerequisite to pinpoint when measuring upset conditions. This may seem obvious but can become more nuanced when considering frequency of data collection, compensating for time lags in process control, or linking cause and effect.

· What: Understanding what your process fluid properties are at varying conditions is paramount in understanding application limitations. A great example is understanding vapor pressure of your fluid to identify temperature, pressure, and flow limits required to avoid cavitation. Imagine that a brewery decides to launch a new product that has a higher concentration of ethanol. The use of higher ethanol levels can lead to cavitation at lower temperatures and higher pressures, therefore, new process limits must be established to prevent premature equipment failure..

The type of pump being used in each location is also critical. Revisit the example above regarding where the pressure is measured for instance. If the same condition is true for a positive displacement pump with 100% volumetric displacement, then the pressure upstream or downstream the throttle valve would not affect pump flow. However, power would continue to increase as pressure upstream of the throttle valve increased. Power would be inversely affected on the PD pump, as it would continue to increase until a mechanical or electrical equipment limit is reached, such as a relief valve blows or motor overload and shuts the pump off. Conversely, a centrifugal pump would experience a lower power draw as the pressure increased at this location.

Secondary variables such as viscosity, pH, conductivity, suspended solids, consistency, dissolved solids, particle size, and lubricity are often overlooked but are extremely valuable in diagnosing and solving equipment reliability issues. These variables provide a more subjective understanding of the process and can give insight into the root cause of equipment failures. For example, pH of a liquid, can have a significant impact on equipment reliability. In many applications, a change in pH can lead to corrosion of equipment or the formation of scale, which can reduce the efficiency of the equipment and/or lead to equipment failure. By monitoring pH, it is possible to detect and correct these issues.

Conductivity, suspended solids, dissolved solids, particle size and lubricity are also valuable in various applications. Conductivity, for instance, is a measure of how well a liquid conducts electricity and can be used to detect impurities in the liquid. Suspended solids, dissolved solids, and particle size are all indicators of the condition of the liquid and can be used to detect contaminants that can damage equipment. Lubricity, which is the ability of a liquid to lubricate, is an important parameter in lubrication systems and process lubricated mechanical components.

Vibration analysis is another powerful tool used for diagnosing issues in pumps and other rotating equipment. By measuring the vibration at various locations of a pump it is possible to detect problems such as unbalanced rotor, misaligned shafts, worn bearings, and other issues. By identifying these problems early, vibration analysis can help prevent downtime and prolong the lifespan of the pump.

In addition to identifying specific problems, vibration analysis can also be used to monitor the health of the pump over time. By comparing vibration data from regular intervals, it is possible to detect changes in the pump's vibration levels that may indicate an impending failure.

Vibration analysis is a non-invasive and cost-effective method for diagnosing pump issues. It can be used on all types of pumps, including centrifugal, positive displacement, and reciprocating pumps. With regular monitoring and analysis, vibration analysis can help to ensure that pumps are running at optimal performance, preventing costly downtime and repairs.

Pump maintenance programs can be greatly improved by using process data in combination with sound mechanical techniques. By leveraging process data to predict equipment failures and optimize maintenance schedules, facilities can build a next level maintenance program.

24 views0 comments


bottom of page