Why does a pump’s hydraulic performance differ from its published curve?
This is one of the most common questions I receive about pump performance.
The answer to the question is often simple but can be difficult to sort out given the myriad possible causes. I offer the following discussion points as a guideline.
First, check for the simple things, such as a restricted suction line, and make sure the pump is primed and not air bound. Remember, you cannot vent a running pump. The lighter air will stay in the middle of the casing and the heavier fluid will move to the outside. Does the pump have the correct rotation direction? Depending on the specific speed of the impeller, a pump running backwards will deliver about half the flow and head.
Air entrainment, even at just two to four percent, will air bind a standard pump. Ascertain adequate submergence (distance from liquid surface to intake centerline) to prevent vortexing (vortices lead to air binding) and a sufficient net positive suction head available (NPSHA) margin. For further explanation of NPSHA and submergence difference, please refer to my column in Pumps & Systems, April 2016 (read it here).
In the case of American National Standards Institute (ANSI) pumps, check to see that the pump impeller clearance is properly set. Note one style of ANSI pump sets the impeller clearance to the stuffing box, while the other style sets to the casing. The pump may simply be worn out and the clearances have opened up, which will manifest as if the pump has a smaller impeller. The volute cutwater (aka “tongue”) will wear, resulting in reduced efficiency, and move the published best efficiency point (BEP) to the right on the curve. The cutwater is a pump flow regulation point and works in concert to match the volute flow to the impeller flow.
In the case of a suction lift (where the fluid source is below the impeller centerline) check for air leaks into the suction system. Remember, the fluid will not leak out, but air will leak in. In a perfect scenario, the maximum suction lift at sea level is less than 34 feet. In the real world with friction and vapor pressure, it is less. Problems are more likely at higher altitudes above sea level, with warmer fluids and with vertical lifts approaching 30 feet. Practical lifts are usually under 26 feet.
When troubleshooting malfunctioning pumps, I often find the wrong impeller has been installed during maintenance, or the correct impeller is installed at the wrong diameter.
A centrifugal pump is “dumb” and will operate where the system curve dictates. If a pump is not operating where it should, review the validity of the system curve. The system curve is an absolute summation of the system’s static head, pressure head, velocity head and friction head. The geometry of the system curve is directly related to the flow rate, pipe size, elevation changes and losses due to friction of all the components in the system. Note the system curve is dynamic and will change with tank elevation and system pressure changes. It will also change with valve position, system age, fouling and corrosion. Reliance on a system print to determine the system curve can also lead to errors. It is best to actually walk the system and see what is really there. I often find unauthorized piping branches, valves and elbows.
When a pump is operated at either end of its curve, there are issues like cavitation, separation and recirculation. A manufacturer’s pump performance curve is based on pumping water at ambient temperature (usually around 68 F), a specific gravity of 1.0, a viscosity of less than 30 centipoise (cP) and the assumption that the fluid is not a slurry. When pumping a fluid other than water, the curve may need to be adjusted/corrected for viscosity. The horsepower requirements (power consumption) will be different for different specific gravities.
Sometimes overlooked, the manufacturer’s curve is typically based on a stated speed, and the actual speed in the field is different. This speed differential will happen with induction motors that are not fully loaded or properly specified, since the motor speed will typically increase at lower loads. Speed differentials can also occur due to voltage variations. Look at the motor nameplate or contact the manufacturer for full load speed and the expected percent of slip. The advent of variable speed drives over the last two decades has made the speed difference issues more prevalent. Even 30 to 50 rpm will make a difference in the pump performance.
Electrical frequency (hertz) variations and settings can also cause issues. These variations are usually from on-site power sources such as diesel engines.
Equipment driven by steam turbine can have issues with the governor/throttle settings. In all cases, the best way to determine the actual speed is with a strobe-tachometer. When using the strobe be careful to pick a singular reference point on the rotor that is unique like a key, keyway or a color-coded coupling bolt. Do not use items like nonspecific coupling bolts, as it is very easy to pick up on multiples/dividers of the actual speed.
This article is based on the assumption that the flow measurement is accurate. Unless there is metering and/or transfer of custody situation, most flow measurement is not accurate. This issue is usually not the fault of the flow meter manufacturer. Incorrect flow meters are installed and/or correct meters are installed incorrectly. Empirical measurement of tank levels or containers is another method that will yield satisfactory results for flow rates if conducted properly.
Sometimes an end user will install a pump and note the discharge head does not correctly match where the pump should be on the curve. In that case, some likely issues to look at are:
- The suction side (absolute) pressure has not been taken into consideration. You must know the differential pressure across the pump, not just the discharge pressure.
- The gauges are not accurate due to age, are out of calibration or the gauge range is incorrect. Additionally, the gauge location/position must be compensated for elevation differences above or below the pump centerline. Gauges can only read pressure and do not read velocity head or the elevation difference.
The fluid velocity in the piping is often ignored or misunderstood. If the system is properly designed/operated, the velocity head can usually be dismissed because it is such a small component. In recent years I have noticed an increase in issues where the wrong size pipe is installed (usually an issue on the suction side). As a general rule, the suction pipe should be one size larger than the pump suction, except in cases of self-primers (suction lift conditions), where they may be the same size. The suction pipe size will depend on length of the run and other friction losses. A general rule for flow rates and velocities is that the pipe diameter squared, multiplied by 20 will yield the normal flow rate for an 8 foot per second velocity. If you know the flow rate, you can use the same equation to solve for the proper pipe size.
The pump curve must be corrected for fluid properties like viscosity. Keep in mind that viscosity changes directly with temperature, and many fluids are also reported at the wrong viscosity. In the event of a higher specific gravity, the horsepower will also increase.
Field test the pump by operating at a few different places on the curve. You can accomplish this by manipulating the discharge valve. Carefully operate the pump at shutoff (maximum head at zero flow) just long enough to record a discharge pressure. Operate at a few more points of increasing flow. If the measured performance does not meet curve expectations at shutoff or anywhere else on the curve, then the issue is likely wear, opened clearances, wrong size impeller or incorrect speed. If the performance meets curve expectations at shutoff but drops off from the curve as the flow increases, then the issue is likely a starved suction, insufficient NPSHA, inadequate submergence or air binding.
Most pump manufacturers in North America test and report pump performance in accordance with Hydraulic Institute / ASME specification 14.6, where depending on the selected test grade the specification will offer some small acceptable tolerance between the published curve and actual pump performance.