Content of PetroWiki is intended for personal use only and to supplement, not replace, engineering judgment. SPE disclaims any and all liability for your use of such content. More information
Centrifugal pumps are the most commonly used kinetic-energy pump. Centrifugal force pushes the liquid outward from the eye of the impeller where it enters the casing. Differential head can be increased by turning the impeller faster, using a larger impeller, or by increasing the number of impellers. The impeller and the fluid being pumped are isolated from the outside by packing or mechanical seals. Shaft radial and thrust bearings restrict the movement of the shaft and reduce the friction of rotation.
Centrifugal pumps are designed with respect to the:
- Number of suctions (single or double)
- Number of impellers (single, double, or multistage)
- Impellers (type, number of vanes, etc.)
Most impellers are arranged from one side only and are called single-suction design. High-flow models use impellers that accept suction from both sides and are called double-suction design.
The efficiency of a centrifugal pump is determined by the impeller. Vanes are designed to meet a given range of flow conditions. Fig. 1 illustrates the basic types of impellers.
Vanes are attached to the central hub, without any form, sidewall, or shroud, and are mounted directly onto a shaft. Open impellers are structurally weak and require higher NPSHR values. They are typically used in small-diameter, inexpensive pumps and pumps handling suspended solids. They are more sensitive to wear than closed impellers, thus their efficiency deteriorates rapidly in erosive service.
Partially open or semiclosed impellers
This type of impeller incorporates a back wall (shroud) that serves to stiffen the vanes and adds mechanical strength. They are used in medium-diameter pumps and with liquids containing small amounts of suspended solids. They offer higher efficiencies and lower NPSHR than open impellers. It is important that a small clearance or gap exists between the impeller vanes and the housing. If the clearance is too large, slippage and recirculation will occur, which in turn results in reduced efficiency and positive heat buildup.
The closed impeller has both a back and front wall for maximum strength. They are used in large pumps with high efficiencies and low NPSHR. They can operate in suspended-solids service without clogging but will exhibit high wear rates. The closed-impeller type is the most widely used type of impeller for centrifugal pumps handling clear liquids. They rely on close-clearance wear rings on the impeller and on the pump housing. The wear rings separate the inlet pressure from the pressure within the pump, reduce axial loads, and help maintain pump efficiency.
Number of impellers
Single stage pumps
The single-stage centrifugal pump, consisting of one impeller, is the most widely used in production operations. They are used in pumping services of low-to-moderate TDHs. The TDH (total dynamic head) is a function of the impeller’s top speed, normally not higher than 700 ft/min. Single-stage pumps can be either single or double suction. The single-stage pump design is widely accepted and has proved to be highly reliable. However, they have higher unbalanced thrust and radial forces at off-design flow rates than multistage designs and have limited TDH capabilities.
The multistage centrifugal pump consists of two or more impellers. They are used in pumping services of moderate-to-high TDHs. Each stage is essentially a separate pump. All the stages are within the same housing and installed on the same shaft. Eight or more stages can be installed on a single horizontal shaft. There is no limit to the number of stages that can be installed on a vertical shaft. Each stage increases the head by approximately the same amount. Multistage pumps can be either single or double suction on the first impeller.
Impeller axial loading
A single-suction, enclosed or semienclosed impeller is inherently subject to continual end thrust. The thrust is directed axially toward the suction because of the low pressures that exist in the impeller eye during pump operation. This thrust is handled with a thrust bearing. The larger the TDH and the larger the impeller-eye diameter, the larger the thrust. Excessive thrust results in bearing and seal damage.
Thrust can be reduced by designing a single-stage impeller for a double suction. In multistage pumps, thrust can be reduced by facing half the impellers in one direction and half in the other. Balancing holes can be used in single-suction, single-stage pumps. The impeller is cored at the rear shroud to allow high-pressure liquid to flow back to the impeller eye.
Impeller radial loading
As the fluid leaves the top of the rotating impeller, it exerts an equal and opposite force on the impeller, shaft, and radial bearings. At the best-efficiency point (BEP), the sum of all radial forces nearly cancels each other out. At capacities below or above the BEP, forces do not cancel out completely because the flow is no longer uniform around the periphery on the impeller. Radial forces can be significant. Heavy-duty radial bearings may be required in lieu of the manufacturer’s standard when pump operation departs significantly from the BEP.
Pump specific speed
Pump specific speed is the speed in revolutions per minute required to produce a flow of 1 gal/min with a TDH of 1 ft, with an impeller similar to the one under consideration but reduced in size. The pump specific speed links the three main components of centrifugal-pump performance characteristics into a single term. It is used to compare two centrifugal pumps that are geometrically similar. Pump specific speed can be calculated from
Ns = pump specific speed
N = pump rotative speed
q = pump capacity
Htd′ = TDH per stage at the BEP.
The pump specific speed is always calculated at the pump’s point of maximum efficiency. The number is used to characterize a pump’s performance as a function of its flowing parameters. Normally, it is desirable to select the impeller with the highest specific speed (smallest diameter). This may be offset by the higher operating cost associated with higher speeds and greater susceptibility to cavitation damage.
Impellers With Low Specific Speeds (500 to 4,000). Radial-flow impellers typically have low specific speeds. Radial-flow impellers are narrow and relatively large in diameter and are designed for high TDHs and low flow capacity. The pumped fluid undergoes a 90° turn from inlet to outlet of the impeller.
Impellers With Median Specific Speeds (4,000 to 10,000). Mixed-flow impellers typically have medium specific speeds and are wider and smaller in diameter than radial-flow impellers. They exhibit medium TDH and medium flow capability. They are typically used in vertical multistage pumps and downhole electrical submersible pumps, which require small diameters.
Impellers With High Specific Speeds (10,000 to 16,000). Axial-flow impellers typically have high specific speeds. In these impellers, the liquid flow direction remains parallel to the axis of the pump shaft. Axial-flow impellers are used for high flow and low TDH applications. They are most commonly used for water irrigation, flood control, pumped storage power-generation projects, and as ship impellers.
Pump performance curves
When a pump manufacturer develops a new pump, the new pump is tested for performance under controlled conditions. The results are plotted to show flow rate vs. head, efficiency, and power consumption. These graphs are known as performance curves. Under similar operating conditions, an installed pump is expected to demonstrate the same performance characteristics as shown on the performance curves. If it does not, this indicates that something is wrong with the system and/or pump. Comparison of actual pump performance with rated performance curves can help determine pump malfunction.
The impeller shape and speed is the primary determinant of pump performance. Fig. 2 illustrates a generalized centrifugal-pump curve. Head, NPSHR, efficiency, horsepower, and brake-horsepower (BHP) requirements vary with flow rate. The TDH is greatest at zero capacity (shutoff head) and then falls off with increasing flow rates. The horsepower curve starts out at some small value at zero flow, increases moderately up to a maximum point, and then tapers off slightly. The pump efficiency curve starts out at zero, increases rapidly as flow increases, levels off at the BEP, and decreases thereafter. The NPSHR is a finite value at zero flow and increases as the square of the increase in flow rate.
It is best to operate the pump at the BEP, but this is not normally feasible. Alternatively, the pump should operate only in the area of the curve closest to the BEP and only in the moderately sloping portion of the head curve. Operating in the flat or steeply sloping portions of the curve results in wasted energy and flow control instability. Pumps that run at or near BEP run smoother and have better run lives. Any time the actual flow drops to less than 50% of the BEP flow, it is wise to consult the manufacturer because shaft deflections may increase dramatically (especially with single-stage overhung-design pumps), which could lead to higher maintenance costs and to failures.
Pumps in parallel
Fig. 3 illustrates the shape the TDH-vs.-capacity curve assumes when identical pumps are operated in parallel and series. Parallel operation occurs where multiple pumps are piped to the same suction and discharge lines. The combined flow rate is the total of the individual pump flows at the TDH. In most cases, the head capacity curves of the parallel pumps are the same, or nearly so. It is not necessary for the curves to be the same as long as each pump operating in parallel can put out the desired TDH.
All centrifugal pumps discharging to an elevated or pressurized vessel and all centrifugal pumps operating in parallel should have check valves in the event of a pump shutdown to keep the pump from spinning backwards. (The danger is a sheared shaft on restart attempt.)
Driver size should be selected so that overloading does not occur at any point across the entire pump curve. Flow orifices or meters should be provided in each pump’s discharge line for verification of flow rates. Suction and discharge piping should be arranged as symmetrically as practical so that all pumps have the same NPSHA.
Series operation is used when a single pump cannot develop the total TDH required. It is also used when a low NPSHR is used to feed a larger pump that requires an NPSHR that cannot be provided from an atmospheric tank or vessel operating at its bubblepoint. In series operation, the combined head is the sum of the individual-pump TDHs at the same flow.
System head curves
The system head curve is a graphical representation of TDH required to be furnished by the pump vs. the flow rate through the piping system. It consists of a constant (static) and an increasing (variable) portion. Fig. 4 illustrates an example of a typical system head curve.
The constant portion represents the static head difference between the suction and the discharge at zero flow and is equal to
The variable portion represents the head required to overcome friction as a result of flow. It varies as the square of the flow and is equal to
pf1 = pressure drop resulting from friction in the suction piping
pf2 = pressure drop resulting from friction in the discharge piping
Pc = discharge flow-control-valve losses.
Regulation of flow rate
It is unusual for a system to require operation at a single fixed flow rate. A pump will deliver only the capacity that corresponds to the intersection of the TDH capacity and system head curves. To vary the capacity, one must change the shape of one or both curves. The head-capacity-curve shape can be changed by altering the pump speed or impeller diameter. The system-head-curve shape can be changed by the use of a backpressure throttling valve (see Backpressure Valves in this page).
The effects of operating at significantly reduced capacity may lead to:
- Operating at much less than the BEP
- Higher energy consumption per unit capacity
- High bearing loads
- Temperature rise
- Internal circulation
These results can be minimized with the use of a variable-speed driver or with the use of several parallel pumps for the total capacity and sequentially shutting down individual units as demand requires.
Higher bearing loads will exist for any flow that departs from the BEP, especially for single-stage, single-suction pumps. This can be anticipated by specifying certain types of heavy-duty and long-life bearings. If the temperature of the pumped fluid rises and the flow rate through the pump decreases, minimum-flow recirculation can be used (see Minimum-Flow Recirculation Valve in this page). The manufacturer generally provides the minimum continuous required flow rate for any pump selection. Operating between the BEP and minimum required flow rate generally avoids all the problems discussed.
The difference between the TDH developed by the pump and the head required by the system head curve represents lost energy. Because most centrifugal pumps are driven by constant-speed electric motors, throttling is the only practical method of regulating capacity. The backpressure valve imposes a variable amount of loss on the system head curve. Closing the valve increases control losses and causes the system head curve to slope up more steeply to intersect the TDH capacity curve at the desired capacity. Opening the valve decreases the control losses and causes the system head curve to slope downward and intersect the TDH capacity curve at a higher capacity. With the valve completely open, the capacity is governed only by the intersection of the two curves.
Minimum flow recirculation valve
The recirculation valve prevents the buildup of excessive amounts of heat within the casing. A minimum-flow recirculation valve should be installed if the pump piping system contains a backpressure valve that could close and result in less than the minimum continuous flow at which the pump can safely operate. A recirculation valve is often used in installations in which the pump piping contains an automatic shutdown discharge valve that could fail in the closed position, or a discharge block valve that can be inadvertently closed. The recirculation valve should be upstream of the first block valve or control valve downstream of the pump. On small pumps, an orifice is usually installed on the recirculation, which continuously recirculates a fixed flow of liquid back to the suction. A control valve costs more but will modulate the recirculation to assure only minimum flow and thus result in less energy loss.
The maximum head that a centrifugal pump can develop is determined by speed, impeller diameter, and number of stages. Thus, to change the head of a pump, one or more of these factors must be changed. Speed can be changed with different gears, belts, or pulleys, or by installing a variable-speed driver. The impeller diameter can be altered for large permanent changes. The number of impellers can be changed by replacing existing impellers with spacers or dummy impellers.
Variable speed control
Most motor-driven centrifugal pumps are operated at constant speed. A direct-current or variable-frequency alternating-current motor control can maintain nearly the same pump efficiency over a larger speed range. Variable-speed control makes it possible to eliminate the backpressure throttling requirements to adjust system head.
Fig. 5 illustrates the head-capacity-curve relationship of a constant-speed and variable-speed pump. The pump is operating at 100% of its capacity, and the TDH is represented by Point 1 on the graph. If it becomes desirable to reduce the capacity to 80% of the rated capacity, the constant-speed-pump operation will move to Point 3. Point 3 requires 110% of the head and 92% of the BHP required at Point 1, and thus, additional backpressure would be required to force the system curve to intersect the pump curve at this point.
A variable-speed driver could, in effect, find a TDH capacity curve that intersects the system curve at Point 2. Point 2 requires only 70% of the head and 73% of the power required at Point 1. Thus, at 80% capacity, the constant-speed pump would operate at Point 3 and the variable-speed pump at Point 2. The potential energy savings is represented by the difference between 92 and 73% of horsepower, or 19%.
The affinity laws are used to predict what effect speed or impeller-diameter changes have on centrifugal-pump performance. The laws are based on dimensional analysis of rotating machines that shows, for dynamically similar conditions, certain dimensionless parameters remain constant. These relationships apply to all types of centrifugal and axial machines.
For a change in pump speed, the following changes in pump performance can be determined:
N1 = old speed
N2 = new speed.
For a change in diameter, the following performance changes can be determined:
D1 = old diameter
D2 = new diameter.
For a change in both diameter and speed, the following changes in pump performance can be determined:
Predictions for speed changes are fairly accurate throughout the range of speed changes. However, predictions for diameter changes tend to be accurate for diameter change of only ± 10% because changing the diameter also changes the relationship of the impeller to the pump casing. Thus, for a 10% increase in either diameter or speed, the flow will increase by 10%, TDH by 21%, and the BHP by 33%.
The efficiency is assumed to be constant in all the previous calculations. Fig. 6 illustrates a graphic example of reduced operating parameters because of speed reducers.
Most centrifugal pumps have a flooded suction. The source is above the pump suction, and atmospheric pressure is sufficient to maintain fluid at the pump inlet at all times. Sometimes the pump must take suction from a source that is below the centerline of the pump. Atmospheric pressure alone will not always keep the suction flooded. Conventional centrifugal pumps are not self-priming. Thus, they are not capable of evacuating vapor from the casing so that fluid from the suction line can replace the vapor. Self-priming pumps are designed so that an adequate fluid volume for repriming is always retained within the pump casing, even if fluid drains back to the source.
A centrifugal pump is a piece of precision machinery that must not be subjected to external strains beyond those it was designed to encounter. It must be installed in the intended position, carefully aligned, and free from piping forces and moments.
Generally, foundation design is not critical. Vibration in a centrifugal pump is minimal unless an engine driver is used. As a general rule of thumb, the foundation should be able to handle three times the weight of the pump, driver, and skid assembly. The manufacturer is the best source for determining the required foundation size.
Poor piping design and installation is a common cause of poor centrifugal-pump performance or failure. Poor piping can result in:
- Performance dropout
- Impeller failure
- Bearing and mechanical seal failures
- Cracked casings
Suction piping is more important than discharge piping.
Fluid source inlet
When the fluid source is above the pump (static head), the source vessel should contain a weir to minimize turbulence, a vortex breaker to eliminate vortexing and vapor entrainment, and a nozzle sized to limit exit velocity to 7 ft/sec or, preferably, less. When the fluid source is below the pump (static lift), the sump, basin, or pit should be designed to provide even velocity distribution in the approach or around the suction inlet and should be sufficiently submerged to prevent vortexing.
Pipe size and elimination of air pockets
Piping should be at least one nominal pipe size larger than the pump suction flange. Velocities should be less than 2 to 3 ft/sec, and the head loss as a result of friction should be less than 1 ft per 100 ft of equivalent piping length. Suction lines should be short and free of all unnecessary turns. For flooded suctions, piping should be continuously sloping downward to the pump suction so that any vapor pockets can migrate back to the source vessel. For static lifts, the piping should be continuously sloping upward with no air pockets (install gate valves in horizontal position). Where air pockets cannot be avoided, the use of automatic vent valves is recommended.
Upstream elbow considerations
When making upstream orientation changes, only long-radius elbows should be used. They should not be connected directly to the pump suction flange, and a minimum of at least two to five pipe diameters of straight pipe should be between the suction flange and the elbow and between successive elbows. This reduces swirl and turbulence before the fluid reaches the pump. Otherwise, separation of the leading edges may occur, with consequent noisy operation and cavitation damage.
Conditions may dictate that permanent strainers be installed in the suction piping. If permanent strainers are not required, temporary cone-type strainers should be installed at least for initial startups. Basket strainers should have at least 150% flow-area screens.
Reducers are required when making a transition from one pipe size to another and in going from the suction-pipe size to the pump flange. Reduction at the pump should be limited to one nominal size change (e.g., 8 to 6 in.). If two or more nominal pipe size reductions are required, it is best to locate any remaining changes several pipe diameters away from the pump inlet. Eccentric reducers should be used, if possible, and should be installed with the flat side up. Concentric reducers should not be used for horizontal suction lines because they could trap vapor that can be pulled into the pump and cause cavitation or vapor lock. Concentric reducers can be used for vertical suction lines and horizontal lines with flooded suction.
Minimum Flow Bypass. The minimum-flow bypass (or “recirculation”) protects the pump from temperature buildup when the pumping rates are low. They should be designed to handle the pump’s minimum flow capacity at minimum discharge pressure with a line restrictor to adjust flow. Small pumps are usually controlled by an orifice or choke tube. For large pumps in which a continuous bypass would consume excessive power, a control valve actuated (opened) by low flow is used.
Check valves are essential to minimize backflow, which can damage the pump. Selection should take into account the effect of water hammer. Water hammer is the transient change in static line pressure as a result of a sudden change in flow. Items that can start the sudden change in flow include the starting or stopping of a pump or the opening or closing of a check valve.
Slow-closing check valves are acceptable on systems with a single pump and long lengths of pipe. Fast-closing check valves are required with multiple pumps operating in parallel and at high heads. As a general guideline, lift (“swing”) check valves are slow unless they are spring loaded. Tilting-disk check valves are fast closing but are more expensive and have a higher pressure drop than swing check valves. When fast-reacting check valves are required, pressure-drop considerations should be secondary.
|Ns||=||pump specific speed|
|N||=||pump rotative speed|
|Htd||=||TDH per stage at the BEP|
|pf1||=||pressure drop resulting from friction in the suction piping|
|pf2||=||pressure drop resulting from friction in the discharge piping|
|Pc||=||discharge flow-control-valve losses|
Use this section for citation of items referenced in the text to show your sources. [The sources should be available to the reader, i.e., not an internal company document.]
Noteworthy papers in OnePetro
Use this section to list papers in OnePetro that a reader who wants to learn more should definitely read
Use this section to provide links to relevant material on websites other than PetroWiki and OnePetro