Sizing, selecting pumps and circulators

By making correct pump and piping selections, a hydronic system can operate at or near peak efficiency, avoiding cavitation and vibration. Here’s a look at relevant codes and best practices.

By Randy Schrecengost, PE, CEM, Stanley Consultants, Austin, Texas October 15, 2014
Learning objectives:
  • Review codes and standards that guide design and specification of pumps and piping systems.
  • Become familiar with different types of pumps and their function as related to an application. 
  • Understand basic information for selecting pumps to meet a distribution loop’s requirements.
  • Understand key equipment and its integration for energy efficiency in HVAC systems. 

There are a number of pumps manufactured today that can be used for different applications. The most typical application is circulating and distributing chilled or hot water for a variety of different building or facility load requirements. The process of selecting and sizing pumps and circulators includes several steps that a designer needs to take to complete his or her task for any given installation. The concepts to consider for a particular design task and the designer’s experience level will determine the complexity of the overall process. Sizing and selecting pumps is not that difficult once the experience has been gained. At the very least, the designer must:
  • Define and understand the system application, and perform a hydraulic or fluid system analysis
  • Determine the basic pump (or circulator) and driver type for the application
  • Determine the pump size and its design operating point
  • Decide on pump design considerations to maximize the system’s energy efficiency.

Designers must understand many basic concepts regarding pumps and hydronic systems, but for this article the fluid systems discussed will be water only. 

Codes, standards, and guides
The design of a hydronic system includes multiple components related to the application that require a review of codes, standards, and/or regulations necessary to complete the design and avoid conflicts that will cost time and money to resolve. Local, state, and federal codes and/or regulations dictate requirements that may affect the design, but there are a few related codes targeting pumps which refer to the specific application. 
Organizations such as ASHRAE, the American Society of Mechanical Engineers (ASME) Boiler and Pressure Vessel Code, the American Petroleum Institute (API Std. 610), and UL  provide several code references or have standards to review for systems, equipment, and testing requirements. ASHRAE Handbook HVAC Systems and Equipment – 2012, Chapter 44 provides information on pumps. ASME PTC 8.2-1990 for Centrifugal Pumps provides information on pump testing and guidance for the application of the code related to centrifugal pumps. UL provides references to the use of pumps related to the systems to which they are applied such as UL 343 – Standard for Pumps for Oil-Burning Appliances, or UL 778. Standard for Motor-Operated Water Pumps. In addition, the Hydraulic Institute (HI) has a vast library of standards related to pumps that the designer will find very helpful, such as Rotodynamic (Centrifugal) Pump Applications. 
A good technical resource for most facility engineers is ASHRAE. ASHRAE has numerous technical sources of information including a series of four handbooks that are updated every 4 years. Each handbook has an entire chapter dedicated to listing “Selected Codes and Standards Published by Various Societies and Associations” relevant to the topics covered within the handbooks. Additionally, ASHRAE Standard 90.1-2013: Energy Standard for Buildings Except Low-Rise Residential Buildings is the reference standard for energy efficiency. This standard illustrates minimum efficiency requirements for building envelope, HVAC, power, lighting, and other equipment. Chapter 6, HVAC, is where designers will find minimum energy efficiency requirements for HVAC system construction with listings for component items such as: water- and/or air-cooled chillers, piping system design flow rates, pumps, insulation, and controls. 
Define and understand the system application
A pump is required in a system composed of piping, fittings, heat exchangers, and other equipment through which a fluid needs to be delivered or transferred. This delivery may be for a long distance, to higher elevations, or circulated within a pressurized loop to assist in a process and perform work. The pumping requirement is immediately affected by the fluid’s nature and its properties of viscosity, density, and specific gravity. The discussion of these topics, although important to the designer, is beyond the scope of this article; however, these terms, along with vapor pressure and the effects of temperatures on a fluid, should be considered in many applications. Additional “pumping” terms will be introduced as we progress and some basic definitions will be provided.
Pumps as discussed in this article are generally considered to be  larger with a larger drivers (i.e., motors), usually pad mounted or otherwise mounted to a floor assembly, and can be used in both closed- or open-circuit systems such as a condenser water loop with cooling towers. A “circulator” is a pump but is typically used in a closed circuit, and is usually smaller with a fractional horsepower motor, although that is not always the case. A circulator only needs to overcome the frictional losses in a piping system without changes required in elevations. These smaller pumps are often sealed units, with the motor rotor, pump impeller, and components such as bearings all sealed within the fluid circuit. Because they are typically smaller, they can be supported entirely by the piping system (flanges). The designer will undergo similar steps in sizing and selecting circulators as with pumps. 
Along with the fluid, other items that affect the overall system design and thus the pump selection include: equipment layout, flow paths, the size and length as well as the type and age of the piping, fittings, valves, piping specialties or appurtenances, noise, and any elevation changes. These parameters establish related system friction losses or pressure drops. Determining this “resistance to flow” in a new or existing hydronic system is probably the most important design task to be completed. All units will be expressed as English units here. This friction pressure drop is referred to as “friction head” loss (Hf), and is usually expressed in height of a liquid column in feet. 
One method of calculating the friction head loss is the Darcy-Weisbach formula: 
with
The length (L in ft), inside diameter (D in ft), and a dimensionless friction factor (f) of the pipe are used, along with the “velocity head” (Hv) in feet using the fluid flow velocity (V in ft/sec) and the acceleration of gravity (g or 32.2 ft/sec2) to calculate this pressure loss. The friction factor of the piping (f) considers the relative roughness of the piping and the “Reynolds Number” based on the pipe diameter and the fluid properties (viscosity, density, specific gravity) and velocity. In some circumstances, the Williams-Hazen formula can be used. The designer should review these formulas, terms, and concepts to fully understand their importance. 
Calculating pressure losses through a piping system for the required head of a pump selection can be done easily enough with an automated spreadsheet using the concept of equivalent lengths of pipe and determining the pressure loss per 100 ft of piping. This method is similar to completing HVAC ductwork friction loss calculations. The system information or items listed above are required for a designer to complete the calculations, plus he or she must determine a factor of safety to use. The Cameron Hydraulic Data suggests 15% to 20% safety factors for commercial piping systems. Head loss per 100 ft of pipe due to friction can also be found in Cameron Hydraulic Data friction loss tables for pure water at 60 F and clean, new pipe. 
Additional adjustments or corrections are required for other temperatures and conditions, and friction can vary with temperature and pipe roughness. The spreadsheet can calculate total dynamic head by multiplying the total equivalent length of a pipe segment by the head loss per 100 ft of pipe. The designer will need to look up equivalent lengths for all the fittings based on the type selected. The loss from each pipe segment is summed and then the safety factor is applied. The total dynamic head can also be rounded up to the next 5 (adjustable) ft of head.
The majority of designers today use a computer program to perform some type of hydraulic modeling of the distribution system to calculate the pressure losses. These head loss calculations, however done, should be completed for any design as they will dictate the selection of all component equipment (chillers, pumps, etc.) as well as the pressure class of the distribution piping, fittings, and valves within the system. These pressures will, in turn, be related to the selected pumping scheme. The purpose is to try to balance the system under design flows including any parallel flow paths, and determine the required pump head to overcome the losses. The overall calculation process is iterative, especially for a newly designed system. An existing system being modified might require several modeling runs as well, but may require only a few changes to fully integrate the system to the modified use. The main point to remember is that every component within the system will affect the pressure and the fluid flow rate and will either fix the pressure at a particular level, increase the pressure, or decrease the pressure. 
After the system is defined, which might also include a simple process flow diagram (Figure 1) or a more detailed piping and instrumentation diagram (PID), and the head losses are determined, the designer should develop “system head curves” (discussed later in the article). These curves will correlate the volumetric flow rates through the various flow paths to the related pressures or hydraulic losses that will be experienced in the piping system.
Additional items to consider for a complete system understanding are: 
  • Whether the system will be operated continuously or in an intermittent mode
  • The system’s flow type (e.g., constant or variable volume)
  • A need to provide continuity of service or redundancy for any equipment (N+1)
  • A need for future capacity or expansion of the system

The eventual wear and tear of the system, which dictates overall material selections. 

Determine pump type, driver 
Pumps are typically grouped into two main categories and referred to as either dynamic or displacement type (see Table 1). These types are separated by how the energy is added to the fluid to make it flow throughout the system. In a dynamic pump, the energy is added continuously to increase the fluid velocities while a displacement pump receives energy in periodic bursts that directly increases pressures. Pumps can be further categorized by physical properties (construction materials, geometries, orientations) or by the fluids with which they work. 
Along with the energy or pressure additions, the capacity or the available flow rates the pumps can generate within the system is important. The basic purpose of the pump is to move a desired flow rate or capacity (typically in gallons per minute, gpm) of a liquid while overcoming the resistance to that movement within the piping system. More specifically, the pump provides a volumetric flow by increasing pressure or developing head (Hd, in feet) on the fluid. This total system head, also called total dynamic head, raises the suction pressure at the pump by the total head required for the system. In other words, the pump will add additional pressure on top of the suction pressure amount, thereby creating the required discharge pressure to overcome the necessary system losses for the desired flow rate. 
The total dynamic head of the system is defined as “equal to the total discharge head minus the total suction head of the pump expressed in feet of water.” Head is considered equivalent to a given vertical height of a column of liquid. Pressure exerted on a base surface from the liquid column depends on the specific gravity of that liquid. The specific gravity (SG) of water is 1.0 at 68 F (use caution with hot water systems) and is the basis for comparison of all other liquids. The formula used to convert between head and pressure (pounds/sq in, psi) is: 
All equipment within the system (chillers, heat exchangers), and all piping, fittings, isolation, and/or control valves, and any other appurtenances will decrease the system pressure (head loss in feet) through friction as the water passes through the system. The pump increases the pressure (head) in the system to provide the required capacity. With all the possible influences described above, the hydronic system requirements and the differences in pump characteristics will typically illustrate that one type of pump is better suited for an application than another. 
Items to consider in selecting a pump include: the overall system layout and building floor space or head room, requirements of the pumping scheme such as flow capacities at design and part loads and if the head changes with flow rate capacities, code issues, the intended life of the system, the up-front costs of the pump verses maintenance costs, and overall energy use (constant versus variable speed). For example, a pump may need to be used for a constant speed and constant capacity application, but does not have a wide range of design pressures available. If the application for the pump requires it to be self-priming, only certain types of pumps would be considered. The piping distribution system may have more than one pump (identified as primary, secondary, or even tertiary) and/or the pumps may operate in series or parallel, all of which will affect the operation of the other pumps within the system.
A parallel arrangement is more common with multiple pumps, and the pumps are typically the same type and size, but not necessarily. The pumps do not need to be sized individually to meet the loop capacity, but can be operated together to do so. In this case, the pump’s flows will travel in parallel paths, usually for a short distance, and are additive to meet total flow with the same head requirement. In a series pump arrangement, the flow through both pumps is the same and the head pressure is additive.
Centrifugal pumps typically have very stable and predictable performances through multiple operating conditions of variable capacity and variable head requirements. A few items can affect their performance, such as changes in impeller size, pump casing geometry, variable liquid properties such as specific gravity and/or viscosity and air entrainment, and pumping loss increase due to mechanical wear. 
The driver for most pumps is typically an electric motor and can be operated at constant speed or variable speed with a variable frequency drive (VFD). Two-speed and multi-speed motors may still be found in service, but have generally been replaced with the more cost-effective VFDs. VFDs allow for pumps to operate along a system head curve and save power during part load operation. Although it is not cost-effective to use a VFD on a motor that will run at full load or 100% speed all of the time, VFDs are valuable when transferring from one pump to another to equalize run times, and provide for preventative maintenance efforts within facilities that operate 24/7. 
Depending on alternative forms of energy available, pumps can be driven with steam (turbines, engines), and gas or diesel fuel (turbines, engines). Thus, the type of driver employed for a pump may cost more than the pump itself. Only motor driven centrifugal pumps are assumed here because these are almost exclusively used in hydronic systems these days. Some common centrifugal pumps are: horizontal split case, vertical split case, vertical turbine (Figure 5), end suction, and a vertical in-line pump.
Cavitation
Vapor pressure is a key liquid property that a designer should be cognizant of. Vapor pressure is defined as “the pressure exerted by a vapor in thermodynamic equilibrium with its condensed phases (solid or liquid) at a given temperature in a closed system.” The equilibrium vapor pressure is an indication of a liquid’s evaporation rate. Another way to look at this is for a fluid to continue to exist in a liquid state, its surface pressure must be greater than or equal to its vapor pressure at the existing temperature. For example, a higher surface pressure is required for maintaining a volatile liquid such as alcohol at the same temperature than for water, because alcohol’s vapor pressure is higher. 
When water flows from the inlet of the pump to the impeller, the pressure drops (suction head changes), and if this drop reduces the absolute pressure to less than or equal to the vapor pressure of water, some of the liquid water will change to a gas and vapor bubbles are formed. Once this mixed flow fluid reaches the higher pressure areas at the entrance of the impellers, the vapor bubbles will collapse. This will cause a concentration of energy, creating large localized forces that can cause mechanical damage (pitting) to the metal surfaces inside the pump. This phenomenon, cavitation, will produce noise and vibration, reduce the efficiency of the pump, cause a loss of total head, and ultimately can lead to equipment failure.
To prevent cavitation, the pump must have the water’s available absolute pressure greater than both the water vapor pressure and the friction loss at this point combined. The net positive suction head available (NPSHA) is the absolute pressure of the water at the pump inlet. This NPSHA is given in psia (pounds/sq in., absolute) and is dependent on the pressure of the water, the water temperature, and the elevation of water leading into the pump suction. The configuration of the system, and where the pump is located within, will affect this value. 
Another term a designer will see is net positive suction head required (NPSHR). This value is determined by the pump manufacturer as it is a function of flow and dependent upon the pump selected. This value does not change for the pump requirements of speed, flow rate, and head; however, it does change with the fluid type used and any wear and tear on the pump over time. The difference of the two values must be positive, and the Hydraulic Institute has applicable NPSH margin ratio guidelines (NPSHA/NPSHR) to implement as necessary. The designer is directed to investigate and become more knowledgeable about cavitation and NPSH to fully understand its importance in selecting a pump. 
Cavitation can especially be a problem in open systems if the designer fails to consider the NPSHA and NPSHR relationship. In addition, note that cavitation can also occur in a closed system if the makeup water pressure is too low, thus causing the pump suction to be too low. This is a rare situation and may indicate makeup water issues or even leaks in the piping system.

Pump affinity laws
Centrifugal pumps impart velocity and convert velocity to pressure. The flow and head may be changed by changing the impeller diameter size or by varying the pump speed (impeller tip speed) through the use of a VFD. This overall relationship is called pump affinity laws, and is limited to centrifugal pumps only. 
Pump affinity laws, as defined by Cameron Hydraulic Data, Section 1:
For small variations in impeller diameter (constant speed)
Flow rate varies with ratio of diameters

Head varies with ratio of the square of the diameters

Brake horsepower varies with ratio of the cube of the diameters

For variations in speed (constant impeller diameter)
Flow rate varies with ratio of speeds

Head varies with ratio of the square of the speeds

Brake horsepower varies with ratio of the cube of the speeds

By substitution, other relationships can be determined such as: 

Determine pump size, design operating point
After the preliminary steps of configuring the piping system layout and calculating the total pump head are complete, the designer needs to select the pump. To specify the pump performance, the designer needs to provide the flow rate in gpm and total developed head in feet. To fully understand how to select the pump, the designer needs to know and develop “system head curves.” As briefly discussed earlier, these curves correlate the volumetric flow rates to the related hydraulic losses in the piping system. The designer should also understand a few more fundamentals.
Assume the following example for system information. 

Note that a pump’s system curve will define the system characteristics, but not the pump’s capability to provide a specific capacity at a given head. This data must be provided by the manufacturers regarding their pump’s performance.  
The designer should note that the selection of a pump based on its performance curve is directly related to its impeller size. The diameter of the impeller is selected to ensure that the pressure (head) is achieved but is not excessive, and that the power required to run the pump is enough to provide the required flow rate but will not overload the motor. In Figure 2, the pump’s system curve (red line) and its capacity curve (blue line) are shown together. Combining these curves on one graphic provides a much better picture of the pump performance as it relates to the system curve. The pump capacity curve relates specifically to a particular impeller size. A corresponding curve (a different impeller size) may be above or below this curve, and traveling in the same fashion (from higher left to lower right) dependent on whether the impeller diameter is larger or smaller. The far left side of the pump curve at the zero flow rate point is called the “shut-off head” of the pump. The far right side of the curve (at the very end) is called the “pump run-out” or maximum flow rate. Within each manufacturer’s pump size, there are always several different impeller sizes that create the different pump curves available. To obtain the system flow rate required, the designer selects the correct impeller diameter, or the manufacturer will need to “trim” the impeller to a custom size for the application. This trimming of the impeller can be done in the field after an installation if the initial pump selection was incorrect for the installed conditions.
Typically, a system design should incorporate a pump to operate at 80% to 115% of its best efficiency point (BEP). Most manufacturers provide more detailed performance curves that typically include information such as maximum and minimum impeller sizes, power required, NPSHR, and the pump’s efficiencies. 
For a constant speed pumping scheme, changes to the piping system that cause an increase in system resistance (i.e., closing a valve) or a decrease in system resistance will affect the system curve and will move the design point left or right along the pump curve as these system changes occur. The pump can run from some minimum continuous flow to pump run-out position. 

Parallel and series pumping schemes also affect the pump-system curve relationship. Pumps in parallel operate at the same pressure but their flow rates are additive, while pumps in series operate at the same flow rates and their pressures (head) are additive. One way to increase flow rate without sizing the system with a bigger pump is to put in two or three pumps in parallel. This might also be advantageous for continuity of service or redundancy (N+1).

If we refer back to the 7500 gpm pump at 364 ft of head and use some other pumps in parallel, the total pump curve is flattened out and becomes more responsive to changes in head (Figure 3).
With a series pump operation, the head is additive. This occurs with the use of booster pumps that might be needed at the far end of a chilled water loop, or to overcome the final pressure in a building (Figure 4).
Finally, system curves are impacted by the use of VFDs. Variable speed drives change the pump speed through adjustments to motor speeds, and these changes can be shown to move along the system curve, not the pump curve, to produce changes in pump flow. 
Many facility operations use VFDs to provide for a softer starting of the pumps, for smoother transitions when transferring pumps from on/off conditions for redundancy, and of course for energy savings attributed to their use during part load conditions. The costs of the VFDs need to be considered but typically are worth the expenditure. 

How to read pump curves
After absorbing the basic information above, how do you read and interpret pump curves? 
  1. Find your known head value developed from your calculations on the vertical left axis of the graphic. Follow the head line over to where it intersects with a pump curve with your desired flow rate or capacity read on the horizontal axis of the graphic when you drop straight down. This curve will be the impeller size necessary to develop the required head and capacity, and should be labeled in inches
  2. Now select your motor size. The motor must drive the impeller without overloading. To do this, observe the horsepower (hp) lines and remember that to the left of the hp line is non-overloading while to the right of the hp line is overloading. You should select the pump motor size large enough that even at the pump run-out condition, the selected impeller size does not cross the selected motor hp line.
  3. The last thing is to determine what the efficiency of the pump will be while operating at the design point. You can observe the U-shaped lines and estimate it by interpolation.

Regarding step 2 above, selecting a “non-overloading” motor size is usually the best method; however, there may be times when the designer might select a motor that could overload in a run-out condition. 

Assume the following example for system requirements.
The system requirements for a constant speed pump were determined to be 2475 gpm and 110 ft of head for a recent project using a vertical inline pump. After consultation with the manufacturer’s representative, the following set of curves was provided with the selection point indicated with a red triangle. 
The pump requirements presented indicated the impeller would be 11.87 in. in this example with NPSHR of 19.7 ft, a BHP of about 85 hp, and an efficiency of about 80%. This selection requires a 100 hp motor to remain non-overloading. 
There are several reasons to consider “a good selection point” for a pump application. In this example, the operating point for the system curve intersects the pump curve in a slightly “sloped” or “steeper” area of the curve. The steepness of the curve will provide some immediate feedback to the designer. A steeper curve allows for relatively larger changes in pressure drop with a smaller effect on desired flow changes (i.e., having to clean a filter in the system on a periodic basis when flow drops). A flatter curve will allow a small change in pressure drop to produce a relatively larger change in flow. This might be applicable in systems with control valves where you may want a large change in flow rate as the control valves close down. There are other times when selecting in a flat region is not advisable. 
Maximizing energy efficiency
The first step in designing any efficient, effective HVAC system for a building or campus is to perform accurate building load calculations and energy modeling. As mentioned earlier, ASHRAE Standard 90.1 provides methods and guidelines for these tasks. The type of HVAC system designed and installed and its configuration will certainly require one or more types of pumping schemes. The constant interaction and changes in HVAC loads within a building, or between multiple buildings on a loop, should be part of the system considerations so all equipment (i.e., pumps) can be sized and controlled properly to account for all the energy impacts.
The designer should become familiar with ASHRAE 90.1, Section 6, which includes various requirements and exceptions that affect pumping design. For example: 
  • In Section 6.4.2.2, the pump differential pressure (head) for the purpose of sizing pumps must include the pressure drop through each device and pipe segment in the critical circuit at design conditions.
  • In Section 6.5.1.2.2, there is a maximum pressure drop allowed for precooling coils and heat exchangers used in water economizers of less than 15 ft of head, or the designer must create a secondary loop so these pressure drops are not seen by the main system pumps when the system is in the normal cooling (non-economizer) mode. 
  • In Section 6.5.4.2, there is a requirement for design efforts to include pump pressure optimization in systems where the total pump system power exceeds 10 hp. The pump control setpoints are varied due to control valve positions in the system to provide for variable fluid flow and are capable of reducing pump flow rates to 50% or less of the design. There are other items to review and there are exceptions.
  • In Section 6.5.4.3, if a hydronic system includes more than one chiller, cooling tower, or boiler, they must be isolated so that all fluid flows through the respective equipment are automatically shut off when the equipment is shut down. Additionally, if there are constant-speed chilled-water, condenser water, or boiler feedwater pumps used to serve multiple units of this equipment (chillers, cooling towers, boilers), the number of pumps shall be no less than the number of units and staged on and off with the individual pieces of equipment.
  • Section 6.5.5.4 requires open-circuit cooling tower flow turndown control when configured with multiple- or variable-speed condenser water pumps.
  • Section 6.7.2.3.3 requires hydronic systems to be proportionately balanced to minimize throttling losses before trimming pump impellers or adjusting pump speeds to meet design flow conditions.
  • Section 7.4.4.4 requires controls on circulating pumps for limiting their operation within water storage tank applications.
  • Section 7.4.5.3 requires simple time switches to be installed on swimming pool heaters and pumps.

Randy Schrecengost is a project manager/senior mechanical engineer with Stanley Consultants. He has extensive experience in design and in project and program management at all levels of engineering, energy consulting, and facilities engineering. He is a member of the Consulting-Specifying Engineer editorial advisory board.