Engineering Manual of
HOT AND CHILLED WATER DISTRIBUTION SYSTEMS CONTROL
Hot and chilled water pumping, distribution, and control systems have similar characteristics. A hot and/or chilled water system distributes heating or cooling energy through a building. The water is pumped from a boiler or chiller to coils or terminal units. Effective control of this energy requires understanding the control loops and related control valves and also an understanding of the pressure/flow relationships between the piping and pumping components of the system.
Water distribution systems used in buildings include:
- LTW. Low temperature water systems supply water at temperatures up to 250F and working pressures up to 160 psi. Although, most LTW boilers have a maximum working pressure of 30 psi.
- MTW. Medium temperature water systems supply water at temperatures between 250 to 350F with pressures up to 160 psi. Maximum medium temperature boiler temperature is 350F.
- HTW. High temperature hot water systems supply water at temperatures over 350F, usually in the 400 to 450F range, with working pressures up to 300 psi.
- CHW. Chilled water systems supply water at temperatures from 40 to 55F with pressures up to 125 psi.
- DTW. Dual temperature water systems supply LTW during the heating season and CHW during the cooling season to the same terminal units.
A typical system (Fig. 51) illustrates the principles of water distribution in a system. The system consists of a heating or cooling source, a pump, distribution piping, and valve controlled coils. The pump provides force to push the water through the system and valves control the flow through the individual coils. The air separator removes entrapped air from the system.
Fig. 51. Typical Water Distribution System.
The expansion tank is charged with compressed air to place the system under the minimum pressure required at the inlet to the pump to prevent pump cavitation and the resultant impeller erosion. The minimum inlet pressure required by a pump is referred to as the net positive suction head (NPSH). Figure 54 indicates the NPSH for a particular pump. The air volume in the tank is sized, based upon the volume of water in the system and the expected water temperature variations, to allow the water to expand and contract as water temperatures vary throughout the year. The expansion tank static pressure does not effect the closed system control valve differential close-off pressure, but must be considered, in addition to the pump head, for valve body and other piping system component pressure rating selection.
The air separator and expansion tank are omitted from the examples in this section for simplicity.
The requirements of a properly applied distribution system are:
1. Maintain controllable
pressure drop across the control valves.
The pump is a key component of a water distribution system. It is essential to understand pump characteristics in order to understand and design distribution systems and to understand pumping system control solutions. Centrifugal pumps are commonly used to distribute hot and chilled water through commercial buildings. Many varieties of centrifugal pumps are available, as shown in Table 2. Figure 52 shows a typical base-mounted pump.
Table 2. Characteristics of Centrifugal Pump Types.
Fig. 52. Typical Cross-Section of an End Suction Pump.
The performance of a given pump is expressed in a curve showing pump head in feet versus gallons per minute (gpm). Figure 53 shows a typical curve. The head is expressed in feet (of water column) which describes pump operation independent of water temperature or density. Pressure losses in piping and components used in HVAC systems are always calculated in feet.
Fig. 53. Typical Pump Head Capacity Curve.
The pump curve in Figure 53 is part of a family of curves for a pump. Each curve of the family represents a different size impeller used with the pump at a specified rpm. It relates to the power input required just to move the water (water horsepower) as follows:
Water hp = (flow x head x SG) ÷ 3960
The motor driving the pump must have a horsepower rating in excess of water horsepower to take care of bearing and seal friction, recirculation within the housing, and impeller efficiency.
NOTE: Water horsepower increases with head and flow. If flow is allowed to increase, the motor may overload.
Commercial pumps have performance curves showing the following data for a given pump speed:
- Total head in ft versus
flow in gpm
Figure 54 is a typical performance curve showing some of the preceding data. Impeller diameters are shown on the left.
Pump efficiency is a comparison of water horsepower developed in the pump and brake horsepower applied by the motor to the shaft and impeller.
Courtesy of Aurora Pump
Fig. 54. 1150 RPM Typical Pump Curve.
Figure 55 illustrates a pump fitted with a 6-1/2 inch impeller, operating at 45 ft of head, and delivering 65 gpm of water.
The curves show that the motor output is 1.17 horsepower. Efficiency = (water horsepower)/(motor horsepower) x 100 = (0.74/1.17) x 100 = 63 percent. This agrees with the efficiency curves shown in Figure 55.
Pump affinity laws (Table 3) show how pump flow, head, and brake horsepower vary as impeller diameter or speed change. These laws help when adjusting an installed pump to changes in the system served. For example, if a pump with an 8-inch impeller delivers 80 ft of head, with a 7.21-inch impeller it would deliver 65 ft of head. It is calculated as follows:
Table 3. Pump Affinity Laws.
Fig. 55. Pump Curve for 1750 RPM Operation.
The pump curves and affinity laws are used to select a pump or pumps for a particular application. The first step is to establish a system head curve. This is calculated from design flow and head loss tables for all the piping, coils, control valves, and other components of the system.
An example is shown in Figure 56. The design point is 65 ft of head at a 515 gpm flow. A system curve (a simple square root curve) can be plotted once the flow and head are known at any particular point, since:
Plot the points (Fig. 56) for flows of 200, 400, and 600 gpm:
The system curve assumes all balancing valves are set for design conditions, that all controls valves are fully open, and that flow through all loads is proportional. The system curve is always the same, even if loading is not proportional, at no load and full (100 percent) load. If the loads are not proportional (such as, some loads off or some valves throttling), the curve rises above that shown for values between full and no load. The system curve in Figure 56 is used in Figure 57 to select the single speed pump. Using this system curve to determine the switching setpoint for dual parallel pumps when the load flow is not proportional can result in damaging pump cycling.
Fig. 56. System Curve for Pump Application.
The design system flow is 515 gpm. Piping, control valve, and equipment losses are calculated at 65 ft. An impeller size and a motor horsepower are selected by imposing the system curve on the pump curve (Fig. 57). The designer has the option of selecting a pump with a 9 1/2 in. impeller (515 gpm at a 76 foot head) or a 8 3/4 in. impeller (500 gpm at a 60 foot head). The smaller impeller requires a 10 horsepower motor and the larger impeller requires a 15 horsepower impeller. Selection of the 9 1/2 in. impeller requires system balancing valves to reduce the system pressure differentials to those matching the design flow of 515 gpm.
When selecting a pump, it is important to remember that:
- With constant speed pumps (and two-way AHU control valves), flow rides the pump curve. The system curve is plotted assuming that the control valves are full open, which in any system, only occurs at the full load. As control valves throttle and loads are turned off the system becomes non-proportional and the system curve rises between no load and design.
- With variable speed pumps, the system control objective is to have the pump curve ride the system curve by keeping at least one control valve open and reducing the pump speed to reduce flow with the diminishing system drop.
Fig. 57. Matching Pump to System.
To better understand system curves, pump curves, and flow control, Figure 58 shows the control valve(s) (the only variable element of a typical system curve) separately from the rest of the system elements. Lines are shown for each of three valve positions; full, 80 percent, and 50 percent flow. These lines, when added to the curve for all other elements of the system intersect the pump curve at the corresponding operating point(s).
Figure 58 shows a system with 500 gpm and 70 foot head at design, 10 ft of which is a full open control valve at the end of the piping run. Line 'A' represents the control valve and connects the pump curve to the static-element system curve. If all control valves positioned to 80 percent flow, the pump head rises, the static-element system resistance drops, and the control valve, represented by line 'B', makes the difference; about 36 ft. Similarly, at 50 percent flow, the valve drop, represented by line 'C', accounts for about 63 ft.
Fig. 58. Pump and System Curves and Control Valves
From the pump affinity laws (Table 3), pump horsepower decreases by the cube of the decreased speed, and flow decreases linearly with speed; so at 80 percent flow, the horsepower is down to nearly 50 percent (80 percent cubed). Since many systems have sharply reduced flow requirements at medium or low loads, pump speed control can provide economical operation for most of the heating (or cooling) season. Figure 59 shows typical performance at reduced speeds. The shaded area of Figure 60 shows the wide range of heads and flows available from a variable speed pump. Variable speed pumps are usually controlled from a differential pressure sensor with either fixed or load reset setpoints.
Control objectives of variable speed pumping systems in networked digital control systems, is to keep the most demanding load control valve full open by varying the pump speed.
Fig. 59. Pump Performance and Efficiency at Various Speeds.
Fig. 60. Typical Variable Speed Pump Performance Range.
Table 4 and Figure 61 show the Figure 58 system with all the control valves remaining full open and the load controlled by varying the pump speed. This is the ideal system wherein the loads of all AHUs vary in unison and the pump speed is controlled to satisfy the valve with the greatest demand. This is usually accomplished via differential pressure control, automatically reset.
Fig. 61. Ideal Variable Speed Pump Control.
Table 4. Variable Speed Pump Relationships
In cooling towers (Fig. 62) and other open systems, static head must be considered when establishing system curves and selecting a pump. Notice that the 90 ft of vertical pipe (static discharge head) is partially offset by the 80 ft of vertical pipe (static suction head) in the suction line. When a system curve is drawn for such a system, static head of the tower must be added to the system curve. The system is designed to operate at 200 gpm against 30 ft of head for piping and valves. The system curve in Figure 63 is drawn through zero head (ignoring the static head of 10 ft) which leads to choosing Pump A.
Fig. 62. Typical Cooling Tower Application.
Fig. 63. System Curve for Open Circuit without Static Head.
Figure 64 shows the system curve adjusted for the 10 ft of static head and the actual operating points of Pump A (selected in Figure 63) and Pump B. Notice Pump A supplies only 175 gpm at 32 ft of head.
Fig. 64. System Curve for Open Circuit with Static Head.
Multiple pumps are used when light load conditions could overload a single pump. These conditions normally occur when two-way control valves are used in the control system. Two-way control valves sharply reduce flow when they begin to close. Figure 65 shows that in a single-pump system, over pressure can result at low flow. At one-third flow, the pump head has increased, the source and piping drop is reduced to one-ninth of the design drop, and the control valve drop has increased greatly. Bypass, variable speed, or throttling valve pressure relief should be used with a single pump. Where the heat exchanger (such as a chiller) requires a high minimum flow rate, a single pump is used, and diversity is not used, three-way load control valves should generally be used.
Fig. 65. System Operation with One Pump, Design and Low Flow Condition.
Multiple pumps may be connected either in parallel or in series into the system. In the dual parallel pump configuration of Figure 66 a single pump can usually handle 75 to 80 percent of the total flow. The system curves show that at design conditions the control valve drop is ten feet (from A to B). At 75 percent flow (375 gpm), the valve drop with both pumps operating increases to over 41 ft (C to E). With one pump and 75 percent flow the valve drop is about 16 ft (C to D). When flow is reduced to 50 Percent, the valve drop is about 55 ft for one pump (F to G) or 63 ft for two pumps (F to H). Dual parallel pumps save energy and provide redundancy for 75 to 80 percent of the flow. They do not provide much relief for high valve pressure drops at low flow.
The pump curves and the system curves indicate possible pump start/stop setpoints. One scenario on a pumping differential fall to 42 ft, energizes the second pump and on a pumping differential rise to 77 ft, switches back to one pump. The 42 ft pumping differential corresponds to a point just before the 1-pump curve intersects the system curve (I), the point at which a single pump no longer can support the system. When the second pump is started, the operating point moves to the 2-pump curve and when the control valves have settled out will be at about Point J. It will vary along the 2-pump curve down to B or up to K. When the operating point reaches K (about 77 ft) the system switches back to a single pump and the operating point is now on the 1-pump curve until the differential pump pressure drops to I, at which time the cycle repeats. See Plotting a System Curve for statement on use of ideal system curve for determining setpoints when coil loading may not be proportional.
Again a reminder to exercise caution when using the ideal system curves for switching pumps on and off. The ideal curves are valid only at full and no load conditions, the rest of the time the actual curve is somewhere above the ideal. Since setpoint determination is not possible without the actual system curve, the lag pump stop setpoint should have a significant margin of safety incorporated. The lag pump start setpoint should be controlled by a differential pressure controller and have the software requirement that one control valve be full open for four minutes before starting.
Time delays must be built in to the control sequence to prevent rapid switching between one pump and two pump operation. With each change in pump operation, all control valves must adjust to new steady-state conditions. The adjustment process often causes overshoot or undershoot until temperature stability returns and no switching should take place during this time. Depending upon the type of temperature control loops, switch-lockout period can vary from 5 minutes for relatively fast discharge air control to over 30 minutes for relatively slow space control.
Fig. 66. System Operation with Two Pumps in Parallel.
Series pumps (Fig. 67), though rarely used in HVAC systems, are useful where both flow and head are sharply reduced at light loads.
Fig. 67. System Operation for Series Pumps.
For pumps in parallel (Fig. 66), assuming two identical pumps, the curve is developed using the following formula:
gpm3 = (gpm1) x 2 for any h1
For pumps in series (Fig. 67), assuming two identical pumps, the curve is developed using the following formula:
h3 = (h1) x 2 for any gpm1
Figure 68 illustrates a closed system where static head (pressure within the system with pump off) does not need to be considered as long as all components are rated for the static head encountered. The pump provides force to overcome the pressure drop through the system and valves control the flow and pressure through the system. Figure 69 shows a graph of the system and pump curves for design load and reduced load conditions. The system curve indicates the pressure drop through the system (with the control valves full open) at various flow rates. The pump curve shows the pump output pressure at various flow rates. Flow always follows the pump curve.
Fig. 68. Simplified Water Distribution System.
In Figure 68 the flow and pressure considerations are:
1. The flow through the heating or cooling source, the supply piping, and the return piping (40 gpm) is the same as the sum of the flows through the three coil circuits:
(10 + 12 + 18 = 40 gpm)
2.Design pressure drop (head loss) includes the drop through the heating or cooling source, supply piping, return piping, and the highest of the three coil circuits:
(23 + 21 = 44 ft)
NOTE: In this example, Coil 1 and 3 balancing valves balance each load loop at the 21 ft design for Loop 2. If the actual coil and control valve drops were less than the design maximum values, the actual balancing valve effects would be greater.
In this example the pump must handle 40 gpm against a total head of 44 ft (19 psi) as shown in Figure 69. (This curve is taken from actual pump tests). The design drop across the valve is 11 ft (4.5 psi) with the valve fully open.
If Figure 68 is a heating system, as the loads reduce valves V1, V2, and V3 start to close. Hot water flow must be reduced to about 15 percent of full flow (6 gpm) to reduce heat output to 50 percent. As flow through the coil is reduced the water takes longer to pass through the coil and, therefore, gives up more heat to the air.
As flow through the system is reduced, a new system curve is established. See the 6 gpm curve in Figure 69. When the flow is reduced, the new head loss in source and supply and return piping can be calculated using the formula:
For example: If gpm1 = 40, gpm2 = 6, and h1 = 33 ft, then h2 = 0.74 ft.
At low flow, the pressure drops through the coils, coil piping, and heat source tend to disappear and nearly all of the now elevated pump head appears across the partially closed valves V1, V2, and V3. This can cause valve noise, poor control, or failure of valves to seat. Control solutions are discussed in following sections.
Fig. 69. System and Pump Curves for a Closed System at Various Loads.
Distribution system control solutions vary dependent upon whether the designer chose a direct or reverse return piping system. Systems are sometimes configured as a combination of both; a high-rise building could, for example be reverse return on the riser and direct return on the floor run-outs. Direct return systems are usually lower cost and used in smaller installations. Reverse return systems are used in both small and large installations.
Fig. 70. Direct Return Piping System
The Figure 70 supply piping runs out to the coils decreasing in size between AHU 1, 2, 3, 4, 5, and 6. The return lines between each AHU are typically sized the same as the respective supply lines. The drop across AHU 6 must be 16 ft in order to get the 200 gpm through the valve (8 ft) and the coil (8 ft). To get the 200 gpm from AHU 5 to AHU 6, the drop across the piping at AHU 5 is 18 ft (16 ft required to get the flow through AHU 6 plus 2 ft to overcome the supply and return piping drops between AHU 5 and AHU 6). The AHU 5 balancing valve is set to prevent the 18 ft drop across AHU 5 from forcing more than 200 gpm to pass through the AHU 5 coil and control valve. The balancing valve B5 then is set to take a 2 ft drop at 200 gpm. Similarly, set a 4 ft, 8 ft, 10 ft, and 12 ft drop respectively across balancing valves B4, B3, B2, and B1.
If this is a variable flow loop with a variable speed pump and the pump is controlled to produce 16 ft across AHU 6, the control issue here is: When AHUs 2, 3, 4, 5, and 6 are off (no flow beyond AHU 1), then, the drop across AHU 1 is only 16 ft, 12 ft of which the AHU 1 balancing valve needs for design flow. Proper solutions are presented later in this section.
Supply piping is the same for a reverse return system (Fig. 71) as for the direct return system (Fig. 70). The return flow is reversed such that the return piping increases in size between AHUs 1, 2, 3, 4, 5, and 6. A full size return line is run back to the source room from AHU 6. In this example, the pump is the same as the direct return, since the return line from AHU 6 also takes a 6 ft drop. If the AHUs and source were positioned in a circular or hex pattern such that Coil 6 is closer to the pump, the run from AHU 6 back to the source would be shorter, and the reverse return piping head would be less than for the direct return, and the piping cost would be similar. In reverse return systems, balancing is usually only a trimming exercise.
Piping systems requiring constant flow through primary equipment (chillers, boilers) and variable flow to AHUs may be coupled or decoupled. See Chiller System Control and Boiler System Control for examples. Primary flow control for coupled systems and secondary flow control for decoupled systems are discussed later.
Fig. 71 Reverse Return Piping System
There are several methods for controlling pressure and flow in water distribution systems. The methods described in this section apply, in general, to both heating and cooling applications.
Coil bypass control uses three-way valves on terminal units and other coil loads in a water distribution system and satisfies the first four of the control system requirements (see Control Requirements for Water Distribution Systems). At reduced loads, the flow bypasses the coils and goes directly to the return main. Figure 72 illustrates the operation of this system as requirements change. Balancing valve (B) is adjusted for equal flow in the coil and the bypass
Fig. 72. Three-Way Valve Control- Coil Bypass.
Two way valves vary both the coil flow and the system flow, thus using less pumping energy at reduced flow.
Consider the following factors when deciding on two-way or three-way control valves.
- Piping Cost. Costs are higher for three-way valves than two-way valves, especially where limited space is available for piping (such as in room air conditioning units and unit ventilators). In addition, balancing cocks must be installed and adjusted in the bypass line.
- Three-Way Valve Cost. A diverting valve is more expensive than a mixing valve and a mixing valve is more expensive than a two-way valve. A mixing valve installed in the leaving water from a coil provides the same control as a diverting valve installed on the inlet to the coil.
- Diversity. If chillers and pumps are selected based upon diversity, three-way valves are inappropriate.
- Flow Characteristics. Three-way valves have linear flow characteristics and two-way valves may be either linear or equal percentage. Obtaining close control with three-way valves requires use of scheduled (reset) hot water temperatures.
- Capacity Index (Cv). Three-way valves for Cvs below 1.0 are often not available, therefore, small three-way valves tend to be oversized. Consider using two-way control valves for all applications of Cv = 4.0 and less where the quantity of two-way valves will have little effect on the total system flow.
- Constant Flow in Mains. Constant flow provides nearly constant pressure differential (drop) across a coil and valve.
- Pumping Cost. A three-way valve system uses full pump capacity even when the system load is very small.
- Part Load Control. Two-way valves allow better control on multiple pump systems during pump failure or part load periods.
- Automatic Control. Distribution control is a manual balancing task for flow loops employing three-way control valves. Automatic distribution controls are usually required to maintain flows and pressures (bypass valve, variable speed pump, pump staging control) for flow loops employing two-way control valves.
For further discussion on control valves, refer to the Valve Selection and Sizing section.
Control solutions for water distribution systems may vary based on:
- Direct return vs reverse
The examples in this section on Flow And Pressure Control Solutions use a distribution system that has six equal loads (coils) as shown in Figure 73. These control solutions are:
1. Single constant speed pump, single chiller system, two-way AHU control valves, and pressure bypass valve to control chiller flow to a minimum of 90 percent full flow.
2.Dual constant speed pumps, dual chiller systems, and pressure bypass valve to control chiller flow to a minimum of 90 percent full flow.
3.High control valve differential pressure control.
4.Decoupled variable speed secondary pumping system with two-way AHU control valves.
Fig. 73 Typical Example Loads.
Figure 74 analyzes Figure 70 pumping system at full flow and at half flow. The flow reduction at half flow is taken evenly across each coil. At half flow with no pressure bypass the control valve pressure drops increase from 8 ft to 44 ft as system friction drops reduce to one-forth of the design values and the pump head rises from 48 to 54 ft.
Figure 75 shows a pressure controlled bypass valve set to maintain 90 percent minimum flow through the chiller to satisfy the chillers minimum flow requirement. At 90 percent flow through the chiller (1080 gpm), the chiller and equipment room piping drops are 81 percent of design (90 percent squared). The pump curve (not shown) indicates a pump head of 50 ft at 1080 gpm.
Valve Location and Sizing
Since the system piping between Loads 1 and 2 is designed for only 1000 gpm and the low load bypass flow could exceed that, the bypass valve is located remotely before Load 1. If necessary to locate the bypass valve after Load 1, redesign the piping to carry the 90 percent flow.
If the differential pressure sensor is located across the main lines at Load 1 as shown in Figure 75 (see Differential Pressure Sensor Location), the best place for the bypass valve is the same location. Because the differential pressure is lower than in the chiller room, valve wear is less.
The valve is sized for approximately 1000 gpm with a 34 foot drop. A double seated valve is appropriate to reduce actuator close off requirements and the inherent leakage will not be a significant factor.
Differential Pressure Sensor Location
As previously stated the chiller design flow is 1200 gpm at 12 ft of head and requires a minimum flow of 1080 gpm. At 1080 gpm the pump curve shows a head of 50 ft.
From the formula:
Calculate the drop across the chiller as 9.6 ft.
Similarly calculate the reduced drop in the supply and return to Load 1 as 3.2 ft.
With the differential pressure controller located across Load 1, the setting is:
50 ft - 9.6 ft - 3.2 ft - 3.2 ft = 34 ft
This location provides a lower head across the load control valves at light loads than if located across the pump and chiller. To ensure the best sensing, be sure that the system strainer is located up stream from the differential pressure controller return pickup, so that a dirty strainer is not sensed as an increasing pressure drop (decreasing flow).
Fig. 74. Single Pump, Pressure Bypass, Direct Return at Full and Half Flow.
Fig. 75. Single Pump, Pressure Bypass System at 90 percent Flow.
The greatest change at the sensor provides the most tolerant and robust control. For this reason the sensor is located, not across the chiller with a setpoint of 9.6 ft, but across Coil 1 where the greatest differential pressure change exists (28 ft at design to 34 ft at 90 percent of design).
NOTE: With the controller pickups in these locations, it does not matter where in the system the load is located, what value it has, or whether it is symetrical or non-symetrical.
Also note that the Load 1 balancing valve takes a full flow 12 foot drop and even with the pressure bypass valve, the control valves will be subjected to drops of near 34 ft for light loads. For larger and more extended systems, both of these values must be considered when evaluating reverse return and control valve high differential pressure solutions.
These values lead to the conclusion that a differential pressure bypass solution may satisfy the light-load flow through a chiller, but may not adequately relieve control valve light-load differential pressures on larger systems. Also be aware that pressure drop changes due to scaled chiller tubes effect the bypass valve operation.
A reverse return system analysis equivalent to the direct return analysis of Figure 74 would show that at half flow and no bypass, the pump head still rises to 54 ft and the control valve drop still rises to 44 ft.
For 90 percent flow control, the pump still operates at a 50 foot head with the sensor located as far away from the pump as practical to take advantage of as many friction drop changes as possible.
The preferred sensor location (Fig. 76) is DP-1 if point A is near Load 1. Sensor location DP-2 is second best and again if point B is near Load 1. If location DP-3 is selected by default, with a setpoint of 40.4 ft as compared with the DP-3 full flow differential pressure of 36 ft. This small 4.4 foot change requires a higher quality sensor and more frequent calibration checks than for locations DP-1 and DP-2. Locate the pressure bypass valve and sensor as far from the chiller/pump as possible , but no closer than DP-3.
In all cases it is still preferable to locate the system strainer outside the control loop as shown in Figure 75.
Fig. 76. Single Pump, Pressure Bypass, Reverse Return at 90 percent Flow.
Dual chiller pressure bypass systems are popular because if a chiller, tower, or pump fails, there is part load redundancy and better part load efficiency.
In Figure 77, the bypass sensor and valve is in the same location as the single chiller system (Fig. 75). The valve is sized for approximately 500 gpm.
The differential pressure setpoint is the same as for the single pump/chiller system (34 ft) when both pumps are running. When only one pump/chiller is operating, the piping between the pump/chiller and differential pressure sensor carry a minimum of 540 gpm (90 percent flow for one chiller) and the piping friction drop (head) falls from 4 ft to 0.8 ft, thus:
With a single pump/chiller operating, the differential pressure setpoint for 90 percent flow is 38.8 ft.
Setpoint = 50 ft (pump drop) - 9.6 (chiller drop) - 0.8 (piping head) - 0.8 (piping drop) = 38.8 ft.
This is up from the 34 ft with both chillers operating.
With digital controls, the differential pressure setpoint offset adjustment when only one chiller/pump is operating is handled by a software routine (dual chiller/pump setpoint plus 4.8) invoked anytime only one pump and one chiller are operating.
One method using pneumatic controls uses two pressure controllers with separate setpoints. The primary controller is set at 38.8 ft with the secondary controller set at 34 foot setpoint and configured as a low limit device during periods of single pump/chiller operation. During periods of dual pump/chiller operation, the 34 foot setpoint controller is used alone.
As noted in the discussion of Figure 75, the differential pressure controlled bypass valve in constant speed pumping systems is adequate to maintain a high flow through chillers, but does little to prevent high differential pressures across the AHU load control valves. In the previous example the load control valve differential pressure varied from 8 ft at design load to 38.8 ft at the single chiller, light-load mode of operation.
If it is expected that the light-load differential pressures will exceed the load control valves close-off rating, locate a throttling valve in the common load piping (valve V8 in Figure 78) to reduce the load pressures while still allowing adequate pressure to maintain chiller flows. Select either a double-seated or balanced cage type valve for the high differential pressure.
Fig. 77. Dual Pumps, Dual Chillers- Pressure Bypass
Fig. 78. High AHU Valve Differential Pressure Control
In these examples, the design differential pressure across load one is 28 ft (Fig. 74), and pressure bypassing occurs at 34 ft. The maximum setpoint for DP-2 (Fig 78) should be about 29 or 30 ft. This initial setpoint is then slowly lowered based upon the percent-open values of load control valves V1 through V6, to a minimum value of 10 to 12 ft.
Anytime either chiller pump starts, DP-2 shall be enabled to control pressure reduction valve V8 at an initial setpoint of 20 ft. Anytime any load control valve is greater than 95 percent open, the DP-2 setpoint shall be incremented at the rate of 0.5 ft every 2.0 minutes up to a maximum of 30 ft. Anytime all control valves are less than 80 percent open the DP-2 setpoint shall be decremented at the same rate to a minimum value of 12 ft. All values shall be user adjustable.
The value of 2.0 minutes in the specification assumes that valves V1 through V6 are controlled from discharge air temperature and should recover in less than 2 minutes from V8-caused changes. If V1 through V6 were controlled from space temperature directly, the time rate for V8 adjustments may need to be extended to 12 to 15 minutes to allow incremental space temperature control recovery from the flow reductions brought about by V8. Valve V8 is normally open and line size to minimize its pressure drop during full load operation. This valve is applicable for direct or reverse return piping configurations where significant piping friction losses migrate out to the control valves upon low flow conditions.
In reverse return systems, locate Valve V8 in the main line piping just before the AHU 1 take-off and the DP-2 sensor across AHU 1 set for approximately 19 ft, this should be adequate for any variation in uneven loading. Allow 19 ft for close-off and good control. Resetting the DP-2 setpoint lower is unnecessary in this constant-speed pumping reverse return example.
Similar to Figure 74, Figure 79 decouples the loads from the source pumps and heat exchangers and uses of a variable speed pump to provide significant energy savings at reduced loads and simpler control. Variable speed pump control matches pump speed to system flow demands. If source devices perform well with variable flow, the primary pumps may be replaced with variable speed pumps and controlled similarly to the decoupled example.
NOTE: The variable speed pump head is only 36 ft since the primary pumps account for the 12 ft chiller head.
Fig. 79. Variable Speed Pump Control
AHU 1 requires a 28 foot differential pressure (8 feet for the coil, 8 feet for the valve, and 12 feet for the balancing valve), for full flow. If 28 feet is available at AHU 1, all other AHUs will have at least the required design differential pressure. Controlling the pump with a sensor positioned as shown for DP-2 set for 28 feet is acceptable.
Locating the sensor (DP-3) at AHU 6 and set for the 16 feet required by AHU 6 will not work when only AHU 1 is operating. With these conditions DP-3 maintains a maximum drop of only 16 feet across AHU 1 which needs a 28 ft differential pressure because of the balancing valve. If the sensor is positioned at AHU 6, the setpoint must still be 28 feet if the system is to operate satisfactorily with non-symetrical loading.
DP-1 located at the variable speed drive (VSD) is the most convenient place. It requires the DP-2 setpoint plus the friction losses between the pump and AHU 1.
Figure 80 shows the operating curve of the system with the differential pressure sensor located in the DP-2 position and set for 28 feet. With each AHU at one-third flow, the speed is 1412 RPM, which produces a 28 foot differential pressure at AHU 1 with 400 gpm system flow. When the coils are equally loaded at one-third flow, each control valve takes a 26 foot drop. In this configuration the pump never operates much below the 1400 rpm speed because of the 28 ft head setpoint.
Fig. 80. Fixed Setpoint with PI, Variable Speed Pumping Control
No pump head control example, so far, takes advantage of both the variable speed pump and a digital control system. The digital control system VSD control algorithm adjusts the differential pressure setpoint based on the demands of all the valves (Fig. 81) and all loads are satisfied with significant savings over any of the three fixed setpoint options. Since, when using valve position load reset there is no difference in performance between the three locations, DP-1 is preferred because of initial cost. Valve position load reset provides adequate control performance whether the sensor is only proportional or is only a static pressure sensor as compared to a differential pressure sensor.
Anytime any AHU chilled water valve is greater than 15% open for greater than one minute, the secondary pump shall be started under EPID control at 20% speed and with a ramp duration of 120 seconds. The pump VSD shall be controlled by a differential pressure sensor located between the supply line leaving the plant room, as far from the pump as practical to avoid hydronic noise that may be present at the immediate pump discharge, and the system return line. At start-up the differential pressure setpoint shall be 30 ft (See Note 1). Anytime any load control valve is greater than 95% open, the differential pressure setpoint shall be incremented at the rate of 0.5 ft every minute up to a maximum value of 38 ft. Anytime all load control valves are less than 80% open, the differential pressure setpoint shall be decremented at the same rate down to a minimum of 7 ft. After 12 minutes, the increment/decrement rate shall be changed from one minute to three minutes (See Note 2). All values shall be user adjustable.
1.From Figure 81, pump
head is 36 feet if all AHUs require full flow, therefore,
the 30 foot value is an arbitrary compromise.
Figure 81 shows the ideal performance of the load reset setpoint control with each AHU demanding one-third flow. All control valves are full open and the differential pressure adjusted to produce a speed of 525 RPM. If the coil loading is non-symetrical to the point that AHUs 1 and 2 are fully loaded while the others are off, the operating point for 1/3 system flow is the same as shown in Figure 80 for 1/3 system flow, since AHU 1 requires a differential pressure of 28 ft for full flow.
Fig. 81. Variable Setpoint, Variable Speed Pumping Control (Ideal Curve).
A pump speed valve position load reset program with over 20 valves can become cumbersome. Also, if any one valve, for whatever reason, stays open most of the time, then the load reset program becomes ineffective. Figure 82 shows an example of the valve position load reset program concept applied to a multibuilding facility with varying differential pressures entering each building, due to varying distances from the pumping plant. The example address two issues, differential pressure control within each building to relieve control valves from extremely high differential pressures and pump speed load reset.
Fig. 82. Multibuilding, Variable Speed Pumping Control.
Each building is provided with a 'choke' valve (V-1), and a load reset loop to maintain water pressure within the building, such that the most demanding AHU valve is always between 80 and 95 percent open. Each building requires a Valve V-1 Control Detail (inset) and a dynamic sequence description of the program. Each control detail includes minimum and maximum differential pressure setpoints, a software MANUAL - AUTO selector, and a setpoint value for the manual position. Ideally the control detail along with the current percent open for each valve within the building is provided graphically for each building.
NOTE: If the choke valve (V-1) is omitted on the most remote building, the choke valves need not be considered for pump sizing.
Pump speed is reset to keep the most demanding Building Valve V-1 between 80 and 95 percent open.
Adjust balancing valves in buildings close to the pumping plant with the choke valve in control, so that high balancing valve differential pressures are not set to negate low load value of the choke valve reset concept.
This concept can be combined with tertiary pumps in remote buildings to control the building differential pressure and choke valves in closer buildings utilizing central pumping.
Figure 80 assumes that all coils are equally loaded (1/3 flow), and that all friction losses are 1/9th [(1/3)2] of their full load value. In symetrical loading and with valve position reset, the balancing valves have no adverse effect. However, if at 400 gpm total flow, AHU 1 and 2 are operating at full flow (200 gpm) and all others are off, the required differential pressure across AHU 1 is 28 ft, 12 ft of which is wasted on the balancing valve.
Balancing Valve Elimination
Elimination of all balancing valves allows the valve position load reset control strategy to satisfy the non-symetrical loading described in Balancing Valve Effects by producing only 16 ft differential pressure at AHU 2 (slightly higher at AHU 1) and save significant pumping energy during most periods of non-symetrical operation. Before eliminating balancing valves consider:
1. Load coil temperature control setpoints must be strictly maintained. In the example, lowering AHU 1 leaving air temperature 5 degrees below the design temperature causes AHU 1 water flow loop to draw significantly more than design flow because it is nearer the pump where the differential pressure is higher. This will slightly starve the other loads.
2. Cool-down periods for other than AHUs 1 and 2 will be extended. With all valves full open, until AHUs 1 and 2 are satisfied, the other AHUs will be starved.
3. Industrial valves may be required to maintain acceptable controllability. At properly controlled full load design conditions, a 28 ft differential pressure drop appears across the AHU 1 control valve. This is the 16 ft differential pressure required at AHU 6 plus the 12 foot piping drop from AHU 1 to AHU 6. With the high differential pressure, some piping configurations will require an industrial valve.
Elimination of or fully open balancing valves might work well in a continuously operating facility with operators who understand disciplined setpoint and self-balancing concepts.
If eliminating balancing valves in a fixed setpoint scheme, position the DP sensor across AHU 6, with a setpoint of 16 ft. If AHU 6 is very remote from the VSD pump controller, it is recommended to put an additional differential pressure sensor at the pump with a maximum set point of 36 ft then reset the setpoint down as required to prevent the AHU 6 differential pressure from exceeding 16 ft. Use of a DDC PID input and output in separate controllers is not recommended because of the communications system reliability.
If balancing valves are removed in a valve position load reset scheme, use the single differential pressure sensor at the pump with a max differential pressure setpoint of 36 ft.
Refer to Figure 79, in summary:
1.If valve position load
reset is employed, the DP sensor may be located in the
pump room for simplicity (position DP-1)
Pumps require a minimum flow to dissipate the heat generated by the pump impeller. A bypass around the pump located out in the system provides the required flow and prevents the heat from building up in the pump. The minimum flow is calculated from the equation:
The minimum flow for 10 horsepower with a 10 degree DT (attributed to the pump heating the water) is only 5 gpm. The bypass may be fixed or if there is a remote AHU in the 12 to 25 gpm size, using a three-way valve on that AHU with the bypass in the bypass leg of the three-way valve prevents bypass water from flowing during full-load periods. Another option is an automatic bypass valve programmed to open anytime all AHU control valves are less than 10% open.
In these examples, the design differential pressure for all AHUs is 16 ft with reverse return, since the balancing-valve drop becomes negligible. If approximately 18 ft is maintained at AHU 1, design flow is available to all AHUs during any non-symetrical flow condition. With a reverse return system locate the differential pressure sensor across AHU 1 with a setpoint of 18 ft. Advantages of valve position load reset are much less with reverse return systems, however, resetting the setpoint from 18 ft to 8 or 10 ft based upon load is worthwhile on large systems.