You are on page 1of 8

ICR0207

ENERGY USE CHARACTERISTICS OF VARIABLE PRIMARY


FLOW CHILLED WATER PUMPING SYSTEMS

William P. BAHNFLETH, Eric B. PEYER*


Dept. of Architectural Engineering, The Pennsylvania State University, 104 Engineering Unit A,
University Park, PA 16802, USA, 814.863.2076, FAX: 814.863-4789, wbahnfleth@psu.edu
*Grumman/Butkus Associates, Evanston, IL, USA

ABSTRACT
The use of variable primary flow pumping in chilled water systems is increasing due to its perceived potential to
reduce energy consumption and initial cost. Neither the conditions under which significant energy savings are
realized nor the likely magnitude of savings are well documented. To address this deficiency, a program of
simulations was conducted that compared variable primary flow system energy use with that of other common
system types. Parameters varied included load type, number of chillers in the central plant, temperature difference
vs. part load characteristics, and climate. Results of these simulations show that for systems without temperature
differential problems substantial reductions in chilled water pumping energy are possible. However, this benefit
decreases as the number of chillers increases. The reduction in annual pumping energy due to variable primary flow
is generally between 25 and 50 percent. Relative to total plant energy use, this is a reduction of 2 to 5 percent.

INTRODUCTION
In simplest conceptual terms, a chilled water system for comfort or process cooling is comprised of three
subsystems: production, distribution, and use. Thermal loads on the use subsystem are met by cooling capacity
created in the production subsystem and delivered by the distribution subsystem. In most systems, chilled water is
produced by packaged chillers and utilized in coils and heat exchangers with water-side control by two-way or
three-way valves. The distribution system is a network of piping through which the distribution medium, typically
water or water-glycol solution, is transported by one or more levels of pumping.

The principal objectives of chilled water pumping system selection and design are to provide the required cooling
capacity to each load, to promote the efficient use of refrigeration capacity in the plant, and to minimize pump
energy consumption subject to whatever budgetary constraints may apply. In the typical design process, such
decisions are made on the basis of economic calculations. Accurate energy use prediction is an essential step in the
development of the operating cost component of such analyses.

(a) (b)

Figure 1: Primary-only chilled water systems. (a) Constant flow. (b) Variable flow.

The simplest chilled water pumping system has a single, primary, level of pumping that connects plants and loads
via a single pumping circuit. When primary-only systems are pumped at constant flow rate, it is typical to have
three-way control valves at loads (Figure 1a). When a primary-only system utilizes variable flow pumping, control

International Congress of Refrigeration 2003, Washington, D.C.


1
ICR0207

valves at loads are typically two-way and a low flow bypass is incorporated to maintain minimum flow through
chiller evaporators under low load conditions (Figure 1b).

A widely used alternative pumping concept is the separation of the chilled water system into two hydraulically
independent circuits: a primary system associated with the plant and a secondary system associated with distribution
and end use (Figure 2). This concept is taken to its logical limit in primary/secondary/tertiary systems, in which
separate pumping is also provided for the load.

(a) (b)
Figure 2: Constant flow primary, variable flow secondary chilled water system.
a) Conventional system b) Check valve in bypass pipe

Typically, primary flow is at a constant flow rate or in increments such that flow through active evaporators of
chillers is constant while secondary and tertiary flows vary in proportion to the cooling load to reduce the part load
energy consumption of pumps. Primary/secondary systems with constant primary flow have been converted into
variable primary flow systems by introducing a check valve into the bypass line to permit flow only from supply to
return (Avery 1998), a practice that have received universal approval from the design community (Rishel 1998).
Multiple chillers are generally connected in parallel as shown in Figures 1 - 3, although some plants employ series
arrangements.

Constant primary flow pumping has its origin in limitations of the controls of packaged chillers manufactured prior
to the mid 1980s. Equipment manufacturers discouraged variable primary flow applications of their chillers and the
habit of using it ingrained. For several decades, constant flow primary with variable flow secondary has been
considered the standard design for systems of any significant size.

However, within the past decade, the position of the primary/secondary system as the preferred system type has been
challenged on several grounds. One is that, in theory, variable primary flow results in minimum chilled water pump
energy usage. A second is that the initial cost of a primary-only system is lower than that of a primary/secondary
system. A third is that primary/secondary systems are subject to a complex of operational problems collectively
referred to as “low ∆T syndrome” (Kirsner 1996). Fourth, and perhaps most importantly, advances in packaged
chiller controls now make it possible to vary evaporator flow sufficiently rapidly and across a sufficiently large
range to make it a practical control approach (Eppelheimer 1996, Redden 1996). Manufacturers with a history of
favoring primary/secondary system structure are increasingly willing to support the application of variable primary
flow (Schwedler and Bradley 2000, 2002).

A comprehensive discussion of the issues associated with variable primary flow and its merits relative to other
systems is beyond the scope of the present paper. Several recent publications provide a broader view (Taylor, et al.
2000, Taylor 2002, Schwedler and Bradley 2002). Instead, attention is focused on the energy consumption
characteristics of variable and constant primary flow systems in order to draw quantitative conclusions regarding the
energy saving that can be expected from the use of variable primary flow. Such data, derived from measurement or
simulation, is very sparse in the published literature (Avery 2001, Schwedler and Bradley 2000, Bahnfleth and Peyer
2001). The present study uses detailed pumping system simulation to evaluate the effect of a wider range of
parameters than these sources and, therefore, supports more general conclusions.

International Congress of Refrigeration 2003, Washington, D.C.


2
ICR0207

1 STUDY PARAMETERS
Parameters considered in the study included:

• Chilled water system type: constant flow primary-only, constant flow primary/variable flow secondary,
variable flow primary-only, constant flow primary/variable flow secondary with bypass check valve
• Climate: hot-humid (Houston, TX), hot-dry (Phoenix, AZ), mild-humid (Syracuse, NY)
• Cooling load type: office building, medical facility, district cooling system
• Number of chillers: two to five equal sized units in parallel for district system, one to four for others
• Chilled water temperature difference (∆T): favorable (linear increase with decreasing load), neutral
(constant), unfavorable (linear decrease with decreasing load)

To limit complexity, only electric motor-driven water-cooled centrifugal chiller plants were considered. Not all
permutations of the above parameters (660 total combinations) were simulated. Preliminary investigations of the
influence of several parameters suspected to be of little significance were used to reduce the total to 348 cases.

2 CHILLED WATER SYSTEM MODEL


Annual quasi-static simulations with an hourly time step were performed using models developed with commercial
equation solving software (Klein and Alvarado 2001) using the same methods as a prior case study by the authors
(Bahnfleth and Peyer 2001). System models included central plant equipment (chillers, primary and secondary
chilled water pumps, cooling towers and condenser water pumps). Cooling loads were not modeled explicitly. The
effect of the piping system was represented by system curves. The selected climates were represented by Typical
Meteorological Year (TMY) weather data (Marion and Urban 1995). Cooling load profiles were generated with
public domain hourly whole-building energy analysis software (DOE 1980).

Polynomial models were used for major components, i.e., chillers, cooling towers and pumps. Chiller models were
of the same type used in the whole-building energy analysis program (DOE 1980). Cooling tower models were
developed from manufacturer’s data using a method described by Stoecker (1989). Pump head characteristic and
efficiency curves were developed from manufacturer’s data. Pump motor and drive models were regressions of
generic characteristics published by ASHRAE (2000). System design conditions used for component modeling are
shown in Table 1. In selecting pump head requirements, an effort was made to reflect typical, rather than best
practice.

Table 1: Chilled water (CHW) and condenser water (CW) system design data.
Load, CHW CHW CHW Leaving CW CW Entering
Load Range,
tons flow, Head, ∆T, CHW Temp. Flow, Head, CW Temp.
type ˚F(˚C)
(kW) gpm(L/s) ft(kPa) ˚F(˚C) ˚F(˚C) gpm(L/s) ft(kPa) ˚F(˚C)
Office 500 1,000 120 12 44 1,500 70 10 85
building (1,758) (63) (36) (6.7) (6.7) (95) (21) (5.6) (29.4)
Medical 1,500 3,000 150 12 44 4,500 100 10 85
facility (5,274) (189) (45) (6.7) (6.7) (284) (30) (5.6) (29.4)
District 4,500 9,000 170 12 44 13,500 100 10 85
plant (15,822) (568) (51) (6.7) (6.7) (852) (30) (5.6) (29.4)

Control algorithms were based on conventional design practices as documented through literature review and
interviews with the design engineers, manufacturers and system operators. All chillers and auxiliaries of a constant
flow primary-only system operated continuously at design flow to approximate three-way valve control. For other
system types, flow to the load was determined by the combined effect of chilled water temperature difference and
cooling load. In a constant flow primary/variable flow secondary system, a chiller was added in response to rising
system supply temperature and staged off when excess flow in the bypass equaled the flow of one chiller. Systems
with check valves could exceed the design evaporator flow rate of the constant speed primary pumps when not
capacity constrained. Chillers in a variable primary flow system were staged on the basis of the more restrictive of

International Congress of Refrigeration 2003, Washington, D.C.


3
ICR0207

available capacity or flow rate limits. The low flow bypass of the variable flow primary-only system operated as
necessary under low part load conditions

3 RESULTS
Simulations with Houston, TX weather are representative of results obtained in other climates and are emphasized in
this summary. Results for constant flow primary only systems are omitted in order to permit a fuller discussion of
primary/secondary and variable primary flow systems, which are of more interest.

Table 2 shows the annual energy consumption for the three study load types with a constant chilled water ∆T and
Houston weather. Values are given for the total plant energy use and chilled water pump energy. Percentage
differences as well as comparisons of energy per unit of peak load are given. The differences are calculated relative
to the constant flow primary/variable flow secondary system’s annual energy consumption.

The variable flow, primary-only system type provided between 2 and 5 percent savings in energy consumption
relative to the constant flow primary/variable flow secondary system. Savings varied mainly with the number of
chillers, and was varied little across load type. The addition of a check valve to the primary/secondary system
resulted in a savings of less than 1 percent for all load types.

Table 2: Effect of load type on the annual energy consumption of the Houston cases.

Constant flow Primary/ secondary Variable flow


primary/ variable with a check valve primary-only
flow secondary
3,157(898) to 3,160(899) to 2,568 3,013(857) to
Total plant energy, kWh/ton(kWh/kW)1
2,587(736) (730) 2,517(716)

Office ∆ plant energy, %


1
Base 0 to -1 -5 to -3
building CHW pump energy, kWh/ton(kWh/kW)1 343(98) to 223(63) 345(98) to 219(62) 199(57) to 170(48)
∆ CHW pump energy, % 1
Base 1 to -2 -42 to -24
∆ chiller aux. energy, % 1
Base 0 to -4 0 to -4
3,340(950) to 3,343(951) to 3,189(907) to
Total plant energy, kWh/ton(kWh/kW)1
2,632(749) 2,611(743) 2,579(734)

Medical ∆ plant energy, %


1
Base 0 to -1 -5 to -2
facility CHW pump energy, kWh/ton(kWh/kW)1 349(99) to 220(63) 351(100) to 217(62) 197 (56) to 187 (53)
∆ CHW pump energy, %1 Base 1 to -1 -44 to -15
∆ chiller aux. energy, % 1
Base 0 to -4 0 to -4
3,420(973) to 3,373(959) to 3,284(934) to
Total plant energy, kWh/ton(kWh/kW)1
3,189(907) 3,148(895) 3,120(887)

District ∆ plant energy, %


1
Base -1 -4 to -2
plant CHW pump energy, kWh/ton(kWh/kW)1 334(95) to 287(82) 322(92) to 276(78) 233(66) to 248(71)
∆ CHW pump energy, % 1
Base -4 -30 to -14
∆ chiller aux. energy, %1 Base -4 to -5 -4 to -5
Note: 1Ranges represent kWh per design ton (kW) for all four study chiller configurations

International Congress of Refrigeration 2003, Washington, D.C.


4
ICR0207

Table 3 shows the annual energy consumption for the office building cases with constant ∆T vs. cooling load model.
As expected, total plant energy consumption was greater for the systems modeled in Houston because of the longer
cooling season. The result was that variable primary flow savings relative to that of constant flow primary/variable
flow secondary systems were greater for Houston cases by roughly a factor of two. However, the qualitative
differences between system types were not significantly impacted by climate. Component energy savings associated
with the variable primary flow systems, increased in proportion to with plant energy savings.

Table 3: Effect of climate on annual energy consumption of the office building cases.

Constant flow
Primary/ secondary Variable flow
primary/ variable
with a check valve primary-only
flow secondary
Total plant energy, kWh/ton(kWh/kW)1 3,157(898) to 3,160(899) to 2,568 3,013(857) to
2,587(736) (730) 2,517(716)
∆ plant energy, kWh/ton(kWh/kW)1 Base 2(1) to -20(-6) -144(-41) to
-70(-20)
Houston
CHW pump energy, kWh/ton(kWh/kW)1 343(98) to 223(63) 345(98) to 219(62) 199(57) to 170(48)
∆ CHW pump energy, kWh/ton(kWh/kW) 1
Base 2(1) to -4(-1) -144(-41) to
-53(-15)
∆ chiller aux. energy, kWh/ton(kWh/kW)1 Base 0 to -19(-5) 0 to -20(-6)
1
Total plant energy, kWh/ton(kWh/kW) 1,589(452) to 1,591(453) to 1,501(427) to
1,211(344) 1,204(342) 1,107(315)
∆ plant energy, kWh/ton(kWh/kW)1 Base 1(0) to -7(-2) -88(-25) to -41(-12)
Syracuse
CHW pump energy, kWh/ton(kWh/kW)1 196(56) to 111(32) 197(56) to 109(31) 107(30) to 76(22)
∆ CHW pump energy, kWh/ton(kWh/kW) 1
Base 1(0) to -2(-1) -88(-25) to -36(-10)
∆ chiller aux. energy, kWh/ton(kWh/kW) 1
Base 0 to -9(-3) 0 to -9(-3)
1
Note: Ranges represent values for all four study chiller configurations

Table 4 shows the effect of varying the number of chillers in the plant on annual energy consumption for the
Houston office building with constant chilled water ∆T. The annual energy consumption for each system type
decreased with increasing number of chillers because staging of chillers in a multiple chiller plant reduced the
energy consumption of auxiliaries.

While auxiliary energy consumption was strongly affected by the number of chillers in the plant, annual chiller
energy consumption varied little. The single chiller case, however provided no opportunity for chiller auxiliary
energy savings, either, because auxiliaries had to operate continuously.

Figure 3 shows the annual energy savings for the variable flow, primary-only and primary/secondary check valve
systems relative to a constant flow primary/variable flow secondary system. Variable flow, primary-only savings
diminished with increasing numbers of chillers. Chilled water pump energy diminished, while the chiller auxiliary
energy remained relatively constant in all multiple chiller cases. Energy savings for variable primary flow with a
single chiller are less than half as large in a plant with four chillers. It is also evident that the true variable primary
flow system is much more effective at reducing pump energy than the check valve retrofit to the primary secondary
system.

International Congress of Refrigeration 2003, Washington, D.C.


5
ICR0207

Table 4: Effect of number of chillers on annual energy consumption of the Houston office building case.

Constant flow
Primary/secondary Variable flow
primary/ variable
check valve primary-only
flow secondary
Total plant energy, kWh/ton (kWh/kW) 3,157 (898) 3,160 (899) 3,013 (857)
1 chiller ∆ plant energy, kWh/ton (kWh/kW) Base 3 (1) -144 (-41)
∆ plant energy, % Base 0 -5
Total plant energy, kWh/ton (kWh/kW) 2,730 (776) 2,708 (770) 2,618 (745)
2 chillers ∆ plant energy, kWh/ton (kWh/kW) Base -22 (-6) -112 (-31)
∆ plant energy, % Base -1 -4
Total plant energy, kWh/ton (kWh/kW) 2,636 (750) 2,613 (743) 2,550 (725)
3 chillers ∆ plant energy, kWh/ton (kWh/kW) Base -23 -86 (-25)
∆ plant energy, % Base -1 -3
Total plant energy, kWh/ton (kWh/kW) 2,587 (736) 2,568 (730) 2,517 (716)
4 chillers ∆ plant energy, kWh/ton (kWh/kW) Base -19 (-6) -70 (-20)
∆ plant energy, % Base -1 -3

Figure 3: Annual energy savings for (a) variable flow, primary-only and (b) primary/secondary check valve
systems relative to a constant flow primary/variable flow secondary system

Table 5 shows the effect of ∆T vs. load characteristics on annual energy consumption for the Houston office
building. The favorable ∆T model reduced the total plant energy consumption by no more than 1 percent. These
small savings were due to a decrease in variable speed pumping energy. Because the chilled water pump energy is a
relatively small portion of the plant, the fractional energy savings were small.

International Congress of Refrigeration 2003, Washington, D.C.


6
ICR0207

Chillers and chiller auxiliaries did not contribute significantly to variable primary flow savings in the favorable and
constant ∆T cases. For this reason the addition of a check valve to the primary/secondary system provided
negligible energy savings for these cases.

Systems experiencing less-than-design ∆T’s consumed more chilled water pump energy and were likely to have
greater chiller and chiller auxiliary energy use. Variable primary flow chilled water pump energy savings relative to
the primary/secondary system was less in the unfavorable ∆T case, while chiller auxiliary savings increased.

The result was that for multiple chiller configurations of the variable flow, primary-only system the total plant
energy savings increased from 3 to 7 percent. However, in the single chiller configuration the total plant energy
savings decreased from 5 to 4 percent due to the resulting drop in chilled water pump energy savings.

Table 5: Effect of ∆T characteristics on annual energy consumption for the Houston office building cases.

Constant flow Primary/


Variable flow
primary/ variable secondary with a
primary-only
flow secondary check valve
3,150 (896) to 3,152 (896) to 3,005 (855) to
Total plant energy, kWh/ton (kWh/kW)1
2,553 (726) 2,554 (726) 2,481 (706)
Favorable
∆T ∆ total plant energy, % 1 Base 0 -5 to -3
∆ CHW pump energy, % 1
Base 0 to 1 -43 to -35
3,157 (898) to 3,160 (899) to 3,013 (857) to
Total plant energy, kWh/ton (kWh/kW)1
2,587 (736) 2,568 (730) 2,517 (716)

Constant ∆T ∆ total plant energy, % 1 Base 0 to -1 -5 to -3


∆ CHW pump energy, % 1 Base 1 to -2 -42 to -24
∆ chiller aux. energy, % 1 Base 0 to -4 0 to -4
3,194 (908) to 3,196 (909) to 3,062 (871) to
Total plant energy, kWh/ton (kWh/kW)1
2,836 (807) 2,725 (775) 2,666 (758)
Unfavorable ∆ total plant energy, % 1 Base 0 to -4 -4 to -7
∆T
∆ CHW pump energy, % 1 Base 1 to -5 -35 to -24
∆ chiller aux. energy, % 1 Base 0 to -13 0 to -13
1
Note: Ranges represent values for all four study chiller configurations.

CONCLUSIONS
Variable flow, primary-only chilled water systems saved 2 to 7 percent of total plant energy relative to
primary/secondary systems depending on the number of chillers in the plant, the ∆T vs. load characteristic of the
system, and the ratios of chilled water pump and chiller auxiliary energy to total plant energy. Savings in chilled
water pumping energy were as large as 50 percent. Increasing the number of chillers in the plant reduced the energy
savings of variable primary flow systems.

For the range of parameters considered, the addition of a check valve in the bypass of a conventional
primary/secondary system reduced total plant energy consumption by no more than 4 percent. This savings resulted
mainly from reduced chiller auxiliary energy during unfavorable chilled water ∆T’s. The check valve yielded little
or no benefit in a single chiller or when chilled water ∆T’s were at or above design values.

International Congress of Refrigeration 2003, Washington, D.C.


7
ICR0207

ACKNOWLEDGMENT
The support of this research through Air-Conditioning and Refrigeration Technology Institute ARTI 21-CR Project
Number 611-20070 Variable Primary Flow Chilled Water Systems: Potential Benefits and Application Issues is
gratefully acknowledged.

REFERENCES
ASHRAE. 2000. ASHRAE Handbook—2000 HVAC Systems and Equipment. Atlanta: American Society of
Heating, Refrigerating, and Air-Conditioning Engineers.
Avery, G. 1998. Controlling Chillers in Variable Flow Systems, ASHRAE Journal, 40(2): 42-45.
Avery, G. 2001. Improving the Efficiency of Chilled Water Plants, ASHRAE Journal, 43(5): 14-18.
Bahnfleth, W. P. and E. Peyer. 2001. A Comparative Analysis of Variable and Constant Primary-Flow Chilled-
Water-Plant Performance. HPAC Engineering, 73(5): 41-50.
DOE. 1980. Department of Energy (DOE) DOE2 Reference Manual, Part 1, version 2.1. Lawrence Berkeley
National Laboratories. Berkeley, CA.
Eppelheimer, D. M. 1996. Variable Flow—The Quest for System Energy Efficiency. ASHRAE Transactions:
Symposia. 102(2): 673-678.
Kirsner, W. 1996. The Demise of the Primary-Secondary Pumping Paradigm for Chilled Water Plant Design.
HPAC Engineering, 68(11): 73-79.
Klein, S. and F. Alvarado. 2002. EES: engineering equation solver. Middleton, WI. F-Chart Software.
Redden, G. H. 1996. Effect of Variable Flow on Centrifugal Chiller Performance. ASHRAE Transactions:
Symposia, 102(2): 684-687.
Rishel, J. B. 1998. System Analysis vs. Quick Fixes for Existing Chilled Water Systems. HPAC Engineering, 70(1):
131-134.
Schwedler, M. and B. Bradley. 2000. Variable-Primary-Flow Systems… An Idea for Chilled-Water Plants the Time
of Which Has Come. HPAC Engineering, 72(4): 41-44.
Schwedler, M. and B. Bradley. 2002. The Saga Continues… Variable Primary Flow Systems Revisited. Trane.
Engineers Newsletter 31(4): 1-8.
Taylor, S., P. DuPont, M. Hydeman, and C. Eley. 2000. CoolToolsTM Energy Efficient Chilled Water Plant Design
and Performance Guide. Pacific Gas & Electric Company.
Taylor, S. 2002b. Piping Chillers to Variable Volume Chilled Water Systems. ASHRAE Journal, 36(7): 43-45.

International Congress of Refrigeration 2003, Washington, D.C.


8

You might also like