You are on page 1of 22

MAIN PAGE

TABLE OF CONTENTS
4223

Analysis of Transfer Pumping Interfaces


for Stratified Chilled Water Thermal Storage
SystemsPart 2: Parametric Study
William P. Bahnfleth, Ph.D., P.E.

Christopher G. Kirchner, P.E.

Member ASHRAE

Associate Member ASHRAE

ABSTRACT
Transfer pumping interfaces are an important component
of chilled water storage systems and other open hydronic
systems. This paper describes a parametric study of seven
interface types applied to a representative chilled water storage application. Interface types considered include indirect
(heat exchanger) interfaces, direct interfaces, direct interfaces
with energy recovery operated at variable or constant speed,
and reversible variable-speed direct interfaces. Annual simulations on an hourly time step were performed using component models described in Part 1 of this paper. Seven levels of
static head, varying from 90 ft to 360 ft (27 m to 110 m), were
considered for each alternative. Construction costs were estimated and economic performance was described in terms of
simple payback period and life-cycle cost. Results of the study
suggest that energy recovery for direct interfaces may be feasible for static head as low as 120 ft (37 m) and that indirect interfaces may not be economically superior except at very high
levels of static head due to high capital cost
INTRODUCTION
Stratified chilled water thermal storage is frequently
incorporated into large chilled water systems because of the
substantial economic and operational benefits it offers. For
reasons of cost, stratified storage tanks are typically unpressurized vessels. For aesthetic reasons, tank height is typically
limited to such an extent that the free surface of water in the
tank is not the highest point in the chilled water system. When
this is the case, a pumping interface is required to move water
from the tank to the system because of the adverse static pressure differential that exists between them. Figure 1 schematically indicates the relationship between a typical open thermal

storage tank, chilled water system, and transfer pumping interface. In addition to controlling pressure and flow rate, a transfer pumping interface also controls flow direction through the
piping that connects it to the chilled water system. A transfer
pumping interface adds capital cost to a chilled water storage
system and can be a major energy consumer if the static pressure differential is large.
The economics of transfer pumping depend strongly upon
the size of the static pressure differential. When this differential is small, the additional energy consumption it causes may
be too small to warrant special energy conservation measures.
However, as the static differential becomes larger, growing
operating cost may justify additional capital expenditures to
limit pump energy usage.
One alternative for large adverse static differential situations is to eliminate the problem by using a heat exchanger to
separate the system into hydraulically independent low- and
high-pressure zones. Another option is the use of hydraulic
turbines in conjunction with a direct interface to recover a
portion of the added pumping power from water flowing back
to the storage tank. Reverse-running centrifugal pumps are
typically used as hydraulic turbines in this application because
they are lower in cost than specially built turbines and have
good performance characteristics when applied with care
(Buse 1981). The heat exchanger interface usually reduces
pumping costs, but it also increases storage tank size. This
trade-off tends to favor the use of energy recovery at moderately large static head differentials and the use of a heat
exchanger interface when static head is very large.
The system designer must select an appropriate interface
type based on reasonable technical and economic criteria.
Tackett (1988) compared the performance of a direct interface
with energy recovery to that of an indirect interface for a high-

William P. Bahnfleth is an assistant professor in the Department of Architectural Engineering at Penn State University, University Park, Pa.
Christopher G. Kirchner is a project mechanical engineer at SHG, Inc., Detroit, Mich.
THIS PREPRINT IS FOR DISCUSSION PURPOSES ONLY, FOR INCLUSION IN ASHRAE TRANSACTIONS 1999, V. 105, Pt. 1. Not to be reprinted in whole or in
part without written permission of the American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc., 1791 Tullie Circle, NE, Atlanta, GA 30329.
Opinions, findings, conclusions, or recommendations expressed in this paper are those of the author(s) and do not necessarily reflect the views of ASHRAE. Written
questions and comments regarding this paper should be received at ASHRAE no later than February 13, 1999.

BACK TO PAGE ONE

Figure 1 Typical chilled water system with open stratified chilled water storage tank and transfer pumping interface.
rise office building with a thermal storage tank located in its
basement. He concluded that energy recovery should be
considered whenever the use of an indirect interface might
also be feasible. Tackett also suggested as a specific guideline
that direct pumping without energy recovery is the most
economical alternative below a twenty-story differential.
Tacketts study, which is the only published analysis of its
type, considering only constant-speed pump and turbine operation, was based only on simple payback analysis and considered a thermal storage system that is atypical in a number of
significant respects. Consequently, a clear need exists to
extend this analysis to include variable-speed pumping, lifecycle cost analysis, and more representative storage systems.
The objective of the present work is to provide a detailed
analysis of transfer pumping economics for a system representative of chilled water storage installations designed in
accordance with current good practice. The performance of
seven interface types is compared in a typical district cooling
application with the adverse static pressure differential varying from 90 ft to 360 ft (27.4 m to 109.7 m) of water. Hourly
time step annual simulations to determine energy consumption and cost characteristics of each interface are performed
using models described in detail in Part 1 of this paper (Bahnfleth and Kirchner 1998). Simulation results and capital cost
estimates are used to determine life-cycle cost and simple
payback periods for each interface type as a function of static
pressure differential. Because of the number of interface types
and static pressure differential levels evaluated, only one
representative load profile and electric rate structure were
considered. However, the results of this analysis are quite clear
in their qualitative implications for interface design and can be
generalized on that basis.
INTERFACE ALTERNATIVES
The following discussion briefly reviews the classification of transfer pumping interfaces and summarizes the alter2

natives considered in this study. A more detailed discussion


may be found in Part 1.
At the most fundamental level, thermal storage transfer
pumping interfaces can be classified as indirect or direct,
depending upon whether the interface does or does not isolate
storage hydraulically from the chilled water system. Direct
interfaces, which permit flow between storage and chilled
water system, may be further classified on the basis of whether
they do or do not incorporate energy recovery using hydraulic
turbines. A typical direct interface with energy recovery is
unidirectional, with dedicated pumps and turbines. However,
reversible interfaces in which pumps and turbines exchange
roles when the storage system mode changes from charging to
discharging have been discussed in the literature and have
been implemented in a large system (Fiorino 1994). For the
purpose of the present study, therefore, four basic interface
types (indirect, direct, direct with energy recovery, and reversible direct with energy recovery) are defined, each of which
may operate at constant or variable speed.
Figures 2 through 5 illustrate typical configurations for
each basic interface type. In each figure, a single pump symbol
may represent a set of parallel pumps.
The indirect interface (Figure 2) thermally connects two
independent hydraulic circuits via a plate-and-frame heat
exchanger. Each circuit has its own constant- or variablespeed pump and four flow-reversing control valves.
The main components of the direct interface (Figure 3)
are its transfer pump, four flow-reversing control valves, and
two pressure-sustaining valves (PSVs). The control valves
operate as in the indirect interface to ensure the proper flow
direction to and from the tank. Pressure-sustaining valves
throttle away excess static pressure of flow returning to the
tank by controlling the static pressure of the chilled water
system.
The direct interface with energy recovery (Figure 4) is
similar, in most respects, to the direct interface. However, the
4223

BACK TO PAGE ONE

Figure 2 Indirect pumping interface.

Figure 3 Direct pumping interface.

4223

BACK TO PAGE ONE

Figure 4 Direct pumping interface with hydraulic energy recovery turbine.

Figure 5 Reversible direct pumping interface.

4223

BACK TO PAGE ONE


pump is connected to an energy recovery turbine through a
double extended motor shaft. The turbine recovers part of the
energy that would have been wasted by the PSV in a direct
interface and transfers it to the motor shaft. Excess static pressure is throttled away by a PSV in series with the turbine. A
clutch prevents the turbine from becoming a load on the motor.
Eight flow-reversing control valves are required. A turbine
bypass valve is provided in order to permit operation with the
turbine off-line. Because the best efficiency point (BEP) of a
pump generally occurs at a lower flow and head than its BEP
as a turbine, the pump and turbine of a direct interface with
energy recovery will tend to differ, with the pump being the
larger of the two (Buse 1981).
The reversible direct interface is shown in Figure 5. Like
the unidirectional interface with energy recovery, the reversible interface has pump and turbine sets coupled mechanically
through the motor shaft. The double-extended shaft electric
motor connecting the pumps is automatically braked and
reversed by a reversible variable-frequency drive. Unlike the
unidirectional case, the reversible pump and turbine are identical if charge and discharge design points are similar. The
pumps selected for this type of interface cannot be optimized
strictly for either pump or turbine duty. Rather, they must
provide an acceptable compromise between performance in
both modes.
Given that each of the four basic interface types can operate at constant or variable speed, eight interface configurations
are possible. However, selection of suitable components for a
constant-speed reversible interface proved very difficult and
preliminary studies indicated that the performance of this type
was poor. Selection is difficult because the pump/turbine must
perform reasonably well in both modes. In the scenario
considered in this study, design flow and head conditions for
both pump and turbine were comparable, as they would be in
many applications. It was found that constant-speed pump
selections with acceptable performance in pump duty recovered very little power when operating as turbines under the
conditions assumed. This problem was not as severe in the
case of variable-speed operation and suitable selections could
be made in most cases. Therefore, only the variable-speed
reversible interface was included in the parametric study. The
seven interface alternatives and the acronyms used to identify
them concisely are listed in Table 1 for convenient reference.
THERMAL STORAGE SYSTEM SIMULATION
The annual performance of each interface alternative was
simulated using spreadsheet programs employing detailed
component models that are described in Part 1 of this paper.
The simulation was driven by an hourly cooling load profile
representing typical district cooling system loads in a coolingintensive climate. Thermal storage was sized on the basis of a
load-leveling partial storage control strategy.
Cooling Load and Thermal Storage Sizing
Annual (8760 hour) and design day cooling load profiles
used to drive the simulation were generated by a public
4223

TABLE 1
Summary of Interface Types
Description
Indirect, Constant Speed

Acronym
IC

Indirect, Variable Speed

IV

Direct, Constant Speed

DC

Direct, Variable Speed

DV

Direct with Energy Recovery, Constant Speed

DCT

Direct with Energy Recovery, Variable Speed

DVT

Reversible Direct with Energy Recovery, Variable Speed

RDVT

domain energy analysis program (BLAST 1991). Loads were


calculated for a prototype building located in Dallas, Texas,
and scaled up to a peak load typical of a large district cooling
system. The design day load profile, which had a peak of
24,226 tons (85,276 kW) and an integrated load of 439,155
ton-h (1,545,825 kWh), is shown in Figure 6. Also shown is
the design day level load chiller profile, which results in a storage tank with a capacity of 42,432 ton-h (149,361 kWh). The
maximum instantaneous storage charge/discharge capacity
was fixed at 6000 tons (21,120 kW), slightly larger than the
peak required on the design day.
Chilled water system return and charging supply temperatures were assumed to be constant at 60F (15.6C) and 39F
(3.9C), respectively. Based on this 21F (12C) temperature
differential and a 0.9 figure of merit (the fraction of theoretical
storage capacity that can be discharged from a fully charged
tank), the tank volume required for the direct interface alternatives was 3,232,950 gal (12,252,881 L). The volume
required for indirect interfaces was 3,993,645 gal (15,135,915
L) based on a reduced storage temperature differential of 17F
(9.4C), resulting from the use of a heat exchanger with a fullload temperature differential between entering cool-side flow
and leaving warm-side flow of 2F (1.1C). The heat
exchanger elevates the charged temperature of storage by 2F
(1.1C) and lowers its discharged temperature by 2F (1.1C),
thereby lowering storage density and increasing storage
volume by 19% relative to a direct interface. The reduced
temperature differential across storage also necessitates
increased peak pump flow rates for indirect interfaces. These
tank size and peak flow rate penalties are serious impediments
to the economic feasibility of an indirect interface.
Thermal Storage Control
The thermal storage discharge algorithm was structured
to utilize storage completely on a daily cycle while keeping the
smallest possible number of chillers on-line and fully loaded
during on-peak hours. The fully charged thermal storage tank
was assumed to have an available capacity equal to its design
day capacity. This is consistent with the assumption that
chilled water system supply and return temperatures remain
essentially constant. It is recognized that temperature differential in many systems varies seasonally, either by design or
5

BACK TO PAGE ONE

Figure 6 Design day cooling load profile and level load thermal storage system operation.
by accident. However, because all interfaces were subjected to
the same loadings and flow rates, it was felt that this assumption would not diminish the significance of pumping energy
comparisons. Discharge capacity needed to serve the system
load could be deducted from storage as long as the inventory
remained positive. For the purpose of implementing the thermal storage operating strategy, it was assumed that the chilled
water plant was composed of multiple 3,000 ton (10,560 kW)
chillers and that the chiller capacity on-line during discharge
was fully loaded.
The thermal storage interface was modeled explicitly, but
the plant was not. Refrigeration plant modeling was not necessary because system-side charging and return temperatures
were assumed to be the same for all interface types. This
implies that, for indirect interface cases, a 43F system-side
discharge temperature is acceptable. For many systems, this
would be satisfactory, but in some instances, use of an indirect
interface would make it necessary to lower the system-side
charging temperature in order to achieve the same discharge
temperature as the direct case. The resulting compressor
energy penalty in such cases should be charged to the indirect
interface, which would reduce its economic attractiveness.
Chiller charge and discharge capacity requirements were
computed on the basis of design thermal storage tank capacity
and daily integrated cooling load. Each interface was assumed
to have sets of three identical parallel transfer pumps. The
minimum increment of storage capacity dispatched was 800
tons (2,814 kW), based on a criterion of limiting minimum
continuous transfer pump flow rate to approximately 40% of
the capacity of one pump.
The sizing and operation of storage were based on the
simplified electric rate structure summarized in Table 2. It is
6

TABLE 2
Electric Utility Rate Summary
On-Peak Rates*
Winter

Summer

Usage ($/kWh)

0.053

0.081

Demand ($/kW

8.07

16.35

Winter

Summer

Usage ($/kWh)

0.042

0.042

Demand ($/kW

0.000

0.000

Off-Peak Rates

On Peak: 8 a.m. to 8 p.m.


Summer: June to August

a composite time-of-use high-voltage rate derived from the


characteristics of a number of large urban utility companies
whose rates were reviewed. This rate incorporates most of the
features commonly found in utility rate structures, including
on-peak and off-peak periods, separate demand and energy
charges, and seasonal variation. Because weekends and holidays are typically defined as off-peak, the thermal storage
system was not discharged on those days.
INTERFACE DESIGN
In order to estimate interface cost and determine frictional
pump head requirements, it was necessary to develop a
detailed design for each interface alternative (Kirchner 1997).
The system layouts and design calculation procedures utilized
for each interface reflect typical practice. As noted previously,
each interface had three horizontal split case pumps or pump/
4223

BACK TO PAGE ONE


turbine sets piped in parallel in order to provide a balance
between system complexity, energy consumption, first cost,
and flow rate turndown capability. Each interface was
assumed to connect to the primary/secondary chilled water
system near a constant pressure point at the bypass, as shown
in Figure 1. This is an advantageous location for a direct interface (Bahnfleth 1995).
Design frictional head loss for direct alternatives was 25
ft (8 m) of water for pumping and 30 ft on the return side for
systems with energy recovery turbines. For indirect alternatives, frictional head loss was 55 ft (17 m) on the system side
and 66 ft (20 m) on the tank side, of which 35 ft (11m) on each
side was due to the heat exchanger. The higher frictional head
loss on the tank side of the indirect interface is due to additional piping length assumed to exist between the storage tank
and the heat exchanger. The static pressure differentials used
to size direct interface pumps represent a range of adverse
differentials from moderately high to very high. The values
considered were 90 ft, 130 ft, 180 ft, 220 ft, 260 ft, 310 ft, and
360 ft (27 m, 40 m, 55 m, 67 m, 79 m, 94 m, and 110 m) of
water, which correspond roughly to elevations of 7, 10, 14, 17,
20, 24, and 28 stories above the tank water level.
Pump and Turbine Selection
Pumps and turbines were selected from the data of a
prominent U.S. manufacturer. Pumps for direct interfaces
were selected to provide 2,300 gpm (145 L/s) each at the
design point, for a total flow rate of 6,900 gpm (435 L/s).
Pumps for indirect interfaces were selected to provide 2,800
gpm (177 L/s) each for a total flow rate of 8,400 gpm (530
L/s). Pump best efficiency point (BEP) characteristics are
shown in Table 3. Energy recovery turbines were selected
from pump data using the BEP method described in Part 1.
The resulting pump selections for turbine operation are
shown in Table 4. Typical pump and turbine performance

curves for direct interfaces with 180 ft and 310 ft (55 m and
95 m) of static pressure differential are shown in Figures 7
and 8. Pump curves shown in these figures are manufacturers data; turbine curves were generated using techniques
described in Part 1.
Selection of pumps/turbines for a reversible interface is
highly constrained. The required component must not only
provide the necessary flow when functioning as a pump, it
must also recover a significant amount of energy across a
range of flow rates when operating as a turbine. Satisfactory
selections were not found in the manufacturers line for the
constant-speed case. Since the equipment line of the manufacturer is quite extensive, it was concluded that the difficulties
encountered are likely to be generic. With the exception of the
90 ft (27 m) case, however, it was possible to select pumps that
would also function adequately as turbines at variable speed.
Pump/turbine selections for reversible interface cases are
shown in Table 5.
Capital Cost
Capital cost estimates were developed to determine the
investment required for each alternative and are summarized
in Table 6. Estimates included the complete interface from the
tank to the connection point on the primary/secondary chilled
water system and the cost of the thermal storage tank. Cost
data for piping and piping system components were obtained
from a standard source (Means 1996). The costs of pumps,
turbines, clutches, heat exchangers, and variable-speed drives
were obtained directly from equipment vendors. The cost of
the clutch included in turbine costs for each of the unidirectional interfaces (Table 6a) was $8,000. Variable-speed pump
costs in Table 6a include variable-frequency drive costs ranging from $13,165 to $22,956 as a function of horsepower. The
reversing option for the drives in the RDVT interface added
$3,900 to $7,500, also depending on the drive horsepower. A

TABLE 3
Pump Best Efficiency Point Data
Static Pressure

Flow Rate

Pressure

Speed
(rpm)

Specific
Speed*

Brake Power

Motor Power

(m)

Eff.
(%)

(Bhp)

(kW)

(Mhp)

(kW)

75

23

88

1780

3506

54

40

75

56

159

88

26

88

1780

3110

64

47

75

56

2070

131

125

38

88

1780

2166

74

55

100

75

(ft of water)

(m)

Interface
Type

Indirect

90

27

130

40

2070

131

170

51

88

1780

1720

101

75

125

93

180

55

2070

131

220

66

88

1780

1418

131

97

150

112

220

67

2070

131

260

78

81

1750

1230

168

125

200

149

260

79

2070

131

300

90

83

1750

1105

189

141

200

149

310

94

2070

131

350

105

80

1750

984

229

171

250

186

360

110

2070

131

400

120

80

1750

890

261

195

300

224

Direct

(gpm)

(L/s)

(ft of water)

2520

159

2520

Defined in IP units (flow in gpm, head in ft, speed in rpm).

4223

BACK TO PAGE ONE

TABLE 4a
Turbine Selection Data for Unidirectional Interfaces (Turbine Best Efficiency Point Data)
Static Pressure

Flow Rate

Pressure

(ft of water)

(m)

(gpm)

(L/s)

(ft of water)

(m)

Eff. (%)

Specific Speed*

90

27

2070

131

55

17

79

4010

130

40

2070

131

90

27

79

2772

180

55

2070

131

135

41

88

2045

220

67

2070

131

170

51

87

1691

260

79

2070

131

200

60

88

1497

310

94

2070

131

240

72

86

1306

360

110

2070

131

275

83

84

1179

Defined in IP units (flow in gpm, head in ft, speed in rpm).

TABLE 4b
Turbine Selection Data for Unidirectional Interfaces
(Pump Best Efficiency Point Selection Data for Turbine Operation)
Static Pressure

Flow Rate

Pressure

Eff. (%)

(ft of water)

(m)

(gpm)

(L/s)

(ft of water)

(m)

90

27

1715

108

41

12

79

130

40

1715

108

68

20

79

180

55

1870

118

116

35

88

220

67

1850

117

144

43

87

260

79

1870

118

172

52

88

310

94

1835

116

200

60

86

360

110

1800

114

223

67

84

TABLE 5
Reversible Interface Pump and Turbine Selections
Static Pressure

Pump Best Efficiency Point


Flow Rate

Pressure

Brake Power

Motor Power

(ft of
water)

(m)

(gpm)

(L/s)

(ft of
water)

(m)

Eff.
(%)

Speed
(rpm)

Specific
Speed*

(Bhp)

(kW)

(Mhp)

(KW)

90

27

130

40

5000

315

230

69

89

1780

2131

326

243

150

112

180

55

5600

353

230

69

91

1780

2255

357

267

200

149

220

67

5150

325

250

75

84

1750

1997

387

289

200

149

260

79

2750

173

220

66

90

1750

1607

170

127

250

186

310

94

2750

173

300

90

84

1750

1273

248

185

300

224

360

110

2750

173

330

99

84

1750

1185

273

203

350

261

Turbine Best Efficiency Point

90

27

130

40

5489

346

265

79

79

1780

2011

180

55

6039

381

258

77

88

1780

2151

220

67

5921

374

308

92

87

1750

1831

260

79

2992

189

250

75

88

1750

1524

310

94

3162

199

370

111

86

1750

1167

360

110

3162

199

407

122

84

1750

1086

Defined in IP units (flow in gpm, head in ft, speed in rpm).

4223

BACK TO PAGE ONE

Figure 7 Direct interface pump and turbine selections180 ft (55 m) static pressure differential.

Figure 8 Direct interface pump and turbine selections310 ft (95 m) static pressure differential.

4223

BACK TO PAGE ONE


TABLE 6a
Cost Summary: Pump and Turbine Cost
Pump and Turbine ($)*

Static Pressure
(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

RDVT

90

27

37,590

123,891

28,500

72,094

63,900

107,494

97,744

130

40

37,590

123,891

34,500

81,896

69,900

117,296

125,969

180

55

37,590

123,891

40,155

95,924

75,555

131,324

140,273

220

67

37,590

123,891

48,900

110,918

86,100

148,118

151,418

260

79

37,590

123,891

48,900

110,918

86,100

148,118

162,266

310

94

37,590

123,891

60,900

128,366

94,800

162,266

191,666

360

110

37,590

123,891

70,800

144,566

104,700

178,466

206,966

All values include installation cost. Variable-speed drive and clutch prices are included where applicable.

TABLE 6b
Cost Summary: Component Cost
Storage Tank ($)*
Item
Storage Tank

IC

IV

DC

DV

DCT

DVT

RDVT

1,477,649

1,477,649

1,196,192

1,196,192

1,196,192

1,196,192

1,196,192

147,765

147,765

119,619

119,619

119,619

119,619

119,619

1,625,414

1,625,414

1,315,811

1,315,811

1,315,811

1,315,811

1,315,811

Mark-Up (10%)
Total

Other Components ($)


Piping and Valves

218,262

218,262

179,316

179,316

208,708

208,708

200,783

Heat Exchanger

500,000

500,000

NA

NA

NA

NA

NA

718,262

718,262

179,316

179,316

208,708

208,708

200,783

Total
*

All values include installation cost.

TABLE 6c
Cost Summary: Total Capital Cost
Total Capital Cost ($)*

Static Pressure
(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

RDVT

90

27

2,822,683

2,959,383

1,644,991

1,714,044

1,747,622

1,816,675

130

40

2,822,683

2,959,383

1,654,495

1,729,570

1,757,126

1,832,202

1,833,385

180

55

2,822,683

2,959,383

1,663,452

1,751,790

1,766,084

1,854,422

1,856,043

220

67

2,822,683

2,959,383

1,677,304

1,775,541

1,782,787

1,881,023

1,873,696

260

79

2,822,683

2,959,383

1,677,304

1,775,541

1,782,787

1,881,023

1,890,880

310

94

2,822,683

2,959,383

1,696,312

1,803,178

1,796,568

1,903,434

1,937,449

360

110

2,822,683

2,959,383

1,711,994

1,828,839

1,812,249

1,929,094

1,961,684

Includes markup on all items other than the storage tank (based on 10% contingency, 20% general conditions, and 20% overhead and profit).

manufacturer quoted a budget price of $475,000 ($118,750


per exchanger) for the four plate-and-frame exchangers in the
indirect interface.
Storage tank costs were based on representative
published cost data (Dorgan and Elleson 1993) and confirmed
by a manufacturer of turnkey stratified chilled water storage
10

tanks (Andrus 1998). Tank cost was calculated using a bare


cost of $0.37/gal ($0.098/L). This cost is representative of the
lower range of cost for surface-mounted, steel chilled water
storage tanks with volumes on the order of 3,000,000 gal
(11,356,000 L). Costs are potentially much higher for volumes
less than 1,000,000 gal (3,790,000 L) and for installations that
4223

BACK TO PAGE ONE


interfaces are 40% to 70% more expensive than direct interfaces for the cases considered.

require special site work and foundations. The use of a relatively low estimate for tank cost casts indirect interfaces in the
most favorable light possible because it minimizes the size
penalty associated with a diminished temperature differential.
The implications of this assumption will be discussed following the presentation of results.

ANNUAL SIMULATION RESULTS


Forty-eight annual simulations were performed. Each of
the seven interface types was simulated at the seven static
pressure differentials, with the exception of the RDVT interface, which could not be modeled at 90 ft (27 m) for reasons
discussed previously. Electric demand and energy results were
compiled and used to calculate annual energy costs based on
the composite rate structure.

Typical budget cost markups were applied to the bare


costs in Tables 6a and 6b in order to obtain the total in Table
6c. Items other than the storage tank were marked up by a
combined factor of approximately 58%, which included a
10% contingency, 20% for general conditions, and 20% for
overhead and profit. Storage tanks, which are likely to be bid
separately from the remainder of the construction in a thermal
storage project, were marked up by 10%.

Energy Analysis
Table 7 summarizes annual electric demand and usage
characteristics for each case. Demand results presented in
Table 7 are the total of twelve monthly on-peak demands, i.e.,
the annual billing demand for each case. These results are
provided to give a gross comparison between alternatives.
Monthly maximum on-peak demands for each interface for
the 90 ft (28 m) and 360 ft (110 m) of water cases are listed in
Table 8. The relatively level monthly on-peak demands listed
are due to the interaction of the control strategy and load
profile. Storage is utilized to nearly its full capacity at some

The cost of direct interfaces varies as a function of static


pressure differential because of the effect of static head on
pump and turbine selections and pump motor size. Indirect
interface costs do not vary with static pressure differential
because there is no effect on pump head in these cases. The
additional components required in the indirect interface
together with increased storage tank size make it a very costly
option. When the impact of tank size is included, indirect

TABLE 7a
Annual Interface Energy Consumption: Usage
Static Pressure

Usage (kW)

(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

RDVT

90

27

788,238

447,190

555,826

481,166

414,850

343,724

130

40

788,238

447,190

738,005

660,836

504,892

428,944

532,320

180

55

788,238

447,190

966,186

885,682

569,741

492,476

673,170

220

67

788,238

447,190

1,249,626

1,161,261

765,489

677,496

791,993

260

79

788,238

447,190

1,423,165

1,326,697

818,665

727,788

919,483

310

94

788,238

447,190

1,727,053

1,627,125

1,022,557

922,576

1,131,728

360

110

788,238

447,190

1,972,752

1,880,756

1,204,902

1,093,656

1,323,420

TABLE 7b
Annual Interface Energy Consumption: Demand
Demand (kW)*

Static Pressure

(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

RDVT

90

27

3,283

3,407

2,133

2,206

1,497

1,543

130

40

3,283

3,407

2,854

2,965

1,842

1,817

2,462

180

55

3,283

3,407

3,767

3,928

2,163

1,901

2,645

220

67

3,283

3,407

4,876

5,072

2,886

2,567

3,289

260

79

3,283

3,407

5,594

5,823

3,139

2,663

3,695

310

94

3,283

3,407

6,767

7,045

3,899

3,367

4,404

360

110

3,283

3,407

7,723

8,061

4,555

3,994

5,247

Sum of twelve months, on peak.

4223

11

BACK TO PAGE ONE


TABLE 8
Selected Maximum Monthly On-Peak Electric Demand
Static Pressure (ft of water)
All
IC

90 (27 m)
IV

DC

DV

Month

DCT

DVT

DC

DV

DCT

DVT

RDVT

Maximum On-Peak Demand (kW)

Jan

274

263

176

177

125

126

632

650

381

333

437

Feb

274

293

178

187

125

130

648

681

378

332

437

Mar

274

290

178

186

125

130

647

678

377

333

438

Apr

274

265

176

178

125

125

633

653

380

333

438

May

274

276

177

181

125

127

639

664

380

333

438

Jun

274

290

178

186

125

130

647

678

380

333

437

July

274

282

178

183

125

128

643

670

380

333

438

Aug

274

287

178

185

125

129

645

675

378

333

436

Sep

274

292

178

186

125

130

648

680

380

333

436

Oct

274

291

178

186

125

130

648

679

380

333

437

Nov

274

289

178

185

125

129

646

677

380

333

438

Dec

274

289

178

185

125

129

646

677

380

333

437

point during on-peak hours of each month. Consequently, only


small variations are observed in the monthly peak flow rate
and demand. This, in turn, causes the peak pumping power for
each interface type to be nearly constant. Percentage differences in annual energy consumption and billing demand relative to a constant-speed direct (DC) interface are listed in
Table 9. The annual usage data are plotted in Figure 9.
Demand and usage are independent of static pressure for
indirect (IC and IV) interfaces. For direct interfaces, however,
both demand and energy increase as static pressure (and,
therefore, total pump head) increases. Direct interfaces with
energy recovery consume significantly less energy and have
lower demand than direct interfaces without energy recovery.
The constant-speed indirect (IC) interface has greater energy
consumption and annual demand than the conventional direct
(DC) interface for static differential less than approximately
130 ft (40 m). This is due to the combined effects of increased
flow rate and heat exchanger pressure drop for the IC alternative. At greater values of static differential, however, indirect
interfaces consume substantially less energy than direct interfaces, even those with energy recovery.
Variable-speed operation reduces the energy consumption of each interface relative to its constant-speed implementation. Variable-speed operation of interfaces with energy
recovery also reduces peak demand relative to constant speed
in higher static head cases because the peak demand associated with a constant-speed pump/turbine set does not occur at
full capacity (Bahnfleth and Kirchner 1998). The difference
between constant and variable-speed energy consumption is
greatest at the lowest level of static differential because a
system curve with little or no fixed head provides the greatest
12

360 (110 m)

opportunity for speed reduction. For example, with static head


of 90 ft (27 m), the variable-speed direct interface consumes
13% less energy than the constant-speed direct interface, but
with static head of 360 ft (110 m), the difference is only 5%.
As the results in Table 9 indicate, unidirectional direct
interfaces with energy recovery (DCT and DVT) significantly
reduce annual energy consumption relative to both constantand variable-speed direct interfaces without recovery (DC and
DV). The DCT case, for example, saves 25% to 42% in annual
energy consumption relative to the DC case, while the DVT
alternative saves 38% to 49%. The reversible (RDVT) alternative achieves significant energy-use reductions relative to
direct interfaces without energy recovery but does not perform
as well as either the DCT or DVT alternatives because its
components do not perform as well in either pump or turbine
mode as do the pump and turbine of the unidirectional interface.
As a function of static pressure differential, energy recovery has a savings trend counter to that of speed control. As
static differential increases, so does the energy saving associated with turbine use. For example, with 90 ft (27 m) of static
head, the variable-speed direct (DV) interface consumes 29%
more energy than the variable-speed direct interface with
energy recovery (DVT), while at 360 ft (110 m) of static the
difference has increased to 42%t. Variable-speed operation
increases the energy savings of energy recovery interfaces but
by a smaller percentage as static differential increases. At 90
ft (27 m) of static, the variable-speed direct interface with
energy recovery (DVT) consumes 17% less energy than its
constant-speed (DCT) analog, but at 360 ft (110 m) it saves
only 9%.
4223

BACK TO PAGE ONE


TABLE 9a
Percent Reduction in Annual Energy Consumption Relative to the DC Interface: Usage
Percent Reduction in Annual Energy Usage *

Static Pressure

(ft of water)

(m)

IC

IV

DV

DCT

DVT

RDVT

90

27

42

20

13

25

38

130

40

39

10

32

42

28

180

55

18

54

41

49

30

220

67

37

64

39

46

37

260

79

45

69

42

49

35

310

94

54

74

41

47

34

360

110

60

77

39

45

33

Negative values indicate increased annual energy usage.

TABLE 9b
Percent Reduction in Annual Energy Consumption Relative to the DC Interface: Demand
Percent Reduction in Annual Energy Demand *

Static Pressure

(ft of water)

(m)

IC

IV

DV

DCT

DVT

RDVT

90

27

54

60

30

28

130

40

15

19

35

36

14

180

55

13

10

43

50

30

220

67

33

30

41

47

33

260

79

41

39

44

52

34

310

94

51

50

42

50

35

360

110

57

56

41

48

32

Negative values indicate increased total annual on-peak demand.

Energy Cost
Annual interface electric costs calculated using the electric rate structure shown in Table 2 are summarized in Table
10. Annual energy cost savings for each interface relative to
the DC alternative are shown in Figure 10. The range of interface operating costs shown provides a striking illustration of
the importance of minimizing adverse static differential in
systems with direct interfaces. From the least operating cost
system (DVT) at 90 ft (28 m) of static to the highest cost (DC)
system at 360 ft (110 m), pumping energy cost increases by
more than a factor of five.
The annual pumping cost per ton-hour of delivered storage is shown in Table 11. These costs are obtained by dividing
the total annual energy cost for the interface by the total capacity delivered to the load. Consequently, they include the cost
of pumping during both charging and discharging and represent the energy cost that would be incurred by a purchaser of
chilled water. Prices for district chilled water in the U.S. range
from less than 15/ton-h (4.3/kWh) to more than 35/ton-h
(10/kWh) with 25/ton-h (7.1/kWh) being a typical value
(Pierce 1997). At the low end of the static head range, unit
4223

pumping costs are less than 0.5/ton-h (0.14/kWh). At the


other extreme, pumping cost per unit of capacity delivered
exceeds 1.5/ton-h (0.4/kWh). These results clearly indicate
that the system owner should not neglect storage pumping
energy cost.
ECONOMIC ANALYSIS
Simple payback and life-cycle cost analyses were
performed using standard techniques (Fuller and Petersen
1995). Life-cycle cost ranks alternatives by their ultimate
economic performance without regard to how rapidly the
initial investment is recovered. Simple payback period gives
an indication of how rapidly the incremental investment associated with an alternative is repaid through operating cost
savings but does not identify the alternative that will produce
the largest life-cycle savings. Other factors being equal, the
least life-cycle cost alternative is the most attractive. In actuality, however, many owners select the lowest life-cycle cost
choice alternative among options that satisfy a particular
maximum payback criterion.
13

BACK TO PAGE ONE

Figure 9 Annual interface pumping energy: (a) constant speed, (b) variable speed.

14

4223

BACK TO PAGE ONE


TABLE 10
Annual Electric Energy Cost Summary
Static Pressure

Demand ($)

(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

RDVT

90

27

35,566

37,028

23,113

23,934

16,216

16,736

130

40

35,566

37,028

30,934

32,166

19,955

19,707

26,676

180

55

35,566

37,028

40,837

42,615

23,433

20,607

28,651

220

67

35,566

37,028

52,859

55,029

31,262

27,822

35,633

260

79

35,566

37,028

60,646

63,184

34,004

28,855

40,036

310

94

35,566

37,028

73,372

76,438

42,243

36,479

47,715

360
110
Static Pressure

35,566

37,028

83,738

87,456
Usage ($)

49,346

43,262

56,835

(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

RDVT

90

27

41,611

23,112

29,399

25,324

21,979

18,102

130

40

41,611

23,112

39,019

34,809

26,758

22,627

28,026

180

55

41,611

23,112

51,080

46,678

30,248

26,029

35,468

220

67

41,611

23,112

66,067

61,225

40,639

35,830

45,188

260

79

41,611

23,112

75,237

69,953

43,504

38,542

48,731

310

94

41,611

23,112

91,310

85,819

54,350

48,877

59,933

360
110
Static Pressure

41,611

23,112

104,303

99,219
Total ($)

64,015

57,946

70,123
RDVT

(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

90

27

77,177

60,140

52,512

49,258

38,195

34,839

130

40

77,177

60,140

69,952

66,976

46,714

42,333

54,702

180

55

77,177

60,140

91,917

89,293

53,681

46,635

64,119

220

67

77,177

60,140

118,926

116,254

71,902

63,652

80,821

260

79

77,177

60,140

135,883

133,137

77,508

67,397

88,767

310

94

77,177

60,140

164,682

162,257

96,593

85,357

107,648

360

110

77,177

60,140

188,041

186,676

113,361

101,208

126,957

Figure 10 Annual energy cost savings relative to the direct constant-speed interface (see Table 1 for acronym definitions).
4223

15

BACK TO PAGE ONE


TABLE 11a
Average Annual Delivered Storage Pumping Energy Costs (IP Units)
Static Pressure

Annual Average Storage Pumping Energy Cost (/ton-h)

(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

RDVT

90

27

0.68

0.53

0.45

0.44

0.34

0.31

130

40

0.68

0.53

0.62

0.59

0.41

0.37

0.48

180

55

0.68

0.53

0.81

0.79

0.47

0.41

0.57

220

67

0.68

0.53

1.05

1.03

0.64

0.56

0.71

260

79

0.68

0.53

1.20

1.18

0.68

0.60

0.78

310

94

0.68

0.53

1.45

1.43

0.85

0.75

0.95

360

110

0.68

0.53

1.66

1.65

1.00

0.89

1.12

TABLE 11b
Average Annual Delivered Storage Pumping Energy Costs (SI Units)
Static Pressure

Annual Average Storage Pumping Energy Cost (/kWh)

(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

RDVT

90

27

0.19

0.15

0.13

0.12

0.10

0.09

130

40

0.19

0.15

0.18

0.17

0.12

0.11

0.14

180

55

0.19

0.15

0.23

0.22

0.13

0.12

0.16

220

67

0.19

0.15

0.30

0.29

0.18

0.16

0.20

260

79

0.19

0.15

0.34

0.33

0.19

0.17

0.22

310

94

0.19

0.15

0.41

0.41

0.24

0.21

0.27

360

110

0.19

0.15

0.47

0.47

0.28

0.25

0.32

Simple Payback
The results of simple payback analysis are presented in
Table 12 and Figure 11. The direct interface without energy
recovery is used as the basis for calculating payback. Payback
relative to the constant-speed DC case is shown in Figure 11a
and payback relative to the variable-speed DV case is shown
in Figure 11b. Payback periods are shorter when the more
expensive DV case is used as the base despite its lower energy
cost.
The payback of the indirect interface is, at best, marginal
over the entire range of static pressure considered. Even with
a static differential of 360 ft (110 m), the simple payback of the
IC and IV alternatives is on the order of ten years. The best
payback achieved is 8.9 years with 360 ft (110 m) of static
head, and payback becomes much worse at lower values. At
90 ft (28 m) of static differential, payback does not occur
because savings are negative. Payback for systems with
energy recovery is relatively longer at lower levels of static
differential but drops below five years before static pressure
reaches 150 ft (46 m). As Figure 11a shows, variable-speed
pumping without energy recovery is not attractive on the basis
of payback at any point within the range of static head differential considered. However, the benefits of variable-speed
pumping for chilled water thermal storage system should not
be evaluated solely in economic terms (Bahnfleth 1995).
16

Prevention of water hammer and shutdowns due to pressure


transients and ease of control are considerations that strongly
favor the use of variable-speed pumps in thermal storage interfaces as a standard practice.
Life-Cycle Cost
Twenty-year life-cycle cost results are presented in Table
13 and Figure 12. Calculations were done on a constant dollar
basis using a real discount rate of 3.5% (corresponding to a
nominal discount rate of roughly 7%). Electric costs were
adjusted using U.S. average commercial sector real escalation
factors published by the U.S. Department of Commerce
(Petersen 1997).
Life-cycle cost increases rapidly with static pressure for
DC and DV interfaces and much more slowly for DCT and
DVT interfaces. Variable-speed systems enjoy a small lifecycle advantage over their constant-speed counterparts. Intersections of the life-cycle cost functions in Figure 12 are points
at which one technology becomes superior to another. Direct
pumping without energy recovery is more costly than direct
pumping with energy recovery over the entire range of static
pressure differential considered for both constant- and variable-speed interfaces. Both constant- and variable-speed
direct pumping interfaces become more expensive than indirect interfaces near 260 ft (80 m) of static differential. Perhaps
4223

BACK TO PAGE ONE

Figure 11 Simple payback period as a function of static pressure: (a) relative to DC interface, (b) relative to DV interface (see
Table 1 for acronym definitions).

4223

17

BACK TO PAGE ONE


TABLE 12a
Simple Payback SummaryBase: Constant-Speed Direct Interface
Static Pressure

Simple Payback (Years): DC Base

(ft of water)

(m)

DV

IC

IV

DCT

DVT

RDVT

90

27

21.2

None

None

7.2

9.7

130

40

25.2

None

132.7

4.4

6.4

11.7

180

55

33.7

78.5

40.7

2.7

4.2

6.9

220

67

36.8

27.4

21.8

2.2

3.7

5.2

260

79

35.8

19.5

16.9

1.8

3.0

4.5

310

94

44.1

12.8

12.1

1.5

2.6

4.2

360

110

85.6

10.0

9.7

1.3

2.5

3.8

TABLE 12b
Simple Payback SummaryBase: Variable-Speed Direct Interface
Static Pressure

Simple Payback (Years): DV Base

(ft of water)

(m)

IC

IV

DCT

DVT

RDVT

90

27

None

None

3.0

7.1

130

40

None

179.5

1.4

4.2

8.5

180

55

88.2

41.3

0.4

2.4

4.1

220

67

26.7

21.1

0.2

2.0

2.8

260

79

18.7

16.2

0.1

1.6

2.6

1.3

2.5

1.2

2.1

310

94

12.0

11.3

360

110

9.1

8.9

Indicates immediate payback.

TABLE 13
Life-Cycle Cost Summary
Static Pressure

Life-Cycle Cost ($)

(ft of water)

(m)

IC

IV

DC

DV

DCT

DVT

90

27

3,912,708

3,808,781

2,386,650

2,409,744

2,287,080

2,308,726

130

40

3,912,708

3,808,781

2,642,475

2,675,514

2,416,893

2,430,104

2,605,985

180

55

3,912,708

3,808,781

2,961,655

3,012,941

2,524,253

2,513,084

2,761,641

220

67

3,912,708

3,808,781

3,356,982

3,417,473

2,798,303

2,780,030

3,015,182

260

79

3,912,708

3,808,781

3,596,473

3,655,924

2,877,482

2,832,910

3,144,589

310

94

3,912,708

3,808,781

4,022,228

4,094,843

3,160,816

3,108,983

3,457,840

360

110

3,912,708

3,808,781

4,367,823

4,465,383

3,413,317

3,358,523

3,754,789

most significantly, the life-cycle cost of direct pumping with


energy recovery was lower than that of indirect interfaces over
the entire range of static pressure differential considered. This
is due directly to the large capital cost penalty associated with
indirect interfaces.
Sensitivity Analysis
A sensitivity analysis was performed to evaluate the
effect of moderate changes in capital and energy cost above
18

RDVT

and below the values assumed for this study. Two scenarios
were considered: (1) 10% increase in capital costs combined
with 10% reduction in energy costs and (2) 10% decrease in
capital costs combined with 10% increase in energy costs. The
net savings, payback period, and life-cycle costs were calculated for each scenario and compared with the results obtained
with baseline assumptions.
The results of the simple payback sensitivity analysis for
the DCT interface are plotted in Figure 13a. It can be observed
4223

BACK TO PAGE ONE

Figure 12 Life-cycle costs as a function of static pressure: (a) constant-speed alternatives (DVT shown for comparison), (b)
variable-speed alternatives (see Table 1 for acronym definitions).

4223

19

BACK TO PAGE ONE

Figure 13 Simple payback and life-cycle cost sensitivityDCT interface: (a) simple payback, (b) life-cycle cost.

20

4223

BACK TO PAGE ONE


that moderate changes in the baseline capital and energy costs
have a very small effect on the payback period. Increases in
capital cost relative to a slight decrease in energy cost extends
payback period as expected. A slight increase in the energy
cost relative to decrease in capital costs had the expected
outcome of increasing energy savings and decreasing the
payback period. The scenarios investigated in this sensitivity
analysis did not produce changes in the relative value of the
various interface alternatives.
The result of the life-cycle cost (LCC) sensitivity analysis
for the DCT interface is plotted in Figure 13b. As the figure
indicates, LCC is relatively insensitive to changes in capital or
energy costs. Therefore, as demonstrated previously with
payback, implementation of either scenario does not favor a
particular interface on the basis of life-cycle cost.
CONCLUSIONS
This study has compared the energy consumption and
operating cost of seven chilled water thermal storage pumping
interface alternatives over a range of static pressure differentials for a typical district cooling system load profile and a
representative time-of-use electric rate structure. Conclusions
based on these results should be qualified in several respects.
Only one load profile was considered, and this load
profile was selected to generate a large number of annual operating hours for the thermal storage system. This results in a
relatively high ratio of savings to investment that tends to
enhance the attractiveness of energy recovery alternatives.
Further, only one thermal storage control strategy was considered, although it is representative of the algorithm used in
many systems.
Sensitivity analysis is based on the simple rate structure
and first cost used in this study and does not encompass the full
range of potential variations in electric utility rates and capital
costs. However, the electric rate employed in this analysis is
typical of many tariffs in the U.S. Practitioners grappling with
the issues discussed in this paper in the course of actual system
design should apply similar methods of analysis using the
loads, rates, and costs relevant to their project.
Subject to these qualifications, the following conclusions
and recommendations can be made:

4223

Indirect interfaces are likely to be a poor economic


choice because of their effect on system capital cost,
particularly storage tank cost. Indirect interfaces were
never the least life-cycle cost alternative and had simple
payback periods in excess of ten years.
A direct interface is the preferred type. If five years is
taken as a maximum attractive simple payback period,
the results of the present study indicate that energy
recovery in a direct interface is feasible for static pressure differentials as low as 120 ft (37 m), which is on the
order of nine stories.
Variable-speed pumping combined with energy recovery in a direct interface is beneficial on a life-cycle cost

basis but increases payback period slightly, relative to


constant-speed operation. Variable-speed operation
reduces the amount of energy recovered by the turbine
but produces net energy savings because of substantial
improvement in the performance of the pump. The control benefits of variable pumping can, therefore, be
incorporated in a direct interface without adversely
affecting system economics.
The variable-speed reversible direct pumping interface
with energy recovery has an acceptable payback relative to
a direct variable-speed interface without energy recovery for
large static head differentials. A five-year payback was
obtained in the present study for static differential greater
than 180 ft (55 m). However, neither the simple payback
period nor the life-cycle cost of this alternative was lower
than that of a unidirectional direct interface with energy
recovery for any value of static differential considered.
Given these results, the value of this configuration appears to
be limited.
ACKNOWLEDGMENTS
The authors wish to extend their appreciation to Mr. Steve
Gavlick of the PACO Pump Corporation for his assistance in
making the pump selections needed in this research. Partial
support of this work through an ASHRAE Graduate Grant-inAid is also gratefully acknowledged.
REFERENCES
Andrus, D. 1998. Matrix Service, Inc., personal communication.
Bahnfleth, W.P. 1995. Hydraulic issues in the design of
chilled water storage systems. Proceedings, International District Energy Association. 86th Annual Conference, 1995.
Bahnfleth, W.P., and C.G. Kirchner. 1998. Analysis of transfer pumping interfaces for stratified chilled water thermal storage systemsPart 1: Model development.
ASHRAE Transactions 105(1).
BLAST. 1991. BLAST user reference, Vol. 2. Urbana:
Department of Mechanical and Industrial Engineering,
University of Illinois at Urbana-Champaign.
Buse, F. 1981. Using centrifugal pumps as hydraulic turbines. Chemical Engineering, January, pp. 113-117.
Fiorino, D.P. 1994. District cooling re-invented. ASHRAE
Journal, Vol. 36, No. 5 (May), pp. 20-28.
Fuller, S.K., and S.R. Petersen. 1995. Life-cycle costing
manual for the federal energy management program.
1995 edition. Washington D.C.: U.S. Department of
Commerce.
Gatley, D.P., and I. Mackie. 1995. Cool storage open
hydronic systems design guide. EPRI, TR-104906.
Kirchner, C.G. 1997. Analysis and modeling of transfer
pumping for stratified chilled water thermal storage sys21

BACK TO PAGE ONE


tems, M.S. thesis, Department of Architectural Engineering, The Pennsylvania State University.
Means. 1996. Means mechanical cost data 1996. Massachusetts: R.S. Means Company, Inc.
Pierce, M.A. 1997. District cooling rates. http://
www.energy.rochester.edu/rates/cooling.htm.

22

Petersen, S.R. 1997. Energy price indices and discount factors for life-cycle cost analysis 1997. Washington D.C.:
U.S. Department of Commerce.
Tackett, R.K. 1988. The use of direct pumping and hydraulic
turbines in thermal storage systems. ASHRAE Transactions 94(1): 1989 - 2007.

4223