Investigation of Cost and Energy Optimization of Drinking Water

Subscriber access provided by CMU Libraries - http://library.cmich.edu

Article

Investigation of Cost and Energy Optimization of Drinking Water Distribution Systems Carla Cherchi, Mohammad Badruzzaman, Matthew Gordon, Simon Bunn, and Joseph G. Jacangelo Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.5b03666 • Publication Date (Web): 13 Oct 2015 Downloaded from http://pubs.acs.org on October 15, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 25

Environmental Science & Technology

1

Investigation of Cost and Energy Optimization of Drinking Water Distribution

2

Systems

3

Carla Cherchi1*, Mohammad Badruzzaman1, Matthew Gordon2, Simon Bunn2, Joseph G.

4

Jacangelo1,3

5

1

MWH Americas, Inc., 300 N. Lake Avenue, Pasadena, CA, USA

6

2

Derceto, Ltd., 63 Albert Street, Auckland 1141, New Zealand

7

3

The Johns Hopkins University Bloomberg School of Public Health, 615 N. Wolfe Street,

8

Baltimore, MD 21205, USA

9

* Corresponding Author: MWH Americas, Inc., 300 N. Lake Avenue, Pasadena, CA, USA,

10

Email: [email protected], Phone: +1 (626)-568-6009, Fax: +1 (626)-568-6101

11 12

Abstract

13

Holistic management of water and energy resources through Energy and Water Quality

14

Management Systems (EWQMS) have traditionally aimed at energy cost reduction with limited

15

to no emphasis on energy efficiency or greenhouse gas minimization. This study expanded the

16

existing EWQMS framework and determined the impact of different management strategies for

17

energy cost and energy consumption (e.g., carbon footprint) reduction on system performance at

18

two drinking water utilities in California, US. The results show that optimizing for cost led to 4%

19

(Utility B, summer) to 48% (Utility A, winter) cost reduction. The energy optimization strategy

20

was successfully able to find the lowest energy use operation and achieved 3% (Utility B,

21

summer) to 10% (Utility A, winter) reduction in energy usage. The findings of this study

22

revealed that there may be a trade-off between cost optimization (dollars) and energy use

23

(kilowatt-hour), particularly in summer when optimizing the system for the reduction of kWh to

24

a minimum incurred 64% and 184% increase in cost compared to the cost optimization scenario.

25

Water age simulations through hydraulic modeling did not reveal any adverse effects on the 1

ACS Paragon Plus Environment


Page 2 of 25

26

distribution system water quality and in tanks from pump schedule optimization targeting either

27

cost or energy minimization.

28 29

Keywords. Energy and Water Quality Management System, Drinking Water Utilities, Cost

30

Optimization, Energy Optimization, Energy Efficiency, Greenhouse Gas Emissions.

31 32

TOC

33 34 35

1

Introduction

36

Over the past several decades, water organizations have been challenged by increasing energy

37

costs and demand and new stringent environmental regulatory requirements.1 An approximate

38

3% share of United States (U.S.) annual electricity consumption is attributable to water utilities

39

operations and future forecasts estimate this percentage to increase due to greater water demand

40

and use of energy intensive treatment solutions.2,

41

requirements for conveyance and distribution system operations vary in relation to the service

42

area topography, raw water sources and adopted management strategies.6 In states where water is

43

moved across large distances and elevations, this energy usage can become substantially higher.

44

For example, in California, the water-related energy consumption accounts for 7.7% of the

45

state’s energy usage.7 More than 90% of GHGs emitted into the atmosphere annually by drinking

46

water utilities in the U.S. is due to electricity usage, which contributes significantly to an

47

increasing carbon footprint estimated to be 45 million tons of CO2-eq per year.8, 9

3, 4, 5

In drinking water utilities, the energy

48

Regional and national regulations have been promulgated or are anticipated in the U.S. and

49

other parts of the world to achieve GHG emissions reduction. Cap-and-trade market mechanisms 2


Page 3 of 25


50

are well-recognized and are viable tools to manage these environmental challenges. In the U.K.,

51

the Carbon Reduction Commitment (CRC), a cap-and-trade market addressing climate change

52

and issued by the Climate Change Act, requires water companies with half-hourly metered

53

electricity consumption equal to or greater than 6,000 MWh per year to reduce CO2 emissions by

54

at least 26% by 2020, against a 1990 baseline.10 In the U.S., the California Air Resources Board

55

has designed a cap-and-trade program that is enforceable and meets the requirements of the

56

Assembly Bill No.32 (AB 32), the California Global Warming Solutions Act of 2006, which is

57

expected to return to 1990 emission levels by 2020 and cover 85% of the state’s GHG

58

emissions.11 Although U.S. water utilities are currently not qualified to participate in cap and

59

trade programs, any future change in regulation will require both water and energy suppliers to

60

consider the GHG implications of their systems.12

61

Energy management programs can assist water utilities in reducing their carbon footprint and

62

meeting current and future cap-and trade requirements, in addition to achieving operating energy

63

cost savings. To facilitate the design of these energy management programs, holistic life cycle

64

approaches have been recently applied as tools to identify the GHG impact of water distribution

65

systems and identify opportunities for emission reduction.13 The majority of the GHG footprint

66

of a water distribution system is related to pumping energy. Electricity and energy use reduction

67

can be achieved at the infrastructure level (e.g., pumps, motors, pipelines, etc.), through energy

68

efficient system designs, rehabilitation or equipment replacement, or by implementing energy

69

optimization strategies, particularly when substantial inefficiencies due to leakage in distribution

70

pipes occurs.13, 14, 15, 16, 17 The need for addressing the reduction of energy use and related GHG

71

emissions in the water sector through system optimization and the importance of evaluating the

72

trade-offs between different optimization objectives and various control strategies (cost vs. kWh

73

minimization) has been emphasized in previous studies.2, 12, 18, 19, 20, 21 Despite this need for multi-

74

objective analysis and optimization, these trade-offs in water conveyance and distribution

75

systems operations remain poorly investigated by the water industry and specific tools for

76

integrated system planning and operations are not widely available. The reluctance of water

77

utilities to implement energy optimization programs and investigate these trade-offs is mostly

78

due to concerns over constraints tied to water quality, the potential impact on production levels

79

and maintenance, in addition to the cultural challenge that such implementation implies at all

80

levels of the organization. 3



Page 4 of 25

81

Energy and Water Quality Management Systems (EWQMS) have been increasingly

82

implemented at drinking water utilities to provide the foundation for a system control

83

management tool to cohesively address water quality, supply and energy management within

84

operational constraints.22 An EWQMS is a collection of individual application software

85

programs that interface with existing monitoring and pre-existing control systems (Supervisory

86

Control and Data Acquisition) in real-time to achieve operating cost reduction through real-time

87

scheduling of pumping during lower cost tariff periods and best efficiency practices within the

88

boundaries of water quality requirements and operating rules.23 Although traditional EWQMS

89

design and operation focuses on achieving energy cost reductions, very little emphasis has been

90

given to better understand the impact in EWQMS performance when the reduction of kilowatt-

91

hours is prioritized. Energy optimization, in fact, can conflict with cost optimization, particularly

92

for utilities, such as those in California, where high differential tariffs result from Time-of-Use

93

(TOU) structures. Furthermore, in current EWQMS systems, the accounting of GHGs emissions

94

associated to pumping system operations is lacking. Therefore, it is of paramount importance

95

that the EWQMS enables water utilities to consider operational options to reduce kWh and GHG

96

emissions and identify the trade-offs between energy management for cost reduction, energy and

97

carbon footprint reduction, and conveyance and distribution system water quality.

98

This study was undertaken to investigate the potential of control strategy optimization for the

99

reduction of energy (e.g., GHG emissions) from water distribution system operations and

100

examine the trade-offs between various control objectives (cost vs. energy minimization). This is

101

achieved by developing and testing an integrated EWQMS modified with a kWh optimization

102

and GHG accounting component that expands the existing and historical EWQMS framework to

103

address the nexus between energy cost management, energy (kWh) use reduction, and system

104

water quality. The functional specifications of the modified EWQMS were appropriately

105

developed and evaluated at two drinking water utilities in California. Off-line simulations were

106

performed on selected areas of these utilities’ distribution networks to reproduce the optimal

107

pumping schedule of different optimization scenarios. Energy costs and kWh minimization were

108

the targeted optimization goals simulated and compared with baseline of historical operations for

109

conditions of high and low water demands. Water age simulations through hydraulic modeling

110

were used to determine the potential impact of these optimization scenarios on tanks and

111

distribution system water age. 4


Page 5 of 25


112

2

Materials and Methods

113

2.1

Site Selection

114

Two water utilities in California, namely Utility A and Utility B, were selected for testing and

115

assessment of the EWQMS modified with a kWh optimization and GHG accounting component.

116

Both water utilities currently have an operational EWQMS system utilizing Aquadapt (Derceto,

117

Ltd). Details on the conveyance and distribution networks controlled by EWQMS at these

118

utilities are presented in Table S1 (Supporting Information).

119

For this study, the analysis was limited to a portion of the distribution network, which did not

120

include the conveyance system and was hydraulically isolated from the remaining areas under

121

EWQMS control. The conceptual schematics of these selected sites at Utility A and Utility B are

122

reported in Figure 1. Each site includes three pumping stations, subjected to different tariff

123

structures by the energy providers, and three water storage tanks. Each pumping station (PS) at

124

Utility A includes three electric pumps with a maximum of two pumps that can run

125

simultaneously. At Utility B, the PS-B1 pump station has two gas pumps (not operated for this

126

study) and one electric pump, whereas PS-B2 and PS-B3 pumping stations only have electric

127

pumps. Graphical representations of the daily and seasonal tariff profiles (summer and winter) at

128

these pumping stations are reported in Supporting Information (Figure S1 and Figure S2). At

129

Utility A, all pumping stations were subject to seasonal TOU tariffs by Pacific Gas & Electric,

130

and the PS-A2 pumping station includes three peak period demand charges. At Utility B, PS-B1

131

pump station was subjected to seasonal TOU tariffs and two peak period demand charges by

132

Southern California Edison, whereas PS-B2 and PS-B3 pump stations were under flat tariffs

133

throughout the year.

134

5



Page 6 of 25

135 136 137

Figure 1: Conceptual schematic of the Utility A (top) and Utility B (bottom) distribution system areas selected for testing.

138 139

2.2

Selection of optimization scenarios

140

Two optimization strategies were implemented in separate evaluations to investigate the

141

potential cost and benefits associated with different goals (e.g., objective functions) and by

142

imposing distinct constraints. The analysis compared historic manual (baseline) operation with

143

the optimization objectives of cost and kWh reduction during the high (summer) and low

144

(winter) demand seasons. Table 1 summarizes the fundamental principles of the optimization

145

goals analyzed in this study, also detailed in the following sections.

146

Minimization of Energy Costs. The pump energy cost depends on the electricity tariff rate

147

applied (on peak, mid-peak and off-peak) and by the related kilowatt (kW) demand charge

148

costs.24 The cost optimization scenario targets the lowest cost for pumping by scheduling the

149

pumps at a lower tariff period and by reducing the demand charges, through the minimization of

150

the following function fc (1):

min ∙ ∙ 1


6

Page 7 of 25


151

where N is the total number of pumps in the distribution system network; Tr = electricity tariff

152

rate ($/kWh); Ep = electricity consumption of a given pump (kWh); Dc is the demand charge

153

($/kW) and D is the peak demand (kW) in the study period. The amount of energy required for

154

pump operations (Ep) depends on the flow through the pump, the head supplied by the pump, and

155

the wire-to-water efficiency, which is the ratio of the energy imparted to the water and the

156

energy delivered to the pumping system by the electrical conductors (ηp·ηm·ηd).16,

157

each pump, the electricity consumption (Ep) is calculated based on the following equation (2)

158

and the total energy consumption (ET) is defined by equation (3): !"

!#

25

Thus, for

∙∙ ∙ 2 ∙ ∙

159 %!

" % 3 %!

160 161

where Q is the flow rate; H is the total pumping head; ηm, ηp and ηd are the efficiency of the

162

motor, pump and drive, respectively; γ is the specific weight of water, t = duration of pumping at

163

a given operating point i; i is the summation index for the operating points; T is the number of

164

operating points within the analysis period; n is the summation index for the number of pumps in

165

the distribution network.16 As a result of the optimization, pump operations under TOU tariff

166

structures are primarily scheduled during off-peak tariff periods. After achieving minimization of

167

the fc function, as secondary objective, the optimizer seeks to minimize the pump energy

168

consumption (EC) through operational pump efficiency strategies described in the following

169

section.

170 171

Minimization of Energy Consumption/GHG Emissions. The optimization targeting kWh

172

reduction limits the amount of energy used to move water, by selecting the lowest hydraulic

173

resistance path and operating near the best efficient point on the pump curve. Thus, the

174

minimization of the total energy consumption, previously calculated in Equation (2), is achieved

175

through the minimization of the following function fe : 7



Page 8 of 25

min ' " 4

176

where N is the number of pumps in the distribution system network and ET is the total energy

177

consumption. The strategies adopted to reduce the system energy consumption include: i)

178

decreasing of the total system head; ii) selecting and operating pumps close to their best

179

efficiency point; iii) selecting efficient pump and pump combinations to meet a given demand;

180

iv) running pumps in series in order to move water with a lower lift; v) properly managing the

181

water storage tanks (e.g., pumping into emptier tanks).

182

The majority of the GHG emissions in water conveyance and distribution systems are from

183

pumping energy and their estimation depends on the kWh consumed as well as on the fuel source

184

mix for their generation. Since the marginal GHG emissions factors from the energy providers

185

are not currently available, the kWh optimization scenario did not account for the energy source

186

providing electricity (e.g., coal, hydropower, nuclear, wind, solar, etc.), which would result in

187

different GHG emissions levels than those obtained using average GHG emission factors.

188

Therefore, for this study, the kWh optimization corresponds to the GHG optimization since

189

constant GHG emission factors were used. The average emission rate used in the GHG

190

calculations at Utility A were those provided by the Pacific Gas & Electric for that utility (0.187

191

metric tons-CO2 per MWh of electricity); at Utility B the fixed emission rate were those assigned

192

per kWh of electricity consumed in California by eGrid for the Western Electricity Coordinating

193

Council sub-region (0.296 metric tons-CO2 per MWh) due to the unavailability of the GHG

194

emission factor specific to the SCE service area. 26

195 196

Table 1. Description of the optimization scenarios Optimization Scenario Optimization Scenario

Description

Baseline

Primary principle: System under no EWQMS operation (manual control by operators)

Cost Optimization

Primary principle: Pump operated when the cost is minimum (selection order: off-peak, mid-peak, on-peak)

8


Page 9 of 25


Optimization Scenario

Description Secondary principle: Pump operated at the lowest specific energy (kWh/ML pumped)

Water Demand

1

197 198

kWh1 Optimization

Primary principle: Pump operated at the lowest specific energy (kWh/ML pumped) Secondary principle: Assumes flat tariff operations

High Demand (Summer)

Utility A: 14 ML/day Utility B: 2.9 ML/day

Low Demand (Winter)

Utility A: 4 ML/day Utility B: 0.9 ML/day

kWh optimization is the GHG optimization when flat GHG emission factors are used and does not assume variability of GHG emissions factors with different energy sources.

199 200 201

2.3

202

The EWQMS optimizer consists of an Excel spreadsheet that runs a Visual Basic for

203

Applications (VBA) program. The VBA program solves the optimization problem (cost and

204

kWh minimization) using linear and nonlinear programming combined with advanced heuristics

205

and includes a GHG accounting component for estimation of the carbon footprint of pumping

206

operations. A simplified representation of the optimizer’s logic, related input and output, is

207

presented in Figure 2. The optimizer requires input such as pump curves and efficiency curves;

208

historical operational data such as pump flows, outflows or pump outflows, tank levels/volumes

209

for each half hourly period and pump total dynamic head values. In addition a number of

210

constraints are applied and influence the convergence and results of the optimization solution.23

211

These constraints ensure compliance with water quality regulations, consumptive use criteria,

212

energy and mass conservation principles in the distribution network and at junction nodes, and

213

optimal pump operations (e.g., minimum run times; maximum starts per hour; minimum flow

214

rates, minimum and maximum plant production rates). Energy rate structures and demand

215

charges, energy efficient measures and asset management practices are other implicit and explicit

216

constraining boundaries typically applied. The model uses these data to forecast water demand

Development of the EWQMS Optimizer Functional Specifications

9



Page 10 of 25

217

for each individual pressure zone for the following 48 hours of operation. The demand forecaster

218

uses either average historical demand for the day type and season for each zone at the starting

219

point of the day and re-calculates water demand every 30 minutes using a combination of the

220

starting point and the last 6 hours of measured values, in order to enable detection of fast

221

changing trends and sudden weather changes. At the start of the day, the optimizer develops a

222

pump station’s schedule in half hourly periods for the upcoming 24 hours based on specific

223

optimization goals (cost savings vs. energy savings), after replacing the historical data with the

224

new water demand prediction profile. The software also improves water quality by seeking to

225

increase turnover of storage and minimize production flow disturbances.

226

The pump schedule generated is then exported into EPANET format and allows the user to run

227

more realistic “extended-period” simulations in order to select a hydraulically feasible schedule

228

with the lowest objective function value.24 The system iterates and continuously adapts to refine

229

its schedule in response to changing conditions including variability in water demand, equipment

230

availability and treatment plant capacity until the hydraulic and water quality results are

231

acceptable.

232

The contribution of leakage loads into pump energy use and operation was not considered as part

233

of the optimization results for this study. The minimization of water losses in water distribution

234

systems and the implementation of pressure management strategies have the potential to improve

235

the energy efficiency of pumping, therefore result in additional GHG emission reduction.27

10


Page 11 of 25


236 237

Figure 2: Input and output from the EWQMS optimizer.

238 11



239

2.4

Page 12 of 25

Water Quality Simulations

240

Calibrated hydraulic models in EPANET format were provided by Utility A and Utility B for

241

the offline simulations. The cost and kWh optimization goals were modeled in the InfoWater

242

model using pump switching derived from the simulations of system operations minimizing cost

243

and kWh. Water age, a useful proxy for water quality, was modeled using the Age of Water

244

quality option in InfoWater. Modeling of water quality based on water age is a recommended

245

practice for water distribution system management to overcome the monitoring constraints of

246

chlorine residuals and DBPs based modeling.28, 29 Water quality modeling results were used to

247

determine whether planned on-line optimization scenarios could materially affect the quality of

248

water being delivered to customers.

249

3

Results

250

The analysis compared historic operations with the optimization objectives of cost reduction

251

and kWh minimization for high (summer) and low (winter) water demand scenarios within the

252

distribution sub-systems at Utility A and Utility B. In addition, water quality predictions were

253

made to foresee the potential impacts of these optimization strategies on the water quality in

254

tanks and that water provided to customers.

255 256

3.1 Impact of Cost and kWh Optimization Strategies on Distribution System Performance at Utility A

257

Figure 3 shows the distribution of pumping to different tariff windows at Utility A, obtained

258

under cost and kWh optimization scenarios. In conditions of high water demand and within a

259

given infrastructure, the system characterized by a fixed storage capacity has less flexibility for

260

pumping higher volumes of water than in low water demand conditions of winter, due to a lower

261

storage to demand ratio. In addition, the tariff system in summer has a higher differential than

262

that of winter.

263

In high water demand conditions, the cost optimization scenario shifted 94% (at PS-A1) and

264

100% (at PS-A2 and PS-A3) of the load to off-peak periods, with no pumping occurring during

265

peak times. Due to the avoidance of pumping during the peak and mid-peak periods at PS-A1

266

and the peak period at PS-A2, the cost optimization simulation operated against a higher average

267

head than any other scenario. At PS-A1 and PS-A3, when the system was optimized for kWh, 12


Page 13 of 25


268

27% and 18% of the pumping occurred during peak hours. As there was less demand in winter

269

all optimization scenarios were able to operate exclusively with the PS-A1 pump station and,

270

with the exclusion of the baseline period, no pumping occurred at PS-A2. The PS-A3 pumping

271

station, mostly avoided the mid-peak periods (

Investigation of Cost and Energy Optimization of Drinking Water

Recommend Documents