Optimal retrofitting of natural gas pressure reduction stations for energy recovery

Accepted Manuscript Optimal retrofitting of natural gas pressure reduction stations for energy recovery

Ermanno Lo Cascio, Marc Puig Von Friesen, Corrado Schenone PII:

S0360-5442(18)30601-7

DOI:

10.1016/j.energy.2018.04.011

Reference:

EGY 12650

To appear in:

Energy

Received Date:

13 December 2017

Revised Date:

17 March 2018

Accepted Date:

03 April 2018

Please cite this article as: Ermanno Lo Cascio, Marc Puig Von Friesen, Corrado Schenone, Optimal retrofitting of natural gas pressure reduction stations for energy recovery, Energy (2018), doi: 10.1016/j.energy.2018.04.011

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

Optimal retrofitting of natural gas pressure reduction stations for energy recovery

2

Ermanno Lo Cascioa, Marc Puig Von Friesenb, Corrado Schenonea,*

3 4

a

DIME - Dipartimento di Ingegneria Meccanica, Energetica, Gestionale e dei Trasporti, Università degli Studi di

5

Genova, via All'Opera Pia 15/A, 16145, Genova.

6

b

SP Technical Research Institute of Sweden, box 857, SE-501 15, Borås, Sweden.

7

*

Author to whom correspondence should be addressed. E-Mail: [email protected]

8 9 10 11 12 13 14

keywords: natural gas pressure reduction stations; energy recovery; optimal design; system thermal integration; turbo expander; district heating Highlights • A novel comprehensive approach for pressure reduction stations retrofitting based on structured procedure.

15

• An optimization model enabling the maximum energy recovery in natural gas pressure reduction stations.

16

• Turbo-expander size defining the recovery potential and leading to non-smooth constrained optimization.

17

• Optimal design of natural gas expansion process for thermal integration with low-temperature processes.

18 19

________________________________________________________________________________________________

20

Abstract

21 22 23 24 25 26 27 28 29 30 31 32 33

In this paper, a structured retrofitting approach (SRA) to the near-optimal design of natural gas (NG) pressure reduction stations (PRSs) is presented. The SRA is designed by considering the waste energy recovery, system integration opportunities and long-term-based objectives to successfully address the entire PRS retrofitting process. The SRA is developed in four phases: pre-retrofit activities, preliminary and executive project design, implementation and commissioning and post-retrofit activities. For design optimization during the preliminary and executive project design phase, a novel mathematical model was developed based on the minimization of the levelized cost of energy (LCOE). The optimization model consists of a non-smooth constrained problem that has been solved by means of different solution methods and has been tested for different thermal peak loads, fuel purchase costs, and natural gas flow rates. Variations of the thermal design conditions from 2,900 kW to 1,300 kW for a constant annual heat demand, fluctuations of the percentage increase of the NG cost by 80-100-120-140%, and reductions of the NG user demand of 30% and 60% were considered. The results highlighted that the proposed optimization technique in PRS retrofitting identifies the best system configuration and turbo expander technology.

34

______________________________________________________________________________________

35 36 37 38 1

ACCEPTED MANUSCRIPT Nomenclature re cov erable Etotal Total energy recoverable for the assessment period (kWh) MAX Euser

Maximum daily thermal energy required by external users (kWh)

Ft

Fuel expenditure in the year t (euros)

hbf1.l .

Boiler 1, average optimal number of operating hours per day at full load (h)

hbf2.l .

Boiler 2, average optimal number of operating hours per day at full load (h)

f .l . hchp

Cogeneration unit, average optimal number of operating hours per day at full load (h)

hTEf .l . It Mt

Turbo expander, average optimal number of operating hours per day at full load (h)

N

re cov erable tot

Investment expenditure in the year t (euros) Operation and maintenance in the year t (euros) Effective amount of energy recovered (kWh)

Na Nb Nc

Potential recoverable energy case a (kWh)

PTE

Turbo expander nominal electric power (kWe)

PTEref PTEI PTEII qTEpreheating Qb1 Qb 2

Turbo expander reference nominal power (kWe)

Potential recoverable energy case b (kWh) Potential recoverable energy case c (kWh)

First stage turbo expander nominal electric power, low temp. confi. (kWe) Second stage turbo expander nominal electric power, low temp. confi. (kWe) Turbo expander preheating need at the reference nominal power (kWe) Gas fired boiler 1 nominal size (kW) Gas fired boiler 2 nominal size (kW)

Qchp

Cogeneration unit nominal thermal size (kW)

Qvlv ,a

Throttling valve preheating energy case a (kWh)

Qvlv ,b

Throttling valve preheating energy case b (kWh)

Qvlv ,c

Throttling valve preheating energy case c (kWh)

p .h. _ MAX Qvalve

Maximum throttling valve preheating thermal need (kW)

Qvlv

Throttling valve preheating energy (kWh)

MAX ext _ user

Q

Maximum external user thermal load (kW)

r Interest rate (%) Greek letters Heat exchanger effectiveness  HE Heat power ratio 

 alt inverter

Alternator efficiency (%)

Inverter efficiency (%) Abbreviations CHP Combined heat and power 2

ACCEPTED MANUSCRIPT COP DHC DHN ESCo RR LCA LCOEel LCOEth NG NPV OEP PI PRS SRA TE WER

Coefficient of performance District heating and cooling District heating network Energy service companies Internal rate of return Lifecycle assessment Levelized cost of electricity, (euros/kWh) Levelized cost of thermal energy (euros/kWh) Natural gas Net present value Optimal energy production Profitability index Pressure regulation station Structured retrofitting approach Turbo expander Waste energy recovery

39 40

1. Introduction

41 42 43 44 45 46 47 48 49 50 51 52 53 54 55

The global demand for natural gas (NG) is expected to grow by up to 50% by 2040 according to the International Energy Agency, and NG is considered the most promising fossil fuel of the 21st century [1]. Therefore, the European natural gas transmission infrastructure has been continuously expanding over recent decades to satisfy new territory needs. For instance, the total Italian infrastructure length has growth approximately 4% in less than 9 years [2]. A similar evolution was noticed worldwide, with the global demand growing by more than 30% over the last 15 years [3]. Regardless of its location, the overall transmission infrastructure is generally composed of different subparts: compression stations, transportation, storage, pressure regulation stations (PRSs) and distribution. The role of PRSs is to ensure a certain pressure drop between the transportation and distribution nodes. NG pressure regulation is commonly achieved by throttling valves. However, it is possible to upgrade this process, enabling energy recovery from the pressure drop by implementing turbo expander (TE) technology. The pressure drop along the NG pipelines is used to generate electricity that can be stored locally or returned to the utility grid. In this case, the NG preheating process, which is fundamental to avoid the formation of methane hydrate, would require higher thermal energy than that of the dissipative process.

56 57 58 59 60 61 62 63 64

The potential energy that might be recovered from PRSs has become an interesting issue for the research community. This aspect is also emphasized by the growing number of PRSs; for instance, in Turkey, according to Neseli et al. [4], the number of PRSs increased from 274 in 2010 to 320 in 2012, and in [4], for a single PRS, a potential electricity recovery of 4.11 GWh/year was estimated for an annual NG mean flow rate of approximately 4.953 kg/s, with a maximum of 6.36 GWh/year for the highest NG flow scenario. Borelli et al. [5] used numerical dynamic simulations to calculate a potential energy recovery of approximately 2.9 GWh/year for a total preheating need of approximately 3.1 GWh/year. In general, the energy recovery depends on the NG flow rate, the component characteristics and the pressure drop, as clearly highlighted in the literature [6], [7], [8] and [9].

65 66 67 68 69 70 71

Technological innovations in NG networks with energy recovery have been increasingly demonstrated by recent studies in which different system configurations have been analysed. Howard et al. [10] investigated a hybrid version of a TE application in which a portion of the preheating energy was provided by a molten carbonate fuel cell working with NG. The results showed that the maximum efficiency of a 12.000 scm per hour TE increased by approximately 10% after implementing the fuel cell. In this case, the total power production was approximately 0.934 GWh/year. To decrease the fuel consumption, Farzaneh-Gord et al. [11] considered the possibility of employing a geothermal heat 3

ACCEPTED MANUSCRIPT 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93

exchanger to boost the preheating process. However, this configuration does not consider the use of TE technology for waste energy recovery (WER). Arabkosarh et al. [12] investigated the possibility of coupling solar thermal collectors and PRSs. Here, simple payback periods of 2.3 years and 6.3 years were estimated with and without implementation of the TE. Kostowski et al. [13] reported the integration of a PRS with a cogeneration unit to propose a thermo-economic analysis, and the results seemed to show economic advantages over boilers. Farzaneh-Gord et al. [14], [15] studied the feasibility of employing uncontrolled and controlled linear heaters with a solar system in PRSs. Another study was presented by Farzaneh-Gord et al. [16], who focused on the possibility of employing vertical ground-coupled heat pumps and modelled the system by writing energy and thermal equations. Economic analysis was then adopted for selecting the most efficient layout of the geothermal system. In [17], with the same method, the authors identified the technical criterion for economic justification of employing combined heat and power (CHP) technologies in PRSs. In a similar manner, Arabkoohsar et al. [18] provided an energetic and economic analysis for PRSs coupled with CHP. The thermodynamic deterministic model was subsequently linked to an economic cost benefit analysis to evaluate the efficiency of the considered system configurations and, therefore, the optimal design. Robust traditional methods proved to be effective in guiding the integration of energy recovery from PRSs with different renewable energy sources, although they were challenging to develop. Finally, in [19], the possibility of reducing the water process temperature was investigated; this is a key aspect to exploit, for instance, with regard to low-temperature waste heat. Kostowski et al. [20] used a thermodynamic model of a PRS to define the thermoecological cost of electricity production; this cost has the advantage of providing a complete picture that includes not only the analysed system but also the production process of the resources consumed by that system, as well as the emission of pollutants.

94 95 96 97 98 99 100 101 102 103 104

The optimization of the system control was also investigated in other studies. A timing optimization model of a single-stage, single-acting, reciprocating expansion engine was presented in [21]. Pellegrino and Villecco [22] proposed an optimal control approach based on a fuzzy logic system; in this research, the proposed widely used optimization and control method was specifically applied for the first time to energy recovery in PRSs. Without delving into the physics of the phenomenon, the fuzzy control achieved good results, with negligible investment costs, a calculated energy recovery larger than 15% and a decrease in the maintenance costs of approximately 15%. In [23], Lo Cascio et al. proposed a multi-objective mathematical model for PRS optimal control where the electrical energy produced by the TE was differentiated based on how the preheating phase was accomplished (renewable sources or fossil-fuel-based production). Their model is based on dynamic modelling through an operational formulation involving costs, environmental impact and the form of energy.

105 106 107 108 109 110 111 112

The optimal design issue of the PRSs was investigated by Sanaye and Nasab in [24], who proposed a relatively fast method to select the size and number of components for PRSs using an objective function defined as the sum of the income and expenses. Subsequently, the optimum values of nine decision variables were obtained by maximizing the objective function using the genetic algorithm optimization technique. According to the authors, this hybrid semi-empirical procedure permits efficiently calculating the optimal configuration for a wide range of operating parameters. However, the variability of the WER potential, which is dependent on the NG flow patterns for a given TE, was not considered in this study.

113 114 115 116 117 118 119 120 121

In general, the abovementioned studies show that different techniques can be adopted to optimize the design of systems for energy recovery from PRSs, although based on the authors’ experience, the current methods for system synthesis problems, clearly described by Frangopolus et al. in [25], do not appear to be completely adequate for PRS retrofitting because of their complexity or inaccuracy. Therefore, this paper introduces the structured retrofitting approach (SRA) that intends to guide system designers and engineers during the entire PRS retrofitting process for energy recovery. While the traditional approach oscillates between heuristic methods and numerical analyses that, with difficulty, lead to a practical project assessment, the new proposed method allows quick and accurate determination of the optimal plant size. 4

ACCEPTED MANUSCRIPT 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141

The proposed SRA envisions a model for component dimensioning, which includes novel key aspects that emerge during the PRS retrofitting, such as the variability of the energy recovery potential and the system thermal integration. For this purpose, a non-smooth constrained optimization problem based on the minimization of the levelized cost of energy (LCOE) was developed. Finally, the SRA promotes a series of post-retrofit activities to emphasize project continuity and dissemination for practical knowledge capitalization. 2. Structured Retrofitting Approach This study has been conducted by considering the methodology in figure 1. In the first phase, a comprehensive retrofitting criterion for the PRS was developed. Then, the optimization problem formulation was formalized based on a reference system configuration. The formulated problem was solved by considering three main solution methods, such as genetic algorithms, pattern search, and the Fmincon solver. At this stage, recall that these methods are not the only ones suitable to find the optimal solution. Different approaches such as particle swarm optimization or simulated annealing could be employed for the same purpose. However, enriching the solution methods was not within the authors’ intended research scope. Further, once a performing solution method was found, to identify uncertainty handling strategies, the optimization problem was solved by considering different external peak loads, fuel purchase costs and NG flow scenarios. Optimal results assessment for different external thermal peak loads Mathematical optimization model design

142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163

Solution method assessment for a reference scenario

Optimal results assessment for different fuel purchase costs

General conclusions

Optimal results assessment for different NG flow scenario

Fig. 1. Study methodology.

Figure 2 illustrates the key steps of a general PRS structured retrofitting approach (SRA). The overall procedure, based on a long-term approach, is divided into four main temporal phases: pre-retrofit activities, preliminary and executive project design, implementation and commissioning and post-retrofit activities. For each phase, a given set of sequential tasks is defined, and the respective information, tools and consideration needs are highlighted. The proposed SRA has been designed to include long-term objectives and collateral activities. In fact, it emphasizes dissemination activities for project continuity and suggests considerations about the opportunity to involve research institutes and universities, of which the purpose is to boost and strengthen different key actions of the process, including dissemination, numerical model setup and validation, operational control model design and tests, forecasting, data analytics, performance assessment, analysis via numerical dynamic simulations, etc. These activities are fundamental for replicability and reference dataset definitions for similar future project implementations. The main task of the pre-retrofit activities consists of local data acquisition. Here, it is necessary to collect different types of specific information related to the PRS and its surrounding conditions: past NG annual flow rates, internal and external thermal and electrical loads, analysis of the potential integration opportunities with neighbouring processes, etc. 5

ACCEPTED MANUSCRIPT 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208

Furthermore, the preliminary and executive project design phase is schematized as a sub-iterative process where the first task consists of identifying a set of potential system configurations. Uncertainty factors (technology supplier long-term reliability, time to delivery, installation and time-to-commissioning) are fundamental to be considered at this stage since they might cause the PRS project to not completely succeed within the planned time schedules. Moreover, considering the magnitude of a single PRS retrofit, the lower contractual strength with respect to technology providers and industries, if compared to larger projects, could emphasize the negative effects of those uncertainty factors. Once the configuration set has been identified, the second step of the process involves the optimal component design. Here, the component dimensioning will be based on the information acquired during the preliminary phase. For this purpose, an optimal design approach will be presented in the following section. Furthermore, once the optimal component size has been defined, it is possible to start the executive PRS project. At this stage, the numerical dynamic simulation represents a powerful tool to understand the system behaviours of Hazardous and Operability (HAZOP) and "What if" analysis, energy performance assessment and transient analysis. An optimal economic design must guarantee that a cost-effective configuration has been achieved. However, an economic analysis and risk assessment are still necessary to properly evaluate component implementation schedules, control strategies and operational planning and, eventually, energy performance contracting, target definition and baseline assessment. As is commonly known, many tools are available for this purpose: levelized cost of energy (LCOE), net present value (NPV), internal rate of return (IRR), profitability index (PI), discounted payback period (DPBP), lifecycle assessment (LCA), etc. Furthermore, during the implementation and commissioning phase, the custom optimal control strategy definition is fundamental to ensure energy savings and carbon emission minimization. At this stage, it is important to highlight that the optimal system design and the optimal control strategy should not be considered as two different and unconnected issues. In fact, the identified optimal design conditions, i.e., the average energy production required to guarantee the validity of the optimal design conditions, must be taken into consideration during the optimal control strategy definition. This is fundamental to keep the optimal design points valid during the system lifecycle. In contrast, the design conditions identified during the preliminary and executive project design phase can no longer be considered optimal in the further phases. In this study, this problem is addressed by defining the OEP index. During the commissioning phase, the system performance data will be acquired, stored and further disseminated. These data are important to proceed with a parallel numerical model validation. Numerical model validation is a fundamental step to be properly achieved to proceed with system performance forecasting. This task is important to consolidate the system operation strategies of optimal control adjustments, operation and maintenance schedule settings, etc. After these activities are completed and the implementation and commissioning phase ends, the postretrofit activities must be conducted. For this purpose, the key task consists of measurement and verification activities (M&V). Finally, once the process has been completed, the retrofit report should be prepared and frequently updated.

6

ACCEPTED MANUSCRIPT SRA Preliminary and executive project design

Tasks

    Preliminary targets and budget set

 

Main activities and considerations

      

Neighbor survey and potential partnership evaluation Neighboring industrial thermal process, pinch analysis and loads Neighboring local electricity usage opportunities DHN type and loads Process temperature reduction opportunities and constrains Project financing strategy Technical and financial out searching opportunities Strategic partnership research

Yearly natural gas flow trends Internal electric loads Heating and cooling loads Solar radiation and availability Local temperatures Geothermal availability

OEP

Main driving forces to consider

PRS configuration identification

Tasks



Post-retrofit activities

Main data

PRS data acquisition

Boundary condition analysis

Implementation and commissioning



PRS preliminary optimal design



Executive project





Detailed dynamic simulation

Preliminary initial budget and expected future financial incomes Technologies supplier reliability, assistance and partnership Preliminary design timeline and human resource availability Other uncertainty factors

  

Main factor to consider Primary energy tariffs trends New technologies availability and maturity Minor mistakes identification

Numerical model validation

Tools Multi-objective operational optimization techniques

Project economic and risk assessment

Forecasting analysis Tasks

Pre-retrofit activities

no

Potential system improvements identification

Satisfied? Joining opportunities and main supporting tools

Custom optimal control strategy definition

Action plan definition

Tasks

Out searching/ partial outsourcing

yes

Involvement of universities and research institutes

System upgrade implementation

Start up and tests

Measurement and monitoring Data storage

Dynamic simulation software: Matlab Simulink, UniSim Design etc

Retrofit report

no Satisfied Experience Dissemination

LCOE, NPV, IRR, PI, DPBP etc.

yes Regular Operation and Maintainance

209 210

Fig. 2. Structured Retrofitting Approach (SRA).

7

ACCEPTED MANUSCRIPT 211

3. Optimal design model

212 213 214 215

The design optimization during the preliminary and executive project design is an essential piece of the SRA and a key element in the proposed methodology, and a relevant part of the SRA success depends on this aspect. For this reason, a novel mathematical model has been developed, which is based on the minimization of the levelized cost of energy (LCOE) defined by Equation 1: n

 (I t 1

LCOE 

t

 M t  Ft ) (1)

(1  r ) t n

E t 1

t

(1  r ) t

I t , M t and Ft are, respectively, the investment expenditure, maintenance and operation costs,

216

where

217

and fuel expenditure in the year t. Then, r and n are, respectively, the discount rate and the expected

218

lifetime of the components or systems, and

219 220 221 222 223 224 225 226

Distinct from Sanaye and Nasab [24], this original model was designed by considering the variability of the WER potential, which mainly depends on the NG flow pattern and the size of the selected TE. For this purpose, the model was structured as a non-smooth constrained optimization problem. Moreover, the optimization model was designed considering the system thermal integration opportunities with the neighbouring users as well. As is known [26], this is a key aspect for strategic PRS retrofitting and for efficient energy exploitation in urban districts. Finally, the model is also easily scalable for system configurations where two TEs are normally implemented in a series due to a higher expansion ratio or for water-process temperature reduction.

227 228 229 230 231 232 233 234 235 236

The proposed optimization approach is given in the system configuration shown in figure 3. This layout represents a reference configuration for energy recovery from natural gas PRSs. The chosen system configuration involves two gas fired boilers, a gas fired CHP unit and a single TE with its heat exchanger, working in parallel with the throttling valves, which operate at very low or high mass flow rates to manage the turbine functioning range. The thermal plant is also connected to a group of buildings through a district heating network, which is supplied by the CHP and boilers. These boilers have an integrative function for peak heat loads and perform a supplementary function during CHP downtime. Though the model is applied to this reference layout, which has a standard energy recovery from the PRSs, it can be extended to diverse configurations. Likewise, the results obtained from the proposed optimization method have a general validity and are suitable for wide generalization.

237

The design approach is based on a few main global constraints, depending on the system considered. The

238

first constraint is related to the amount of total recoverable energy Etotal

239 240 241 242 243 244

a time length t. Generally, the total amount of recoverable energy for a certain PRS and TE is strictly dependent on the NG flow rates, pressure drop, temperature conditions, system configuration, and the type of system control. Another four constraints refer to maximum size, redundancy condition, overall thermal energy balance and TE synchrony (for TEs in series). From the results of the design optimization process, the optimal sizes of the nominal components and the respective average optimal energy production (OEP) will be known for a given time length t and for each component considered. In a future 8

Et is the energy generated by the system.

re cov erable

from the turbo expander for

ACCEPTED MANUSCRIPT 245 246 247 248 249 250 251 252

stage, once the system process and instrumentation diagram (P&ID) has been defined, the OEP conditions should be considered to properly address the system control optimization problem, as previously mentioned. In the following section, the optimization problem will be formulated with reference to the system configuration shown in figure 3. However, it can be extended and adapted to different configurations with minor modifications. As shown in figure 3, the redundancy condition is generally satisfied by two boilers that work in conjunction with an internal combustion cogeneration unit. These three components will satisfy internal (pre-heating process) and external (DHN) needs. One single TE is considered in the reference optimization problem. B1

B2

Twin pump

Local users

Recirculating pump Cogeneration unit Flow control valve

NG 24 bar

National grid

Local grid

TE

E-1

Natural gas load

V-1

E-3

E-7

254

City users

V-4

High pressure NG grid Thermal grid

253

Let down sub-stations

V-3

E-6

Medium voltage electric grid

Medium pressure NG grid

TE pressure controller

Throttling valve pressure controller

Fig. 3. Reference system configuration [23].

255 256

3.1. Problem statement

257

The problem objective function is expressed as:

minLCOE 

(2)

MAX p .h. _ MAX Qb1  Qb 2  Qchp  PTE  Qext _ user  Qvalve

(3)

258

subjected to the following constraints.

259

Component size

260

The nominal size balance is defined as:

261

Qb1 , Qb 2 and Qchp are, respectively, the nominal sizes of the two boilers and the CHP, and PTE

262

where

263

represents the turbo expander nominal electric power. For simplicity, the heat power ratio  is considered 9

ACCEPTED MANUSCRIPT 264 265

constant for this model. Normally, the heat power ratio varies between 0.95 and 1.1 depending on the operating conditions and the system characteristics [19]. The heat power ratio is defined as: preheating  qTE   HE  ref  PTE   altinverter

  

(4)

266

 alt

inverter

267

where  HE ,

268

the effectiveness of its generator and the DC/AC inverter efficiency. The

269

preheating need at the reference nominal power

270

considered as constant for this model.

271

The terms Qext _ user and Qvalve

272 273 274 275

(DHN or other integrated process) and the maximum throttling-valve-preheating needs. The over dimensioning by considering these values as constant before and after the TE implementation will further ensure safe system operations for the case scenarios in which the system has to work under certain extreme operating conditions of NG flow peaks or temporary TE faults.

276

Redundancy

277 278

Heat production must always be ensured in case of component faults. Otherwise, methane hydrates will suddenly occur. Thus, the redundancy condition is defined as:

MAX

and

are, respectively, the effectiveness of the heat exchanger linked to the TE,

p .h. _ MAX

qTEpreheating , instead, is the TE

PTEref . This ratio, as a first approximation, will be

in Equation 3 represent the maximum expected external thermal load

Qb1  Qb 2  0

(5)

f .l . MAX Qb1 hbf1.l .  Qb 2 hbf2.l .  Qchp hchp  PTE hTEf .l .  Qvlv  E user

(6)

279

Overall thermal energy balance

280

The thermal energy balance is defined as:

281 MAX

Qvlv

282

where Euser represents the average daily maximum thermal energy required by external users, and

283 284 285

represents the average throttling-valve-preheating need expected after the retrofit. For this reason, it will depend on the type and size of the implemented TE, which is a variable of the optimization problem, for instance, considering the Honeywell TE models [27] and given the PRS conditions:

550  PTE  450  Qvlv  Qvlv ,a  450  PTE  160  Qvlv  Qvlv ,b  P  160  Q  Q vlv vlv ,c  TE 286

10

ACCEPTED MANUSCRIPT 287

Therefore, Qvlv ,a , Qvlv ,b and Qvlv ,c are the values of throttling-valve-preheating needs for different TE sizes

288

and the given PRS. The terms hb1 , hb 2 , hchp and

289 290 291

number of operating hours per day at full load for boilers 1 and 2, CHP and TE. For a given time length, the product between the optimal nominal sizes, the respective optimal average number of operating hours per day and the number of expected operating days will result in the OEP.

292

Natural gas balance

293

The overall recoverable energy balance is defined as:

f .l .

f .l .

f .l .

hTEf .l . , respectively, represent the average optimal

nPTE hTEf .l .  N totre cov erable

(8)

294 295

where n is the number of days per year that the TE is expected to work considering system stops for

296

maintenance, and N tot

297

n days of work considered. The

298 299

strictly related to the NG flow and varies with the TE size, since each TE has different technical constraints upon the admissible peak NG volume flow. Then:

re cov erable

represents the amount of average energy that might be recovered within the

Qvlv cannot be considered as a continuous function. The N totre cov erable is

re cov erable 550  PTE  450  N tot  Na  re cov erable  Nb 450  PTE  160  N tot  re cov erable  Nc  PTE  160  N tot

300

N a , N b and N c are the quantities of the average recoverable energy for different TEs under certain

301

where

302

PRS conditions and time lengths.

303

TE synchrony

304 305 306

In the case of a low-temperature system configuration where a double-stage expansion process is used, the power and operation synchrony constraints between the TEs must be added to the mathematical problem formulation, such as:

hTEf .l1.  hTEf .l2.  0

(7)

307 I TE

II TE

(8)

P - P =0 308

PTEI and PTEII are the nominal powers of the first- and second-stage expansion turbines in the low-

309

where

310

temperature configuration, respectively.

311

3.2. Model limitations and enhancements 11

ACCEPTED MANUSCRIPT 312 313 314 315

Some minor limitations must be underlined for the proposed approach. The abovementioned OEP does not consider the effect of efficiency at partial loads for each component. For this purpose, the efficiency must be considered as a reference value to properly address the system optimal control. Finally, the application of this model might be time consuming for non-expert users since it requires a preliminary effort to assess

316

the

317 318

which is necessary to properly identify what the future NG flow rates are for the considered period of assessment.

319

4. Case study

320 321 322 323 324 325 326 327 328 329 330 331

The optimization approach has been tested for the case study shown in figure 3 that mirrors the PRSs energy recovery system developed within the CELSIUS project [26]. The NG inlet pressure and temperature are, respectively, 25 bars and 11ºC with a fixed pressure drop of about 20 bars. Since the upstream pressure and the pressure drop through the TE are assigned, as in the real operation, the outlet temperature is controlled by the isentropic expansion efficiency, which in turn depends on the mass flow rate, i.e., on the turbine operating conditions. Though the formation temperature of methane hydrates is a function of the gas composition, the TE inlet temperature in the test case ensures that no separation of a solid phase occurs during the expansion. Considering the NG composition range and isentropic expansion efficiency values, the correlations of Motiee et al. [28], Towler et al. [29] and Kidnay et al. [30] have been used to determine the upstream temperature to avoid methane-hydrate formation. Under the given conditions, a water process temperature of 85ºC prevents methane hydrate formation and ensures safe operating conditions.

332

First, the optimization scenarios were assessed to estimate the values of

333 334 335 336

based on the given conditions. Furthermore, the model was implemented in a Matlab environment. The results of the genetic algorithm, pattern search and Fmincon methods are shown in the following. Then, the optimization results for different thermal design conditions, fuel purchase cost and different NG flow scenarios are presented. The Honeywell TE solutions [27] were considered for this case study.

337

4.1. Calculation conditions and optimization scenario assessment

338 339 340 341 342 343 344 345

The calculation period n for the LCOE estimation was set to 20 years, which is considered as the average component lifetime in [31]. An interest rate r of about 5% was considered. Initially, for industrial users, a total linear increase of 60% on gas price tariffs was considered for the calculation period. Furthermore, the optimization results for different economic projections are presented. The injected electricity tariffs for the CHP and TE were set at 0.28 and 0.11 euros/kWh, respectively, and were considered constant throughout the assessment period. Investment cost functions are reported in table 3, and the component lower and upper bounds are in table 4. Operation and maintenance costs are calculated as 2% of the investment cost per year [31].

346 347 348 349 350 351 352

Different system operating scenarios were analysed in normalized form for different TE sizes and NG flow rates (figure 4). Here, the standard scenario ("standard flow") is derived from real plant monitoring with its values normalized to a reference value here called scm*. Lower value scenarios ("30% lower" and "60% lower") are, instead, hypothetical scenarios where the NG flow rates were reduced, respectively, by about 30% and 60% with respect to the standard scenario. The gap shown in figure 4 for time values of about 6,100-6,500 hours, corresponds to the temporary system stop for maintenance purposes. These values are referred to as a period length of 1 year. Then, they were considered the same for all 20 years for this 12

N a , N b and N c values, as well as the Qvlv ,a , Qvlv ,b and Qvlv ,c . This requires long-term estimation,

N a , N b , N c Qa , Qb , and Qc

ACCEPTED MANUSCRIPT 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368

assessment. It is convenient to highlight that these scenarios are under a pure pneumatic system control, in which the pressure regulator (throttling valves and TE) set points are defined to give operation priority to the TE. Moreover, these scenarios use the boundary conditions reported in table 1 (pressure drop, inlet and outlet temperatures, etc.), and the following results were calculated by means of hourly simulations through a calculation spreadsheet.. Thus, throttling-valve openings will follow the trends shown in figure 5 (for a 550 kW TE). Consequently, considering a nominal size of about 550 kWe, the electricity shown in figure 6 can be produced. Instead, figure 7 shows the WER for different TE sizes with the standard NG flow scenario. From figure 8a and 8b, it is possible to observe that for a TE of about 550 kWe nominal, the total recovered energy proportionally decreases together with the throttling-valve-preheating need from the "standard flow" scenario to the "30% lower" scenario, However, if the NG flow is about 60% with respect to the "standard scenario", the throttling-valve-preheating need increases. This is because the TE technical lower flow limit is more frequently overstepped with respect to the higher NG flow scenarios, as shown in figure 6. For the same reason, in the 60% lower NG flow scenario (figure 8a), the energy recovered by implementing a nominal 450 kWe TE would be higher than by implementing a nominal 550 kWe TE. This means that an energy recovery increase of about 4.5% could be achieved by implementing a component with a cost 16.3% cheaper. Thus, the need to formulate a proper optimization model is evident.

369

Table 1. Calculation conditions.

Inlet pressure [bar] Oulet pressure [bar] Inlet NG temperature [°C] Inlet NG pressure [bar] Outlet NG pressure [bar] TEs characteristics

24 5 15 24 5 MTG550-450-160, Honeywell®

Flow rate (scm/scm*)

370 4 3 2 1 0 0

800

1600

2400

3200

Standard scenario

371 372

(h) 4000Time 4800

5600

30 percent lower

6400

7200

8000

8800

60 percent lower

Flow rate (scm/scm*)

Fig. 4. Normalized NG flow scenarios considered.

4 3 2 1 0 0

373 374

800

1600

2400

3200

Standard scenario

4000 4800 Time (h) 30 percent lower

5600

6400

7200

8000

8800

60 percent lower

Fig. 5. Throttling valve normalized flows for the nominal 550 kW TE.

13

Energy (kWh/kWh*)

ACCEPTED MANUSCRIPT 3 2 1 0 0

800

2400

3200

Standard scenario Energy (kWh/kWh*)

375 376

1600

4000 4800 Time (h)

5600

30 percent lower

6400

7200

8000

8800

60 percent lower

Fig. 6. Normalized recovered electrical energy for the nominal 550 kW TE for different NG flow scenarios.

3 2 1 0 0

800

1600

2400

3200

4000 4800 Time (h)

550 kW

377 378

450 kW

5600

6400

7200

8000

8800

160 kW

1.2

5

1

4

0.8 0.6

550 450 160

0.4 0.2 0 Standard scenario

30 % lower

60 % lower

Energy (kWh/kWh*)

Energy (kWh/kWh*)

Fig. 7. Normalized recoverable energy for different TE sizes for the fixed standard NG flow scenario.

3 550 450 160

2 1 0 Standard scenario

30 % lower

60 % lower

379 380

Fig. 8. Yearly total recoverable energy (a); total yearly throttling-valve-preheating need (b).

381 382

4.2. Comparison of solution methods

383 384 385 386 387 388 389 390

The standard flow scenario in figure 4 was considered as a reference case and for testing purposes. Therefore, three different methods were considered: genetic algorithm (GA), pattern search (PS) and Fmincon (find minimum of constrained nonlinear multivariable function) solver. For the GA, a population size of approximately 200 individuals was considered. Table 2 shows the initial point used for the optimization calculation. From figure 9a, the optimal solution found using the GA suggests a system configuration where the TE size is relatively too high compared with the OEP (figure 9b). Practically, this situation would increase the investment cost, while sensibly limiting the energy production. This would obviously lead to obtaining the highest LCOEel for this case study, as shown in figure 10. On the other hand, 14

ACCEPTED MANUSCRIPT 391 392 393 394 395 396 397 398 399

using the PS solution method, the thermal production would be addressed mainly by the CHP unit, which would lead to achieving the highest LCOEth. This is due to the higher O&M costs of the CHP in comparison with gas-fired boiler units. Finally, the Fmincon solver, seems to find the best performing optimal solution. This is because the solution foresees the best balance between the size of the components and energy production. As results for this case (figure 10), the LCOEth is slightly higher than that found for the GA case, but for the Fmincon case, the overall LCOE is at its minimum because of achieving the lowest LCOEe.. Finally, for the case scenario considered and for the non-smooth constraint function, the Fmincon solver turns out to be the best performing solution method and, for this reason, was used to identify the optimum design conditions.

400

Table. 2. Initial calculation conditions.

401

Qb1 [kW] 800

hb1 [h] 10

Qb2 [kW] 800

402

hb2 [h] 10

Qchp [kW] 169

hchp [h] 10

PTE [kW] 170

hTE [h] 10

a

b

R2

1552

-0.5791

0.9367

1349

-0.06584

0.9779

2602

-0.115

0.99

Table 3. Investment cost curves [26],[27] and [31].

Component

f(x)

Standard gas-fired boilers

𝑎 × (𝑄 𝐵 )

𝑏

CHP unit

𝑎 × (𝑄𝐶𝐻𝑃)

Turbo expander

𝑎 × (𝑄𝑇𝐸)

403

𝑏

𝑏

Table 4. Components lower and upper bounds.

Component Standard gas-fired boilers (kW) CHP unit power (kW) CHP unit thermal power (kW) Turbo expander power (kW)

Upper bound 2000 550 630 550

Lower bound 25 160 160 160

404 3 2.5

1500 GA PS Fmincon

1000 500

406

2

GA PS Fmincon

1.5 1 0.5

0

405

OEP (GWh/y)

Nominal power (kW)

2000

0 Qb1 Qb2 Qchp P TE

Qb1 Qb2 Qchp P TE

Fig. 9. Optimal solution methods comparison: a) components size; b) OEP.

15

ACCEPTED MANUSCRIPT 0.08

LCOE

0.06 GA PS Fmincon

0.04 0.02 0

407 408

LCOE th

LCOE el

Fig. 10. LCOE comparison between different optimization methods.

409

4.3. Assessing optimization results

410 411 412 413 414 415 416 417 418 419

Figures 11a-b, and 12 show the effect of varying the thermal design conditions from 2,900 kW to 1,300 kW on the optimal results for a constant annual heat demand (lower thermal design conditions correspond to smoother heat production with relatively low peak loads). These values are typical of small-sized DHN. On the one hand, a peak value of 2,900 kW would correspond to the case in which the thermal loads are characterized by more variable loads as compared with the 1,300 kW case (smooth production). From figure 11a, the direct effect of this variation on the optimal results is a proportional reduction in the sizes of the two redundant gas-fired boilers. The CHP and TE units seem to maintain a stable trend. The same consideration is valid for the average optimal energy production calculated with a slight difference for the 1,800 kW case, where it is possible to observe a slight increase in the boiler production compared with the previous cases (figure 12).

420 421 422 423 424

Figures 13a-b and 14 show the effect of varying the percentage increase in the fuel cost (natural gas in this case) by 80-100-120-140%. This fuel cost increase was considered linear for the assessment period. From figures 13a and 13b, it is clear that there is no direct effect on the optimal size or optimal average energy production. The only difference observed is related to the values of the LCOEth and LCOEel, which proportionally increase with higher fuel purchase cost scenarios.

425 426 427 428 429 430 431 432 433

The effect on the optimal results of varying the NG user demand when keeping the outlet pressure at 5 bars is shown in figures 15a-b and 16. It is possible to observe that the component sizes, thermal energy production and LCOEth remain the same for all NG flow conditions. The only direct effect is observed on the optimal solutions related to the CHP and TE units. On the one hand, figures 15a and 15b show that both the optimal sizes and the optimal average energy production proportionally decrease at lower flows. On the other hand, the LCOEel proportionally increases (figure 16). This is because at lower NG flow scenarios, the electricity produced by the TE is obviously lower. For this reason, for a fixed annual thermal energy demand (DHN) user, the total thermal production will be lower, since the preheating need is lower in comparison with the first cases.

16

ACCEPTED MANUSCRIPT 4

1500

2900 kW 2300 kW 1800 kW 1300 kW

1000 500

OEP (GWh/y)

Nominal power (kW)

2000

3

2900 kW 2300 kW 1800 kW 1300 kW

2 1 0

0

Qb1

Qb1 Qb2 Qchp P TE

434 435

Qb2

Qchp

P TE

Fig. 11. Optimal results comparison for different user thermal design conditions a) components size; b) OEP.

0.04

LCOE

0.03

2900 kW 2300 kW 1800 kW 1300 kW

0.02 0.01 0 LCOE th

436 437

LCOE el

Fig. 12. LCOE comparison between different user thermal design conditions.

3 2.5

1500

80% 100% 120% 140%

1000 500 0

439

2

80% 100% 120% 140%

1.5 1 0.5 0

Qb1

438

OEP (GWh/y)

Power (kW)

2000

Qb2 Qchp P TE

Qb1

Qb2

Qchp

P TE

Fig. 13. Optimal results comparison for different fuel purchase cost increases: a) Components size; b) OEP.

0.05

LCOE

0.04 80% 100% 120% 140%

0.03 0.02 0.01 0

440 441

LCOE th

LCOE el

Fig. 14. LCOE comparison for different fuel purchase cost increases.

17

ACCEPTED MANUSCRIPT 3 2.5

1500

Standard scenario 30 % lower

1000 500

60 % lower

0

443

2

Standard scenario 30% lower

1.5 1 0.5

60% lower

0 Qb1 Qb2 Qchp P TE

442

OEP (GWh/y)

Power( kW)

2000

Qb1 Qb2 Qchp P TE

Fig. 15. Optimal results comparison for different NG flow scenarios: a) components size; b) OEP.

0.04

LCOE

0.03 Standard scenario 30% lower 60% lower

0.02 0.01 0

444 445

LCOE th

LCOE el

Fig. 16. LCOE comparison for different NG flow scenarios.

446 447

5. Discussion and recommendations

448 449 450 451 452 453 454

In general, in a PRS retrofitting process, for a given NG flow rate, it is necessary to identify the ideal TE size to enable maximum exploitation of the WER potential to ensure the minimum investment expenditure. This issue is caused by the technical upper and lower flow limits typical of those machines [27]. Thus, appropriate optimization techniques, such the one proposed in this study, are needed to identify the optimal design of the component. With reference to the optimization method, the case study presented suggests that the Fmincon solver can properly address the optimization problem to ensure the lowest overall LCOE. However, other optimization methods might be suitable as well.

455

Furthermore, it is critical to highlight an important aspect of the optimal design related to the forecasting

456

uncertainty. For the application of the proposed model, the preliminary step consists of estimating,

457

N b and N c and Qa , Qb , and Qc , which determine the WER. For this reason, the obtained optimal

458 459 460 461 462 463 464 465

solution is valid if the scenario occurs in reality. However, the identified scenario (fixed a priori) could change for uncertain reasons, which could affect the WER. To address this problem, in the previous section, the effect of varying the NG flow was evaluated. As a result, for this specific system configuration, it is clear how the effect of this variation will primarily impact the OEP of the TE and CHP rather than the component sizes. Then, to maintain the validity of the optimal design results, the NG flow uncertainty can be easily handled by adjusting the operating number of hours for the components based on the NG scenario. In the second stage, for the plant control, this process would be possible by considering the constraint relative to the OEP index for the control optimization strategy definition.

Na ,

18

ACCEPTED MANUSCRIPT 466 467 468 469 470 471 472 473 474 475

The performed simulations and their results support the effectiveness of the proposed retrofit on the process performance. Although it is currently impossible to field test the design indications obtained from the SRA, the elements collected during the work converge to confirm the efficiency of the optimization process. Both the comparison between the results obtained with the different optimization methods (Fig. 9, Fig. 10) and the sensitivity analysis performed with the variation of the main process parameters (Fig. 11, Fig. 13, Fig. 15) prove that the results steadily converge towards the best solution. However, if the design optimization problem ensures that the most cost-effective retrofit solution is selected, then the real dynamic functioning of the system may be sensibly affected by elements such as uncertainty factors and/or human factors, which can alter the overall final performance of the system and then the capacity of the model to predict it. This aspect cannot be known a priori during the optimization.

476 477 478 479 480 481 482 483 484 485 486 487 488 489

The optimal result assessment instead showed that by lowering the thermal peaks and maintaining the average user thermal demand (smooth thermal production) as constant, it is possible to observe slight decreases in the LCOE values until the thermal design condition reaches a certain threshold. In fact, passing from a thermal design condition of approximately 1,800 kW to 1300 kW, it is possible to observe a slight increase in the LCOEth and LCOEel of approximately 3.7% and 7.5%, respectively. This finding is because the optimal solution foresees a higher production for the CHP-IC unit to balance the lower thermal production of the boilers with respect to the previous cases. Practically, the CHP-IC unit presents higher operating costs than those of the boiler units. Therefore, this result seems to suggest that for relatively low thermal needs and under these specific conditions, the CHP-IC unit might not be the most cost-effective retrofitting solution when it works in conjunction with the boiler units. This conclusion was also found in [17] and [18], where a different approach to component sizing was used. However, under different conditions, the CHP-IC unit might become a cost-effective retrofitting solution, as shown in [13]. Therefore, the implementation of the CHP unit in PRS retrofits should be accurately evaluated case-by-case to achieve cost-effective retrofitting interventions.

490 491 492 493

Furthermore, the effect of the linear fuel purchase cost increase was analysed for the different scenarios of an 80-100-120-140% increase with respect to the current fuel costs. As a result, the increase in the fuel costs neither influences the optimal design results nor the OEP in this model. The only difference can be observed in the LCOE, which will proportionally increase with the fuel costs.

494

6. Conclusions

495 496 497 498 499 500 501 502 503

Energy recovery from a pressure reduction in natural gas networks was addressed through a structured retrofitting approach aimed at optimal design. The scope of this methodology is to drive the entire retrofitting process and enable project success. The developed method was demonstrated to maximize energy savings and reduce the complexity and calculation time. The architecture of the proposed method is quite different from the current design techniques that are either heuristic or computationally demanding. The novel formulation of a non-smooth constrained optimization problem based on the LCOE minimization allows managing variable boundary conditions. Moreover, the innovative technological impact of the SRA method enables the exploitation of situations for which the current design does not allow and for which only an optimized size can recover waste energy.

504 505 506 507

The proposed case study shows that the SRA optimization technique permits identifying the best system configuration in the PRS retrofit, particularly with regard to the turbo expander model and heat supplier technology. All the different issues of the integrated system design are addressed with the optimization technique. The turbo expander design provides helpful recommendations concerning the number and size 19

ACCEPTED MANUSCRIPT 508 509 510 511

of the machinery. Moreover, the effect of varying the natural gas flow patterns on the optimal results can be evaluated to address the uncertainty. Optimization can be maintained along the system life for variable operating conditions by proper system operation management, which can be easily addressed using the OEP index.

512 513 514 515 516 517 518 519

Similar to previous studies, it has been confirmed that the combined heat and power technologies are not suitable for all retrofitting scenarios. The actual benefits are intimately linked to the natural gas flow in the station. If the flow is too low, the implementation of the technology is not justifiable. Furthermore, the repercussions of increasing fuel costs on the optimal results were evaluated. The results for the case study proposed in which all thermal supply technologies (boilers and cogeneration unit) were fed by natural gas showed that the eventual costly operating scenarios are disconnected from the optimization problem in the sense that no consequence on the diminishing component was observed. However, this situation might be different for multi-fuel-based systems, and for those cases, the optimization results might be different.

520 521 522 523 524 525 526

Finally, this study highlights that optimal retrofit interventions are challenging to design due of their sensitivity to calculation scenarios and boundary conditions. The lack of appropriate design tools leads to incomprehension and underestimation of the actual benefits in terms of energy savings and carbon emission reductions of PRS retrofitting, thus limiting the diffusion of this smart, energy-saving technology. Instead, through an effective optimization tool such as the SRA, it is possible to achieve the effective exploitation of the waste energy recovery potential from natural gas transportation and distribution networks.

527

Acknowledgements

528 529 530 531 532

The present work has been developed in the framework of a collaboration with the C. M. LERICI Foundation (www.cmlerici.se) and with the support of the Grant 2015M8S2PA "Clean heating and cooling technologies for an efficient energy smart grid" by the Italian MIUR.

533 534

[1]

W. E. Outlook 2016, available online: http://www.worldenergyoutlook.org/publications/weo-2016/ (Accessed 23/03/2017).

535 536

[2]

Capacita' di trasporto, available online: http://www.snamretegas.it/it/servizi/Annotermico_2016_2017/Capacita_trasporto/ (accessed 06/04/2017).

537 538 539

[3]

A. Toscano, F. Bilotti, F. Asdrubali, C. Guattari, L. Evangelisti, C. Basilicata. Recent Trends in the World Gas Market : Economical , Geopolitical and Environmental Aspects. Sustainability 2016; 1:242030.

540 541

[4]

M. Alparslan, O. Ozgener, L. Ozgener. Energy and exergy analysis of electricity generation from natural gas pressure reducing stations. Energy Convers Manag 2015; 109:120-93.

542 543

[5]

Borelli, F. Devia, E. Lo Cascio, C. Schenone, and A. Spoladore. Combined Production and Conversion of Energy in an Urban Integrated System. Energies 2016; 1:17-9.

544 545 546

[6]

E. Ashouri, F. Veysi, E. Shojaeizadeh, M. Asadi. Journal of Natural Gas Science and Engineering The minimum gas temperature at the inlet of regulators in natural gas pressure reduction stations ( CGS ) for energy saving in water bath heaters. J Nat Gas Sci 2014; 230:240-21.

547 548

[7]

I. Andrei, T. Valentin, T. Cristina,T. Niculae. Recovery of Wasted Mechanical Energy from the Reduction of Natural Gas Pressure. Procedia Eng 2014; 986:990-69.

References

20

ACCEPTED MANUSCRIPT 549 550 551

[8]

D. Borelli, F. Devia, M. Marré Brunenghi, C. Schenone, A. Spoladore. Waste energy recovery from natural gas distribution network: Celsius project demonstrator in Genoa. Sustainability 2015; 16703:16719-7.

552 553

[9]

I. Fletcher,C.S. Cox, W.J.B. Arden, A. Doonan. Modelling of a two-stage high-pressure gas reduction station. Appl Math Model 1996; 741:749-20.

554 555 556

[10]

C. Howard, P. Oosthuizen, and B. Peppley. An investigation of the performance of a hybrid turboexpander-fuel cell system for power recovery at natural gas pressure reduction stations. Appl Therm Eng; 2011; 2165:2170-31.

557 558 559

[11]

M. Farzaneh-gord, R. Ghezelbash, A. Arabkoohsar, L. Pilevari, L. Machado. Employing geothermal heat exchanger in natural gas pressure drop station in order to decrease fuel consumption. Energy 2015; 164:176-83.

560 561 562

[12]

A. Arabkoohsar, M. Farzaneh-gord, M. Deymi-dashtebayaz, L. Machado. A new design for natural gas pressure reduction points by employing a turbo expander and a solar heating set. Renew Energy 2015; 239:250-81.

563 564

[13]

W. J. Kostowski and S. Usón. Thermoeconomic assessment of a natural gas expansion system integrated with a co-generation unit. Appl Energy 2013;58:66-101.

565 566 567

[14]

M. Farzaneh-gord, A. Arabkoohsar, M. D. Dasht-bayaz, V. Farzaneh-kord. Feasibility of accompanying uncontrolled linear heater with solar system in natural gas pressure drop stations. Energy 2012; 420:428-41.

568 569 570

[15]

M. Farzaneh-gord, A. Arabkoohsar, M. D. Dasht-bayaz, and L. Machado. Energy and exergy analysis of natural gas pressure reduction points equipped with solar heat and controllable heaters. Renew Energy 2014; 258:270-72.

571 572 573

[16]

M. Farzaneh-gord, R. Ghezelbash, M. Sadi, A. Jabari. Integration of vertical ground-coupled heat pump into a conventional natural gas pressure drop station : Energy , economic and CO 2 emission assessment. Energy 2016; 998:1014-112.

574 575 576

[17]

V. Farzaneh-kord, A. B. Khoshnevis, A. Arabkoohsar, M. Deymi-dashtebayaz. Defining a technical criterion for economic justi fi cation of employing CHP technology in city gate stations. Energy 2016; 389:401-111.

577 578 579

[18]

A. Arabkoohsar, Z. Gharahchomaghloo, M. Farzaneh-gord, R. N. N. Koury. An energetic and economic analysis of power productive gas expansion stations for employing combined heat and power. Energy 2017; 737:748-133.

580 581 582

[19]

D. Borelli, F. Devia, E. Lo Cascio, C. Schenone.Energy recovery from natural gas pressure reduction stations : Integration with low temperature heat sources. Energy Convers Manag 2018; 274:283159.

583 584

[20]

W. J. Kostowski, S. Usón, W. Stanek, P. Bargiel. Thermoecological cost of electricity production in the natural gas pressure reduction process. Energy 2014; 10:18-76.

585 586 587

[21]

M. Farzaneh-gord and M. Jannatabadi. Timing optimization of single-stage single-acting reciprocating expansion engine based on exergy analysis. Energy Convers Manag 2015; 518:529105.

588 589

[22]

Arcangelo Pellegrino and Francesco Villecco. Design Optimization of a Natural Gas Substation with Intensification of the Energy Cycle. Math Prob Eng 2010; Article ID 294102; 21

ACCEPTED MANUSCRIPT 590

doi:10.1155/2010/294102.

591 592 593

[23]

E. Lo Cascio, D. Borelli, F. Devia, C. Schenone.Future distributed generation : An operational multi-objective optimization model for integrated small scale urban electrical , thermal and gas grids. Energy Convers Manag 2017; 348:359-143.

594 595

[24]

S. Sanaye and A. M. Nasab. Modeling and optimizing a CHP system for natural gas pressure reduction plant. Energy 2012; 358:369-40.

596 597 598

[25]

C. A. Frangopoulos, M.R. Von Spakovsky, E. Sciubba. A brief review of methods for the design and synthesis optimization of energy systems. International Journal of Thermodynamics 2002; 151:1605.4.

599

[26]

CESLIUS project, available online: http://celsiuscity.eu/ (accessed 10/10/2017)

600 601 602

[27]

Gas expansion Turbine, available online: https://www.honeywellprocess.com/enUS/explore/products/gas-measurement-and-regulation/gas-expansion-turbines/Pages/default.aspx (accessed 06/08/2017).

603

[28]

Motiee M. Hydrocarbon Process. Int Ed 1991.

604 605

[29]

B.F. Towler, S. Mokhatab. Quicky estimate hydrocarbon formation conditions in natural gases", Hydrocarb Process 2005; 61:62-84.

606

[30]

Kidnay AJ, Parrish WR,McCartney DG. Fundamentals of natural gas processing. CRC Press 2011.

607 608 609

[31]

A. Herbst, S. Hirzel, M. Ragwitz, M. Rehfeldt, M. Reuter, J. Steinbach. EUROPEAN COMMISSION DIRECTORATE-GENERAL FOR ENERGY Directorate C. 2 – New energy technologies, innovation and clean coal. March, 2016.

610 611 612 613 614

22