Assessment of the use of Vanadium Redox Flow Batteries for Energy Storage & Fast Charging of EV by XRG Inc.

This is a personal pre-press copy (for non-commercial use) of the paper that has been published in Energy (currently still on-line only). The published version is available at: http://www.sciencedirect.com/science/article/pii/S0360544216301803. To cite this article: Á. Cunha, F. P. Brito, J. Martins, N. Rodrigues, V. Monteiro, J. L. Afonso, P. Ferreira, Assessment of the use of Vanadium Redox Flow Batteries for Energy Storage and Fast Charging of Electric Vehicles in Gas Stations Energy (published online March 2016, in-press), DOI:10.1016/j.energy.2016.02.118 1 2

Assessment of the use of Vanadium Redox Flow Batteries for Energy

Storage and Fast Charging of Electric Vehicles in Gas Stations

Álvaro CUNHA* a

F. P. BRITO*, b

Jorge MARTINS* c

Nuno RODRIGUES** d

Vitor MONTEIRO*** e

João L. AFONSO*** f

Paula FERREIRA****g

11 12

* Department of Mechanical Engineering, Universidade do Minho, Azurém 4800-058

Guimarães, Portugal.

** Petrotec, Inovação e Indústria SA., Parque Industrial de Ponte, Pavilhão C2, S. João

de Ponte, 4805-661 Guimarães, Portugal.

*** Centro Algoritmi, Universidade do Minho, Azurém, 4800-058 Guimarães, Portugal.

**** Research Centre for Industrial and Technology Management (CGIT),

Universidade do Minho, Azurém, 4800-058 Guimarães, Portugal

19 20

a – e-mail: alvarogcunha@hotmail.com

b – e-mail: francisco@dem.uminho.pt

c – e-mail: jmartins@dem.uminho.pt

d – e-mail: nuno.rodrigues@petrotec.pt

e – e-mail: vmonteiro@dei.uminho.pt

f– e-mail: jla@dei.uminho.pt

g – e-mail: paulaf@dps.uminho.pt

Keywords:



Flow Battery



Vanadium Redox Flow Battery



Energy Storage



Peak Shaving



Electric Vehicle



Fast Charging

ABSTRACT A network of conveniently located fast charging stations is one of the possibilities to

facilitate the adoption of Electric Vehicles (EVs). This paper assesses the use of fast

charging stations for EVs in conjunction with Vanadium Redox Flow Batteries

(VRFBs). These batteries are charged during low electricity demand periods and then

supply electricity for the fast charging of EVs during day, thus implementing a power

peak shaving process. Flow batteries have unique characteristics which make them

especially attractive when compared with conventional batteries, such as their ability to

decouple rated power from rated capacity, as well as their greater design flexibility and

nearly unlimited life. Moreover, their liquid nature allows their installation inside

deactivated underground gas tanks located at gas stations, enabling a smooth transition

of gas stations' business model towards the emerging electric mobility paradigm. A

project of a VRFB system to fast charge EVs taking advantage of existing gas stations

infrastructures is presented. An energy and cost analysis of this concept is performed,

which shows that, for the conditions tested, the project is technologically and

economically viable, although being highly sensitive to the investment costs and to the

electricity market conditions.

NOMENCLATURE Greek letters ε ηCh ηAC-DC η ηtotal VRFB ηsystem μ v ρ

Roughness CHAdeMo charger efficiency VRFB charger with an efficiency (AC/DC) Efficiency of the VRFB without pumping losses VRFB efficiency with pumping losses Overall system efficiency Dynamic viscosity Cinematic viscosity of the fluid Specific mass of the liquid electrolyte

Acronyms DoD EV G1 G2 GHG ICE PHEV PSB RFB SoC VRFB ZBB ZCB

Depth of discharge Electric Vehicle All vanadium redox flow batteries Vanadium bromide redox flow battery Greenhouse gases Internal Combustion Engine Plug-In Hybrid Electric Vehicles Polysulfide/Bromide technology Redox Flow Battery State of Charge Vanadium Redox Flow Batteries (G1 and G2 technologies) Zinc/Bromine technology Cerium/Zinc technology

Variables A Acs Ar Cin C CF Cout D DpY E E EBITDA Er Ea Esold F f g H I Inv i i’ k L Le l le MARR N Nd

Amortizations Permeated cross section area of the electrodes Amortization rate per year Concentration of vanadium in the solution before the cell Costs associated Cash-Flows Concentration of vanadium in the solution after the cell Internal diameter of the pipe operating days per year Energy density of the liquid electrolyte Equilibrium potential Earnings before interest, taxes, depreciation and amortization Real discharged energy Available stored energy Energy supplied/sold to charge a car Faraday constant Coefficient of friction Gravitational acceleration Head loss Current Investment Inflation rate Real interest rate Coefficient of head loss Length of the section Equivalent piping length Permeated specimen length of the electrode Thickness Minimum acceptable rate of return Number of cells Number of cars per day

mm N.s/ m2 m2/s kg/m3

€ m2 mol/L € € mol/L m Wh/L V € W W W 9.6485 x 104 C.mol-1 m/s2 m A € m m m m -

NPV OCVdisch OCVchg Ppump Δp Δpstack Δppipe P PLeaf Pr Pout Pin Pt p R r Re re RBT LR S SoCmin s Tax T TOG t tchg tdisch V V Vdisch Vchg

Net present value Open circuit voltage of the stack during discharge Open circuit voltage during charge Pumping power Total pressure loss Pressure loss in the Stack Pressure loss in the pipes Permeability of the electrodes Power consumption of the Nissan Leaf during charge Real discharge power Power output Input power Payback time Electrical energy purchasing price Electrical resistance Universal constant of ideal gases Reynolds number Resistivity Results before taxes Liquid result Gain from sales Minimum state of charge of VRFB during cycle Electrical energy selling price Taxes Temperature Taxes over gain Time Defined time to charge VRFB Defined time to discharge VRFB Volume of liquid stored Velocity of the fluid Voltage output of stack during discharge Intput voltage of the stack during charge

Subscripts chg disch in n out

Charge Discharge Input Year Output

€ V V kW Pa Pa Pa m2 kW kW kW kW years € Ω 8,3145 J.mol-1.k-1 Ωm € € € € € K s s s m3 m/s V V

1. INTRODUCTION

The disruptive increase of urban traffic along the last decades is posing serious

sustainability concerns, mainly those related to urban air quality and greenhouse gases

(GHG) emissions, as well as the excessive dependency of developed economies on

fossil fuels. It is expected that in 2030 the transportation sector will be responsible for

55% of total oil consumption [1]. It is also expected that the population will grow 1.7

times and the number of cars even more (3.6 times) between 2000 and 2050 [2]. In this

context, the current policies promoting emissions reduction and the improvement of the

energy efficiency of Internal Combustion Engines (ICE) are contributing to palliate

these issues [3]. Various strategies have been explored along time to address these

issues, such as engine downsizing achieved with turbo-charging [4], the strategy of over

expansion explored by the authors [5, 6] and used in several efficient hybrid powertrains

or waste energy harvesting such as exhaust thermal energy recovery in form of Organic

Rankine Cycle or Seebeck effect thermoelectric generators [7, 8].

Nevertheless, the increase of the overall efficiency of conventional powertrains does

not seem sufficient by itself to achieve the efficiency and emissions goals set by

national and international agreements, nor does it improve the desired diversity of

energy sources. Nowadays, the main alternatives to the traditional ICE are the Plug-In

Hybrid Electric Vehicles (PHEVs) and the full Electric Vehicles (EV) [9]. These

alternatives allow the reduction of the global fossil fuels consumption that is allocated

to the traditional transports systems and are a key technology to the future smart grids

[10]. Some of these alternatives are now available in the market with substantial success

[11], such as Toyota Prius (PHEV) or the Nissan Leaf (EV). These vehicles are globally

more efficient than ICE vehicles, mainly under urban traffic since they have no idling

losses, they have good low end torque without the need for inefficient clutching, and

they can recover some of the kinetic energy lost during the braking [3, 11]. In [12] a

comparative environmental life cycle comparison between conventional and electric

vehicles has been presented. As an example, using EVs, the global GHG emissions can

decrease from 10% to 24% when compared with conventional diesel or gasoline

vehicles. In [13] a study highlighted the EV as a means to contribute to the overall

reduction of the fossil sources and energy used for transportation, although certainly this

will depend on the electricity production performance.

Unfortunately, the success of PHEVs and EVs is currently hampered by some

notable disadvantages, mostly related with energy storage and power grid charging [14].

The main disadvantages are their typically low autonomy (usually up to 150 km) which

results from the low energy density of current battery technologies and the long time

required to perform standard battery charging processes (typically, a full charge will

require around 8 hours to complete) [11, 15]. The combination of these two factors is

known to induce the so-called range anxiety phenomenon which, along with the high

cost of batteries is preventing the wide adoption of electric mobility [16]. A range

extender unit may be added to the powertrain to prevent this, and in fact the authors

have confirmed the merits on a Life Cycle basis, of efficiency-oriented range extenders

[17], but the use of such systems increases design complexity and cost, as the price tag

of some existing models incorporating range extenders indicate.

In order to minimize some of the aforementioned shortcomings related to energy

storage, some EVs allow to perform a fast battery charging. The CHAdeMo (CHArge

de MOve) protocol [18] is one of the most popular DC fast charging protocols in

100

electric mobility, normally displaying a maximum power output of 50 kW. Fig. 1 shows

101

an example of a CHAdeMO fast charging station developed by a partner (PETROTEC)

102

of the team. With this charging mode the battery of many existing models can be

103

charged up to 80% of their State-of-Charge (SoC) in less than half an hour [19]. This

104

substantially reduces the inconveniences associated with small range, provided that fast

105

charging stations are available along the main roads. Of course, BEVs are not practical

106

for frequent long trips due to the need for frequent charging stops. Nonetheless, it would

107

be highly valuable for electric mobility that these long trips would be possible to do if

108

necessary. The range of mass market Battery Electric Vehicles is often around 100-150

109

km, so the suitable distance between two consecutive charging stations should be lower

110

than that distance to allow for occasional long trips.

111

Unfortunately, the high power output required by these chargers is demanding in

112

terms of local infrastructure. A high power consumption plan must be contracted with

113

the electric grid service provider, representing a substantial fixed cost even without any

114

energy consumption. Moreover, EV charging demand will normally occur at daytime,

115

coinciding with costly electrical peak demand periods.

116

Fortunately, many of the aforementioned disadvantages of fast charging may be

117

averted by decoupling grid consumption and the consumption due to vehicle charging

118

by means of stationary energy storage systems. In fact, the energy needed for high

119

power vehicle charging may be stored previously and more gradually (with lower

120

average power) at off-peak demand schedules than in the case of direct grid vehicle

121

charging. This allows reducing both the installed power consumption limit and the

122

average cost of electricity. Also, power quality problems associated with power grid

123

voltage, stability and frequency are minimized [20]. In this context, the present work

124

explores the use of a specific energy storage technology to perform EV fast charging

125

during daytime using electricity previously stored during low demand periods.

126

Moreover, the proposed energy storage technology could also be integrated into

127

microgrids, to store the energy produced from renewable power sources contributing to

128

smooth their intermittent production and adapt it to power demand [21].

129

The load levelling process and the peak shaving process rely on the storage of

130

energy during low demand periods, releasing that energy when the electrical load is

131

high [22, 23]. The main goal of the load levelling process is to stabilize the electrical

132

load, avoiding fluctuations in the consumed power, while in the case of peak shaving

133

process the main goal is to use the stored energy solely to remove the load peaks

134

consumption. For both processes, the energy stored during the night is equal to the

135

energy supplied by the storage system during the day. The comparison between peak

136

shaving and load levelling is illustrated in Fig. 2.

137

Typically, these two processes are implemented in low output power applications,

138

such as domestic grids or small factories with a few kW of power. They have several

139

advantages, the first of all being the reduction of the maximum power consumed from

140

the power grid and consequently the reduction of the installed power, which results in

141

lower prices [24]. Secondly, it permits a better management of the energy demanded

142

from the power grid, taking into account the different energy prices depending of the

143

schedule, because it is possible to buy cheaper power during off-peak periods, such as

144

during night-time [24]. Thirdly, it permits a greater incorporation into the grid of energy

145

derived from renewable sources like solar and wind, which are unpredictable sources,

146

often with the peak power generation occurring in counter-cycle with demand. This

147

means that the availability of an energy storage buffer will avoid wasting the energy

148

produced during low demand periods storing it and releasing it later during high

149

demand events. This will enable a real substitution of electricity obtained from fossil

150

fuel by electricity from renewable energy sources [25, 26].

151

Nowadays, reversible hydroelectric power plants are being often used since they can

152

use the excess of energy produced by renewables (generally the wind energy produced

153

during night hours) to pump water back to the hydroelectric dam, which creates a

154

gravitational energy storage. However, this resource is not always available or sufficient

155

to solve the problem and so, the integration of large scale batteries systems in the

156

electrical power grid seems to be a good solution for complement this energy storage

157

system.

158

There are several energy storage technologies that can be used for load levelling and

159

peak shaving processes besides the pumped hydro storage. They are compressed air

160

storage and batteries. Regarding for batteries, many research groups have been studied

161

the use of lead acid [27, 28], sodium sulphur (NaS) [22], lithium ion [29] and also

162

Redox Flow Batteries (RFB) [30] for these applications.

163

Many of the aforementioned systems have requirements not easily achieved for the

164

application proposed in this work. Among the various battery technologies, the RFB

165

have several advantages over the remainders, namely because their energy capacity is

166

uncoupled from their rated power [31].This is so because energy depends mainly on the

167

amount of electrolyte stored, while rated power is a function of the cell stack. Other

168

advantages of these batteries are related with their liquid nature and their storage (in

169

tanks), which can be of any shape. In [32] and [33]the recent developments and studies

170

of RFB concerning electrolytes, electrodes, membranes, and aqueous and non-aqueous

171

systems have been reviewed. There are many types of RFB with various redox couples

172

used, however, the Vanadium Redox Flow Battery (VRFB) is currently among the most

173

studied and promising technologies of this kind. These batteries have the advantage of

174

using the same chemistry in both half cells. The main advantage of this is that the cross-

175

contamination of the electrolytes will not render the resulting mix unsuitable for the

176

function as would be the case of electrolytes with different chemistries [34]. This is one

177

of the main reasons for their fairly extended life even when compared with the latest

178

Lithium ion battery chemistries. As main disadvantage, complete VRFB systems are

179

still expensive, although the growing maturity of this technology and its attractiveness

180

as an enabler for the wide adoption of intermittent renewable sources is likely to

181

decrease its cost in the midterm [35-37].

182

The recent economic crisis affecting several western economies was accompanied

183

by a reduction in the demand of transportation fuel [38], this reduction is showed in Fig.

184

3 and it can be seen that in the European Union the fuel consumption dropped by almost

185

4% between 2007 and 2011. In the same period, some of the most pronounced drops

186

occurred in Ireland and Spain, which reduced around 23% and 15%, respectively. Other

187

sharp reductions can be also observed, like the one occurred in Greece, which reduced

188

about 17% between 2009 and 2011, and in case of Portugal there was a sharp decline of

189

about 7% between 2010 and 2011.

190

As an alternative to the costly and laborious deactivation/disposal of surplus large

191

fuel storage tanks in gas stations, a retrofit of these deposits could be performed,

192

adapting them for VRFB electrolyte storage and using the storage system for EV fast

193

charging with the strategy explained before. One merit of such an approach would be to

194

easily obtain EV fast charging spots in places which are already strategically located for

195

vehicle traffic, optimizing otherwise wasted space and infrastructures and

196

complementing the ICE vehicle fuel supply business with the emerging plug-in vehicle

197

charging business in one place.

198

Following previous work by the authors in electric mobility and gas station equipment,

199

including a review article on the VRFB technology and its prospects to the electric

200

mobility area [33], the present work assesses this philosophy by carefully describing the

201

operating principles of a VRFB, the use of energy storage technologies for load levelling

202

and peak shaving and by performing a draft design of a VRFB system capable of fast

203

charging simultaneously two electric vehicles. This design is then analysed in terms of

204

energy performance and economic viability.

205

206

Operating Principles of a Vanadium Redox Flow Battery

207

The operating principle of RFBs is partly similar to the operation of a conventional

208

battery, but a major distinction is the fact that the energy storage unit (the active materials)

209

is physically separated from the energy production unit (the cell stack) [39]. So, in a

210

VRFB the active materials are not permanently sealed inside the cell (like in a

211

conventional battery), but are stored separately in tanks and pumped into the cell

212

according to the energy demand [34].

213

This process is represented in Fig. 4, which represents the two tanks, one for positive

214

electrolyte (cathode) and other for negative electrolyte (anode), the cell and the pumps

215

[39]. When the liquid electrolytes are injected into the cell an electrochemical reaction

216

(oxidation-reduction or redox) occurs, with movement of electrons along the electric

217

circuit, as there is an exchange of ions through the membrane to maintain charge

218

neutrality between the different ionic solutions [31].

219

The positive and negative electrodes in vanadium redox flow batteries are typically

220

carbon based materials, such as carbon or graphite felts, carbon cloth, carbon black,

221

graphite powder and so on [40]. These electrodes have shown a good potential in terms

222

of operation range, a good stability and a high reversibility. Similarly to other battery

223

technologies, the electrodes are a very important component on the performance of the

224

vanadium redox flow batteries.

225

There are many types of redox flow batteries, such as: the zinc-bromine (ZBB) [41]; the

226

polysulfide-bromide (PSB) [42]; the Cerium-Zinc (ZCB) [43]; and the Vanadium Redox

227

Flow Batteries (VRFB), which include the first generation (G1 - the all vanadium system,

228

normally called VRB in the literature) and the second generation (G2 - the vanadium

229

bromide system) [33, 44].

230

The G1 is now the most studied technology and involves solely vanadium species in both

231

half cells at different valence states. It operates in an electrochemical couple based on two

232

different reactions of vanadium ions in a dilute acid solution. This is possible because

233

vanadium oxide is a stable material in four different valence states [45].

234

The cathodic and anodic reactions can be represented as follows [46]:

235 +

236

VO2 + 2H + e â&#x2020;&#x201D; VO + H2O

237

V â&#x2020;&#x201D;V +e

238

E+ = 1.00 V E- = - 0.26 V

239

And the overall reaction is [46]:

240 +

241

VO2 + 2H + V ↔ VO + H2O + V

242 243

The G1 has an advantage relatively to the other redox flow batteries, which is that in the

244

event of a cross mixing between the two liquid electrolytes, the regeneration of the

245

solution may be performed simply by recharging the fluids, unlike systems with different

246

metals in which the mixed liquids would have to be replaced or removed and treated

247

externally or disposed [34].

248

Unfortunately, VRFBs still have a low energy density when compared with conventional

249

batteries. This is due to the maximum concentration of Vanadium that can currently be

250

dissolved in the supporting electrolyte. In the case of the G1 technology, typically the

251

maximum vanadium ion concentration is 2 M or less, which corresponds to an energy

252

density of 25 Wh/kg or 33 Wh/L, and that concentration is limited by the stability of the

253

V5+ ions at temperatures above 40 ºC and the solubility limit of V2+ and V3+ ions in

254

supporting electrolyte at temperatures below 5 ºC. Nonetheless, several studies have been

255

made in the last few years testing the incorporation of additives in the positive electrolyte

256

in order to inhibit the vanadium precipitation and enable the use of higher concentrations

257

and thus increase the energy density.

258

The G2 technology employs a vanadium bromide solution in both half-cells and shares

259

all the benefits of the G1 technology, including the fact that the cross contamination is

260

eliminated [47]. Since the bromide/polyhalide couple has lower positive potential than

261

the V(IV)/V(V) couple, the bromide ions will preferentially oxidize at the positive

262

electrode during the charging [48].

263

The cathodic reactions

264

ClBr2-+ 2e ↔ 2Br-+ Cl-

of this technology are as follows [49]:

265

266

BrCl2-+ 2e â&#x2020;&#x201D; Br -+ 2Cl-

267

And the anodic reaction is as follows [49]:

268

V â&#x2020;&#x201D;V +e

269

However, the G2 technology has a disadvantage, which is the propensity for the formation

270

of bromine vapours during charging. This requires the use of expensive bromine

271

complexing agents which limit the attractiveness of the G2 technology [48].

272

Since the G2 technology is not yet in a mature state and the G1 technology is the most

273

studied and there are already several companies commercializing it [50-55], the present

274

work has focused on this technology. In Fig. 5 a typical charge/discharge cycle of a VRFB

275

at 40 mA/cm2 is shown [56]. It can be seen that the maximum voltage during charge is

276

1.74 V, while during discharge the voltage varies from 1.42 V down to 0.8 V. However, it

277

can also be seen that below approximately 1.15 V, the slope of the curve becomes very

278

sharp, which means that the SoC is close to zero.

279

Currently there are already several VRFB battery manufacturers and some operating

280

plants worldwide. A previous revision article by the authors provides some insight into

281

this subject [33].

282

VRFB System for Charging EVs In Gas Stations

283

This paper proposes an energy storage technology to be used by electric vehicle fast

284

charging stations to make a peak shaving process, enabling the simultaneous charging of

285

several EVs without having to incur into excessive power availability (e.g. supply

286

capacity) charges.

287

Following several decades of vehicle traffic, existing gas stations are now generally quite

288

conveniently located and evenly distributed for vehicle access. Therefore, the location of

289

EV charging spots at those places is highly attractive since it will not be necessary to

290

captivate additional real estate to build a network of new charging spots suitably located.

291

An EV owner could charge his vehicle using, for instance, a fast charging CHAdeMo

292

station located at such places in roughly half an hour, enabling the use of EVs for

293

occasional long travels, should these stations be conveniently distributed along the main

294

driveways. Furthermore, gas stations could adapt their business size as a function of the

295

increase of EV market penetration.

296

System Layout

297

Fig. 6 shows a typical steel fuel tank used in gas stations [57], among the various fuel

298

tanks available in the market. Capacities vary from 1 000 litres of capacity, to the giant

299

tanks with 100 000 litres of capacity [57, 58], but one of the most commonly used in gas

300

stations is the 20 000 litre tank with around 2.5 m of diameter and around 4.7 m in length.

301

Typically, the gas station underground tanks that are deactivated are not removed and are

302

left unused. It would therefore be advantageous to retrofit these leftover tanks to store the

303

liquid electrolytes. It would inclusively be a way of gradually migrate from a fossil fuel-

304

based business to an electric mobility services business as the market slowly adapts from

305

one paradigm to the other. The present work has considered these tanks for the storage of

306

the liquid electrolytes, but surface tanks are also possible to use and they will facilitate

307

the installation and service of the system.

308

The shape and material of the tanks raises some difficulties. Since they are typically made

309

from steel [57, 58], the liquid electrolytes cannot be in direct contact with them due to

310

their acid nature. To avoid that contact, the tanks may be coated with an acid resistant

311

material. In such case, each fuel tank can only be used for one electrolyte (positive or

312

negative). Alternatively, one or several smaller flexible tanks made from an acid resistant

313

may be installed inside the steel tanks. In this latter case, two smaller flexible tanks, each

314

one containing a different liquid electrolyte, may be accommodated inside one steel tank.

315

As can be seen in Fig. 6, the diameter of the main tank opening (only 60 cm) may hamper

316

the insertion of the flexible tanks inside it. Several small solid tanks made from PVC or

317

other acid resistant material could also be used. However, the best solution seems to be

318

the use of flexible rubber tanks as those firstly proposed by Sumitomo Electric Industries

319

Ltd. These were made specifically to take advantage of the fluidic form of these batteries

320

and to allow their insertion in unused spaces such as underground cisterns, through

321

manholes [59].

322

Rubber tanks with a shape which reasonably conforms to the interior of the steel tanks

323

should be made, as illustrated in Fig. 7 (a), including a support structure to separate both

324

tanks and leaving free space below the manhole to allow the entry of installation and

325

service staff. This structure can be done simply with Landsquare PVC beams and mounted

326

in loco with stainless steel screws.

327

Another alternative, is to use four tanks (two for positive and two for negative electrolyte).

328

This facilitates their introduction into the fuel tank. On the other hand, this configuration

329

allows the existence of two VRFBs using the same gas tank so it is possible to have one

330

battery to charge EVs and the other one being charged at low power input from the electric

331

grid. The four tank configuration is represented in Fig. 7 (b).

332

To make these tanks, the appropriate rubber should be selected. It must be highly

333

resistant to the corrosion with sulphuric acid under the prescribed concentration.

334

Typically, the two liquids electrolytes (anolyte and catholyte) of a G1 are prepared by

335

dissolving around 2M VOSO4 (vanadyl sulphate) into around 5M H2SO4 aqueous

336

solution, to form tetravalent vanadium ions [34], which means that the volume

337

concentration of sulphuric acid may be between 0.005% and 27%.

338

Table 1 which has been compiled from several references [60-62], shows the chemical

339

resistance of the most common rubber types, and it can be seen that there are various

340

types of rubber which are resistant of acid sulphuric for the concentrations needed, like

341

the EPDM, Butyl and Teflon.

342

However, the Vanadium oxides are also corrosive and their presence within the electrolyte

343

must be considered, so a rubber which is resistant to both substances is needed. After

344

analysing Table 1, as a first approximation it seems that Butyl rubber will be a good

345

candidate material for the rubber tanks.

346

2. MODELLING

347

The modelling of the system under analysis is presented in this section. The aim is

348

to evaluate its technological viability in order to characterize the charge/discharge

349

cycles, the flow rate and the required pumping power. This will allow the estimation of

350

the overall efficiency of the system. These calculations do not exclude the importance of

351

building and testing a physical prototype to compare the theoretical and real conditions

352

of operation in order to confirm the accuracy of the present simulation, but show an

353

initial approach to the design of such a system.

354

This project involves the use of a VRFB to store energy from the electrical grid

355

during low demand schedules (and also from renewable sources, if required). The stored

356

energy is then supposed to supply two CHAdeMo fast chargers to charge EVs similar to

357

the ones manufactured by Petrotec [63]. Fig. 8 outlines the architecture of the proposed

358

system. Renewable energy sources could be present at the gas station (e.g., solar or

359

wind energy) in order to reduce the energy consumed from the grid and to benefit from

360

the energy storage facility. However, all the energy included in the present calculations

361

is considered to come from the grid.

362

Electric Model (VRFB discharge)

363 364

During the fast charge of an EV the voltage, current and power output from charger

365

will vary, and these parameters were monitored by Bai et al. [64] for a Nissan Leaf. The

366

evolution of the power output is represented in Fig. 9. This charging cycle was

367

performed by a Terra 51 charger manufactured by ABB and the Nissan Leaf was

368

charged up to 80% SoC. By integrating the power against time it can be concluded that

369

in this particular cycle about 14.8 kWh were supplied to the car (actually, due to the

370

EVâ&#x20AC;&#x2122;s battery efficiency, the energy effectively stored in the vehicle will be lower, but

371

this is not relevant for the present analysis). This confirms that the vehicle was not fully

372

discharged when the charge began. In a real case each EV will have a different initial

373

SoC, but for the present analysis it seems fit to consider that all EVs will be in the same

374

initial conditions and will be submitted to the same charging cycle as the car monitored

375

in Fig. 9.

376

The power output (Pout) can be calculated for each time step by equation 1:

377

đ?&#x2018;&#x192;đ?&#x2018;&#x153;đ?&#x2018;˘đ?&#x2018;Ą =

đ?&#x2018;&#x203A;đ?&#x2018;?đ?&#x2018;&#x17D;đ?&#x2018;&#x;đ?&#x2018; đ?&#x2018;&#x192;đ??żđ?&#x2018;&#x2019;đ?&#x2018;&#x17D;đ?&#x2018;&#x201C;

Îˇđ??śâ&#x201E;&#x17D;

(1)

378 379

The voltage output of the stack during discharge (Vdisch) can be calculated as a

380

function of the open circuit voltage of the stack during discharge (OCVdisch), its internal

381

resistance (R) and the discharge current (I) in each time step by the equation 2: đ?&#x2018;&#x2030;đ?&#x2018;&#x2018;đ?&#x2018;&#x2013;đ?&#x2018; đ?&#x2018;?â&#x201E;&#x17D; = đ?&#x2018;&#x201A;đ??śđ?&#x2018;&#x2030;đ?&#x2018;&#x2018;đ?&#x2018;&#x2013;đ?&#x2018; đ?&#x2018;?â&#x201E;&#x17D; â&#x2C6;&#x2019; đ?&#x2018;&#x2026;đ??ź

(2)

382 383 384

The instantaneous discharge current (I) of the stack can be calculated as a function of the output power and the output voltage through equation 3:

đ??ź= 385

đ?&#x2018;&#x192;đ?&#x2018;&#x201A;đ?&#x2018;˘đ?&#x2018;Ą Vđ?&#x2018;&#x2018;đ?&#x2018;&#x2013;đ?&#x2018; đ?&#x2018;?â&#x201E;&#x17D;

(3)

The open circuit voltage of the stack is a function of the number of cells (N), the

386

equilibrium potentials E+ and E-, the universal constant of the ideal gases (r), the

387

temperature (T), the Faraday constant (F) and the state of charge, through equation 4

388

[65]:

389

đ?&#x2018;&#x201A;đ??śđ?&#x2018;&#x2030;đ?&#x2018;&#x2018;đ?&#x2018;&#x2013;đ?&#x2018; đ?&#x2018;?â&#x201E;&#x17D;

đ?&#x2018;&#x;đ?&#x2018;&#x2021; đ?&#x2018;&#x2020;đ?&#x2018;&#x201A;đ??ś 2 = đ?&#x2018; (đ??¸ â&#x2C6;&#x2019; đ??¸ + ln ) đ??š (1 â&#x2C6;&#x2019; đ?&#x2018;&#x2020;đ?&#x2018;&#x201A;đ??ś)2 +

â&#x2C6;&#x2019;

(4)

390 391

However, due to the internal resistances, the output voltage will always be lower

392

that the open circuit voltage of the stack, which means that there are losses associated

393

and the real discharge power (Pr) is considered to be the sum of the power output and

394

the losses as represented in Fig. 10, and can be calculated by equation 5:

395

đ?&#x2018;&#x192;đ?&#x2018;&#x; = đ?&#x2018;&#x201A;đ??śđ?&#x2018;&#x2030; đ?&#x2018;&#x2018;đ?&#x2018;&#x2013;đ?&#x2018; đ?&#x2018;?â&#x201E;&#x17D; đ??ź

(5)

396 397 398

So, for a given time lapse (đ?&#x203A;Ľt=tf-t0), the real discharged energy (Er) of the stack can be calculated through:

399 đ?&#x2018;Ąđ?&#x2018;&#x201C;

đ??¸đ?&#x2018;&#x; = â&#x2C6;Ť đ?&#x2018;&#x192;đ?&#x2018;&#x; đ?&#x2018;&#x2018;đ?&#x2018;Ą

(6)

đ?&#x2018;Ą0

400

On the other hand, the open circuit voltage of the stack depends on the SoC of the

401

VRFB, which will vary during the discharge according to the real discharged energy

402

(Er).

403

The SoC at a given moment can be calculated as a function of the energy density of

404

the liquid electrolyte (E), the volume of liquid stored (V), the maximum state of charge

405

of the VRFB (SoCmax) and the real energy discharged up to that moment, through

406

equation 7:

407

đ?&#x2018;&#x2020;đ?&#x2018;&#x201A;đ??ś =

đ??¸ đ?&#x2018;&#x2030; đ?&#x2018;&#x2020;đ?&#x2018;&#x201A;đ??śđ?&#x2018;&#x20AC;đ?&#x2018;&#x17D;đ?&#x2018;Ľ â&#x2C6;&#x2019; đ??¸đ?&#x2018;&#x; đ??¸ đ?&#x2018;&#x2030;

(7)

408 409 410 411

The computation of the results will be iterative and the output power (Pout) over time will be considered to be a known input (profile of Fig. 9). When the electrodes are carbon felts (as proposed in the present work), they must be

412

compressed in order to prevent the leakage of the liquid electrolyte and, at the same

413

time, reduce the electric resistance. Their electric resistance varies with the compression

414

ratio as illustrated in Fig. 11 for a specific model. For instance, for a compression ratio

415

of 20%, the electric resistance reduces by around 70% [66].

416

Often, the available information regarding electrodes and bi-polar plates is, rather than

417

resistance, their resistivity (re). Therefore, the electrical resistance (R) must be

418

calculated as a function of their section area (A) and thickness (le) through equation 8: đ?&#x2018;&#x2026; = đ?&#x2018;&#x;đ?&#x2018;&#x2019;

đ?&#x2018;&#x2122;đ?&#x2018;&#x2019; đ??´

(8)

419

420

2.1.1. Electric Model (VRFB charge)

421 422

With the knowledge of the discharge profiles of Fig. 9 it is possible to define the

423

charge cycle for the VRFB. Firstly, the available stored energy, Ea, (shown in Fig. 12)

424

coincides with the real discharged energy, calculated through equation 9. So, for a given

425

SoC during charge, the available energy stored is calculated as follows:

426

đ??¸đ?&#x2018;&#x17D; = đ??¸ đ?&#x2018;&#x2030; đ?&#x2018;&#x2020;đ?&#x2018;&#x201A;đ??ś â&#x2C6;&#x2019; đ??¸ đ?&#x2018;&#x2030; đ?&#x2018;&#x2020;đ?&#x2018;&#x201A;đ??śđ?&#x2018;&#x161;đ?&#x2018;&#x2013;đ?&#x2018;&#x203A; 427 428 429

(9)

The open circuit voltage during charge (OCVchrg) can be calculated as a function of the SoC, as it was done previously for the discharge phase with equation 4. Considering the real power (equal to the input power minus the losses) during a

430

given charge time lapse (tch), the evolution of OCVchrg as a function of time can be

431

estimated. In this way, considering a constant power charge, it is then possible to

432

calculate the current during each time step:

433 â&#x2C6;&#x2018; đ??¸đ?&#x2018;&#x17D;

đ??ź=

đ?&#x2018;Ąđ?&#x2018;?â&#x201E;&#x17D;

đ?&#x2018;&#x201A;đ??śđ?&#x2018;&#x2030;đ?&#x2018;?â&#x201E;&#x17D;đ?&#x2018;&#x;đ?&#x2018;&#x201D;

(10)

434 435 436

With this information it is possible to calculate the output voltage of the stack during charge (Vchrg):

437

đ?&#x2018;&#x2030;đ?&#x2018;?â&#x201E;&#x17D;đ?&#x2018;&#x;đ?&#x2018;&#x201D; = đ?&#x2018;&#x201A;đ??śđ?&#x2018;&#x2030;đ?&#x2018;?â&#x201E;&#x17D;đ?&#x2018;&#x;đ?&#x2018;&#x201D; + đ?&#x2018;&#x2026;đ??ź

(11)

438 439

Note that in this case the input voltage will be the sum of the open circuit voltage

440

and the losses. This means that the input voltage is always higher that the open circuit

441

voltage. This is the opposite of what happens during the discharge of the VRFB.

442

So, the input power (Pin) can be calculated as follows:

443

đ?&#x2018;&#x192;đ??źđ?&#x2018;&#x203A; = đ?&#x2018;&#x2030;đ?&#x2018;?â&#x201E;&#x17D;đ?&#x2018;&#x;đ?&#x2018;&#x201D; đ??ź 444 445 446

And the efficiency of the VRFB (Îˇ) can be obtained through:

(12)

đ?&#x2018;Ą

đ?&#x153;&#x201A;=

â&#x2C6;Ť0 đ?&#x2018;&#x192;đ?&#x2018;&#x153;đ?&#x2018;˘đ?&#x2018;Ą đ?&#x2018;&#x2018;đ?&#x2018;Ą đ?&#x2018;Ą

đ?&#x2018;?â&#x201E;&#x17D; â&#x2C6;Ť0 đ?&#x2018;&#x192;đ??źđ?&#x2018;&#x203A; đ?&#x2018;&#x2018;đ?&#x2018;Ą

(13)

447 448

An important factor in the viability study of a VRFB system is the required mass

449

flow rate, as it represents an extra energy consumption which will reduce the overall

450

system efficiency. Typically, higher flow rates induce higher voltages due to lower

451

concentration of the reaction products within the cells in each side of the membranes

452

during operation [67]. However, there is an extra consumption of energy which reduces

453

the overall energy efficiency due to the increase of flow rate. Therefore, it is useful to

454

find the optimal value which will result in the maximum efficiency of the system.

455

It is necessary to divide the number of moles of vanadium oxidized per second by

456

the molarity of vanadium ions in the solution in order to calculate the minimum

457

required flow rate [68]. So, knowing the number cells (N), the current (I), and the

458

concentration of vanadium in the solution (Cv) and the Faraday constant (F) it is

459

possible to calculate the stoichiometric flow rate (Q) through equation 13 [68]:

460

đ?&#x2018;&#x201E;=

đ??źđ?&#x2018; (đ??żâ &#x201E;đ?&#x2018; ) đ??š đ??śđ?&#x2018;Ł

(14)

461 462

Since the flow rate depends on the concentration of vanadium, it is necessary to take

463

into account the variation of the concentration of the vanadium reactant in the solution

464

due to the variation of the SoC of the liquid stored in tank during the charge and

465

discharge cycles. The total concentration of vanadium is constant. However, during the

466

discharge, the concentration of reactants decreases, while the concentration of products

467

increases, and vice-versa.

468

Knowing the vanadium reactant and product concentration in the solution, the SoC

469

can be calculated in two ways, regarding to the concentration at the anode or at cathode

470

respectively [69, 70]:

471

đ??śđ?&#x2018;&#x2030; 2+ đ??śđ?&#x2018;&#x2030;đ?&#x2018;&#x201A;2+ đ?&#x2018;&#x2020;đ?&#x2018;&#x201A;đ??ś = = đ??śđ?&#x2018;&#x2030; 2+ + đ??śđ?&#x2018;&#x2030; 3+ đ??śđ?&#x2018;&#x2030;đ?&#x2018;&#x201A;2+ + đ??śđ?&#x2018;&#x2030;đ?&#x2018;&#x201A;2+

(15)

472 473

And the Depth-of-Discharge (DoD) can be calculated as follows [69]:

474

đ??ˇđ?&#x2018;&#x201A;đ??ˇ = 1 â&#x2C6;&#x2019; đ?&#x2018;&#x2020;đ?&#x2018;&#x201A;đ??ś =

(16)

Electrolyte Flow Model

475 476

đ??śđ?&#x2018;&#x2030; 3+ đ??śđ?&#x2018;&#x2030;đ?&#x2018;&#x201A;2+ = đ??śđ?&#x2018;&#x2030; 2+ + đ??śđ?&#x2018;&#x2030; 3+ đ??śđ?&#x2018;&#x2030;đ?&#x2018;&#x201A;2+ + đ??śđ?&#x2018;&#x2030;đ?&#x2018;&#x201A;2+

Equation 14 does not take into account the variation of the SoC through the

477

membrane. This variation would be negligible for very high flow rates. Although it is

478

not quite the case in the present work, this effect has been neglected for the sake of

479

simplicity.

480

Note that when the VRFB is charging Cv of equation 14 refer to the concentration of

481

V2+ or VO2+ . On the other hand, when the VRFB is discharging, Cv refer to the

482

concentration of V3+ or VO2+. This means that the flow rate will have to vary during the

483

operation of the VRFB. It will increase during the increase of the SoC (when charging)

484

and during the increase of DoD (when discharging). This is consistent with the study

485

made by Ma et al. [71].

486

Note that the flow rate calculated by equation 14 is the flow rate for each electrolyte

487

(i.e., anode and cathode solutions). However, this value represents the minimum flow

488

rate necessary to produce the desired current if all vanadium ions existing within the

489

solution were oxidized while flowing through cells. In practice, this does not happen

490

and some authors suggest the use a flow rate which is higher than the one theoretically

491

calculated.

492

Unfortunately, to the authorsâ&#x20AC;&#x2122; knowledge, there is not sufficient information

493

available in the literature concerning the difference between the theoretical and the real

494

flow rate needed, so in this paper it will be considered that the VRFB will operate with

495

the stoichiometric flow rate calculated from equation 14. In this way, the calculated

496

efficiency with pumping power might be somewhat overestimated.

497 498

So, knowing the flow rate, the pumping power (Ppump) can be estimated through equation 17:

499 đ?&#x2018;&#x192;đ?&#x2018;?đ?&#x2018;˘đ?&#x2018;&#x161;đ?&#x2018;? = đ?&#x2018;&#x203A;đ?&#x2018;?đ?&#x2018;&#x17D;đ?&#x2018;&#x;đ?&#x2018; đ?&#x2018;&#x201E; â&#x2C6;&#x2020;đ?&#x2018;?

(17)

500 501

With Î&#x201D;p representing the total pressure loss in the system, calculated through:

502

â&#x2C6;&#x2020;đ?&#x2018;? = â&#x2C6;&#x2020;đ?&#x2018;?đ?&#x2018;?đ?&#x2018;&#x2013;đ?&#x2018;?đ?&#x2018;&#x2019;đ?&#x2018; + â&#x2C6;&#x2020;đ?&#x2018;?đ?&#x2018; đ?&#x2018;Ąđ?&#x2018;&#x17D;đ?&#x2018;?đ?&#x2018;&#x2DC;

(18)

503 504

The pressure loss in the Stack (Î&#x201D;pstack) is due to the flow of the liquid through the

505

graphite felt electrodes and can be calculated based on the flow rate, the dynamic

506

viscosity (Îź), the permeated specimen length (l), the permeability (P) and the permeated

507

cross section area (Acs) [66]:

508

â&#x2C6;&#x2020;đ?&#x2018;?đ?&#x2018; đ?&#x2018;Ąđ?&#x2018;&#x17D;đ?&#x2018;?đ?&#x2018;&#x2DC; = 509

đ?&#x2018;&#x201E;đ?&#x153;&#x2021;đ?&#x2018;&#x2122; đ?&#x2018;&#x192; đ??´đ?&#x2018;?đ?&#x2018;

(19)

510

It is necessary to consider that the permeability of the electrodes will vary with the

511

compression of the electrodes, according to the chart represented in Fig. 13. Of course, a

512

trend-off must be made here due to the dependence of the resistivity on this parameter

513

(recall Fig. 11).

514

The pressure loss in the pipes (Î&#x201D;ppipe) should also be calculated according to the

515

specific piping geometry chosen. To calculate the total pressure drop, it is necessary in

516

the first place, to calculate the Reynolds number (Re) in each pipe section [72]:

517

đ?&#x2018;&#x2026;đ?&#x2018;&#x2019; =

4đ?&#x2018;&#x201E; đ?&#x153;&#x2039;đ??ˇđ?&#x2018;Ł

(20)

518 519

where v represents the kinematic viscosity of the fluid (resulting from the division of the

520

dynamic viscosity by the density) and D the inner diameter of the pipe. The coefficient

521

of friction (f) can be calculated from the Colebrook expression, which is the basis for

522

the Moody Diagram [72]:

523

đ?&#x153;&#x20AC;â &#x201E; 1 2,51 = â&#x2C6;&#x2019;0,86 ln ( đ??ˇ + ) đ?&#x2018;&#x201C; 3,7 đ?&#x2018;&#x2026;đ?&#x2018;&#x2019; â&#x2C6;&#x161;đ?&#x2018;&#x201C;

(21)

524 525 526

The localized head losses can be considered as an equivalent piping length (Le) calculated as a function of the coefficient of head loss (k) [72]:

527

đ??żđ?&#x2018;&#x2019; =

â&#x2C6;&#x2018;đ?&#x2018;&#x2DC; đ??ˇ đ?&#x2018;&#x201C;

(22)

528 529 530

So, the total head loss (H) for each section can be calculated function of the parameters calculated previously by equation 23 [72]:

531

đ??ť=đ?&#x2018;&#x201C;

đ??ż + đ??żđ?&#x2018;&#x2019; đ?&#x2018;&#x2030; 2 đ??ˇ 2đ?&#x2018;&#x201D;

(23)

532 533

where V corresponds to the velocity of the fluid, L corresponds to the length of the

534

section and g is the gravitational acceleration. The pressure drop for each section can be

535

calculated as follows [72]:

536

â&#x2C6;&#x2020;đ?&#x2018;?đ?&#x2018;?đ?&#x2018;&#x2013;đ?&#x2018;?đ?&#x2018;&#x2019; =

đ??ť đ?&#x153;&#x152;đ?&#x2018;&#x201D;

(24)

537 538 539 540

541

542

where Ď is the density of the liquid electrolyte. The total pipe pressure loss will be the sum of the pressure losses in each section and the pumping power can be calculated using equations 17 and 18.

Economic Analysis The technological analysis made would be incomplete without a suitable economic

543

analysis. Firstly, it is necessary to roughly estimate the total cost of the proposed

544

system, even if there is a substantial uncertainty degree concerning its real cost.

545

Nevertheless, estimations based on the information obtained from the manufacturers are

546

valuable for this purpose.

547

Firstly, it is necessary to calculate the overall system efficiency using this VRFB

548

connected to two CHAdeMo chargers each with a given efficiency (ÎˇCh ) and one VRFB

549

AC/DC charger with a given efficiency (ÎˇAC-DC). So the overall system efficiency

550

(Îˇsystem) can be calculated by equation 25.

551

đ?&#x153;&#x201A;đ?&#x2018;&#x2020;đ?&#x2018;Śđ?&#x2018; đ?&#x2018;Ąđ?&#x2018;&#x2019;đ?&#x2018;&#x161; = đ?&#x153;&#x201A;đ??´đ??śâ&#x2C6;&#x2019;đ??ˇđ??ś đ?&#x153;&#x201A;đ??śâ&#x201E;&#x17D; đ?&#x153;&#x201A;đ?&#x2018;Ąđ?&#x2018;&#x153;đ?&#x2018;Ąđ?&#x2018;&#x17D;đ?&#x2018;&#x2122; đ?&#x2018;&#x2030;đ?&#x2018;&#x2026;đ??šđ??ľ

(25)

552 553

For the economic evaluation of the project, both the NPV (net present value) and the

554

payback time (PT) criteria will be computed. The economic evaluation is then based on

555

a traditional discounted Cash-Flow (CF) approach, including overall financial

556

assumptions on the depreciation rate, taxes, inflation rate and discount rate and also

557

estimation of the project expected sales and operational costs.

558

So, firstly, the calculation of the gains from sales (S) for the first year (n = 1) is

559

computed as a function of the energy sold for the charging of each vehicle (Esold), the

560

number of EVs charged per day (Nd), the number of operating days per year (DpY) the

561

price of the electricity sold (s) and the average inflation rate (i). Sales for the first year

562

(n = 1) can then be calculated by equation 26.

563

S n=1 = (đ??¸đ?&#x2018; đ?&#x2018;&#x153;đ?&#x2018;&#x2122;đ?&#x2018;&#x2018; Nd s DpY) (1 + i)

(26)

564 565

Considering that all the assumed parameters remain constant during the whole life

566

cycle considered, sales for the following years can be calculated as described in

567

equation 27.

568

S n = S nâ&#x2C6;&#x2019;1 (1 + i)

(27)

569 570 571

The associated costs (C) for the first year (n = 1) are calculated according to equation 28, depending on the price of the electricity bought to charge the batteries (p).

572

C n=1 =

(đ??¸đ?&#x2018; đ?&#x2018;&#x153;đ?&#x2018;&#x2122;đ?&#x2018;&#x2018; Nd DpY) p (1 + i) ÎˇSystem

573 574

The costs for the following years are calculated by equation 29.

(28)

575

C n = C n−1 (1 + i)

(n > 1)

(29)

576

577

3. RESULTS AND DISCUSSION

578

Simulation Conditions

579

It is assumed that there will be always two vehicles charging simultaneously (two

580

chargers connected to the VRFB) and that the CHAdeMo charger efficiency (ηCh) will

581

be around 95%, so the power output (Pout) can be calculated for each time step by

582

equation 1.

583

Table 2 and Table 3 show the specific input values used to simulate the VRFB

584

system proposed for this project, but the analysis proposed may be used for different

585

values.

586 587

The permeability and resistance of the electrodes depends on the compression of the

588

electrodes (recall Fig. 11 and Fig. 13). The chosen compression of 20% therefore yields

589

a permeability of the electrodes around 4.7x10-11 m2 and a 70% drop in the original

590

electrode resistance [66].

591

Therefore, the internal resistance (R) of this system can be calculated by the sum of

592

the single components. In the case of electrodes and bi-polar plates it is calculated by

593

equation 8 and the total resistance of the system will be around 4.897 mΩ.

594

Regarding the hydraulic calculations, it is considered that each liquid electrolyte has

595

a height difference between the tank and the stack of 1 m. The tubes used are considered

596

to be made of PVC with low roughness (ε = 0.005 mm). The first section of the piping

597

is a circular tube with 40 mm of inner diameter, a length of 12 m and three 90º bends,

598

which carries the liquids from the tanks to the cell. The second section of the piping is a

599

collector with one output for each cell, with 3 mm of inner diameter and 20 cm of

600

length. The third section is another collector attached to each output of section 2. This

601

collector has 5 outputs, which means that the entry of the liquid into the cell is made

602

through 5 different locations in order to uniformize the flow. Each output is also a

603

circular tube with 3 mm of inner diameter and 20 cm of length. The return circuit

604

displays an identical circuit, symmetrical to the one just described. The pressure losses

605

in collectors have been neglected. Regarding the localized head losses a coefficient of

606

head loss (k) of 0.9 has been considered for each one of the bends [72]

607

608

System Performance

609

Fig. 14 (a) and (b) represent the variation of the voltage and the current,

610

respectively, during a simulation of a VRFB used for 14 consecutive fast charging

611

cycles of 2 EVs simultaneously. The strong variations of voltage and current are due to

612

the variable power profile of each vehicle charge according to Fig. 9.

613 614

The vehicles simulated have the characteristics of the Nissan Leaf, while the

615

charging profile was the one represented in Fig. 9. This means that a total of 28 vehicles

616

were charged in a 7.5 hour period, with the final state of charge of the VRFB having

617

dropped to around 1%, with no energy left to perform another vehicle charge. The rise

618

in current observed in Fig. 14 (b) is a natural consequence of the drop in voltage

619

observed in Fig. 14 (a), to ensure the programmed power cycle. The slight ripple

620

observed in the SoC curve is due to the non-constant power profile (Fig. 9).

621 622

Considering a system with 250 cells and a membrane area of 0.5 m2, the internal resistance can be calculated using the information from manufacturers (as the

623

information gathered in Table 2) along with the membrane area and the number of cells.

624

The present work considers that the membranes have a square geometry, but a

625

rectangular shape could also have been adopted.

626

The voltage and current, as a function of charge time, are showed in Fig. 15 (a) and

627

Fig. 15 (b) respectively. The calculated theoretical energy efficiency was 93.7% without

628

considering the pumping losses.

629

Fig. 16 (a) and (b) display the variation of the flow rate for each liquid electrolyte

630

during discharge and charge cycles, respectively. It can be seen that there is an

631

exponential increase of the flow rate along time during both charge and discharge.

632

It can be seen that this system will have very high flow rates during the ending of

633

the charge and it very high pumping power will be needed, resulting in a very low

634

efficiency, around 9%.

635

So, in order to reduce the pumping power and increase the system efficiency, a

636

reduction of the SoC interval used is proposed. A good compromise seems to be

637

choosing a maximum SoC obtained during charge of 97% (which reduces the total

638

number of cycles to 13). The result is shown in Fig. 17 (a) and (b) for the discharge and

639

charge respectively.

640

Under these conditions the minimum SoC achieved during discharge will be 4.7%,

641

and the VRFB efficiency considering pumping losses (ÎˇtotalVRFB) will be 91.7% (93.8%

642

without pumping losses). However the reduction of the SoC range used also resulted in

643

a reduction of the system capacity used, from 462 kWh to only 426 kWh. On the other

644

hand, these 426 kWh available are affected by the discharge efficiency with means that

645

in these conditions the VRFB will only provide to the fast charger around 405 kWh.

646

This energy is sufficient to charge 26 EVs.

647 648

Economic Analysis

The cost of the system has been considered to be that provided by a manufacturer

649 650

for a complete system (324 000 €) with the power and capacity similar to the values

651

required by this project (100 kW and 405 kWh). This commercial system includes

652

tanks, pumps, control system, cell stack and liquids electrolytes. However, this is a

653

typical system in which all the components are sealed in a box. The present work

654

considers that switching the solid tanks by the rubber tanks will not impact the final

655

price of the system.

656

A 20 year life cycle is considered for the project, with 26 EVs being charged per

657

day, during all 365 days of the year. For the electricity prices, the values were estimated

658

according to the Portuguese electricity time-differentiated tariffs that can reach at nigh a

659

value close to 0.08 €/kWh. This was assumed as the price of electricity bought to charge

660

the batteries (p). As for the selling price, the estimation departed from the values

661

reached during day for the Portuguese household consumers added of a premium value.

662

This premium can be justified by the fast charging possibility offered to the EV user and

663

also by the strategic location of the charging stations (e.g., motorways). The selling

664

price is then assumed to be 0.4 €/kWh. The sales per vehicle depend on the energy

665

supplied to charge the vehicle (Esold) (14.8 kWh according to the charge profile of Fig.

666

9). The inflation rate (i) is considered to be 1.8%, a value corresponding to the average

667

Harmonized Index of Consumer Prices for the EU1. The tax over gain was assumed to

668

be (TOG) 25%, and the minimum acceptable rate of return was settled as (MARR) 5%

669

for the base case analysis. The savage value of the equipment was considered to be null

670

at the end of the 20 years life time. The efficiency of the CHAdeMo chargers and VRFB

671

AC/DC charger considered are showed in Table 3. Under the assumed conditions the

672

global NPV reaches 109 906 €, which demonstrates the potential financial interest of the

Average value for the euro area for the period 1996-2015. Information obtained from the European Central Bank on https://www.ecb.europa.eu/stats/prices/hicp/html/inflation.en.html, consulted on September 2015. 1

673

project for private investors. As for the simple payback period, it can be reached

674

between the 8th and 9th years.

675

However, the present approach represents an innovative project and the uncertainty

676

of the financial and even technical assumptions should be recognized. In particular,

677

aspects such as the number of EVs charging per day, technical assumptions on

678

efficiency, the investment cost, the electricity prices differential or the discount rate are

679

important parameter whose values can have a high impact on the financial viability of

680

the investment. In fact, the assumed required investment for this system is still high

681

because it is not yet a mature technology and there are still not a lot of manufacturers

682

worldwide. It is expected that during the next few years the price of this technology will

683

decrease. The electricity prices, the access or not to differentiated tariffs or even the

684

existence of different electricity tariff schemes are country or region dependent, which

685

turns important the computation of the minimum prices differential to ensure the project

686

interest. The number of EVs charged and the technical efficiency of the system have a

687

direct impact on the sales and as such have also a major role on this analysis. Finally the

688

discount rate (which is the MARR affected by the inflation rate) can influence

689

significantly the obtained values and it is on the other hand highly influenced by the

690

perceived risk of the investment.

691

692

Sensitivity Analysis

693

In the previous subchapters the system viability was analysed both technologically and

694

economically. However the results of the analysis performed are specific for the input

695

parameters assumed and presented in Table 2 and Table 3. So in this subchapter a

696

sensitivity analysis is made in order to evaluate the effect of the variation of the main

697

input parameters on the viability of the project.

698

The technological viability is mainly dependent on the efficiency of the system, which

699

is its most important characteristic. It can be seen in Fig. 18 thru Fig. 21 that this system

700

has a high efficiency and it is still possible to increase it by varying, on one hand, some

701

of the construction parameters, like the number of cells and the membrane area and on

702

the other hand, it is also possible to adjust the operational parameters, like the number

703

of EVs in simultaneous charging and the VRFB charging time. In another way, it can be

704

seen in these figures that the efficiency does not change abruptly with the variation of

705

any parameter studied, which means that this system has a good flexibility of design and

706

operation.

707

Another important parameter that must be studied is the maximum current in the

708

system, since if too high currents are present, they might require the use of excessive

709

cable sections.

710

The first parameter studied is the number of EVs charged simultaneously. When

711

comparing the original two EVs against only one car, higher currents and flow rates will

712

result, yielding lower efficiencies. In practice, however, the average number of EVs

713

being charged at a given moment will be somewhere between one and two, with the

714

system efficiency being located somewhere between these two cases. This is illustrated

715

in Fig. 18, where a total of 26 EV charges were simulated for both situations.

716

Another important parameter that influences the system efficiency is the number of cells

717

used. The use of more cells will result in higher output voltages and lower currents,

718

which will result in lower losses. On the other hand, the use of more cells will mean

719

higher system costs. Estimating the variation of system cost as a function of the size of

720

the system is out of the scope of the present work, but the size chosen seemed to be a

721

good compromise between a reasonable system size, efficiency and maximum current

722

during the discharge.

723

Analysing Fig. 19 it can be seen that for a system with 125 cells the efficiency is low

724

(around 86%) and the current is around 800 A, which is very high and would require the

725

use of cables with high section. 250 cells seems to be a good choice because it results in

726

reasonable values of maximum current and efficiency.

727

The membrane area is also an important parameter. It can be seen in Fig. 20 that the

728

efficiency increases with the increase of the membrane area. On the other hand, the

729

maximum current density during the discharge also decreases. However, bigger

730

membranes means higher costs. Therefore, like in case of the choice of the number of

731

cells, it would be also necessary to take this into account. A membrane area of 0.5 m2

732

was chosen as being to be a good compromise.

733

The last parameter analysed is the VRFB charging time. Higher charging times

734

represent a lower charging power and therefore a higher efficiency due to the lower

735

required flow rates and currents. But it also increases the period of unavailability of the

736

system for EV charging. However, the charging time can be reduced by increasing the

737

charging power but it also increases the maximum current needed. This is a key part of

738

the economic viability of the project. When analysing Fig. 21 it can be seen that the

739

variation of system efficiency with varying charging time is not that significant. This

740

work considers that the system will be charged in 12 hours at constant power, during the

741

night, when both the demand and the electricity costs will be lower.

742

Regarding to the economic viability, it will depend on the NPV and the payback time.

743

Although a positive NPV the project can be considered as economically viable, long

744

payback times might render the project unattractive.

745

The relationship between the technological and the economic viability is provided

746

by the VRFB system efficiency (with pumping) which is the main parameter directly

747

affecting the total efficiency of the system (equation 25). This efficiency will be

748

translated into the relationship between the energy consumed from the grid and the

749

energy sold to charge the electric vehicles. Therefore, the economic viability of this

750

project will be strongly affected by the reduction of the efficiency of the system. Fig. 22

751

shows the variation of the NPV and payback time as a function of the VRFB efficiency

752

(with pumping).

753

The 80% efficiency is the value announced by a manufacturer, as previously referred.

754

Also, the efficiency calculated in section 4.3 for two EVs (91.7%) is also presented. It is

755

possible to see that the pumping influences greatly the NPV, so in order to improve the

756

efficiency it is possible to reduce the pumping losses by using a lower compression rate,

757

which can be seen in Fig. 12, increasing the permeability. However, it can be seen in

758

Fig. 11 that reducing the compression rate will also increase the resistivity, which will

759

reduce the efficiency, so an optimal trade-off between these two effects should always

760

be pursued in a real application.

761

Another important factor that affects the economic viability of this project is the

762

MARR, which has been considered to be 5% in the previous section.

763

The NPV almost increases 50% when decreasing the MARR from 5 to 4%. However, it

764

can be seen in Fig. 23 that for values of 9% and below the NPV is already negative,

765

which means that this project would only be economically viable for lower values of

766

MARR. Recognizing this as an innovative project, the value of the MARR seems to be

767

a critical aspect demonstrating the need to reduce costs or resource to public support

768

schemes at least in the initial demonstrative phase of the technology.

769

The number of EVs charged per day is also an important factor and it can be seen in

770

Fig. 24 that NPV is positive if more than 19 EVs are charged per day during 20 years.

771

The NPV nearly doubles when increasing from 22 EVs to 26 EVs per day.

772

As already mentioned, it is expected that the price of this kind of system will tend to

773

decrease along time. Therefore, it is useful to analyse the variation of the NPV with the

774

variation of the system cost, as represented in Fig. 25. It can be seen that it is possible to

775

nearly double the NPV with a 40% reduction in the system initial investment cost.

776

However, it is also clear that if the investment increases by more than around 30% the

777

project will no longer be financially viable. Once more, this demonstrates the need to

778

further develop the technology reducing the risk for the investors.

779

Finally it is worth to analyse how the electricity selling prices and their differential

780

represents a fundamental parameter for the project financial attractiveness. Fig. 26

781

shows the obtained NPV for different electricity selling prices, showing that a price

782

differential above 0.22â&#x201A;Ź/kWh (0.30â&#x201A;Ź/kWh - 0.08â&#x201A;Ź/kWh) must be ensured for the project

783

to be considered attractive.

784

785

4. CONCLUSIONS

786

The proliferation of Electric Vehicles (EVs) will bring a higher demand for battery

787

fast charging locations. Also, the high power demanded for fast charge stations is a

788

disadvantage. The use of energy storage systems, and in particular, Vanadium Redox

789

Flow Batteries (VRFBs) seems to be a good solution for reducing the installed power

790

with a peak shaving strategy. Existing or recently deactivated gas stations are privileged

791

locations for this purpose and many of them have available space and unused fuel

792

storage tanks. Furthermore, flow batteries also provide the possibility of taking

793

advantage of the availability of empty fuel storage tanks.

794

The present work details a preliminary project of a Vanadium Redox Flow Battery

795

of first generation (G1) to be used in gas stations for supplying electric energy for two

796

CHAdeMo chargers (50 kW each) working simultaneously. The VRFB is charged for

797

12 h during off-peak power demand (at night). The same cell stack is used for charging

798

and discharging the liquid electrolytes. This preliminary project was conceived using

799

commercially available system components.

800

A method for storing the liquid electrolytes of VRFBs in rubber tanks (installed

801

inside the fuel storage tanks normally used in gas stations) has been proposed. This will

802

prevent corrosion of the fuel tanks, which are normally made of steel, while allowing

803

the storage of both liquid electrolytes (anode and cathode) in the same fuel tank without

804

mixing. Furthermore, the flexibility of the rubber enables the use of large rubber tanks

805

that can still be inserted through the manhole entry of these fuel tanks.

806

The preliminary project was assessed in terms of voltage, current, power and

807

efficiency (both with and without pumping losses). An efficiency around 92%

808

(including pumping losses) has been predicted when using the VRFB system to fast

809

charge 26 EVs per day. Nevertheless, a more detailed assessment will require the

810

manufacture and testing of a prototype. Although specific input parameters have been

811

used, the methodology proposed in this work is sufficiently detailed to allow the

812

assessment of diverse configurations.

813

A cost analysis of the preliminary project was also performed in terms of Net

814

Present Value (NPV) for 20 years (the life time considered for the system) and the

815

Payback Time. It has been concluded that, for the input parameters considered and for a

816

Minimum Acceptable Rate of Return (MARR) of 5%, the project is economically

817

viable, and the investment is recovered in 9.5 years with a NPV of 109.9 kâ&#x201A;Ź. However,

818

the sensitivity analysis clearly shows the fragility of the obtained values, which should

819

be looked with caution given the initial stage of the project. The sensitivity analysis

820

illustrated that slight variations in the input conditions might substantially affect the

821

viability analysis. It is particularly important to recognize the potential for technology

822

development and even reach economies of scales that would make the project much

823

more attractive in the short term. However, the viability largely depends on the

824

electricity market operating conditions, namely electricity time-differentiated tariffs and

825

adequate regulatory conditions for the development of this business model supported on

826

free electricity market trading. Future work should then address both technology

827

development and regulatory and market environment as key factors for the commercial

828

success of the project. Additionally, issues related to the impacts of these projects on

829

the management of the electrical power grid, and their contribution for facilitating the

830

incorporation of intermittent renewable energy sources (such as wind power and solar

831

photovoltaic), with the benefits of reducing fossil fuel dependency and associated

832

emissions must also be further explored.

833

In the midterm, if the energy density of flow batteries substantially increases, the

834

philosophy proposed in the present work will be easily adapted in order to charge the

835

Electric Vehicles by substituting their discharged liquid electrolytes by charged

836

electrolytes, instead of indirectly charging the vehicle electrically via a fast charging

837

station. This would eliminate one of the major current disadvantages of EVs, which

838

consists on their long charging times. The time required to charge an EV would then be

839

similar to the time required to supply a conventional fuel vehicle.

840 841 842

ACKNOWLEDGEMENTS The authors would like to thank Petrotec S.A. for the support given to the present

843

work. This project was funded by Project MOBI-MPP (MIT-Pt/EDAM-

844

SMS/0030/2008) supported by the MIT Portugal Program and FEDER funds through

845

the Programa Operacional Factores de Competitividade - COMPETE and National

846

Funds through FCT - Foundation for Science and Technology. Francisco P. Brito

847

benefited from post-doctoral grants SFRH/BPD/51048/2010 and

848

SFRH/BPD/89553/2012 supported by the MIT Portugal EDAM and FCT, respectively.

849

850

REFERENCES

851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908

[1] International Energy Outlook 2009. Energy Information Administration Office of Integrated Analysis and Forecasting U.S. Department of Energy Washington DC; May 2009. [2] Zdenek C, Pavel M. Electric, hybrid electric and combustion engine driven cars and their impact on environment. Conference Electric, hybrid electric and combustion engine driven cars and their impact on environment. p. 1-5. [3] Martins J, Brito F, Pedrosa D, Monteiro V, Afonso JL. "Real-Life Comparison Between Diesel and Electric Car Energy Consumption". “Grid Electrified Vehicles: Performance, Design, and Environmental Impacts”. New York: Nova Science Publishers; 2013. [4] Silva C, Ross M, Farias T. Analysis and simulation of “low-cost” strategies to reduce fuel consumption and emissions in conventional gasoline light-duty vehicles. Energy Conversion and Management. 2009;50(2):215-22. [5] MARTINS JJG, UZUNEANU K, RIBEIRO BS, JASANSKY O. THERMODYNAMIC ANALYSIS OF AN OVER-EXPANDED ENGINE SAE 2004-01-0617 2004. [6] Ribeiro B, Martins J. Direct Comparison of an Engine Working under Otto, Miller and Diesel Cycles: Thermodynamic Analysis and Real Engine Performance. SAE SP. 2007;2093:75. [7] Brito FP, Martins J, Goncalves LM, Sousa R. Temperature Controlled Exhaust Heat Thermoelectric Generation. SAE International Journal of Passenger Cars - Electronic and Electrical Systems. 2012;5(2):561-71. [8] F.Brito, Martins J, Goncalves LM, Antunes N, Sousa D. Influence of Heat Pipe Operating Temperature on Exhaust Heat Thermoelectric Generation. SAE International Journal of Passenger Cars - Mechanical Systems 2013-01-0559. 2013;6(2). [9] Boulanger AG, Chu AC, Maxx S, Waltz DL. Vehicle Electrification: Status and Issues. Proceedings of the IEEE. 2011;99(6):1116-38. [10] Galus MD, Vayá MG, Krause T, Andersson G. The role of electric vehicles in smart grids. Wiley Interdisciplinary Reviews: Energy and Environment. 2013;2(4):384-400. [11] Martins J, Brito F. "Carros Elétricos". Porto, Portugal: Publindústria, Dez. 2011. [12] Hawkins TR, Singh B, Majeau-Bettez G, Strømman AH. Comparative Environmental Life Cycle Assessment of Conventional and Electric Vehicles. Journal of Industrial Ecology. 2013;17(1):53-64. [13] Camus C, Farias T. Electric vehicles as a mean to reduce, energy, emissions and electricity costs. Conference Electric vehicles as a mean to reduce, energy, emissions and electricity costs. p. 1-8. [14] Cadoux F, Gross G. Integration of Vehicles with Rechargeable Batteries into Distribution Networks. Smart Grids: John Wiley & Sons, Inc.; 2013. p. 243-61. [15] Ribeiro B, Brito F, Martins J. "A Survey on Electric/Hybrid Vehicles": Edited by Ronald K. Jürgen, SAE International, Warrendale, USA, 2010. [16] Faria R, Moura P, Delgado J, de Almeida AT. A sustainability assessment of electric vehicles as a personal mobility system. Energy Conversion and Management. 2012;61(0):19-30. [17] Ribau J, Silva C, Brito FP, Martins J. Analysis of four-stroke, Wankel, and microturbine based range extenders for electric vehicles. Energy Conversion and Management. 2012;58(0):120-33. [18] Role of Chademo: http://www.chademo.com/wp/role/challenge/ ( last acessed in 11/10/2013) [19] ChadeMo Association: http://www.chademo.com/ (last acessed in 15/07/2013) [20] Sekyung H, Soohee H, Sezaki K. Development of an Optimal Vehicle-to-Grid Aggregator for Frequency Regulation. Smart Grid, IEEE Transactions on. 2010;1(1):65-72. [21] Saber AY, Venayagamoorthy GK. Plug-in Vehicles and Renewable Energy Sources for Cost and Emission Reductions. Industrial Electronics, IEEE Transactions on. 2011;58(4):1229-38. [22] Kerestes RJ, Reed GF, Sparacino AR. Economic analysis of grid level energy storage for the application of load leveling. Conference Economic analysis of grid level energy storage for the application of load leveling. p. 1-9. [23] Kerestes RJ. Economic Analysis of Grid Level Energy Storage for the Application of Load Leveling: University Of Pittsburgh, 2011. [24] EDP energia: http://www.energia.edp.pt/ ( last acessed in 15/07/2013) [25] Notre Dame Open Course Ware - environmental philosophy - Chapter 9: Replacing fossil fuel with clean energy - Last acessed in 07/11/2013: http://ocw.nd.edu/philosophy/environmentalphilosophy/eduCommons/philosophy/environmental-philosophy/unearthed/chapter-9-replacing-fossilfuel-with-clean-energy [26] Department of Alternative Energy Development and Efficiency - AEDP 2012-2021: http://www.dede.go.th/dede/images/stories/dede_aedp_2012_2021.pdf ( Last acessed in 07/11/2013) [27] Carr DS. Ten-megawatt load-levelling lead/acid battery project. Journal of Power Sources. 1988;23(1– 3):183-92.

909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970

[28] Nakayama Y, Takahashi S, Hirakawa K, Yamaguchi Y. Development of a long life 35 Ah capacity VRLA battery for load-leveling applications. Journal of Power Sources. 2004;125(1):135-40. [29] Iwahori T, Ozaki Y, Funahashi A, Momose H, Mitsuishi I, Shiraga S, et al. Development of lithium secondary batteries for electric vehicles and home-use load leveling systems. Journal of Power Sources. 1999;81–82(0):872-6. [30] Shibata A, Sato K. Development of vanadium redox flow battery for electricity storage. Power Engineering Journal. 1999;13(3):130-5. [31] Hagedorn NH, Thaller LH. Design Flexibility of Redox Flow Systems. National Aeronautics and Space Administration Lewis Research Center, DOE/NASA/12726-16, NASA TM-82854. 1982:19. [32] Wang W, Luo Q, Li B, Wei X, Li L, Yang Z. Recent Progress in Redox Flow Battery Research and Development. Advanced Functional Materials. 2013;23(8):970-86. [33] Cunha Á, Martins J, Rodrigues N, Brito FP. Vanadium redox flow batteries: a technology review. International Journal of Energy Research. 2015;39(7):889-918. [34] Skyllas-Kazacos M, Rychick M, Robins R. All-vanadium redox battery. US Pat, 4 786 567 1988. [35] Kear G, Shah AA, Walsh FC. Development of the all-vanadium redox flow battery for energy storage: a review of technological, financial and policy aspects. International Journal of Energy Research. 2012;36(11):1105-20. [36] International Renewable Energy Agency - Energy Storage and Renewables for Island Power - A guide for decision makers Last acessed in 07/11/2013: http://www.irena.org/DocumentDownloads/Publications/Electricity%20Storage%20and%20RE%20for% 20Island%20Power.pdf [37] Cleantech Conference & Showcase 2013, Washington DC - Last Acessed in 07/11/2013: http://www.techconnectworld.com/Cleantech2013/a.html?i=1574 [38] Eurostat - Energy consumption of transport - Last acessed in 07/11/2013: http://epp.eurostat.ec.europa.eu/tgm/table.do?tab=table&plugin=1&language=en&pcode=tsdtr250 [39] SHIGEMATSU T. Redox Flow Battery for Energy Storage. SEI Technical Review. 2011;73:4-13. [40] Li W, Liu J, Yan C. Graphite–graphite oxide composite electrode for vanadium redox flow battery. Electrochimica Acta. 2011;56(14):5290-4. [41] Lai Q, Zhang H, Li X, Zhang L, Cheng Y. A novel single flow zinc–bromine battery with improved energy density. Journal of Power Sources. 2013;235(0):1-4. [42] Remick RJ, Ang PGP. Electrically rechargeable anionically active reduction-oxidation electrical storage-supply system US Patent 4485154, November 271984. [43] Clarke RL, Dougherty B, Harrison S, Millington PJ, Mohanta S. Cerium Batteries. US Patent 2004/0202925 A1 2004. [44] Skyllas-Kazacos M, Kazacos G, Poon G, Verseema H. Recent advances with UNSW vanadium-based redox flow batteries. International Journal of Energy Research. 2010;34(2):182-9. [45] Vanadium Redox Flow Batteries: An In-Depth Analysis EPRI, Palo Alto, CA: 2007 1014836. [46] TASSIN N. Investigation on Storage Technologies for Intermittent Renewable Energies: Evaluation and recommended R&D strategy. 2006. [47] Helen Prifti AP, Suminto Winardi , Tuti Mariana Lim , Maria Skyllas-Kazacos. Membranes for Redox Flow Battery Applications. Membranes. 2012;2:32. [48] Parasuraman A, Lim TM, Menictas C, Skyllas-Kazacos M. Review of material research and development for vanadium redox flow battery applications. Electrochimica Acta. (0). [49] Skyllas-Kazacos M. Vanadium/polyhalide redox flow battery. US Patent 7 320 844 B22008. [50] Golden Energy Century Ltd: http://www.gec.cn/en/ (last acessed in 12/06/2013) [51] Golden Energy Fuel Cell Co., Ltd.: http://www.gefc.com/en/ ( last acessed in 12/06/2013) [52] Gildemeister: http://www.gildemeister.com/de ( last acessed in 12/06/2013) [53] Prudent Energy: http://www.pdenergy.com/ (last acessed in 12/06/2013) [54] REDT - Renewable Energy Dynamics Technology Ltd.: http://www.poweringnow.com/ ( last acessed in 12/06/2013) [55] V-Fuel: http://www.vfuel.com.au/ (last acessed in 12/06/2013) [56] Peng Qian HZ, Jian Chena , Yuehua Wen , Qingtao Luo ,, Zonghao Liu DY, Baolian Yi. A novel electrode-bipolar plate assembly for vanadium redox flow battery applications. Journal of Power Sources. 2007;175:8. [57] Henriques & Henriques S.A: http://www.heh.pt/ (last acessed in 15/07/2013) [58] Citergaz: www.citergaz.com ( last acessed in 15/07/2013) [59] Tokuda N, Kanno T, Hara T, Shigematsu T, Tsutsui Y, Ikeuchi A, et al. Development of a Redox Flow Battery System. SEI Technical Review. 2000;50:88-94. [60] Mykin Inc.: http://mykin.com/ (last acessed in 07/08/2013) [61] http://www.customadvanced.com/. Custom Advanced Connections, Inc. - last acessed in 07/08/2013. [62] http://www.ingersollrandproducts.com/_downloads/ChemGuide_8677-P.pdf. Ingersoll Rand - last acessed in 07/08/2013.

971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992

[63] Petrotec, PFAST MULTI: http://www.petrotec.eu/products-andservices/dispensers/solu%C3%A7%C3%B5es-ambientais/p-fast-multi.aspx ( last accessed in 9/2/2016) [64] Sanzhong B, Lukic SM. Unified Active Filter and Energy Storage System for an MW Electric Vehicle Charging Station. Power Electronics, IEEE Transactions on. 2013;28(12):5793-803. [65] You D, Zhang H, Chen J. A simple model for the vanadium redox battery. Electrochimica Acta. 2009;54(27):6827-36. [66] SGL Group - SIGRACET and SIGRACELL Components for Flow Batteries: http://www.sglgroup.com/cms/_common/downloads/products/productgroups/nm/Mobility/NMM_Components_for_Flow_Batteries.pdf (last acessed in 12/06/2013) [67] Spellman K, Stiles K, Little I. Economic Report on Vanadium Redox Flow Battery with Optimization of Flow Ratw 2013. [68] Moore MA, Counce RM, Watson JS, Zawodzinski TA, Kamath H. A Step by Step Methodology for a Base Case Vanadium Redox Flow Battery 2012. [69] Vetter M, Dennenmoser M, Schwunk S, Smolinka T, Dötsch C, Berthold S, et al. Redox flow batteries – Already an alternative storage solution for hybrid PV mini-grids? Proceedings 5th European Conference PV-Hybrid and Mini-Grid, pp 100-109. Tarragona, Spain(2010). p. 100-9. [70] Dumanic D, Wallin F. Flow Batteries Status and Potential 2011. [71] Ma X, Zhang H, Sun C, Zou Y, Zhang T. An optimal strategy of electrolyte flow rate for vanadium redox flow battery. Journal of Power Sources. 2012;203(0):153-8. [72] Streeter VL, Wylie EB. Mecanica dos Fluidos. 7ª ed1982.

FIGURES

993 994

995

Fig. 1 â&#x20AC;&#x201C; Example of a CHAdeMo fast charging station (62) (courtesy Petrotec)

996 997 998

(a) Load Leveling Energy Stored

Energy Supplied by the Storage System

Power

Energy Supplied by the Storage System

Power

Energy Stored

(b) Peak Shaving

Power Demanded by Loads

999 1000

1001

9 time (h)

Power Demanded by Loads

9 time (h)

Fig. 2 - Comparison between the processes of peak shaving (a) and load levelling (b)

Increase of Road Transport Energy Consumption (base year 1990)

2,5

1,5

EU (27 countries) Ireland Spain Portugal

Germany Greece Italy United Kingdom

0,5 1990

1002

1995

2000 Year

2005

2010

1003

Fig. 3 - Variation of road transport energy consumption between 1990 and 2011 in some

1004

countries of European Union (adapted from [38])

1005

1006 1007

1008

Fig. 4 - Operating principle of a Redox Flow Battery (RFB)

1009 1010

Fig. 5 - Typical charge/discharge cycle of a single cell (G1 technology) for a current

1011

density of 40 mA/cm2 (adapted from [56]).

1012

1013 1014

Fig. 6â&#x20AC;&#x201C; Typical fuel tank used in gas stations (courtesy of Henriques & Henriques S.A.)

1015

[57].

1016

1017

1018

Fig. 7 - Scheme of rubber tanks for VRFBs (outline view of fuel tanks with support

1019

structure): (a) Two tank configuration; (b) Four tank configuration.

1020 vi ii

v o io

Power Grid

v o io

Power Converter t

Vanadium Redox

Power Converter t

Flow Batteries

Power Converter

Renewables

1021

Fig. 8 â&#x20AC;&#x201C; Proposed system architecture.

1022

1023

1024 1025

Fig. 9 â&#x20AC;&#x201C; Typical evolution of the charging power output of a Nissan Leaf, as monitored

1026

by Bai et al. [64].

1027 1028

Fig. 10 - Discharging process of a VRFB.

1029

1030 1031

Fig. 11 - Electrical resistance of the GFA6EA graphite felt electrodes of SGL GROUP

1032

as a function of the compression rate [66].

1033

1034 1035

1036

Fig. 12 - Charging process of a VRFB.

1037

Fig. 13 - Permeability of the graphite felt electrodes SGL GFA6EA [66].

1038

1039

80%

400 60% 350 40% 300 20%

250 200

(a)

350

80%

300 250

60%

200

40%

150

20% 100 50

0% 0

1040

100

200

300

100%

Current SOC

0% 0

400

(b)

Time (minutes)

SOC

450

400

100%

Discharging Current (A)

Voltage SOC

SOC

Discharging Voltage (V)

500

100

200

300

400

Time (minutes)

Fig. 14 - Discharging Voltage (a) and Current (b) cycles for VRFB system.

1041

1042

350

60%

300

40%

250

20%

0% 0

1044

(a)

200

400

SOC

80%

140

60% 120 40% 100

20%

0% 0

600

Time (minutes)

100% Current

80%

200

1043

160

(b)

200

400

600

Time (minutes)

Fig. 15 - Charging Voltage (a) and Current (b) cycles for the VRFB system.

SOC

400

100%

Charging Current (A)

Voltage SOC

SOC

Charging Voltage (V)

450

1045

80%

1200 60%

800

40%

400

20%

0 100

(a)

200

300

100% Flow rate SOC

80%

120

60% 80 40% 40

20%

0% 0

1046

SOC Flow rate (L/min)

Flow rate (L/min)

1600

160

SOC

100% Flow rate (L/min) SOC

0% 0

400

200

400

600

(b)

Time (minutes)

Fig. 16 - Variation of flow rate during discharging (a) and charging (b) cycles.

1047

1048

80%

60%

40%

20%

0 100

200

300

100%

80%

60%

40%

20%

0% 0

Pumping power SOC

0% 0

400

(b)

200

400

600

1049

(a)

1050

Fig. 17 - Variation of pumping power during discharging (a) and charging (b) cycles.

Time (minutes)

1051

96% 94,8%

94%

92%

93,8%

93,4%

91,7% Efficiency with pump

90% 2 cars 1052

Efficiency without pump

1 car

SOC

100%

Pumping power SOC

SOC Pumping power (kW)

Pumping power (kW)

1053

Fig. 18 - Efficiency comparison between 1 and 2 EVs charging simultaneously.

1054 Efficiency without pumping

Efficiency with pumping

750

Maximum current

600 93%

450 300

89%

Current (A)

Efficiency

97%

900

150

85%

0 125 150 175 200 225 250 275 300 325 350 375 400

1055

Number of cells

1056

Fig. 19 - Variation of efficiency and maximum current as a function of the number of cells.

1057

99%

Efficiency

95%

250 200

91%

150

87%

100 83% 50

79% 75%

Current density (mA/cm2)

Efficiency without pumping Efficiency with pumping Maximum Current Density

0 0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

Membrane Area (m2) 1058 1059

Fig. 20 - Variation of the efficiency and maximum current density as a function of membrane

1060

area.

1061

300

95%

250 200

90%

150 85%

100

80%

75%

Maximum current (A)

Efficiency

100%

Efficiency without pumping Efficiency with pumping

Maximum current

0 6

VRFB charging time (hours)

1062 1063

Fig. 21 - Variation of the efficiency and maximum current (during charge) as a function of

1064

VRFB charging time.

1065

VRFB efficiency with pumping

NPV

150 000 €

100 000 €

50 000 €

0€

1066

NPV Payback time (yearss)

60.0% 47 183 9.72

70.0% 73 103 9.10

80.0% 92 544 8.69

91.7% 109 906 8.35

95.0% 114 030 8.27

1067

Fig. 22 - NPV and Payback time as a function of VRFB efficiency with pumping (with 91.7%

1068

corresponding to the efficiency obtained by the analysis with 2 EVs in simultaneous charging).

1069

MARR 150 000 €

NPV

100 000 €

50 000 €

0€ 4%

10%

-50 000 €

1070 1071

Fig. 23 – Net Present Value as a function of the Minimum Acceptable Rate of Return (MARR).

1072

Average number of EVs per day 150 000 €

NPV

100 000 € 50 000 € 0€

-50 000 €

1073 1074

1075

NPV Payback time (years)

18 -4 720 € 11.3

20 23 936 € 10.4

22 52 593 € 9.6

24 81 250 € 8.9

26 109 906 € 8.3

Fig. 24 - NPV and Payback time as a function of the average number of EVs charged per day.

System cost variation 300 000 €

NPV

250 000 € 200 000 € 150 000 € 100 000 € 50 000 €

0€ -50 000 €

1076

50% 60% 70% 80% 90% 100% 110% 120% 130% Série1 241 221 214 958 188 695 162 432 136 169 109 906 83 643 € 57 380 € 31 117 € Série2 4.5 5.4 6.1 6.9 7.6 8.3 9.0 9.7 10.4

1077

Fig. 25 - NPV and Payback time as a function of the change in the initial investment (100%

1078

means no change in cost).

1079

Influence of selling price (€/kWh) 250 000 €

NPV

200 000 € 150 000 € 100 000 € 50 000 € 0€ 0.30

1080 1081

1082 1083

0.35

0.40

0.45

0.50

-50 000 €

Fig. 26 - Influence of selling price on the Net Present Value

TABLES

1084 1085 1086

Table 1 - Compatibility of various types of rubber with sulphuric acid and vanadium oxides [60-62].

Natural Rubber Sulphuric acid C dilute (10%) Sulphuric acid B 25% Sulphuric acid B 25 – 50% Sulphuric acid 50 – 98% Sulphuric acid D 98% Vanadium Oxide Vanadium Pentoxide

EPDM Nitrile Neoprene Viton SBR

Silicone Teflon Butyl rubber rubber

1087 1088

A - Recommended

1089

effect

B - Minor to moderate negative effect

C - Moderate to severe negative

D - Not recommended

1090 1091

Table 2 - VRFB stack components selected for the analysis

1092

Membrane

Eletrodes

Bi-polar plates

Nafion 117

SGL GFA6EA [66]

SGL PPG86 [66]

Resistivity

15 Ωmm

1 Ωmm

Thickness

6 mm

3 mm

Compression ratio

20%

Resistance reduction

70%

0.1 Ωcm2

0.018 Ω

0.000297 Ω

250

500

249

Model

Electric Resistance Number of Elements 1093 1094

1095

1096 1097 1098

Table 3 - Input values used to simulate the VRFB system. Number of cells (N)

250

Cell equilibrium potentials (E+/ E-)

1.004 / -0.255 V

Energy density (E)

33 Wh/L

Volume of liquid stored (V)

14000 L

Dynamic viscosity of liquid electrolyte (Îź)

4.93x10-3 Pa.s

Density of liquid electrolyte (Ď )

1320 kg/m3

Membrane area (for each cell)

0.5 m2

Internal resistance (R)

4.897 mÎŠ

Efficiency of VRFB AC/DC charger (đ?&#x153;&#x201A;đ??´đ??śâ&#x2C6;&#x2019;đ??ˇđ??ś )

95%

Efficiency of ChadeMo fast charger (ÎˇCh)

95%

Initial (maximum) State of Charge (SoCmax)

100%

Operating Temperature (T)

25 ÂşC

Vanadium Concentration

2 M/L

Number of EVs charging simultaneously

VRFB Charging time (tch)

12 h