Optimization of autonomous hybrid systems with hydrogen storage_ Life cycle assessment.pdf

INTERNATIONAL JOURNAL OF ENERGY RESEARCH

Int. J. Energy Res. 2012; 36: –

Published online 28 February 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/er.1830

749 763

Optimization of autonomous hybrid systems with

hydrogen storage: Life cycle assessment

Geovanni Hern ´ ndez Galvez 1 , Oliver Probst 2 , O. Lastres 3 , Airel N ´ ˜ ez Rodr´guez 3 ,

Alina Juantorena Ug ´ s 4 , Edgar Andrade Dur ´ n 1 and P. J. Sebastian 1, ,y

1 Centro de Investigaci ´ n en Energ ´ a-UNAM, Temixco, Morelos 62580, M ´ xico

2 Instituto Tecnol ´ gico del Estudios Superiores de Monterre, Ave. Eugenio Garza Sada 2501 Sur, Col. Tecnol ´ gico 64849, Monterrey,

Nuevo Le ´ n, M ´ xico

3 Instituto de Estudios de la Energ´a, Universidad del Istmo, Oaxaca, M ´ xico

4 Universidad Polit ´ cnica del Estado de Morelos, Jiutepec, Morelos, M ´ xico

SUMMARY

The design of autonomous systems for the rural electriﬁcation is a complex task due to the diversity of variables

involved in such processes. The absence of programs and methods that carry out this task in a clear and precise

manner constitutes a barrier to the dissemination of these systems, although some tools have been developed that

present other possible limitations. The exclusion of the environmental dimension in the design and evaluation

process of hybrid systems means that the true beneﬁts are not evaluated in terms of quality and quantity. In an

attempt to overcome such deﬁciencies, this work presents a new method of design; approached from the multi-

objective optimization of systems. The multi-objective optimization by means of enumerative search implemented

by the Hybrid Optimization Model for Electric Renewable program is used to generate a set of solutions optimized

economically by the value of the net present cost (NPC). The analysis of greenhouse gas emissions (in tCO 2 -eq.) in

the life cycle of each one of the system components is carried out and a set of solutions with the values of the two

objective functions is generated, namely NPC and NAE SLC (net avoided emissions in the system life cycle). The

method is applied to a case study in a Cuban rural community. The compromise solution obtained by means of the

proposed algorithm includes a wind turbine (WT) of 25.4 and 8 kW of photovoltaic panels, while that of the HOGA

includes a WT of 76 and 21 kW of photovoltaic panels. Both commitment solutions consider hydrogen storage

instead of storage in batteries, as a better option for the energy storage. Copyright r 2011 John Wiley & Sons, Ltd.

KEY WORDS

stand-alone wind energy system; multi-objective optimization; life cycle analysis; fuel cells; electrolyzers

Correspondence

*P. J. Sebastian, Centro de Investigaci ´ n en Energ´a-UNAM, Temixco, Morelos 62580, M ´ xico.

y E-mail: sjp@cie.unam.mx

Received 7 October 2010; Revised 28 December 2010; Accepted 30 December 2010

1. INTRODUCTION

technologies with low carbon emission levels, accom-

panied by massive investment in next generation

technologies, without which growth cannot be

achieved with low carbon emissions.

In its 2009 publication of World Energy Outlook

(WEO), the International Energy Agency (IEA)

addresses issues of special relevance to the world

energy situation and forecasts for this area until 2030.

According to the report, expanding access to modern

energy for the world’s poor remains a priority. An

estimated 1.5 billion people (more than a ﬁfth of the

world population) still lack access to electricity.

Approximately 85% of these people live in rural areas,

mainly in sub-Saharan Africa and South Asia. It

is estimated that by 2030 the total number will be

Poverty reduction and sustainable development remain

a top priority at international level. Climate change

threatens the entire world, but developing countries are

the most vulnerable. They are estimated to bear

approximately 75–80% of the cost of damage caused

by climate change. Even warming of 21C above pre-

industrial temperatures could result in a permanent

reduction in gross domestic product between 4 and 5%

in Africa and South Asia. For the temperature not to

deviate from the 21C above pre-industrial levels

(probably the best outcome that can be achieved), a

true revolution is needed in the energy sector. This

entails the rapid dissemination of energy efﬁcient

749

G. H. Galvez et al.

Optimization of autonomous hybrid systems

decreased by only 200 million, but it will

increase

Mathematical models are developed for each compo-

nent of the system and compared with experimental

results. It was concluded that there are no

insurmountable technical difﬁculties associated with

hydrogen production by hybrid systems. Field

observations showed that hybrid systems are feasible

and reliable enough, and require less maintenance. In

addition, electrochemical effects caused by the inter-

mittent nature of wind and solar resources can be

decreased through the use of batteries as short-term

storage systems.

Several works have been published that investigate

the technical and economic feasibility of hydrogen sto-

rage in autonomous systems [3–8]. Khan and Iqbal [9]

performed the modeling, simulation and analysis of an

isolated wind system with hydrogen storage. MATLAB-

Simulink is used for system dynamics modeling. The

simulation results showed the proper functioning of

each system component and their mutual interactions.

Abdullah et al. [10] conducted a comparative study

between different power schemes for electriﬁcation of a

rural information and telecommunications center

located in a remote area of Sarawak, East Malaysia. This

case study demonstrates that combined photovoltaic and

hydrogen systems are more reliable, although currently

more expensive than stand-alone PV systems.

Dufo and Bernal [11] made a multi-objective triple

design of an isolated hybrid system, simultaneously

minimizing the total cost over the lifetime of the sys-

tem, pollutant emissions and the unmet load. They

used a multi-objective evolutionary algorithm and a

generic algorithm to ﬁnd the best combination of hy-

brid system components and control strategies. Of the

research papers that have been published on the sub-

ject, this is one of the most innovative, given the ad-

vantages of multi-objective optimization compared

with the mono-objective method.

Kasheﬁ et al. [12] designed a hybrid wind/photo-

voltaic/fuel cell system, under the criterion of mini-

mizing the annual cost over 20 years of operation. The

problem of optimization is subject to a reliable ful-

ﬁlling of demand, including the analysis of the ﬂaws of

WTs, of photovoltaic arrangements and of the DC/AC

converter. Particle Swarm Optimization, an optimiza-

tion algorithm based on particle clouds, is used in this

study. The results demonstrate the inﬂuence of com-

ponent failures on reliability and cost of the system.

The optimization of particle clouds is also used by

Hakimi and Moghaddas-Tafreshi [13] in the design of

an autonomous system for a residential area of

southeastern Iran. The system consists of fuel cells,

WTs, electrolyzers, a reformer, an anaerobic reactor,

and some tanks of hydrogen. Biomass is used as an

available energy source. In the system, the hydrogen

produced in the reformer is delivered directly to a fuel

cell. When the energy produced by the WT and the fuel

cell (fed by the reformer) is higher than the demand,

the excess is delivered to the electrolyzer. Otherwise, it

in Africa.

The energy sector will continue to be subjected to

profound changes and will face new challenges in the

coming years. The need for energy supply in isolated

areas with difﬁcult access to utility distribution net-

works, or areas where these are not economically

viable, will remain a challenge for many years to come

for various governments in order to ensure the devel-

opment of rural areas.

There have been many alternatives that are used in

different countries to provide electricity for isolated

consumers. Traditionally, diesel generators were used

with a corresponding environmental cost. Then, with

the development of renewable technologies, such as

hydroelectric systems, photovoltaics and small wind

turbines (WTs), there appeared an alternative for elec-

triﬁcation. Depending on available energy resources,

these technologies have been used as independent sys-

tems (only one source of energy) or as hybrid systems

(which involve more than one source of energy),

ensuring the autonomy of the electricity supply.

One of the intrinsic characteristics of these renewable

energy resources, which substantially differentiate them

from fossil fuels, is their variability. The intermittent

nature of these resources means that the exploitation by

autonomous systems requires the existence of modes of

energy storage. Traditional battery banks that have

been widely used have demonstrated technical and

environmental constraints. Battery storage requires a

high cost. In addition, the batteries are very sensitive to

unexpected operating conditions. Another downfall is

that most users cannot replace them due to a lack of

ﬁnancial resources, which means total loss of a func-

tioning system. Furthermore, the use of heavy metals

and more aggressive electrolytes in batteries can be

disadvantageous from the ecological point of view if

there is no careful recycling process in place.

An alternative to storing energy in batteries is the

integration of electricity generation technologies (wind

in our case) with hydrogen technologies (electrolyzers,

storage tanks and fuel cells). In the research that has

been done, the storage of energy in hydrogen in

autonomous wind systems has been approached from

several perspectives. The use of various simulation and

optimization tools (Hybrid Optimization Model for

Electric Renewable (HOMER), HYBRID2, TRNSYS,

HYDROGEMS, INSEL, ARES, RAPSIM, SOMES,

SOLSIM, CARE and HOGA) predominates in most

studies [1].

Sensitivity analyses carried out by several authors

[2,3] have shown that lowering the costs of hydrogen

storage systems to expected values in the medium and

long term will make them competitive with battery

storage systems, even without taking into account

externalities.

Sopian et al. [3] described the behavior of an in-

tegrated wind/photovoltaic/hydrogen/battery system.

750

Int. J. Energy Res. 2012; 36 :749–763

2011 John Wiley & Sons, Ltd.

DOI: 10.1002/er

Optimization of autonomous hybrid systems

G. H. Galvez et al.

uses an additional fuel cell that is fed by the stored

hydrogen. The optimized objective function is the net

present cost (NPC) of the system.

On the other hand, the literature review analysis did

not show the use of multi-criteria decision analysis to

optimize integrated wind energy systems with hydro-

gen storage, although such tools have been widely used

in various branches of engineering. Some of the main

developments in this regard are as follows:

Ramanathan and Ganesh [14] present a multi-ob-

jective programming model for the optimal allocation

of energy resources to various energy end uses. The

normative model was applied for the households sector

of Madras city. The model is solved using non pre-

emptive goal programming. A multi-objective pro-

gramming model with eight objectives is discussed. The

energy allocation process is viewed within a system

framework encompassing the economy and the en-

vironment. Objectives that are consistent with the

broad aims of economic costs, energy conservation,

balance of payments, employment generation and en-

vironmental friendliness were considered.

In Shi et al. [15], the optimal design of the hybrid

energy system has been formulated as a multi-objective

optimization problem. The techno-economical perfor-

mance of the hybrid energy system was optimized; the

trade-offs between the multi-objectives were analyzed

using multi-objective genetic algorithms. The proposed

method is tested on the widely researched hybrid PV

wind power system design problem. The optimization

seeks the compromise system conﬁgurations with re-

ference to three incommensurable techno-economical

criteria, and uses an hourly timestep simulation pro-

cedure to determine the design criteria with the

weather resources and the load demand for one re-

ference year. The well-known efﬁcient multi-objective

generic algorithm, called NGAS-II is applied.

Chedid et al. [16] provide a methodology for the

optimization of an existing electrical distribution net-

work when upgraded by renewable energies. The pro-

posed problem was formulated using multi-objective

linear programming in conjunction with fuzzy logic. It

will be shown that optimization using fuzzy logic can

provide decision makers with more ﬂexibility which

would assist them in the allocation of various energy

resources to optimally meet the various end uses and

solve the problem of renewable energy connection to

existing distribution networks.

In Barzegar Avval et al. [17], the gas turbine power

plant with pre-heater is modeled and the simulation re-

sults are compared with one of the gas turbine power

plants in Iran. In the optimization approach, the ex-

ergetic economic and environmental aspects have been

considered. In multi-objective optimization, the three

objective functions, including the gas turbine exergy

efﬁciency, total cost rate of the system production

including cost rate of environmental impact and CO 2

emission, have been considered. The thermoenvironomic

objective function is minimized while power plant exergy

efﬁciency is maximized using a generic algorithm.

Baghernejad and Yaghoubi [18] developed a multi-

objective optimization scheme which was applied to an

Integrated Solar Combined Cycle System that pro-

duces electricity to ﬁnd solutions that simultaneously

satisfy exergetic as well as economic objectives. This

corresponds to a search for the set of Pareto optimal

solutions with respect to the two competing objectives.

The optimization process is carried out by a particular

class of search algorithms known as multi-objective

evolutionary algorithms.

Sayyaadi and Amlashi [19] performed the thermo-

dynamic and thermoeconomic optimization of a hor-

izontal geothermal air conditioning system. The

objective functions based on the thermodynamic and

thermoeconomic analysis are developed. An artiﬁcial

intelligence technique known as evolutionary algorithm

has been utilized for optimization. This approach has

been applied to minimize either the total levelized cost

of the system product or the exergy destruction of the

system. Three levels of optimization including thermo-

dynamic single objective optimization, thermoeconomic

single objective optimization and multi-objective opti-

mization (with simultaneous optimization of thermo-

dynamic and thermoeconomic objectives) are

performed. In multi-objective optimization, both ther-

modynamics and thermoeconomic objectives are con-

sidered, simultaneously. In the case of multi-objective

optimization, an example of decision-making process

for selection of the ﬁnal solution from available optimal

points on Pareto front is presented.

Despite research on the subject, some limitations

remain to be overcome:

Only Weibull’s probability density function is

used to adjust the distribution of frequencies of

observed wind speeds, which does not always

represent the best ﬁt. This brings out errors in the

estimation of diverse parameters such as available

wind power, the energy produced by WTs and

others. This fact is accentuated when the observed

wind histogram is bimodal.

An adequate analysis of adjustment by the WT to

wind resource available, in order to choose

between different models to achieve the best

energy performance expected in the speciﬁc site,

is lacking. This would allow a better approach to

overall system optimization.

Ideally, calculations of greenhouse gas (GHG)

emissions should be estimated for the system’s

operating period and not for the life cycle of each

of its components. This would give a more

accurate environmental dimension.

With this in mind, this research presents the multi-

objective optimization of autonomous systems with

751

Int. J. Energy Res. 2012; 36 :749–763

2011 John Wiley & Sons, Ltd.

DOI: 10.1002/er

G. H. Galvez et al.

Optimization of autonomous hybrid systems

2. MATERIALS AND METHODS

hydrogen storage. This was done using the enumera-

tive optimization carried out by HOMER in order to

carry out the optimization of net avoided emissions in

the life cycle derived from the calculation of the

equivalent GHG emissions for each of the components

of the system. Compromise programming is used to

select the best alternative (closer to the ideal) based on

multiple criteria.

The research was conducted by means of a case

study involving the electriﬁcation of a rural community

of 40 homes with approximately 200 inhabitants,

namely the Cuban coastal town of Playa Caletones

located in the Gibara Municipality of Holguı´ n

Province (Figure 1).

Figure 1. Location of the Playa Caletones rural community on the island of Cuba.

Figure 2. Average daily consumption and wind speed proﬁles for Playa Caletones rural community.

Figure 3. Average daily wind speed proﬁles for Playa Caletones rural community.

752

Int. J. Energy Res. 2012; 36 :749–763

2011 John Wiley & Sons, Ltd.

DOI: 10.1002/er

Optimization of autonomous hybrid systems

G. H. Galvez et al.

The economic optimization of the system was done

using the free software HOMER. This optimizes the

system according to the NPC by means of an enu-

merative search. For the study, data on wind speed

recorded over two years on the study site as well as

consumption proﬁles estimated for the type of rural

community in question were used. These were based on

studies conducted in similar communities (Figures 2

and 3). The average daily load is 120 kWh, with a peak

power of 16 kW.

At 10m above ground level the average wind speed

is 4.16m s 1 , increasing to 4.51 and 5.17m s 1

speed design that the WT must (start-up wind speed

V s , rated wind speed V r and cut-out wind speed V o )

better match the site was based on the Weibull para-

meters estimated by Windographer through the

method of maximum likelihood (k51.09 and

c 55.24m s 1 ). Given the value of k, we use the fol-

lowing equation to determine the optimal value of V r ,

as shown in Figure 4, which is derived from the results

published by Geovanni et al. [20]:

V r ¼ð k13:1 Þ c;

k p 1:8

ð 1 Þ

The following equations were used to determine the

values of V s and V o [21]:

V s ¼ 0:275 V n

for 20

and 50m, respectively (as logarithmic proﬁle).

Prior to the economic optimization of the system,

optimization on the WT was carried out according to

the best coupling between the power curve and the

theoretical distribution of frequencies of wind speeds

at the study site. The calculation of the optimum wind

ð 2 Þ

V o ¼ 1:850 V n ð 3 Þ

The conﬁguration of the system under study is shown

in Figure 5.

The system includes a WT, a fuel cell, hydrogen

tank, an electrolyzer, batteries and a converter. The

system, including photovoltaic modules and a diesel

generator, was also analyzed. Capital costs used in the

simulation of the system are shown in Table I [2].

The calculation of emissions in the life cycle was

performed in equivalent tons of carbon dioxide, using

the emission rates reported in system components [22].

Table II reports the values used, based on the energy

produced during the lifetime of each of the components.

For each of the combinations of components gen-

erated by HOMER, equivalent emissions in the life

cycle were calculated, then the net avoided emissions,

NAE SLC (Equation (4), where the subscript means

Figure 4. Optimal rated wind speed (normalized with respect

to the c scale factor of Weibull) for different intervals of variation

for the shape factor k.

Table I. Capital costs for components.

Components

Capital cost

1500 $ kW 1

3000 $ kW 1

Fuel cell

2000 $ kW 1

Electrolyzers

1300 $ kg 1

Hydrogen tank

1.3$ Ah 1

Batteries

1000 $ kW 1

Converter

800 $ kW 1

Diesel generator

6900 $ kW 1

Photovoltaic modules

Table II. CO 2 equivalent emissions in the life cycle system

components.

GHG (kg CO 2 -eq. kWh 1 )

Components

0.011

Photovoltaic module (mono-Si)

0.045

Diesel generator

0.88

Fuel cell (H 2 by electrolysis)

0.02

Electrolyzer and hydrogen tank

0.011

Pb-acid Battery

0.028

Figure 5. The sketch of the autonomous system that involves

all the components.

Converter

753

Int. J. Energy Res. 2012; 36 :749–763

2011 John Wiley & Sons, Ltd.

DOI: 10.1002/er

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: