Master's Program in Renewable Energy Systems, 60 credits
Nickyar Ghadirinejad
Energy technology, 15 credits
Halmstad 2018-05-29
M AST E R
Design of an off-grid renewable-energy hybrid
system for a grocery store: a case study in
Malmö, Sweden
ii
Abstract:
On planet Earth, fossil fuels are the most important sources of energy. However, these resources are limited and being depleted dramatically throughout last decades. Finding feasible substitutes of these resources is an essential duty for humanity. Fortunately, Mother Nature is providing us a number of good solutions for this crucial threat against our planet. Solar irradiance, wind blowing, oceanic and maritime waves are natural resources of energy that are capable of completely covering the annual consumption of all inhabitants on the Earth. In this research a set of components including “Northern Power NPS 100-24” wind generators, “Kyocera KD 145 SX-UFU” PV arrays, “Gildemeister 10kW- 40kWh Cellcube” battery bank and HOMER bi-directional converter system were considered and successfully applied on HOMER tool and Particle Swarm Optimization (PSO) method. The main design goals of the presented hybrid system are to use 100% renewable energy resources in the commercial sector, where all power is produced in the immediate vicinity of the business place, adding strong advertising values to the setup. In order to supply hourly required load for a grocery store (1000 ) in Malmö city with 115 kW peak load and 2002 kWh/d with maximum 0.1% unmet, the system was optimized to achieve minimum Levelized Cost of Energy (LCOE) and the lowest Net Present Cost (NPC). The HOMER simulation for quantitative analysis, along with a Particle Swarm Optimization (PSO) solution method is proposed and the results are compared. The results show that an optimized hybrid system with 3.12
LCOE, and power production of 28.5% by PV arrays and 71.5% by wind generators, is the best practice for this case study.
Keywords: Renewable energy, PV arrays, Wind turbine, Optimization, HOMER, PSO, Grocery store
iii
Sammanfattning
De fossila bränslena är idag de viktigaste energikällorna på jorden. Dessa resurser är dock begränsade och har utarmats i en allt högre takt under de senaste decennierna. Att hitta möjliga ersättare för dessa resurser är därför viktigt. Lyckligtvis tillhandahåller naturen ett antal bra lösningar för detta avgörande hot mot vår planet. Solstrålning, vind, havsströmmar och -vågor är naturliga resurser av energi som kan täcka hela den årliga globala förbrukningen. I den här rapporten studeras ett hybridsystem bestående av Northern Power NPS 100-24 vindkraftverk, Kyocera KD 145 SX- UFU solcellerspaneler, Gildemeister 10kW-40kWh Cellcube batteribank och HOMER dubbelriktad växelriktare. Detta modellerades och optimerades dels i mjukvaran HOMER, dels via optimeringsmetoden Particle Swarm Optimaization (PSO). Det övergripande designkravet för det presenterade hybridsystemet är att använda 100% förnyelsebar energi i en kommersiell verksamhet, där all elektricitet produceras i närhet av verksamheten, vilket kan ge tydliga marknadsföringsvärden till installationen. För att kunna möta energibehovet varje timme för en livsmedelsbutik (1000 ) i Malmö med 115 kW toppförbrukning och 2002 kWh/dag, med maximalt 0,1% ej mött behov, optimerades systemet för att uppnå minimal energikostnad (Levelized Cost of Energy, LCOE) och lägsta nettonuvärde (Net Present Cost, NPC). En HOMER-simulering för kvantitativ analys, tillsammans med en PSO-optimering, har genomförts och resultaten har jämförts. Resultaten visar att ett optimerat hybridsystem med LCOE på 3,12 SEK/kWh, där solceller står för 28,5% av kraftproduktionen och vindkraftverk för 71,5%, är den bästa lösningen för denna fallstudie.
Nyckelord: Förnyelsebar energi, Solcellerspaneler, Vindkraftverk, Optimering, HOMER, PSO, Livsmedelsbutik
iv
Acknowledgement
I would like to express my sincere gratitude to my advisor Dr. Fredric Ottermo (PhD) for his continuous support of my MSc study and research, for his patience, motivation, enthusiasm, and immense knowledge.
I would like to dedicate this Master’s thesis to my Parents and express them my gratitude and appreciation for their continuous support and encouragements during my whole life.
Hereby, I heartedly thank my friends, Mr. Hoshangswamy Kale and Mr. Lameck Niyomugabo, for their precious help throughout doing my Master’s thesis and my all friends who have made my stay in Sweden more enjoyable and memorable.
With especial thanks to my elder brother, Dr. Mazyar Ghadiri Nejad.
Halmstad, May.2018
Nickyar Ghadirinejad
ghadirinejad@gmail.com
+46-790-100934
v
* The cover photo, “Land Of The Giant Lollipops” is taken by Matt Molloy, Canadian
time-lapse photographer. The Times Stack technique presents passing of time in a
single photograph instead of a moving video.
vi
Table of Contents
1 Chapter one: Introduction ... 1
Introduction... 2
1.1 Electricity supply and usage in Sweden... 2
1.2 Renewable Energy and obstacles in electrical distribution ... 3
1.3 Renewable Energy and environmental impacts... 4
1.4 Hybrid power generator systems ... 4
1.5 Reliability of the RE hybrid systems ... 5
1.6 Optimization of hybrid system... 5
1.7 Thesis definition ... 5
1.8 2 Chapter 2: Literature review... 6
Introduction: ... 7
2.1 Objective and Multi Objective Optimization (MOP): ... 7
2.2 PSO ... 8
2.3 HOMER Pro ... 8
2.4 Applied optimization methods on hybrid systems... 8
2.5 Conclusion ... 10
2.6 3 Chapter three: Material and methods ... 11
Material and methods: ... 12
3.1 Load demand calculation: ... 12
3.2 Photovoltaic array ... 13
3.3 3.3.1 Wind turbine ... 14
3.3.2 Battery bank... 15
3.3.3 Converter ... 16
System architecture... 17
3.4 All possible scenarios ... 18
3.5 Objective function ... 18 3.6
vii
Constraints ... 18
3.7 Hybrid system architecture ... 19
3.8 Levelized Cost of Energy (LCOE): ... 19
3.9 Net Present Cost (NPC): ... 20
3.10 Particle swarm optimization (PSO)... 20
3.11 Conclusion... 21
3.12 4 Chapter four: Results and Discussions ... 22
Results and discussion: ... 23
4.1 System input data:... 23
4.2 Homer optimal results... 25
4.3 Electrical summary ... 26
4.4 PV arrays summary... 28
4.5 4.5.1 Wind Turbine summary... 29
4.5.2 Bi-directional converter summary ... 29
4.5.3 Battery bank... 30
Impact of selecting controller types... 32
4.6 Minimum SOC of battery bank ... 32
4.7 PSO optimal results ... 33
4.8 5 Chapter five: Conclusion and recommendations... 34
Conclusions... 35
5.1 Recommendation ... 35
5.2 Reference: ... 36
Appendix A: ... 39
viii List of Figures
Figure 3.1 Electricity consumption profile in 2010 for a 1000 m2 prototype grocery store located in
Malmö... 12
Figure 3.2 Power Curve of NPS 100C-24 Class III wind turbine ... 15
Figure 3.3 Price list of typical roof mounted PV system including inverter price [31] ... 17
Figure 3.4 Hybrid system arrangement ... 18
Figure 3.5 Proposed hybrid system on HOMER ... 19
Figure 4.1 Monthly average solar global horizontal irradiance in 2010 [34] ... 23
Figure 4.2 Monthly average wind speed data at 37 meter height in 2010 [35] ... 23
Figure 4.3 Monthly average temperature data in 2010 [36] ... 24
Figure 4.4 Monthly average load demand in 2010 for a typical 1000 m2 grocery store in Malmö ... 24
Figure 4.5 AC load profile... 24
Figure 4.6 Cost summary of run #1 ... 26
Figure 4.7 Kyocera KD 145 SX-UFU output (kW) ... 29
Figure 4.8 System Converter Inverter Output (kW)... 29
Figure 4.9 System Converter Rectifier Output (kW) ... 30
Figure 4.10. Annual electric production vs. total NPC (kr) (at optimal number of PV, WG& converter) ... 30
Figure 4.11 Frequency of battery State of Charge ... 30
Figure 4.12 Average AC Primary load ... 31
Figure 4.13 Annual State of Charge of battery bank ... 31
ix List of Tables
Table 1-1 Annual supply and usage of electricity in Sweden from 2006 till 2016 (in GWh) [1] ... 3
Table 2-1 Literature review on hybrid system optimization methods [11] ... 9
Table 3-1 Weighted average turnkey prices of typical PV panels (excluding VAT) ... 13
Table 3-2 Specifications of Kyocera KD145SX-UFU PV array... 14
Table 3-3 Specifications of Northern Power NPS 100C-24... 15
Table 3-4 Specifications of Gildemeister 10kW-40kWh CELLCUBE® ... 16
Table 3-5 Specification of system converter ... 17
Table 3-6 constraints descriptions ... 19
Table 3-7 PSO algorithm [10] ... 21
Table 4-1 Impacts of operating reserve factors on annual unmet load amout (using LF controller) ... 25
Table 4-2 Net Present Costs of run #1... 26
Table 4-3 Electrical summary of run #1 ... 28
Table 4-4 Kyocera KD 145 SX-UFU Electrical Summary of run #1 ... 28
Table 4-5 Northern Power NPS 100C-24 Electrical Summary ... 29
Table 4-6 Date and time of unmet loads ... 32
Table 4-7 Comparison of data collected: Minimum 30% SOC vs. proposed hybrid system... 33
Table 4-8 PSO optimal results... 33
Table 5-1Comparison between reported results of HOMER and PSO method ... 35
1
1 Chapter one: Introduction
2
Introduction 1.1
In recent decades, with the growing population and industrialization, the need for access to new energy resources proceeds increasingly. The risk of ending fossil fuels, and worries about endangered environment and ecology (i.e.
fossil fuels depletion, crucial threat against environment and future of our green planet) have been some of main scientists’ concerns. One of the valuable solutions which have been found is using renewable energy sources (RES).
Industrial and commercial sectors are the main electric users in most countries. One of the main priorities for every manufacturer and working producer company is how to reduce the operation costs and also providing their needed power to run their machines in lower price. For several centuries, incinerating fossil fuels and burning materials were the main source of generating power to run the industrial wheels. Polluting the environment, causing natural disasters (such as global warming and increasing in CO2 emission), had lead a plethora of scientists and green activists to find a new method to solve the mentioned problems. Trying to benefit from renewable resources, not only helped to reduce the detrimental effects, but also encourage stakeholders to approach to use these cost-effective and less pollutant resources.
By changing the human lifestyle of man in accordance to the technology progresses, human needs further energy production. If the traditional power supply continues in its same way, the air pollution crisis will have uncontrollable impacts on the environment. To name some of positive aspects of the use of sustainable and clean energies e.g. wind and solar power, it is worth mentioning their minimal dependence on the use of manpower and water resources.
Moreover, to reduce the cost of electricity transmission, specifically to remote areas, (such as considering less road construction costs and managing large network), most developed countries are providing great part of their electrical needs from RES. The raised issues were some of the most undeniable factors towards necessity of RES development which accelerated this evolution to happen.
Electricity supply and usage in Sweden 1.2
In this section, annual supply and usage of electricity in Sweden is depicted in Table 1-1 across 17 year period, starting in 2001 and proceeding through 2016. By increasing, annual electricity energy usage (157,861 GWh in 2006 to 166,786 in 2016), It is clear from the Table 1-1 that the losses are quite large which can be decreased by using Distributed Generation (DG).
.
3
Table 1-1 Annual supply and usage of electricity in Sweden from 2006 till 2016 (in GWh) [1]
Production 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Hydro power
61,192
65,591
68,550
64,973
66,773
66,609
78,412
60,935
63,334
74,806
61,713
Wind power
988
1,432
1,996
2,485
3,502
6,101
7,165
9,842
11,234
16,268
15,479
Solar .. .. .. .. ..
13
19
35
47
97
143
Nuclear power
64,983
64,279
61,266
49,987
55,626
58,026
61,393
63,597
62,185
54,347
60,542 Conventional
thermal power
13,151
13,406
14,150
15,839
19,056
16,779
15,456
14,789
13,155
13,419
14,621 CHP in
industry
5,142
5,707
6,063
5,894
6,242
5,790
6,111
5,640
5,583
5,613
5,527 CHP in public
steam ant hot water works
7,249
7,163
7,402
9,484
12,276
10,180
9,015
8,839
7,151
7,568
8,803 condensing
steam power
749
510
666
444
517
801
318
300
411
227
278 gas turbines
and others
12
26
20
17
21
9
12
11
11
11
14 T otal
production 140,314
144,708
145,962
133,200
144,912
147,528
162,444
149,198
149,956
158,937
152,499
Imports
17,547
16,051
12,754
13,771
14,932
12,481
11,680
12,674
13,852
9,294
14,287 Total supply
157,861
160,759
158,716
146,971
159,844
160,009
174,124
161,872
163,808
168,230
166,786
Use
Manufacturing industries, mines &
quarries
57,406
57,944
56,558
50,657
53,359
53,843
52,981
50,935
49,552
48,784
49,506
Services
40,039
40,964
40,868
40,667
41,352
40,041
40,682
40,717
39,855
41,023
42,291
Agriculture
3,252
2,967
2,648
3,045
3,184
2,993
3,150
3,109
2,998
3,149
3,227
Households
34,807
33,457
33,470
33,934
37,282
33,702
35,086
34,431
32,636
33,841
35,071
Losses
10,860
10,691
10,456
9,583
11,813
9,703
10,966
10,003
9,292
9,539
10,669 T otal use
within Sweden
146,364
146,023
144,000
137,886
146,991
140,282
142,864
139,195
134,333
136,336
140,764
Exports
11,497
14,736
14,716
9,085
12,853
19,714
31,254
22,676
29,475
31,894
26,022
Total use
157,861
160,759
158,716
146,971
159,844
159,996
174,124
161,872
163,808
168,230
166,786
Renewable Energy and obstacles in electrical distribution 1.3
Increase in population and fast industrialization process over the previous years has prompted an expansion in electrical energy utilization. Power quality issues and voltage unsteadiness are its outcomes. Likewise, issues for example, voltage drop and high distribution losses in transmission lines in electrifying rural areas, demo nstrate a poor system execution [2]. On the other hand, electric power research organizations, due to their financial allocations, facing it difficult to reinforce and improve their grids. Distributed Generation (DG) can be one of the favored alternatives. Distributed Generation, because of its eagerness to utilize sustainable power sources, and creating cogeneration units, has encountered a solid development as of late [2].
4
Increase in greenhouse gases has showcased a danger to the ecosystem and environment and will have drastic effects like global warming, melting of glaciers in arctic area, massive droughts, imbalance in the ecosystems in large number of areas etc. Thus, using distributed generation resources, precisely which operates on clean energies, has been considered the most.
Considering all the aspects discussed above, using renewable energy sources is widely suggested to evade high transmission costs bared due to power production by traditional power plants to rural loads and minimize price of power generation.
Today most common renewable energy technologies are related to harness of wind and solar energy. In this research Wind turbines and Solar PV cells exhibits are used. It ought to be noticed that DG either gives the power as an off-grid source, isolated from national distribution grids (stand-alone systems), or as an incorporated piece of a network to give some portion of grid power.
Renewable Energy and environmental impacts 1.4
To minimize the use of traditional energy sources, wind and solar power plays a vital role but also, they cannot be the only source of electricity in a grid system of distribution for base- load system. Power generated from fossil fuels and conventional form energy production, produces significant quantity of pollutant per kWh of electricity generated. Wind-solar hybrid system on the other hand emits no such pollutants hence giving clean energy which affects environment in no way or in very minimal way. Out of all type of emissions or pollutants, only CO2 emission was considered in this thesis work. In recent decades, and coming future governments and organizations are more and more aware of the possibilities of renewable energies, depending on the resources and technologies they possess they can take advantage of this source.
Europe Union countries, for example in 2010 produced 12% of its required electricity through renewable sources [3]. Apart from considering the environmental impact, the other important contribution or reason for considering renewables are costs. For DG, in rural and farfetched areas, to reduce the transmission costs and improving efficiency, renewable sources have an advantage over convention sources.
Hybrid power generator systems 1.5
In general hybrid system mean combination of various power generation with various sources of energy, which complement each other and act as a single unit [4] [5].
Storage is an integral part of this system; it acts as a mean to store energy during excess power generation. This extra power can be retreated when there is peak demand at load side in national grid. Storage systems also play an important role in dealing with fluctuations and intermittencies in the grids and generation section.
5
Reliability of the RE hybrid systems 1.6
The most vital reason of using hybrid systems is to increase and improve power generation reliability. By reliability, we imply that the hybrid system ought to have the capacity to meet e nergy demands, and this raises an ampleness issue. Ampleness implies that the system ought to have the capacity to give enough power. Then again, the likelihood of losing power or absence of energy supply ought to be at the least level. The goal of this project was design of a system which can supply the required power neither with shortage nor interruption or disconnection.
Optimization of hybrid system 1.7
In hybrid system optimization, some factors like reliability, Net Present Cost (NPC) and Levelized Cost of Energy (LCOE) and capital cost should be taken into account. Optimal values for each parameter, such as system performance and capacity and number of the components must be estimated and evaluated as well as the storage system regarding objective functions and constraints.
Thesis definition 1.8
In this research a set of wind generators and PV arrays and a set of storage battery banks, and their related power production models were considered. The goal of this method is providing the hourly required load for a grocery store in Malmö, Sweden.
In order to find optimal values (e.g. number of all included equipment and all probable costs) for designed system working for 20-year lifespan; the minimum LCOE and NPC are the objectives; in condition that no shortage occurs. Hence, by merely harnessing RE resources no carbon footprint and CO2 will be generated. To reach this aim, HOMER optimizer software will be applied.
6
2 Chapter 2: Literature review
7
Introduction:
2.1
There is a considerable agreement among environmental experts and scientists who share this view that insisting on consuming fossil fuel and oils has detrimental effects on our planet in terms of air pollution, soil erosion, and unsustainability. These days, it is an urge to almost all governments to find sustainable replacements for nonrenewable fuels and allocate more budgets for running related researches. Via growth of knowledge about renewable energy, the biggest hurdle in design of RE systems and/or hybrid systems is system optimization.
There are a big number of optimization methods available, and new methods which are presented every day.
The speed of finding optimal values, method precision, number of population needed in search space, type of learning and regenerations, having continued and discrete steps, etc. are some of critical, competitive parameters to prefer a single method rather than the others.
Objective and Multi Objective Optimization (MOP):
2.2
For decision making in problematic cases including multiple criteria, several objective functions should be considered simultaneously; especially, whenever more than two conflicting objectives (e.g. trade-off cases, maximum production or highest peak load with low LCOE) impact the main goals of projects. In intricate cases, classical optimization methods cannot effectively converge to the most efficient results. In 2015, Desale et al. [6]
categorized optimization methods into Exact and Approximate approaches. Approximate approach included Heuristic and Meta-Heuristic methods which are being used mostly whenever classical optimization methods cannot quickly converge to expected results, or in cases they can never find the optimal solutions. It should be borne in mind that none of optimization methods are able to guarantee that the highly- reliable, optimal values can be obtained through iterative generation processes. Hence, that is the main reason why it is always mentioned as an approximate method not accurate one.
Heuristic methods (based on local Search) boost the solving processes with higher precision, fewer needed variables, and simpler approach through complex problems in comparison with classic ones. Divide and Conquer, Branch-and-bound, Cut and Plane, are some of popular Heuristic search techniques [6] [7].
Meta-Heuristics Algorithms (via guided local search, and scatter search) can develop non-deterministic optimal solutions more quickly and more efficiently than Heuristic algorithms. Meta-Heuristics methods are classified into Population-based (e.g. Genetic Algorithm, Tabu Search, Simulated Annealing) and Trajectory-based (e.g.
Hill Climbing) [6]. Inspired by nature behaviors such as ant colony’s network communication, a swarm of bats’
random and dynamic flight patterns, highly elaborated meta- heuristic algorithms have been established. Ant Colony Optimization (ACO) (developed by Marco Dorigo in 1992), Particle Swarm Optimization (PSO) (by James Kennedy & Russell Eberhart, 1995), Harmony Search (HS) (by Zong Woo Geem, 2001), Improved Harmony Search (IHS) (by Mahdavi, Fesanghary & Damangir, 2007), Artificial Bee Colony Algorithm (ABC)
8
(by Karaboga, 2005), Firefly Optimization (FA) (by Xin-She Yang, 2007), Cuckoo Search Algorithm (CSA) (by Xin-She Yang & Deb, 2009), Bat Algorithm (BA) (by Xing-She Yang, 2010) [8] and Migrating Birds Optimization (MBO) (by Duman, Uysal and Alkaya, 2012) are some of renowned Meta-Heuristic algorithms [6].
In 2014, Mirjalili et al. [9] presented a new population based meta-heuristics optimizer algorithm inspired by grey wolves’ leadership and hunting pattern. This method was used in 2018 to optimize an off-grid hybrid system including wind/photovoltaic/fuel cell [10].
PSO 2.3
The PSO method is inspired by general artificial life, for instance by have a glance at flying patterns of flocks of birds, movements of school of fish, and also social human behavior, and group activities without any collision of each group member. In this method, the least effort for food or/and shelter exploration (in search space) is desired. It is a population-based evolutionary technique which works far more effectively than Genetic Algorithm (GA), regarding using CPU and number of required parameters. In comparison with classical approaches, PSO is more stable, requires fewer time steps, and can be parallelized easily. Moreover, rather than GA, PSO is simpler to understand and be applied including fewer adjustable parameters and less computational costs. But still it is a slow method comparing to classical approaches. One of the main cons of PSO method is its unrepeatability in terms of computations. This issue would badly affect decision making process. It is worth to mention that the great advantage of HOMER Pro is its advanced optimization method. In this software optimal values are repeatable and processes are less time-consuming.
HOMER Pro 2.4
HOMER Pro software is an American computer modeling tools which its abbreviation stands for “Hybrid Optimization Model for Electric Renewables”. HOMER Pro is one of the most applicable modeling software which is used in order to simulate and optimize different HRES systems, considering wide variety of options and objectives.
Applied optimization methods on hybrid systems 2.5
In the last decades, a number of hybrid, off-grid or connected to the grid, have been designed and optimized.
Table 2-1 depicts a short list of related literature review.
9
Table 2-1 Literature review on hybrid system optimization methods [11]
Re se arche rs Syste m C ompone nts O bje ctive
Function
O ptimization approach
Model span
W ind
Turbine PV Fuel
C ell Biomass Hydro
Power Ge other. Storage
Diesel
&
othe r
Maleki et al.
[12] ● ● ● PSOAIW Modified
PSO 1 year
Maleki &
Rosen [13] ● ● Natural
gas ● minimize
system total cost
Modified
PSO 1 year
Fadaee &
Radzi [14] ● ● ● Multi-objective
optimization
evolutionary algorithms
20 years
Isa et al. [15] ● ● ●
Lowest total net present cost/
lowest levelized cost of energy/
Low pollutant gas
HOMER 25
years
Diaf et al. [16] ● ● ●
Low loss of power supply
probability/
Lowest levelized cost
of energy
PSO 1 year
T rivedi [17] ● ●
MPO/ Minimizes fuel cost/
Minimizes SO2 and NOx
emission
GA 1 day
Elliston et al.
[18] ● ● ● ● ● Minimize
annualized cos GA 1 year
Ugirimbabazi
[19] ● ● ● ● ● Minimum
LCOE & NPC HOMER 25
years
Eke et al. [20] ● ● Minimizes total
cost LP 1 year
Wang et al.
[21] ● ● ● ● multi-objective
problem NSGA-II 1 year
Garyfallos et
al. ● ● ● ● ● Minimizes total
cost SA 10
years
Akella et al. ● ● ● ● Minimizes total
operation cost LP 1 year
Hanane et al. ● ● ● ●
Minimizes difference between hydrogen demand and
supplied
MINLP 30
days
Kashefi et al.
[22] ● ● ● ● Minimize
annualized cost PSO 20
years
Lagorsea et al. ● ● ● Minimizes total
cost Simulation 1 year
Orhan et al. ● ● ● Minimizes total
cost SA 20
years Raquel and
Daniel ● ● ● ● Minimizes the
LEC LP + heuristic 1 year
Iniyan et al. ● ● ● ● Minimize
cost/efficiency ratio
LP 11
years
Juhari et al. ● ● ● ● Minimizes cost
of energy Simulation 1 year Katsigiannis
& Georgilakis ● ● ● ● Minimizes cost
of energy T abu search 20 years Budischak et
al. ● ● ● ● ● Minimizes cost
of energy
Enumerative method
20 years
Elliston et al. ● ● ● ● ● Minimize
annualized cost GA 1 year
T rivedi ● ●
Minimizes fuel cost, Minimizes SO2 and Nox
GA 1 day
10
Re se arche rs Syste m C ompone nts O bje ctive
Function
O ptimization approach
Model span
W ind
Turbine PV Fuel
C ell Biomass Hydro
Power Ge other. Storage
Diesel
&
othe r
Lorestani &
Ardehali [23] ● ● ●
T o meet thermal and electrical loads with minimum
total cost
PSO 1 year
Abedi et al. ● ● ● ● ●
Minimizes total cost Minimizes unmet load Minimizes fuel
emission
Fuzzy
technique 1 year
Bernal et al. ● ● ● ● ● ● Minimizes total
cost GA 1 year
Ahmarinezhad
et al. [24] ● ● ● ● ● ● Minimizes total
cost PSO 20
years Mohammadi
et al. [25] ● ● ●
Minimum NPC with different
unmet load
HOMER 20
years Yuan et al.
[26] ● ● ● Lowest NPC &
LCOE HOMER 10
years Vatankhah et
al. [10] ● ● ● ● Lowest NPC &
LCOE Grey Wolf
In Table 2-1 “LP”, “ILP” and SA stand for Linear Programming and Interval Linear Programming, and simulated annealing, respectively. “TSP” stands for Two-stage Programming, “GA” for Genetic Algorithm, and
“PSO” for Particle Swarm Optimization method. “NSGA-II” was defined by Wang et al. [21] as the abbreviation of “Non-dominated Sorting Algorithm II.
Conclusion 2.6
Due to claimed statements, in this study, HOMER Pro software and PSO method were selected to be applied for evaluating feasibility of our proposed hybrid system.
11
3 Chapter three: Material and
methods
12
Material and methods:
3.1
After selection of optimization method and applicable optimization software, set of input data and mathematical equation models should be sought and precisely selected. In this chapter hourly required load (in two categories : on weekdays and weekends), PV arrays’ expenses and output power as well as wind generators costs and outputs along with time dependent charging and discharging stored energy in battery bank have been described.
The exchange rate of US dollar to Swedish Krona was taken 8.22.
Load demand calculation:
3.2
The first set of data required for running each optimization and simulation of hybrid system is accessing to load demand profile. According to Noren and Pyrko’s [27] proposed non-linear correlation, Hourly Electricity Consumption Indicator (HECI) can be defined as below mentioned equation.
) (3.1)
Where , and are constant coefficients (for weekdays and weekends separately, given in Appendix A) which were found via regression for each month. T is the average daily temperature (°C) in the selected region (in this study, Malmö air daily temperature). SFA is sales floor area (in m²). They assumed that grocery stores include sales floor area (SFA) and gross floor area (GFA). Through data analyzing, it was obvious that electricity consumption in SFA is highly greater than electricity usage in GFA. The main reason of this difference is via existence of several refrigerators and fridges in SFA. In this project, a 1000 m2 prototype grocery store, located in Malmö, was selected. After coding the abovementioned equation on MATLAB, by application of hourly outdoor temperature in 2010, load profile was collected successfully (Figure 3.1).
Figure 3.1 Electricity consumption profile in 2010 for a 1000 m2 prototype grocery store located in Malmö 0
20 40 60 80 100 120 140
1 1001 2001 3001 4001 5001 6001 7001 8001
Laod Demand in 2010 (kW)
time (hr)
13
Photovoltaic array 3.3
Many researchers have offered various models of PV generated power. Looking for easy-to-converge optimization model, the best practice is choosing a simple and linear model using PV efficiency and solar irradiance [22] [16] (neglecting temperature variations). Others scientist may utilize mo re elaborated models, mingled with temperature impact [28]. For instance, Rui Wang et al. applied latitude and longitude parameters, maximum power tracing efficiency in their calculations [29]. Finally, the equation 3.2 [30] was offered to calculate output power at tth time step:
{ (
)
(3.2)
In the above mentioned equation, is the provided power output by each photovoltaic panel, is the PV rated power, r is the solar radiation factor, is a certain radiation point set usually as 150 ( ), and is the solar radiation in the standard environment set usually as 1000 ( ). In this study, Kyocera KD145SX- UFU PV panels were opted to produce (DC current) electricity. Regarding maximum power point (MPP) voltage output of each Kyocera panel ( ), in order to acquire about 1 kW output power per each PV array, it was assumed that each PV array should include seven PV panels.
In this research, roof- mounted horizontal plates were used although it is obvious that on tilted photovoltaic modules, panels have higher levels of solar ray absorption. Based on Table 3-1 [31] initial investment would be estimated about 11.6 .
Table 3-1 Weighted average turnkey prices of typical PV panels (excluding VAT) Grid-connected, roof mounted,
above 29 kWp (commercial)
Systems installed to produce electricity to grid- connected industrial buildings
11.6
In order to estimate more realistic operation and maintenance (O&M) expenses large numbers of restoration and upkeep activities should be born in mind. To name a few, one can mention annual electrical inspection, panel cleaning (including dusting and washing panels), vegetation control, invertor preservation, etc. Annual O&M of solar PV is about 10.00-45.00
[32]1.
1 Prices in the research are in Swedish krona [SEK] currency: 1 US$ = 8.22 SEK
14
Table 3-2 Specifications of Kyocera KD145SX-UFU PV array
Parameter Value
of each PV array 145 W × 7 PV panels = 1,015 W
150
1,000
Initial investment 10,000 SEK Replacement cost 10,000 SEK
O&M cost 150 SEK/year design life 20 years
3.3.1 Wind turbine
{
(3.3)
(3.4)
Where is swept area of the turbine, is wind velocity at anemometer height (10 m above the ground).
In this study, it was considered that there is a long distance between obstacles. Hence, “d” was assumed as zero.
The terrain class in selected location in Malmö for system installation was near the suburb area therefore, roughness length was considered as (z0)= 1. is power coefficient of the wind turbine which is the maximum value of kinetic energy captured by each wind turbine cannot exceed
(59.3%) according to Betz’s law.
Standard value of for modern wind turbines is about 0.40 which was double checked with NPS 100C-24 power curve Figure 3.2, and was fairly correct.
The hub height of Northern Power NPS 100C-24 wind turbine, based on vendor recommendation is 37 m above the ground ( . By using power law of equation 3.4 ( ) (which was found to give an identical conversion factor from 10 m to 37 m as compared to a logarithmic law with roughness length 1 m), collected wind velocity data were converted for 37 m hub heath. In Table 3-3, capital investment, replacement cost and other specifications of the chosen wind turbine are listed:
15
Table 3-3 Specifications of Northern Power NPS 100C-24
Parameter Value
3 m/s
11 m/s
25 m/s
Rotor Diameter 24.4 m AWG Swept area 467.6
Hub Height 37 m
95 kW
Initial investment 2,337,000 SEK [33]
Replacement cost 2,337,000 SEK O&M cost 10,000 SEK/year design life 20 years
Figure 3.2 Power Curve of NPS 100C-24 Class III wind turbine
3.3.2 Battery bank
Charging and discharging scenarios of proposed model of battery bank (equations 3.5 and 3.6) were derived based on [30] suggested models and were derived successfully. Specifications of Gildemeister 10kW- 40kWh CELLCUBE® is presented in Table 3-4.
3.3.2.1 Charging scenario:
[ ( 3.5)
16 3.3.2.2 Discharging scenario:
[
] (3.6)
where and is charge efficiency and discharge efficiency of battery bank respectively. is hourly self- discharge rate. is the amount of stored electricity at tth time step.
Table 3-4 Specifications of Gildemeister 10kW-40kWh CELLCUBE®
95%
95%
95%
0.0002
Nominal Capacity 40 kWh Initial investment 32880 SEK Replacement cost 32880 SEK O&M cost 1,000 SEK/year design life 10
It should be mentioned that for each battery there is lowest and the highest operational storage limit. Below a certain limit, it is not possible to discharge the battery (in this study the lowest limit was considered as zero and the highest 40 kWh of electricity energy). Obviously, exceeding the rated battery capacity is not possible as well.
Hence, for storing higher amount of electricity, more batteries are needed to be prepared for the system design.
3.3.3 Converter
In this project, based on a suggestion provided by HOMER Pro the System Converter was selected. Regarding PV cost estimation and Figure 3.3, Table 3-5 is presented for bi-directional converter.
17
Figure 3.3 Price list of typical roof mounted PV system including inverter price [31]
Table 3-5 Specification of system converter
95%
Initial investment 1600 SEK Replacement cost 1600 SEK O&M cost 100 SEK/year design life 20
Maleki et al. [30] considered =80%, but, after comparing with HomerPro recommended System Converter”
(Inverter input= 95% and Rectifier input=90%), =95% was selected (converter includes rectifier and inverter system).
System architecture 3.4
In the presented design, wind generator is connected to AC bus bar. PV arrays and a battery bank are electrically connected to DC bus bar where converters are the connectors among AC and DC bus bar in two directions.
18
Figure 3.4 Hybrid system arrangement
All possible scenarios 3.5
Three different scenarios can be assumed for our suggested hybrid system, defined as blow:
No charging & discharging
Charging Scenario
Discharging Scenario
Objective function 3.6
The hybrid system should operate by profiting from renewable resources only (100% renewable fraction). Hence, CO2 emission would be zero, as the main goal of present research. The lowest amount of LCOE and Net Present Cost (NPC) of the hybrid system in SEK are two main objectives in our optimization design which would be met. In order to avoid any food corruption in the grocery store, all in-use refrigerators and freezers should always be switched on. In this case, no power outages and unmet load are allowed. It is assumed that all component availabilities must be 100% and loss of power-supply probability must be zero.
Constraints 3.7
In Table 3-6 all constraints of PSO algorithm and the hybrid system applied on HOMER are defined.
19
Table 3-6 constraints descriptions
No. of constraints Description
#1
#2
#3
#4
#5 Energy stored in the operational battery bank is limited to:
The state of charge (SOC):
If → If →
#6 Minimum Net Present Cost (NPC)
#7 Minimum LCOE
#8 To supply all load needed (Max. 0.1 % shortage)
Hybrid system architecture 3.8
PV arrays and battery bank are connected to DC busbar, wind generators and electric load are connected to AC busbar as well. As it is shown in Figure 3.5, two busbars are connected with a set of bi-directional converters.
Figure 3.5 Proposed hybrid system on HOMER
Levelized Cost of Energy (LCOE):
3.9
Annuity factor is:
(3.7)
20
(3.8)
Where q = (1+ p) is equal to” interest rate +1”, is initial cost, operation and maintenance cost and is replacement cost in SEK, n is number of system design life.
(3.9)
In above mentioned equation, and are total annualized cost and annual energy yield [kWh].
Net Present Cost (NPC):
3.10
∑
(3.10)
In NPC, is the amount of annual net cost for the designed project (in SEK).
Particle swarm optimization (PSO)
3.11
Starting off by generating a random population, PSO locates each swarm. Then it continues to converge to the best solutions regarding velocity of each and every individual. Best location will be stored in PSO memory.
Important steps in PSO are initialization, encoding, parent selection, and update of the positions.
Initialization step includes considering “population size” (PS), “constriction factor” in particle swarm in order to limit velocity (CF), upper bound for “number of iterations” (MI), and acceleration coefficients, and .
In Encoding step evaluation of fitness function and initial population is performing.
In the step of parent selection, PSO sustain and save the best position of each found swarm through iterations and inscribe them as the best records as global positions. In this algorithm, after each steep of iteration, the list of global positions will be checked and velocity parameter will be updated.
In the step of update of the positions, shown in equation 3.11 and 3.12, velocity of swarm ) and its related position ( ) will be updated considering coefficient of control ( ) acceleration coefficients ( and ).
( ) (3.11)
(3.12)
Where is the best position at time step and is the best recorded global position as well. In the equation 3.11, and are uniformly generated random numbers (in an interval from 0 to 1), and in the equation 3.12, χ is the coefficient to limit the velocity.
As it is mentioned, in each step, comparing to the best global position, this parameter will be replaced with new position or maintained as the best one so far.
21
PSO algorithm will cease to estimate; once it accomplishes all iterations (based on defined maximum number of iterations in PSO) [10]. As Vatankhah Barenji et al. described, PSO method is executed as it presented in
Table 3-7.
Table 3-7 PSO algorithm [10]
Step Description
0 Initialization of PSO parameters 1 Generating of population
2 Fitness function calculation for initial individuals 3 Initial individuals recorded in
4 The best individual registered in 5 No. of Iteration as 0
6 While (Iteration < MI) do
7 By using eq. 3.11 & 3.12 update recorded positions 8 Calculate fitness value for new members
9 if value in step #8 is less than at
10 Position of that individual must be replaced by 11 if value in step #8 is less than at
12 Position of that individual must be replaced by 13 End
14 End
15 No. of Iteration + 1 restore in No. of Iteration 16 end while
17 Return
Conclusion 3.12
Now, all the parameters, constraints, and costs, as well as PSO algorithm are meticulously defined. The next chapter reports final results and optimal values for components included in RE hybrid system.
22
4 Chapter four: Results and
Discussions
23
Results and discussion:
4.1
In this chapter, actual solar irradiance, wind velocity and ambient temperature data were collected from data bank, reported by a Swedish statistical organization [34]. The hybrid system was run for 20 years of project lifetime, 4% discount and no annual shortage, with in different operating reserve (i.e. various percentages for
“load in current time step” and “annual peak load”).
System input data:
4.2
As it is shown in Figure 4.1 Monthly average solar global horizontal irradiance in 2010 Figure 4.1 in Malmö in July maximum daily irradiance occurs and in December minimum radiation. Annual average global irradiance in this city is 2.73
.
Figure 4.1 Monthly average solar global horizontal irradiance in 2010 [34]
Figure 4.2 depicts average annual wind velocity in Malmö. The annual average is 4.88 (m/s)
Figure 4.2 Monthly average wind speed data at 37 meter height in 2010 [35]
Flowingly, Figure 4.3 presents 7.37 (˚C) as the annual average temperature.
24
Figure 4.3 Monthly average temperature data in 2010 [36]
Load profile of the grocery store via embedding hourly ambient temperature data on Noren and Pyrko’s formula [27], meticulously calculated and are shown in Figure 4.4 as well as monthly alternative current load profile in Figure 4.5.
Figure 4.4 Monthly average load demand in 2010 for a typical 1000 m2 grocery store in Malmö
Figure 4.5 AC load profile
