- Email: [email protected]
The performance of a grid-tied microgrid with hydrogen storage and a hydrogen fuel cell stack
Energy Conversion and Management 87 (2014) 421–427
Contents lists available at ScienceDirect
Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman
The performance of a grid-tied microgrid with hydrogen storage and a hydrogen fuel cell stack Linfeng Zhang a,⇑, Jing Xiang b,1 a b
Department of Electrical Engineering, University of Bridgeport, CT 06604, USA Anhui Polytechnic University, Anhui, PR China
a r t i c l e
i n f o
Article history: Received 5 February 2014 Accepted 14 July 2014
Keywords: Hydrogen fuel cell Hydrogen storage PV system Energy management
a b s t r a c t In a heat-power system, the use of distributed energy generation and storage not only improves system’s efficiency and reliability but also reduce the emission. This paper is focused on the comprehensive performance evaluation of a grid-tied microgrid, which consists of a PV system, a hydrogen fuel cell stack, a PEM electrolyzer, and a hydrogen tank. Electricity and heat are generated in this system, to meet the local electric and heat demands. The surplus electricity can be stored as hydrogen, which is supplied to the fuel cell stack to generate heat and power as needed. The performance of the microgrid is comprehensively evaluated and is compared with another microgrid without a fuel cell stack. As a result, the emission and the service quality in the first system are higher than those in the second one. But they both have the same overall performance. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction In order to improve the efficiency, reliability, economics, and sustainability of power generation and distribution, the traditional power grid is being upgraded to the smart grid (SG). A SG includes new technologies in information and communications, distributed energy generation (DEG) and energy storage (DES), advanced measurement and sensing, controls, cyber security, and consumer-side energy management [1–4]. DEG is small-scale power generation with power less than 50 kW. It includes micro turbines (lturbine), micro combined heat and power (lCHP) systems, photovoltaic systems (PV), wind turbines, and solar thermal systems [5,6]. For the local thermal and electrical loads, lCHP systems, such as internal combustion engines (ICEs), stirling engines, and fuel cells (FCs), are attractive because electricity and heat can be produced near the point of demand with high efficiency and avoidance of the transmission and distribution losses in the conventional centralized generation model [7–10]. The lturbine, without waste-heat recovery, can be regarded as a special case of lCHPs. For a lCHP, heat and electricity generation are coupled and they have three operating strategies: heat-led, electricity-led and cost-led. In the first strategy, the lCHP should supply heat to meet the onsite demand as much as possible with electricity as a byproduct. The second one is to operate the lCHP ⇑ Corresponding author. Tel.: +1 203 475 4249. 1
E-mail addresses: [email protected] (L. Zhang), [email protected] (J. Xiang). Tel.: +86 13855321034.
http://dx.doi.org/10.1016/j.enconman.2014.07.045 0196-8904/Ó 2014 Elsevier Ltd. All rights reserved.
to meet the local electricity demand and the objective of the third strategy is to minimize the cost. If electricity and heat generation can be decoupled in the presence of heat storage, the control strategy will be different. For DEG, scheduling freedom is important in its optimal operation. lturbine and lCHP are controllable but with some limits. For instance, different fuel technologies have start-up times ranging from minutes to hours. Therefore, the operation scheduling for these generators must be made with the consideration of their dynamic operating constraints. Different to the lCHP, the generation from renewables strongly depends on the weather and the operation of wind turbine, PV system, and solar thermal system has no scheduling freedom. Thus, the widespread emergence of such DEG on the consumer side significantly increases the variability of generation and the intermittent nature DEG from renewables has significant negative impacts on grid voltage and frequency stability. To balance supply and demand and to minimize the DEGinduced power fluctuations in the grid, compensating changes are required in the DES, loads, and output from flexible generation sources [11]. DES includes rechargeable battery, super capacitor, and hot water tank technologies. They have different ratings for power and discharge time. Other energy storage technologies, like compressed air energy storage, hydrogen, and pumped-hydro storage are normally deployed in large scale with long discharge time up to days [10,12]. The control strategies for DES will be based on the DES behavior, efficiency, and the energy conversion efficiency. For electrical and thermal loads, they have distinct characteristics, and they can be categorized into different types based on whether
422
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
they can be trimmed in the peak period or shifted to the off-peak period. Thus, they can adjust their consumption levels to correct voltage sags and flickers or to help stabilize the system frequency. Local DEG, DES, and loads can be aggregated as a microgrid (MG) with a grid connection for import and/or export of electricity. With two-way communication and advanced metering infrastructure, customer service is improved especially in power outage detection, grid operation, disaster recovery, and accuracy of load estimation [13,14]. Basically, one MG can run in two modes, grid-tied or islanded [15]. In the first mode, power flows in two ways between a MG and the grid, a MG may consume the power from the other MGs or it may supply power to them for credits according to the agreement between this MG and a local utility company. In an islanded mode, there is no power flow between the MG and the grid. This intentional ‘‘islanding’’ under certain circumstances, provides local reliability, stability, and security. However, this operation does not change or disrupt the integrity of the transmission grid as a whole [15,16]. In a SG, most energy management for a MG as a consumer is to minimize the electricity cost. Current electricity price on the grid is dynamic and it is established by the market to balance sellers’ supply and buyers’ demand. The balancing process is continuous and instantaneous since electricity must be produced at nearly the same time it is consumed. The market trades in electricity can be bilateral transactions in the form of a day-ahead market and spot market trades in real-time. The bilateral transaction is short-term forward market trading and is conducted between wholesale consumers and power plants and it provides price certainty. But MGs with DEG, DES, and loads must pursue a combination of short-term trading and spot trading. Power flows in two ways between the grid and the MGs. Thus, in order to minimize their cost from the generation and consumption, MG portfolios can be balanced and the generation, storage, or demand would be managed or scheduled under some short-term predictable and unpredictable circumstances [11,17–19]. The effective energy management can be market-based controls or agent-based controls. The first one emphasizes the creation of a competitive market environment for MGs with DEG, DES, and demand and the regulation of the production and consumption with direct market signals in order to meet the perspectives on cost and effectiveness. Depending on the control structure, there are three different operations: direct control, price signal control, and internal exchange [10]. The second one, agent-based control, emphasizes the importance of MG coordination and reliable communications with its neighbors. This improves the power system scalability and the optimization objectives include minimum cost, maximum profit, or highest efficiency [11]. In this paper, the performance of a MG with a hydrogen FC stack, as a CHP system, is comprehensively evaluated with the consideration of cost, emission, and service quality. The mathematical model of a MG is presented in Section 2. Besides a fuel cell stack, this MG includes a PV system, a PEM electrolyzer, and a hydrogen tank. Moreover, this MG is connected to a grid with a dynamic price and limited power availability in peak period. To operate the MG, an energy management is developed with the FC stack under a total-power-led scenario. In Section 3, the performance of the MG is analyzed and it is compared with the performance of a similar MG but without a CHP system.
2. Model 2.1. General structure Fig. 1 shows a basic structure of a microgrid. One agent in this sub-power system collaborates with the other agents on the grid to manage local energy generation, consumption, and flow. To
minimize the communications overhead, a regional manager, an agent selected from multiple microgrids, will exchange information with the managers in a wide area. Therefore, the energy system is a distributed system while the communication is a hierarchy system. The structure of agent-based energy management is shown in Fig. 2. The agent can be the weather forecast or the estimated energy generation/consumption pattern for a local DEG and loads for the upcoming time period. This agent communicates with its regional manager which is also an agent but voted by its neighbors to be the manager. The regional manager then aggregates the energy profiles and trade with the other managers through bids and auctions in order to balance the supply and demand and determine a price. A two-layer hierarchical structure will ensure scalability and reduce communication overhead. Thus, with the information from the regional manager, an agent can make a decision to optimize the operation of the system through real-time control with AMI. In this work, the structure of the grid-tied MG is shown in Fig. 3. The electricity generated from the PV system is first supplied to the local load and then to the electrolyzer. The spare electricity, if there is, goes to the grid. In this system, an electrolyzer is used to covert electricity to hydrogen, which is stored in the tank. The stored H2 can be supplied to the hydrogen FC stack as needed. The running of the FC stack is based on the total-power-led scenario and the outputs, electricity and heat, go to the loads and part of heat can be stored in the thermal tank. Since there is no CO2 or NOx as byproduct, this microgrid is emission free. For comparison, a similar MG without hydrogen tank and a FC stack is also simulated. In the system running, an agent module collects all information from the MG components and predicts PV system output, electricity price on the grid, heat load, and electric load. Then, it makes a plan for the power flow and schedule for the electrolyzer and the FC stack in the next hour. 2.2. PV system With the consideration the daily electricity and heat load, the area of the solar array is set as 200 m2. According to the location, Bridgeport, Connecticut, USA, the daily ambient temperature profile is shown in Fig. 4 and the temperature effect on the PV system open circuit voltage can be calculated as [20,21]:
V oc;ambient ¼ aðT STC T ambient Þ þ V oc;rated
ð1Þ
Here, a is a temperature coefficient as 0.12, Voc,ambient the open circuit voltage at ambient temperature, TSTC temperature at standard test conditions, 25 °C, 1000 W/m2 solar irradiance, Tambient the module temperature, and Voc,rated open circuit voltage at STC. Thus, there is a profile of the PV system output in Fig. 4 and the maximum power output is 85 kW at 12 PM. 2.3. Hydrogen FC stack The hydrogen proton-exchange-membrane fuel cell stack consists of 300 fuel cells in series with the parameters in Table 1. The main material is stainless steel and heat insulator. The operating temperature is 343 K, the optimum temperature this kind of fuel cells. For the cold start in a winter, the heat generated at the beginning is used to increase its temperature. At 373 K, part of the heat is transferred to the heat load or heat tank. In winter, the stack is insulated to minimize heat transfer and the heat conductivity used in this work is one hundreds of that in stainless steel. Fig. 5 shows the outputs from the fuel cell stack at three different temperatures, 293, 313, and 343 K. For the VI curve, it shifts lower as temperature increases. So does the power–current curve.
423
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
Fig. 1. The schematics of a microgrid.
Table 1 Parameters of a FC stack.
Agent
Forecasting and estimation
Market interface
Management interface
Market
Control and measurement
Parameter
Value
Number of cells Active area per cell Operating temperature Cathode pressure Anode pressure
300 (in series) 200 cm2 70 °C 2 atm 2 atm
70 293K 313K 343K
350
DEG
Load
60 heat
300
Voltage (V)
Electricity Heat H2
grid Electrolyzer
PV system
50
voltage
Fig. 2. The structure of an agent-based energy management system.
H2 tank
40 250
electricity
200
30 20
Power (kW)
DES
10
150
0 100
Heat load
temperature power
Temperature (K)
60
280
40 270
Power (kW)
80
290
20 260 0 10
150
200
On a power–current curve, there is a maximum power point, 30 kW, corresponding to the current density as 850 mA/cm2. The heat-current curve shows an opposite trend and it shifts higher as temperature increases. Compared with the electric power– current curve, heat generated is high than the electricity if the current is higher than 110 A. To run the stack, the power demand ranges from 5 kW to 22 kW, corresponding to 75% of the maximum power output.
100
300
5
100
Fig. 5. The voltage, electric power, and heat from the hydrogen FC stack at 293, 313, and 343 K.
Fuel cell
Fig. 3. The structure of the microgrid.
0
50
Current (A)
Electric load
250
0
Heat tank
15
20
Time (hour) Fig. 4. The ambient temperature and the power output from the PV system in one day.
2.4. Electricity from the grid Residential electricity rate from the local utility company is break down as electricity 7.7 ¢/kW h, transmission, distribution, and delivery 11.2 ¢/kW h, so total is 18.9 ¢/kW h with the basic service charge 16.5 $/month [22]. With the consideration of the hourly price pattern [23], Fig. 6 shows the tariff used in this work. For solar electricity, under a range of financing assumptions and locations, the levelized cost is estimated by the U.S. DOE ranges from 10 ¢/kW h to 18 ¢/kW h 11 for utility-scale, 16 ¢/kW h to 31 ¢/kW h for commercial systems and 16–25 ¢/kW h for residential PV systems [24]. Thus, the price of electricity from PV system is set as 20 ¢/kW h in this calculation. About hydrogen FC stack,
424
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
its average installed cost is 5600 $/kW, the electricity cost is between 45 and 55 ¢/kW h but 14 ¢/kW h for a CHP system [20]. With the consideration of the cost on electrolyzer, the price is assumed to be 30 ¢/kW h. In addition to the dynamic price in this simulation, there are upper limits, 30 kW and 20 kW, set for the power from the grid in the day and night time, respectively. 2.5. The operation of the hydrogen FC stack For a FC stack as a power generator, there are two operating points if the electricity load is less than the maximum power. The selection of the operating point is based on the system requirements, such as lowest cost, lightest unit, and highest power density. For example, a point at high current density make the cell small to lower down the capital cost, but its efficiency will be low with more heat generated. For a FC stack as a lCHP system, both the desired heat and power outputs are given, the operating point can be determined by the power/heat ratio if both points are allowed by the system. With enough hydrogen, a FC stack as a lCHP supply supplies electricity and heat as:
hchp þ hother ¼ hd
ð2Þ
echp þ eother ¼ ed
ð3Þ
where hchp is the heat from the lCHP, hother can be the heat from boiler, heat storage, and solar thermal system, echp is the electricity from the lCHP, eother can be from a wind turbine, PV system, grid, and electricity storage, and ed is the demand. Electricity from the grid is positive when power flows from the grid and negative when power flow into the grid. The constraint on the lCHP’s generation is
PHcap1 < PH < PHcaph
ð4Þ
where PH is power or heat and PHcapl and PHcaph are the lower and upper limits. Moreover, a stack should be planned to run at least certain time because it has to raise its temperature first from a cold start, especially in winter. In this paper, this time interval is at least 1 h and the stack runs under a total-power-led scenario. 2.6. Performance evaluation For the economic return on the investment of an individual MG, a performance metric Q for each MG is [25]:
Q ¼ w1 F þ w2 E þ w3 S1 þ w4 S2
ð5Þ
F¼
cGridav g c av g
15 Electricity Electricity+T&D
10
5
Here, cGrid-avg is the daily average electricity price from grid. cavg is the average electricity cost for a MG and it does not include the heat cost from the FC stack. Moreover, loads can be categorized into different types based on whether they can be shifted, interrupted, decreased, or cancelled. Loads also have different priorities in the presence of limited supply. Generally, demand is a random variable with a probability distribution in an operation time window, and it can be regarded as a series of separated and fine-grained tasks, which means each task can be completed in a sequence but not in continuous time slots. The electric service quality can be calculated as:
S1 ¼
S2 ¼ 0
0
5
10
ð6Þ
T X 3 Psupply;i 1X wi T t¼0 i¼1 P demand;i
ð7Þ
Here, T is a period of time, and i is the priority of the load (high, normal, or low). Psupply,i is the energy supplied to the load demand with a priority i, and Pdemand,i is the load demand with a priority i. wi (i = 1, 2, 3) is a weighting factor for the load with priority i, and P wi = 1. In this simulation, w1, w2, w3 are set as 0.5, 0.3, and 0.2 for the electrical loads with high, normal, and low priorities, respectively. The heat load is mainly for space heating and the priority of such load is fixed. Heat supply can be from the heat tank and electricity, whose priority is set only lower than that of the critical load. No matter where is the heat supply from, heat service quality is:
20
Price (cents/kWh)
F is a price index of electricity, E is an environmental effect or emission index due to atmospheric emissions, S1 is the service quality of electricity, and S2 is the service quality of heat. Q, F, E, and S are normally but not limited between 0 and 1. wi (i = 1, 2, 3, 4) are weightP ing factors and wi = 1. The ultimate objective for each MG is to maximize its overall performance index and the values of wi can be set according to the requirement of the consumer. In this simulation, the electricity price and emission are a little bit more important than the service quality of electricity and heat and w1, w2, w3 and w4 are 0.3, 0.3, 0.2, and 0.2, respectively. On the grid, the price of electricity is dynamic, based on the supply and demand. The day-ahead price can be forecast using wavelet transforms and related simulations [12,26]. Although the price is mainly determined by the supply from the power plants and the major demand, it is can be affected by DER and AD. The agent only participates in day-ahead market or spot market. For each device, based on the historical energy pattern, a predicted energy profile is generated for the day-ahead trading using neural network techniques. The Mont Carlo simulation is utilized to generate scenarios for real-time uncertainties. In the management process, optimization objectives are different for the shareholders of the power system versus the consumers. For residents or microgrids, there is a tradeoff between energy cost, energy security, and environmental protection. For utilities, the objective is the maximization of profit. Utilities are more likely to generate electricity when the price is high. In each case, the performance index, Eq. (5), is applicable but with different weighting factors. The combination of marketing, prediction, and real-time control in one subsystem exploits all potential benefits derived from each period of time. The price index of electricity, F, can be calculated as:
15
Time (hour) Fig. 6. Residential electricity rate.
20
Hsupply Hdemand
ð8Þ
Finally, the environmental effect index is:
E¼1
e av g eplant
ð9Þ
425
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
(b) 40 Electricity (kW)
Electricity (kW)
40 30 20 Electricity Load Supply
10 0 40
0
10
20
30
40
50
60
Heat (kW)
30 20 Heat
10
Load Supply
30 20 Electricity Load Supply
10 0 0 40
70
Heat (kW)
(a)
10
20
30
40
50
60
30 20 Heat Load Supply
10 0
0 0
10
20
30
40
50
60
70
70
0
10
20
30
Time (hour)
40
50
60
70
Time (hour)
340
FC Ambient
320 300 280 260 0 20
Power (kW)
1500
Hydrogen (moles)
Temperature (K)
Fig. 7. The electricity and heat power for the system with a FC stack (a) and without a FC stack (b).
10
15
20
30
40
50
60
70
Electricity Heat
10
1000
500
0
5 0
0
10
20
-5 0
10
20
30
40
50
60
30
40
50
60
70
Time (hour)
70
Time (hour)
Fig. 10. The hydrogen in the tank.
Fig. 8. the temperature (upper) and the supply (lower) from the fuel cell stack.
3. Results and discussion where eavg is the average atmospheric emission for electricity consumed in a MG, and eplant is the emission from a thermal power plant. In this paper, the emission is set as 3 units for electricity from the fire power plant and 0 unit for electricity generated from a solar panel or a fuel cell stack in a MG.
(a)
100
100 PV Grid
80 60
60
Electricity (kW)
Electricity (kW)
(b)
Solar Panel FC ENG from Grid Electrolyzer
80
Fig. 7(a) and (b) shows the electricity and heat profiles for the system with a FC stack and without one, respectively. In the day time, most electricity load is met in both systems. After sunset, these two systems obtain limited electricity from the grid.
40 20
40 20 0
0 -20
-20
0
5
10
Time (hour)
15
20
-40
0
5
10
Time (hour)
Fig. 9. The profile of the MGs with a FC stack (a) and without a FC stack in the first day.
15
20
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
(a)
Price Envir. Elec. Qual. Heat Qual. Overall
1.4 1.27
Index
1.2
(b) 1.29
1.27
1.2
1.0 0.90
0.90
0.8
0.78
0.74 0.72 0.78
0.740.72
1.29
1.0 0.86
0.86
0.8
0.72 0.67
0.68
0.71
0.64
0.64
0.6
0.6
0.4
Price Envir. Elec. Qual. Heat Qual. Overal
1.4
Index
426
1
2
0.4
1
2
Day
Day
Fig. 11. The performance indices of the systems with a FC stack (a) and without a FC stack (b).
In addition to this, the FC stack in the first case can generate electricity and heat for around 5 h till its hydrogen tank becomes empty. Thus, more electricity is supplied to the load. For the heat load and supply, they show the similar result as that in the electricity. Fig. 8 shows the FC stack temperature increase from the cold start. The output from the FC stack is electricity only at the beginning. After 40 min, the temperature rises to the optimum value, 343 K, it maintains its temperature and starts to supply heat to the heat load or the heat tank. When the FC stack stops running, its temperature drops and eventually equals to the surrounding temperature. Fig. 9(a) shows the electrical power profiles of the MG from the PV system, the FC stack, grid, and electrolyzer, respectively. The power from the PV system starts to increase at 7:30 AM and its maximum output is 85 kW at 12 PM. The output from the PV is not only supplied to the load, but also to the electrolyzer for energy storage. The rest goes to the grid for credit from the local utility company. When the output from the PV system drops at 5 PM, additional electricity comes from the grid and the FC stack to meet the local demand. In Fig. 9(b), the electric power from the PV system goes to the local loads and the grid. If it is insufficient to meet the demand, electricity from the grid will be transferred to the MG. The amount of H2 fluctuates between 0 and 1600 mol in the tank as shown in Fig. 10. All hydrogen is generated by the electrolyzer. The hydrogen from the tank can be supplied for 5 h after the sunset. The peak of the amount appears at around 4 PM and it is used up at the beginning of the next day. Here, hydrogen is a kind of storage method of solar energy. No electricity from the grid contributes to the hydrogen generation due to the upper limits of electrolyzer input power, tank capacity, and availability of grid power input. Fig. 11 shows different performance indexes for the two power systems. In Fig. 11(a), although the prices for the electricity from the PV system and FC stack are higher than that from the grid, the price index is 1.27 due to the heat supply counted in the index calculation. In this work, the cost for the heat supply from the FC stack is negligible. Moreover, some credits are brought by the small amount of electricity transferred from the system to the grid in the day time. The environmental effect index is less than 1 and this is due to the use of electricity from the grid. The service qualities of electricity and heat are both between 0.72 and 0.8. Most electricity and heat demand is met. In Fig. 11(b), the price index of the system without a FC stack is around 1.29 because of the significant credits for electricity transfer from the system to the grid, shown in Fig. 9(b). The environmental effect index is lower than that in the first case even they have almost the same inputs from the grid,
as shown in Fig. 9. But, less usage of power from renewable energy resource will decrease this index according to Eq. (9). The electric quality index is close to 0.7 while the heat quality index is low as 0.64. Finally, the overall indexes for both systems are close to 0.9 with the weighting factors set for Eq. (5). 4. Conclusions In this work, with the dynamic electricity price and the constraint on electricity from the grid, the performance is evaluated on a grid-tied MG with DES and a hydrogen FC stack and it is also compared with that of a similar MG but without DES or a hydrogen FC stack. With the assumed weighting factors on the consideration of price, environment, and service quality, both MGs show almost the same overall performance index. Except the price index, the environmental and service quality indexes are higher in the MG with DES and a FC stack. Current grid electricity price fluctuates in a narrow range, less than 10 ¢/kW h and the price from the PV system and the FC stack is much high. Further work will be focused on the effect of the electricity price change on the optimal MG structure. Moreover, with more and more restrict legislation on the environment protection, the method for the performance evaluation, such as the weighting factors in the calculation, will be updated. References [1] Recovery act-smart grid demonstration U.S.Department of Energy, 2009. [2] Grose TK. The cyber grid. ASEE Prism; 2009. p. 26–31. [3] Hawkes AD, Leach MA. Modelling high level system design and unit commitment for a microgrid. Appl Energy 2009;86:1253–65. [4] Hledik R. How green is the smart grid? Electr J 2009;22(3):29–41. [5] Wu Z, Gu W, Wang R, Yuan X, Liu W. Economic optimal schedule of CHP microgrid system using chance constrained programming and particle swarm optimization. In: Power and energy society general meeting, 2011 IEEE; 2011. [6] Pedrasa MAA, Spooner TD, MacGill IF. Coordinated Scheduling of Residential Distributed Energy Resources to Optimize Smart Home Energy Services. Smart Grid, IEEE Transactions on, vol. 1(2); 2012. p. 134–43. [7] Alsayegh O, Alhajraf S, Albusairi H. Grid-connected renewable energy source systems: challenges and proposed management schemes. Energy Convers Manage 2010;51:1690–3. [8] Lasseter RH. MicroGrids. In: IEEE Power engineering society winter meeting; 2002. [9] Lopes JAP, Moreira CL, Madureira AG. Defining control strategies for MicroGrids islanded operation. IEEE Trans Power Syst 2006;21(2):916–24. [10] You S. Developing virtual power plant for optimized distributed energy resources operation and integration. In: Department of Electrical Engineering, Technical University of Denmark: Lyngby; 2010. [11] Molderink A, Bakker V, Bosman MGC, Hurink JL, Smit GJM. Management and control of domestic smart grid technology. In: Smart grid, IEEE transactions on, vol. 1(2); 2010. p. 109–19.
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427 [12] Conejo AJ, Plazas MA, Espinola R, Molina AB. Day-ahead electricity price forecasting using the wavelet transform and ARIMA models power systems. In: IEEE Transactions on, vol. 20(2); 2005. p. 1035–42. [13] Bennett C, Highfill D. Networking AMI smart meters. In: Energy 2030 conference. ENERGY 2008. IEEE2008; 2008. p. 1–8. [14] Hart DG. Using AMI to realize the smart grid. In: Power and energy society general meeting – conversion and delivery of electrical energy in the 21st century, IEEE; 2008. p. 1–2. [15] Lasseter RH, Piagi P. Microgrid: a conceptual solution. In: PESC2005: Aachen, Germany. p. 20–5. [16] Zhong Z, Huo H, Zhu X, Cao G, Ren Y. Adaptive maximum power point tracking control of fuel cell power plants. J Power Sources 2008;176:259–69. [17] http://www.iso-ne.com/nwsiss/grid_mkts/how_mkts_wrk/whlsle_elec_trad/ index-p2.html. [18] Rahman MM, Mto A. Technologies required for efficient operation of a smart meter network. In: Industrial electronics and applications (ICIEA), 2011 6th IEEE conference on; 2011. p. 809–14. [19] Mohsenian-Rad A-H, Wong VWSJ, Jackson J, Schober R, Leon-Garcia A. Autonomous demand-side management based on game-theoretic energy
[20] [21] [22]
[23]
[24]
[25]
[26]
427
consumption scheduling for the future smart grid. In: Smart grid, IEEE transactions on, vol. 1(3); 2010. p. 320–31.. http://www.fao.org/docrep/x0490e/x0490e07.htm. http://www.wcc.nrcs.usda.gov. Sechilariu M, Wang B, Locment F. Building-integrated microgrid: advanced local energy management for forthcoming smart power grid communication. Energy Build 2013;59:236–43. Chen Y, Wu Y, Song C, Chen Y. Design and implementation of energy management system with fuzzy control for DC microgrid systems. IEEE Trans Power Electron 2013;28(4):1563–70. Nunna K, Doolla S. Multiagent-based distributed-energy-resource management for intelligent microgrids. IEEE Trans Ind Electron 2013;60(4):1678–87. Gu W, Wu Z, Bo R, Liu W, Zhou G, Chen W, Wu Z. Modeling, planning and optimal energy management of combined cooling, heating and power microgrid: a review. Int J Electr Power Energy Syst 2014;54:26–37. Chen Z, Wu L, Fu Y. Real-time price-based demand response management for residential appliances via stochastic optimization and robust optimization. In: Smart grid, IEEE transactions on. 99. p. 1–9.
Contents lists available at ScienceDirect
Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman
The performance of a grid-tied microgrid with hydrogen storage and a hydrogen fuel cell stack Linfeng Zhang a,⇑, Jing Xiang b,1 a b
Department of Electrical Engineering, University of Bridgeport, CT 06604, USA Anhui Polytechnic University, Anhui, PR China
a r t i c l e
i n f o
Article history: Received 5 February 2014 Accepted 14 July 2014
Keywords: Hydrogen fuel cell Hydrogen storage PV system Energy management
a b s t r a c t In a heat-power system, the use of distributed energy generation and storage not only improves system’s efficiency and reliability but also reduce the emission. This paper is focused on the comprehensive performance evaluation of a grid-tied microgrid, which consists of a PV system, a hydrogen fuel cell stack, a PEM electrolyzer, and a hydrogen tank. Electricity and heat are generated in this system, to meet the local electric and heat demands. The surplus electricity can be stored as hydrogen, which is supplied to the fuel cell stack to generate heat and power as needed. The performance of the microgrid is comprehensively evaluated and is compared with another microgrid without a fuel cell stack. As a result, the emission and the service quality in the first system are higher than those in the second one. But they both have the same overall performance. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction In order to improve the efficiency, reliability, economics, and sustainability of power generation and distribution, the traditional power grid is being upgraded to the smart grid (SG). A SG includes new technologies in information and communications, distributed energy generation (DEG) and energy storage (DES), advanced measurement and sensing, controls, cyber security, and consumer-side energy management [1–4]. DEG is small-scale power generation with power less than 50 kW. It includes micro turbines (lturbine), micro combined heat and power (lCHP) systems, photovoltaic systems (PV), wind turbines, and solar thermal systems [5,6]. For the local thermal and electrical loads, lCHP systems, such as internal combustion engines (ICEs), stirling engines, and fuel cells (FCs), are attractive because electricity and heat can be produced near the point of demand with high efficiency and avoidance of the transmission and distribution losses in the conventional centralized generation model [7–10]. The lturbine, without waste-heat recovery, can be regarded as a special case of lCHPs. For a lCHP, heat and electricity generation are coupled and they have three operating strategies: heat-led, electricity-led and cost-led. In the first strategy, the lCHP should supply heat to meet the onsite demand as much as possible with electricity as a byproduct. The second one is to operate the lCHP ⇑ Corresponding author. Tel.: +1 203 475 4249. 1
E-mail addresses: [email protected] (L. Zhang), [email protected] (J. Xiang). Tel.: +86 13855321034.
http://dx.doi.org/10.1016/j.enconman.2014.07.045 0196-8904/Ó 2014 Elsevier Ltd. All rights reserved.
to meet the local electricity demand and the objective of the third strategy is to minimize the cost. If electricity and heat generation can be decoupled in the presence of heat storage, the control strategy will be different. For DEG, scheduling freedom is important in its optimal operation. lturbine and lCHP are controllable but with some limits. For instance, different fuel technologies have start-up times ranging from minutes to hours. Therefore, the operation scheduling for these generators must be made with the consideration of their dynamic operating constraints. Different to the lCHP, the generation from renewables strongly depends on the weather and the operation of wind turbine, PV system, and solar thermal system has no scheduling freedom. Thus, the widespread emergence of such DEG on the consumer side significantly increases the variability of generation and the intermittent nature DEG from renewables has significant negative impacts on grid voltage and frequency stability. To balance supply and demand and to minimize the DEGinduced power fluctuations in the grid, compensating changes are required in the DES, loads, and output from flexible generation sources [11]. DES includes rechargeable battery, super capacitor, and hot water tank technologies. They have different ratings for power and discharge time. Other energy storage technologies, like compressed air energy storage, hydrogen, and pumped-hydro storage are normally deployed in large scale with long discharge time up to days [10,12]. The control strategies for DES will be based on the DES behavior, efficiency, and the energy conversion efficiency. For electrical and thermal loads, they have distinct characteristics, and they can be categorized into different types based on whether
422
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
they can be trimmed in the peak period or shifted to the off-peak period. Thus, they can adjust their consumption levels to correct voltage sags and flickers or to help stabilize the system frequency. Local DEG, DES, and loads can be aggregated as a microgrid (MG) with a grid connection for import and/or export of electricity. With two-way communication and advanced metering infrastructure, customer service is improved especially in power outage detection, grid operation, disaster recovery, and accuracy of load estimation [13,14]. Basically, one MG can run in two modes, grid-tied or islanded [15]. In the first mode, power flows in two ways between a MG and the grid, a MG may consume the power from the other MGs or it may supply power to them for credits according to the agreement between this MG and a local utility company. In an islanded mode, there is no power flow between the MG and the grid. This intentional ‘‘islanding’’ under certain circumstances, provides local reliability, stability, and security. However, this operation does not change or disrupt the integrity of the transmission grid as a whole [15,16]. In a SG, most energy management for a MG as a consumer is to minimize the electricity cost. Current electricity price on the grid is dynamic and it is established by the market to balance sellers’ supply and buyers’ demand. The balancing process is continuous and instantaneous since electricity must be produced at nearly the same time it is consumed. The market trades in electricity can be bilateral transactions in the form of a day-ahead market and spot market trades in real-time. The bilateral transaction is short-term forward market trading and is conducted between wholesale consumers and power plants and it provides price certainty. But MGs with DEG, DES, and loads must pursue a combination of short-term trading and spot trading. Power flows in two ways between the grid and the MGs. Thus, in order to minimize their cost from the generation and consumption, MG portfolios can be balanced and the generation, storage, or demand would be managed or scheduled under some short-term predictable and unpredictable circumstances [11,17–19]. The effective energy management can be market-based controls or agent-based controls. The first one emphasizes the creation of a competitive market environment for MGs with DEG, DES, and demand and the regulation of the production and consumption with direct market signals in order to meet the perspectives on cost and effectiveness. Depending on the control structure, there are three different operations: direct control, price signal control, and internal exchange [10]. The second one, agent-based control, emphasizes the importance of MG coordination and reliable communications with its neighbors. This improves the power system scalability and the optimization objectives include minimum cost, maximum profit, or highest efficiency [11]. In this paper, the performance of a MG with a hydrogen FC stack, as a CHP system, is comprehensively evaluated with the consideration of cost, emission, and service quality. The mathematical model of a MG is presented in Section 2. Besides a fuel cell stack, this MG includes a PV system, a PEM electrolyzer, and a hydrogen tank. Moreover, this MG is connected to a grid with a dynamic price and limited power availability in peak period. To operate the MG, an energy management is developed with the FC stack under a total-power-led scenario. In Section 3, the performance of the MG is analyzed and it is compared with the performance of a similar MG but without a CHP system.
2. Model 2.1. General structure Fig. 1 shows a basic structure of a microgrid. One agent in this sub-power system collaborates with the other agents on the grid to manage local energy generation, consumption, and flow. To
minimize the communications overhead, a regional manager, an agent selected from multiple microgrids, will exchange information with the managers in a wide area. Therefore, the energy system is a distributed system while the communication is a hierarchy system. The structure of agent-based energy management is shown in Fig. 2. The agent can be the weather forecast or the estimated energy generation/consumption pattern for a local DEG and loads for the upcoming time period. This agent communicates with its regional manager which is also an agent but voted by its neighbors to be the manager. The regional manager then aggregates the energy profiles and trade with the other managers through bids and auctions in order to balance the supply and demand and determine a price. A two-layer hierarchical structure will ensure scalability and reduce communication overhead. Thus, with the information from the regional manager, an agent can make a decision to optimize the operation of the system through real-time control with AMI. In this work, the structure of the grid-tied MG is shown in Fig. 3. The electricity generated from the PV system is first supplied to the local load and then to the electrolyzer. The spare electricity, if there is, goes to the grid. In this system, an electrolyzer is used to covert electricity to hydrogen, which is stored in the tank. The stored H2 can be supplied to the hydrogen FC stack as needed. The running of the FC stack is based on the total-power-led scenario and the outputs, electricity and heat, go to the loads and part of heat can be stored in the thermal tank. Since there is no CO2 or NOx as byproduct, this microgrid is emission free. For comparison, a similar MG without hydrogen tank and a FC stack is also simulated. In the system running, an agent module collects all information from the MG components and predicts PV system output, electricity price on the grid, heat load, and electric load. Then, it makes a plan for the power flow and schedule for the electrolyzer and the FC stack in the next hour. 2.2. PV system With the consideration the daily electricity and heat load, the area of the solar array is set as 200 m2. According to the location, Bridgeport, Connecticut, USA, the daily ambient temperature profile is shown in Fig. 4 and the temperature effect on the PV system open circuit voltage can be calculated as [20,21]:
V oc;ambient ¼ aðT STC T ambient Þ þ V oc;rated
ð1Þ
Here, a is a temperature coefficient as 0.12, Voc,ambient the open circuit voltage at ambient temperature, TSTC temperature at standard test conditions, 25 °C, 1000 W/m2 solar irradiance, Tambient the module temperature, and Voc,rated open circuit voltage at STC. Thus, there is a profile of the PV system output in Fig. 4 and the maximum power output is 85 kW at 12 PM. 2.3. Hydrogen FC stack The hydrogen proton-exchange-membrane fuel cell stack consists of 300 fuel cells in series with the parameters in Table 1. The main material is stainless steel and heat insulator. The operating temperature is 343 K, the optimum temperature this kind of fuel cells. For the cold start in a winter, the heat generated at the beginning is used to increase its temperature. At 373 K, part of the heat is transferred to the heat load or heat tank. In winter, the stack is insulated to minimize heat transfer and the heat conductivity used in this work is one hundreds of that in stainless steel. Fig. 5 shows the outputs from the fuel cell stack at three different temperatures, 293, 313, and 343 K. For the VI curve, it shifts lower as temperature increases. So does the power–current curve.
423
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
Fig. 1. The schematics of a microgrid.
Table 1 Parameters of a FC stack.
Agent
Forecasting and estimation
Market interface
Management interface
Market
Control and measurement
Parameter
Value
Number of cells Active area per cell Operating temperature Cathode pressure Anode pressure
300 (in series) 200 cm2 70 °C 2 atm 2 atm
70 293K 313K 343K
350
DEG
Load
60 heat
300
Voltage (V)
Electricity Heat H2
grid Electrolyzer
PV system
50
voltage
Fig. 2. The structure of an agent-based energy management system.
H2 tank
40 250
electricity
200
30 20
Power (kW)
DES
10
150
0 100
Heat load
temperature power
Temperature (K)
60
280
40 270
Power (kW)
80
290
20 260 0 10
150
200
On a power–current curve, there is a maximum power point, 30 kW, corresponding to the current density as 850 mA/cm2. The heat-current curve shows an opposite trend and it shifts higher as temperature increases. Compared with the electric power– current curve, heat generated is high than the electricity if the current is higher than 110 A. To run the stack, the power demand ranges from 5 kW to 22 kW, corresponding to 75% of the maximum power output.
100
300
5
100
Fig. 5. The voltage, electric power, and heat from the hydrogen FC stack at 293, 313, and 343 K.
Fuel cell
Fig. 3. The structure of the microgrid.
0
50
Current (A)
Electric load
250
0
Heat tank
15
20
Time (hour) Fig. 4. The ambient temperature and the power output from the PV system in one day.
2.4. Electricity from the grid Residential electricity rate from the local utility company is break down as electricity 7.7 ¢/kW h, transmission, distribution, and delivery 11.2 ¢/kW h, so total is 18.9 ¢/kW h with the basic service charge 16.5 $/month [22]. With the consideration of the hourly price pattern [23], Fig. 6 shows the tariff used in this work. For solar electricity, under a range of financing assumptions and locations, the levelized cost is estimated by the U.S. DOE ranges from 10 ¢/kW h to 18 ¢/kW h 11 for utility-scale, 16 ¢/kW h to 31 ¢/kW h for commercial systems and 16–25 ¢/kW h for residential PV systems [24]. Thus, the price of electricity from PV system is set as 20 ¢/kW h in this calculation. About hydrogen FC stack,
424
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
its average installed cost is 5600 $/kW, the electricity cost is between 45 and 55 ¢/kW h but 14 ¢/kW h for a CHP system [20]. With the consideration of the cost on electrolyzer, the price is assumed to be 30 ¢/kW h. In addition to the dynamic price in this simulation, there are upper limits, 30 kW and 20 kW, set for the power from the grid in the day and night time, respectively. 2.5. The operation of the hydrogen FC stack For a FC stack as a power generator, there are two operating points if the electricity load is less than the maximum power. The selection of the operating point is based on the system requirements, such as lowest cost, lightest unit, and highest power density. For example, a point at high current density make the cell small to lower down the capital cost, but its efficiency will be low with more heat generated. For a FC stack as a lCHP system, both the desired heat and power outputs are given, the operating point can be determined by the power/heat ratio if both points are allowed by the system. With enough hydrogen, a FC stack as a lCHP supply supplies electricity and heat as:
hchp þ hother ¼ hd
ð2Þ
echp þ eother ¼ ed
ð3Þ
where hchp is the heat from the lCHP, hother can be the heat from boiler, heat storage, and solar thermal system, echp is the electricity from the lCHP, eother can be from a wind turbine, PV system, grid, and electricity storage, and ed is the demand. Electricity from the grid is positive when power flows from the grid and negative when power flow into the grid. The constraint on the lCHP’s generation is
PHcap1 < PH < PHcaph
ð4Þ
where PH is power or heat and PHcapl and PHcaph are the lower and upper limits. Moreover, a stack should be planned to run at least certain time because it has to raise its temperature first from a cold start, especially in winter. In this paper, this time interval is at least 1 h and the stack runs under a total-power-led scenario. 2.6. Performance evaluation For the economic return on the investment of an individual MG, a performance metric Q for each MG is [25]:
Q ¼ w1 F þ w2 E þ w3 S1 þ w4 S2
ð5Þ
F¼
cGridav g c av g
15 Electricity Electricity+T&D
10
5
Here, cGrid-avg is the daily average electricity price from grid. cavg is the average electricity cost for a MG and it does not include the heat cost from the FC stack. Moreover, loads can be categorized into different types based on whether they can be shifted, interrupted, decreased, or cancelled. Loads also have different priorities in the presence of limited supply. Generally, demand is a random variable with a probability distribution in an operation time window, and it can be regarded as a series of separated and fine-grained tasks, which means each task can be completed in a sequence but not in continuous time slots. The electric service quality can be calculated as:
S1 ¼
S2 ¼ 0
0
5
10
ð6Þ
T X 3 Psupply;i 1X wi T t¼0 i¼1 P demand;i
ð7Þ
Here, T is a period of time, and i is the priority of the load (high, normal, or low). Psupply,i is the energy supplied to the load demand with a priority i, and Pdemand,i is the load demand with a priority i. wi (i = 1, 2, 3) is a weighting factor for the load with priority i, and P wi = 1. In this simulation, w1, w2, w3 are set as 0.5, 0.3, and 0.2 for the electrical loads with high, normal, and low priorities, respectively. The heat load is mainly for space heating and the priority of such load is fixed. Heat supply can be from the heat tank and electricity, whose priority is set only lower than that of the critical load. No matter where is the heat supply from, heat service quality is:
20
Price (cents/kWh)
F is a price index of electricity, E is an environmental effect or emission index due to atmospheric emissions, S1 is the service quality of electricity, and S2 is the service quality of heat. Q, F, E, and S are normally but not limited between 0 and 1. wi (i = 1, 2, 3, 4) are weightP ing factors and wi = 1. The ultimate objective for each MG is to maximize its overall performance index and the values of wi can be set according to the requirement of the consumer. In this simulation, the electricity price and emission are a little bit more important than the service quality of electricity and heat and w1, w2, w3 and w4 are 0.3, 0.3, 0.2, and 0.2, respectively. On the grid, the price of electricity is dynamic, based on the supply and demand. The day-ahead price can be forecast using wavelet transforms and related simulations [12,26]. Although the price is mainly determined by the supply from the power plants and the major demand, it is can be affected by DER and AD. The agent only participates in day-ahead market or spot market. For each device, based on the historical energy pattern, a predicted energy profile is generated for the day-ahead trading using neural network techniques. The Mont Carlo simulation is utilized to generate scenarios for real-time uncertainties. In the management process, optimization objectives are different for the shareholders of the power system versus the consumers. For residents or microgrids, there is a tradeoff between energy cost, energy security, and environmental protection. For utilities, the objective is the maximization of profit. Utilities are more likely to generate electricity when the price is high. In each case, the performance index, Eq. (5), is applicable but with different weighting factors. The combination of marketing, prediction, and real-time control in one subsystem exploits all potential benefits derived from each period of time. The price index of electricity, F, can be calculated as:
15
Time (hour) Fig. 6. Residential electricity rate.
20
Hsupply Hdemand
ð8Þ
Finally, the environmental effect index is:
E¼1
e av g eplant
ð9Þ
425
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
(b) 40 Electricity (kW)
Electricity (kW)
40 30 20 Electricity Load Supply
10 0 40
0
10
20
30
40
50
60
Heat (kW)
30 20 Heat
10
Load Supply
30 20 Electricity Load Supply
10 0 0 40
70
Heat (kW)
(a)
10
20
30
40
50
60
30 20 Heat Load Supply
10 0
0 0
10
20
30
40
50
60
70
70
0
10
20
30
Time (hour)
40
50
60
70
Time (hour)
340
FC Ambient
320 300 280 260 0 20
Power (kW)
1500
Hydrogen (moles)
Temperature (K)
Fig. 7. The electricity and heat power for the system with a FC stack (a) and without a FC stack (b).
10
15
20
30
40
50
60
70
Electricity Heat
10
1000
500
0
5 0
0
10
20
-5 0
10
20
30
40
50
60
30
40
50
60
70
Time (hour)
70
Time (hour)
Fig. 10. The hydrogen in the tank.
Fig. 8. the temperature (upper) and the supply (lower) from the fuel cell stack.
3. Results and discussion where eavg is the average atmospheric emission for electricity consumed in a MG, and eplant is the emission from a thermal power plant. In this paper, the emission is set as 3 units for electricity from the fire power plant and 0 unit for electricity generated from a solar panel or a fuel cell stack in a MG.
(a)
100
100 PV Grid
80 60
60
Electricity (kW)
Electricity (kW)
(b)
Solar Panel FC ENG from Grid Electrolyzer
80
Fig. 7(a) and (b) shows the electricity and heat profiles for the system with a FC stack and without one, respectively. In the day time, most electricity load is met in both systems. After sunset, these two systems obtain limited electricity from the grid.
40 20
40 20 0
0 -20
-20
0
5
10
Time (hour)
15
20
-40
0
5
10
Time (hour)
Fig. 9. The profile of the MGs with a FC stack (a) and without a FC stack in the first day.
15
20
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427
(a)
Price Envir. Elec. Qual. Heat Qual. Overall
1.4 1.27
Index
1.2
(b) 1.29
1.27
1.2
1.0 0.90
0.90
0.8
0.78
0.74 0.72 0.78
0.740.72
1.29
1.0 0.86
0.86
0.8
0.72 0.67
0.68
0.71
0.64
0.64
0.6
0.6
0.4
Price Envir. Elec. Qual. Heat Qual. Overal
1.4
Index
426
1
2
0.4
1
2
Day
Day
Fig. 11. The performance indices of the systems with a FC stack (a) and without a FC stack (b).
In addition to this, the FC stack in the first case can generate electricity and heat for around 5 h till its hydrogen tank becomes empty. Thus, more electricity is supplied to the load. For the heat load and supply, they show the similar result as that in the electricity. Fig. 8 shows the FC stack temperature increase from the cold start. The output from the FC stack is electricity only at the beginning. After 40 min, the temperature rises to the optimum value, 343 K, it maintains its temperature and starts to supply heat to the heat load or the heat tank. When the FC stack stops running, its temperature drops and eventually equals to the surrounding temperature. Fig. 9(a) shows the electrical power profiles of the MG from the PV system, the FC stack, grid, and electrolyzer, respectively. The power from the PV system starts to increase at 7:30 AM and its maximum output is 85 kW at 12 PM. The output from the PV is not only supplied to the load, but also to the electrolyzer for energy storage. The rest goes to the grid for credit from the local utility company. When the output from the PV system drops at 5 PM, additional electricity comes from the grid and the FC stack to meet the local demand. In Fig. 9(b), the electric power from the PV system goes to the local loads and the grid. If it is insufficient to meet the demand, electricity from the grid will be transferred to the MG. The amount of H2 fluctuates between 0 and 1600 mol in the tank as shown in Fig. 10. All hydrogen is generated by the electrolyzer. The hydrogen from the tank can be supplied for 5 h after the sunset. The peak of the amount appears at around 4 PM and it is used up at the beginning of the next day. Here, hydrogen is a kind of storage method of solar energy. No electricity from the grid contributes to the hydrogen generation due to the upper limits of electrolyzer input power, tank capacity, and availability of grid power input. Fig. 11 shows different performance indexes for the two power systems. In Fig. 11(a), although the prices for the electricity from the PV system and FC stack are higher than that from the grid, the price index is 1.27 due to the heat supply counted in the index calculation. In this work, the cost for the heat supply from the FC stack is negligible. Moreover, some credits are brought by the small amount of electricity transferred from the system to the grid in the day time. The environmental effect index is less than 1 and this is due to the use of electricity from the grid. The service qualities of electricity and heat are both between 0.72 and 0.8. Most electricity and heat demand is met. In Fig. 11(b), the price index of the system without a FC stack is around 1.29 because of the significant credits for electricity transfer from the system to the grid, shown in Fig. 9(b). The environmental effect index is lower than that in the first case even they have almost the same inputs from the grid,
as shown in Fig. 9. But, less usage of power from renewable energy resource will decrease this index according to Eq. (9). The electric quality index is close to 0.7 while the heat quality index is low as 0.64. Finally, the overall indexes for both systems are close to 0.9 with the weighting factors set for Eq. (5). 4. Conclusions In this work, with the dynamic electricity price and the constraint on electricity from the grid, the performance is evaluated on a grid-tied MG with DES and a hydrogen FC stack and it is also compared with that of a similar MG but without DES or a hydrogen FC stack. With the assumed weighting factors on the consideration of price, environment, and service quality, both MGs show almost the same overall performance index. Except the price index, the environmental and service quality indexes are higher in the MG with DES and a FC stack. Current grid electricity price fluctuates in a narrow range, less than 10 ¢/kW h and the price from the PV system and the FC stack is much high. Further work will be focused on the effect of the electricity price change on the optimal MG structure. Moreover, with more and more restrict legislation on the environment protection, the method for the performance evaluation, such as the weighting factors in the calculation, will be updated. References [1] Recovery act-smart grid demonstration U.S.Department of Energy, 2009. [2] Grose TK. The cyber grid. ASEE Prism; 2009. p. 26–31. [3] Hawkes AD, Leach MA. Modelling high level system design and unit commitment for a microgrid. Appl Energy 2009;86:1253–65. [4] Hledik R. How green is the smart grid? Electr J 2009;22(3):29–41. [5] Wu Z, Gu W, Wang R, Yuan X, Liu W. Economic optimal schedule of CHP microgrid system using chance constrained programming and particle swarm optimization. In: Power and energy society general meeting, 2011 IEEE; 2011. [6] Pedrasa MAA, Spooner TD, MacGill IF. Coordinated Scheduling of Residential Distributed Energy Resources to Optimize Smart Home Energy Services. Smart Grid, IEEE Transactions on, vol. 1(2); 2012. p. 134–43. [7] Alsayegh O, Alhajraf S, Albusairi H. Grid-connected renewable energy source systems: challenges and proposed management schemes. Energy Convers Manage 2010;51:1690–3. [8] Lasseter RH. MicroGrids. In: IEEE Power engineering society winter meeting; 2002. [9] Lopes JAP, Moreira CL, Madureira AG. Defining control strategies for MicroGrids islanded operation. IEEE Trans Power Syst 2006;21(2):916–24. [10] You S. Developing virtual power plant for optimized distributed energy resources operation and integration. In: Department of Electrical Engineering, Technical University of Denmark: Lyngby; 2010. [11] Molderink A, Bakker V, Bosman MGC, Hurink JL, Smit GJM. Management and control of domestic smart grid technology. In: Smart grid, IEEE transactions on, vol. 1(2); 2010. p. 109–19.
L. Zhang, J. Xiang / Energy Conversion and Management 87 (2014) 421–427 [12] Conejo AJ, Plazas MA, Espinola R, Molina AB. Day-ahead electricity price forecasting using the wavelet transform and ARIMA models power systems. In: IEEE Transactions on, vol. 20(2); 2005. p. 1035–42. [13] Bennett C, Highfill D. Networking AMI smart meters. In: Energy 2030 conference. ENERGY 2008. IEEE2008; 2008. p. 1–8. [14] Hart DG. Using AMI to realize the smart grid. In: Power and energy society general meeting – conversion and delivery of electrical energy in the 21st century, IEEE; 2008. p. 1–2. [15] Lasseter RH, Piagi P. Microgrid: a conceptual solution. In: PESC2005: Aachen, Germany. p. 20–5. [16] Zhong Z, Huo H, Zhu X, Cao G, Ren Y. Adaptive maximum power point tracking control of fuel cell power plants. J Power Sources 2008;176:259–69. [17] http://www.iso-ne.com/nwsiss/grid_mkts/how_mkts_wrk/whlsle_elec_trad/ index-p2.html. [18] Rahman MM, Mto A. Technologies required for efficient operation of a smart meter network. In: Industrial electronics and applications (ICIEA), 2011 6th IEEE conference on; 2011. p. 809–14. [19] Mohsenian-Rad A-H, Wong VWSJ, Jackson J, Schober R, Leon-Garcia A. Autonomous demand-side management based on game-theoretic energy
[20] [21] [22]
[23]
[24]
[25]
[26]
427
consumption scheduling for the future smart grid. In: Smart grid, IEEE transactions on, vol. 1(3); 2010. p. 320–31.. http://www.fao.org/docrep/x0490e/x0490e07.htm. http://www.wcc.nrcs.usda.gov. Sechilariu M, Wang B, Locment F. Building-integrated microgrid: advanced local energy management for forthcoming smart power grid communication. Energy Build 2013;59:236–43. Chen Y, Wu Y, Song C, Chen Y. Design and implementation of energy management system with fuzzy control for DC microgrid systems. IEEE Trans Power Electron 2013;28(4):1563–70. Nunna K, Doolla S. Multiagent-based distributed-energy-resource management for intelligent microgrids. IEEE Trans Ind Electron 2013;60(4):1678–87. Gu W, Wu Z, Bo R, Liu W, Zhou G, Chen W, Wu Z. Modeling, planning and optimal energy management of combined cooling, heating and power microgrid: a review. Int J Electr Power Energy Syst 2014;54:26–37. Chen Z, Wu L, Fu Y. Real-time price-based demand response management for residential appliances via stochastic optimization and robust optimization. In: Smart grid, IEEE transactions on. 99. p. 1–9.