NaS Technology Allocation for Improving Reliability of DG-Enhanced Distribution Networks Ehsan Naderi, Student Member, IEEE, Iman Kiaei, Student Member, IEEE, M.R. Haghifam, Senior Member, IEEE
[email protected]
[email protected]
[email protected]
Faculty of Electrical and Computer Engineering Tarbiat Modares University Tehran, Iran Abstract—Reliability calculation is the most important concern in designing and planning of distribution systems, which is considered, in terms of the economical concerns, with minimal interruption of customers. Storage devices are usually used for storing the electric energy while the electric price is low, and deliver their reserved energy when it has financial profit. Also, it could be effective for decreasing the average duration and frequency of outages in load points. In this paper, a method for evaluation of average annual energy not supplied (ENS) for a radial distribution system with DG units and storage devices is presented. Besides, the genetic algorithm (GA) is used to find the optimal capacity and the best location for storage device in order to minimize ENS and installation cost simultaneously. The effectiveness of this method is examined on a real distribution network.
INTRODUCTION
The main responsibility of the power system is to provide the customers with the electrical energy in an economical way and with a specified level of reliability. The distribution system, as a part of power systems, sets the final connection between the customers and the utility. Normally, the distribution networks have a main feeder which has a radial configuration. A radial feeder of the distribution network comprises a set of underground cables, lines, and equipments which connect every load point to the energy resource. Because of the low establishment and maintenance costs and its simple design, the radial distribution systems are utilized pretty widely. In case of any unexpected event in any of these parts, the breaker is activated and the load is disconnected; hence, the low reliability of this configuration. The analysis of the customer failure statistics shows that the distribution network plays the major role in the failure of the load supply [1]. Therefore, reliability enhancement has become the center of attention for utilities and researchers.
2.
Replacing overhead lines with underground cables and insulated lines.
3.
Using automatic network sectionalizing, automatic network configuration, and maneuver points in order to use alternative resources [4, 5].
4.
Adding alternative resources like DGs and storages [6, 7].
In this paper, the effect of capacity and location of the storages on the ENS is calculated using a deterministic method. In this calculation, the ideal operation of the protection devices is presupposed and the loads could have three priorities which are predetermined; the distribution network can efficiently supply the whole load; the DGs and storages can operate as island and they remain acceptable voltage and frequency in the network. Finally, a method is presented to optimize the locations and capacities of the storages in the network in order to achieve the minimum of the ENS. II.
PROBLEM FORMULATION
By increasing the importance of reliability of the load points as a consequence of raising the interruption cost of loads, utilities and other system authorities are persuaded to improve the reliability indices of network. Among the different ways of increasing the reliability of power systems in this paper usage of storage devices in distribution systems have been
There have been many methods used to enhance the reliability which some important ones are as follows [2]:
978-1-4244-5721-2/10/$26.00 ©2010 IEEE
Adding additional substations in order to shorten the lines [3].
Losing electrical energy imposes some important social and economical effects on the customers and energy suppliers. The failure time in an area could increase the power interruption cost of that very area to a noticeable degree. Where there are commercial and industrial loads, this amount could rise up to millions of dollars. These conditions have led the distribution companies to use alternative power resources with high reliability characteristics in the distribution networks. The DGs and storages increase the reliability of the networks. By use of storages in the networks, the interruption duration and, in consequence, the ENS could be decreased. Factors like the capacity of the storages and their locations on the network could affect the indices of load point reliability.
Keywords—Reliability Assessment, Storage Devices, Distribution Systems, Distributed Generation, Restoration Time, Genetic Algorithm, Energy Not Supplied.
I.
1.
148
PMAPS 2010
technology, because of its adequate technological maturation, high charging and discharging efficiency, and less environmental impacts. Because of these advantages, Japanese network has only more than 70 MW installed capacity of NaS battery system— the largest facilities at 8 MW/8 h [8].
studied. In this research effect of placing a storage device in a radial distribution network will be calculated. Storage technologies, similar to DG units, when are used for the purpose of reliability improvement, decrease the repair time of network sections. In this situation when a fault is occurred in the network, assuming the protection system is perfect, the faulted zone of network will be isolated and other parts of system could be operated. In a radial distribution system, parts of grid that placed upstream of faulted section, after the detection and isolation time will be supplied by distribution transformer, and their repair time will be reduced to the time of disconnecting and reconnecting the grid. For parts which are located downstream of the faulted section in a network without any enhancement resources, the reliability parameters are not changed. But in a grid that equipped with DG units or storage devices, after isolating the defected zone, depending on section of installation and capacity of additional resources, the repair time of those loads that could be supplied by DG or storage units will be decreased to isolation time and for the loads that could not be supplied the repair time remains at the time of removing the fault.
NaS battery based on the internal temperature, has an output limit, where the feasible discharging times are specified according to the limit called NaS pulse limit. Fig. 1 shows NaS pulse limit versus the different discharging duration. It is apparent that increment of the NaS pulse factor leads to the increase in power loss [9]. The parameters of the studied NaS battery plant are proposed in Table I [10]. The salvage value of NaS battery plant is assumed to be 15% of its capital cost. B. Reliability Calculation Reliability indices are statistical aggregation of reliability data for a well-defined set of loads, components, and customers. Most reliability indices are average values of a particular reliability characteristic for an entire system, operating region, substation service territory, or feeder. The following are referred to as reliability indices of load points: average failure rate λ (f/yr), average outage time r (h), average annual outage time U (h/yr), average energy not supplied ENS (kWh/yr). In order to achieve a more concrete view of the network’s condition, the reliability assessment indices which are related to the system and show the behavior of the entire feeder are used. However, because of calculating the interruption cost, only the ENS index is studied in this paper. Formula for this index is:
Aim of this paper is to evaluate the effect of a storage device in radial distribution networks, with three load priorities. It is assumed that the power supply from the grid is adequate that means after isolating the faulted part of network the parts that connected to distribution transformer could be supplied. The load points are integrated and could not be supplied partly. Protection system is ideal, namely the minimum area of network will be disconnected if a fault occurred in system. DG units and storage device can be operated in islanded mode and can control the voltage and frequency in acceptable limits while operating separately. By these assumptions depending on the section of installation and capacity of storage device the ENS of loads will be affected. By using the deterministic approach for evaluating the ENS of system, the effect of additional resources compared to original grid. In this paper genetic algorithm is used to find the best capacity for storage device considering the investment cost and improvement in cost exposed to customer for energy not supplied. For each of the sections of network the best capacity of storage is calculated and at last the optimal place and capacity is chosen by comparing the results.
(1) ENSi= Average energy not supplied at load point i C. Genetic Algorithm Genetic algorithm (GA) is search method based on principles of natural selection and genetics [11]. GA encodes the decision variables of a search problem into finite-length strings of alphabets of certain cardinality. The strings which are candidate solutions to the search problem are referred to as chromosomes, the alphabets are referred to as genes, and the values of genes are called alleles. In contrast to traditional optimization techniques, GA works with coding of parameters, rather than the parameters themselves. To compare the solutions, fitness function is defined such that reflexes desired characteristics of the problem. Using fitness value of each solution the best ones are chosen to evolve toward better solutions.
A. Storage Technology Nowadays, importance of access to reliable distribution system leads to increase the usage of alternative resources. Electric storage devices are categorized in alternative resources, which have some advantages like network reliability increase, peak load shifting, grid voltage support, frequency regulation, and emergency power. Mandates for renewable energy raise grid stability and dispatchability issues. Energy storage can also control the peaks and valleys of renewable generation and assure grid stability and reliability.
TABLE I.
Recently, several new electrical storage technologies have been presented like NaS (Natrium Sulfur) batteries, redox flow batteries, flywheels, superconducting magnetic energy storage (SMES), and compressed air energy storage (CAES). In the present paper, NaS battery system is selected as a storage
149
ECONAMIC PARAMETERS OF STUDIED NAS BATTERY PLANT
Pmax (MW)
10
Emax (MWh)
70
η (%)
90
Life time (years) Capital cast ($/kW) Utilization factor (%)
15 2250 95
Figure 1. NaS pulse trend versus the different discharge durations
faults could be calculated by using the analytical method. Each of the sections could have DG or storage units or both. All the storage devices, like NaS batteries, are energy limited and can supply a certain load in a limited time interval. By the increase or decrease of the load, the duration that storage device can supply will be reduced or increased. In this approach, DG units could be energy limited or not.
Here capacity of storage device is selected as chromosome and by converting it to binary code each of the bits will be the alleles. To select the good solution for crossover stage a roulette wheel is used which works with incremental probability of solutions that result from fitness values. In each iteration, the best answer is saved and new solutions are the best ones of the previous population. After performing the defined number of iterations the optimal capacity for storage unit will be obtained.
After a fault occurred in a particular subsection, all load points in that section experience an interruption. For each of the other sections, the actual generation or storage active power
Mutation is designed in order to add diversity to the population and ensure that it is possible to explore the entire search space. Mutation is often the secondary operator in GA, performed with a low probability. One of the most common mutations is the bit-flip mutation that is used here. III. CALCULATION OF RELIABILITY OF THE RADIAL DISTRIBUTION SYSTEM IN PRESENCE OF DGS AND STORAGES This section describes details of the proposed method for calculation of the average annual energy not supplied, (ENS), for a radial distribution system. It is assumed that after a detection and restoration time, the fault area will be disconnected from grid by proper operation of protection devices and other parts of network could be provided by the remaining supply systems. By operation of breakers, distribution feeder is divided into sections such that in each section, costumers are exposed to same reliability indices. In a radial feeder, placement of N protection devices results in N+1 sections. If feeder is not radial, the number of sections may decrease. For example, Fig.2 shows the formation of five reliability sections after the allocation of five breakers on a non-radial distribution feeder. By occurring a fault in any part of a section, all the loads in that section will be disconnected, and also for a fault occurring outside the section, the customers of the section may be connected (if they are connected to a supply system) or disconnected. Thus, the loads in each section will be exposed to the same number and duration of outages, and have same reliability indices. Assuming the constant values for failure rates (λ) and repair times (r), the annual duration of permanent
Figure 2. Distribution feeder equipped with six capacity-constrained distributed generators and five breakers (the corresponding reliability sections defined by dashed lines)[12].
150
capacity that have been installed could be used to provide the individual first priority loads in the section. After supplying all inside first priority loads the extra capacity (if there is any) will be used to supply other non fault sections. After providing all first priority loads, in the same manner, the other lower priority loads will be supplied by the remaining power sources.
Iran. The network has 16 load points and 2138 customers. There are three kinds of customers with different costs of onehour interruption that specifies three load priorities. The first load priority has 19.38 ($/kW) interruption cost and the second and the third ones have 8.4 and 2.1 ($/kW). The authors developed a computer program to calculate the reliability Index, ENS, using system data shown in Table II. The purpose for developing this program was to determine the best location and optimal capacity of storage units in order to reduce the interruption cost as much as possible. The power capacity of
The duration of customer's outage that could connect to supply systems will be reduced to restoration time plus time interval that energy resources cannot provide customers’ required energy. IV.
OPTIMIZATION PROCEDURE
Start
Separation of distribution feeder by protection devices to different sections will bring the question that where additional equipment should be put to achieve the greatest improvement. In this paper, we consider the improvement of reliability index, ENS, by adding a storage device to a radial network.
Read the network configuration and system data such as failure and repair rates, load parameters, and DG units and NaS information.
For investing on storage systems, utility should be sure about the fact that the installation will reduce the interruption cost of the customers.
Select the first section network i=1. Set the parameters of GA including the number of iterations and preliminary population, limit for random generation.
The problem is how to put storage device in a radial distribution system to achieve minimal cost. Therefore, the objective function of this problem is the sum of the interruption cost, that is a mixture of a storage life time value and the capital cost of storage device minus Salvage value. Interruption cost is computed by multiplying each load level ENS value by its outage cost. The annual interruption cost is converted to a value along with the storage life time using (2) described in [13]: 1
1
Generate the preliminary population (capacity).
Calculate the fitness function j=1.
(2)
Use roulette wheel to select parents for regeneration
1 Where
Perform the crossover and mutation on parents.
R=Annual Value
Calculate the fitness function for new children.
i=Difference between interest rate and inflation Rate
Save the best answer and select the best ones for new population j=j+1.
Capital cost of storage devices is a function of their active power and energy capacity.
j reached to iteration limit
(3)
No
Yes Save the best answer for section i, i=i+1.
All of the network sections are evaluated.
In the objective function, the composite cost is formulated as a function of capacity and location of the storage device. For more clarification, the algorithm associated with the proposed method is illustrated in Fig. 3.
No
Yes Chose and show the greatest improvement.
A. Studied network The application of the proposed method to a distribution system with several load points and three load levels is performed on the radial distribution system shown in Fig 4. This network is an under-operation system in Tehran utility,
Stop Figure 3.
151
Solution procedure flowchart
Figure.4.
single line diagram of studied network
each DG unit is assumed to be 650 kW and energy is limited (the energy required for one hour in full load) or unlimited for different cases. Because DGs are assumed as standby units, they only affect outage time and do not affect interruption frequency.
1) Case 1 The effect of DG unit placement on determination of location and capacity of storage device is analyzed by changing the DG locations and solving the problem for the best location and capacity of storage systems.
TABLE II. TEST SYSTEM DATA Average Load (kw)
Load Priority
Section Number
Length (km)
Failure Rate (failure/yr.km)
Repair Time (hr)
1
0.42
1.49
3
90
1
2
0.25
1.49
3
—
—
3
0.294
1.49
3
100.1
1
4
0.411
1.49
3
18.7
3
5
0.190
1.49
3
—
—
6
0.190
1.49
3
90
1
7
0.340
1.49
3
269.5
2
8
0.110
1.49
3
—
—
9
0.124
1.49
3
50.6
1
10
0.03
1.49
3
87.6
2
11
0.124
1.49
3
90
1
12
0.200
1.49
3
85
3
13
0.240
1.49
3
200
2
14
0.104
1.49
3
51.04
1
15
0.167
1.49
3
26.4
1
16
0.189
1.49
3
—
—
17
0.033
1.49
3
45.1
2
18
0.138
1.49
3
16.5
3
19
0.481
1.49
3
—
—
20
0.200
1.49
3
374
2
21
0.290
1.49
3
90
1
The results that are shown in Table III indicate that location of storage device depends on location of DGs in system. In this network configuration section 4 is the best point for placement of equipment, and the minimal cost obtained from the installation of a DG unit in section 4 with no storage investment. In this case minimal interruption cost is calculated by placement of a storage device in section 4 and a DG unit in section2. 2) Case 2 The effect of high priority load penetration is investigated by assuming all load points in the first priority and the previous experiment is repeated. The results, shown in Table IV, reveal that in this situation by increasing load interruption costs, investment on storage devices is more beneficial. The greatest effect on ENS was achieved by placing storage in section 4 while having a DG TABLE III.
152
TEST SYSTEM DATA FOR CASE 1
DG section number
Optimal Storage section number
Optimal Storage capacity (kW)
Cost of ENS ($)
Total Cost ($)
No DG No DG 2 3 4
— 4 4 3 —
0 64 54 53 0
1.46 1.36 1.13 1.41 1.23
— 1.8714 1.8523 1.8887 1.7490
5 6
4 4
53 53
1.15 1.15
1.8503 1.8503
unit in section 5. Also, the minimal cost was obtained by operating a single storage in section 5.
REFERENCES [1]
3) Case 3
[2]
In this situation effect of equipment quality is analyzed by examination of changes in failure rate and repair time of system instruments.
[3]
This case in performed for three conditions that the outcomes are depicted in Table V. Results from this case indicates that the capacity installation is tightly depends on system equipment parameter and as the parts of grid are going worse the need for investment on additional resources increases. This investment is justified by the improvement in costs of ENS. V.
[4]
[5]
CONCLUSIONS [6]
In the present paper allocation and capacity determination of storage devices in radial distribution systems for reducing the reliability Index, ENS, was investigated. In this research, the presence of energy limited DG units in network was considered.
[7]
[8]
The method was applied to a distribution network in Iran and ENS was computed for 3 cases with the developed computer program. Numerical results showed that utilization of storage device in distribution systems has a significant effect on the reliability of these systems. Also, the improvement of reliability depends on the location and characteristics of the installed equipment.
[9]
[10]
In this paper, unavailability of DG units and storage systems and its impact on the reliability indices was not studied. This issue could be analyzed in further studies. TABLE IV.
[11]
[12]
TEST SYSTEM DATA FOR CASE 2
DG section number
Optimal Storage section number
Optimal Storage capacity (kW)
Cost of ENS ($)
Total Cost ($)
No DG No DG 2 3 4 5 6
— 5 5 2 — 4 5
0 52 52 43 0 266 52
4.18 4.09 4.44 4.14 4.39 4.05 4.28
— 5.3604 5.8082 5.4132 5.6486 5.7253 5.6081
[13]
BIOGRAPHIES Ehsan Naderi was born in Shirvan, Iran, 1984. He received the B.Sc. degree in power systems engineering from the Shahrood University of Technology, Shahrood, Iran, in 2008. He is currently pursuing the M.Sc. degree in power systems engineering at Tarbiat Modarres University (TMU), Tehran, Iran. His research interests include planning and static and dynamic performance of Distributed Generation (DG) and stand alone power systems. Iman Kiaei (S’09) was born in Iran in 1986. He received the B.Sc. with honour in Electrical Engineering from Ferdowsi University of Mashhad, Mashhad, Iran in 2008. He is currently the M.Sc. degree student in power systems engineering at Tarbiat Modarres University (TMU), Tehran, Iran. His main research interests are power system planning, power system restructuring, electricity market and power system reliability.
TABLE V. TEST SYSTEM DATA FOR CASE 3
Failure Rate
Repair time r
Optimal Storage section number
Optimal Storage capacity (kW)
1.49 2 4 5 1
3 3 5 7 2
4 4 4 5 —
53 64 79 127 0
Cost of ENS ($)
Total Cost ($)
1.36 1.83 5.01 7.96 0.781
1.8503 2.4700 6.5997 10.467 1.0030
R. Billinton , and R.Allan, “Reliability evaluation of power systems,” Plenum, New York, 2nd edition, 1996. R.E.Brown, “Electric power distribution reliability,” Marcel Dekker Inc., New York, Basel, 2002. P. L. Cinquemani, A. Maunder, A. Bareggi, and L. Caimi, “SelfRepairing secondary underground residential distribution cable,” in Proc. IEEE Transmission and Distribution Conference and Exposition, vol.2, pp.797-802, Nov. 2001. A. Skoonpong, and S. Sirisumrannukul, “Network reconfiguration for reliability worth enhancement in distribution systems by simulated annealing,” in Proc. IEEE International Conference on Electronics, Computer, Telecommunications and Information Technology, vol. 2, pp.937-940, May 2008. Y.Qudaih, and T.Hiyama, “Reconfiguration of power distribution system using multi agent and hierarchical based load following operation with energy capacitor system,” in Proc. IEEE International Power Engineering Conference, pp.223-227, Dec. 2007. A. A. S. Algarni, and Kankar Bhattacharya, “Disco Operation Considering DG Units and Their Goodness Factors,” IEEE Trans. Power Syst., vol. 24, no. 4, pp. 1831–1840, Nov. 2009. Y.M.Attwa, and E.F. E1-Saadany, “Reliability based analysis for optimum allocation of DG,” in Proc. IEEE International Electrical Power Conference, pp.25-30, Oct.2007. I. Gyuk, P. Kulkarni, J. H. Sayeer, J. D. Boyes, G. P. Corey, and G. H. Peek, “The United States of storage,” IEEE Power & Energy Magazine, pp. 31-39, March-April 2005. T. Ohtaka, A. Uchida, and S. Iwamoto, "A voltage control strategy with NaS battery systems considering interconnection of distributed generations," in Proc. IEEE Power System Technology Conference, vol. 1., pp. 226-231, Nov 2004. S. J. Kazempour, and M. Parsa Moghaddam, “Economic viability of NaS battery plant in a competitive electricity market,” in Proc. IEEE International Conference on Clean Electrical Power , pp 453-459, 2009. E. Bueke, and G. Kendall, “Search methodologies introductory tutorials in optimization and decision support techniques,” New York, Springer, 2005 . A. Pregelj, M. Begovic´, and Ajeet Rohatgi, “Recloser allocation for improved reliability of DG-enhanced distribution networks,” IEEE Trans. Power Syst., vol. 21, no. 3, pp. 1442–1449, Aug.. 2006. Handbook of the european educational tool on cogeneration, 2nd edition, Dec 2001.
Mahmood-Reza Haghifam (M’95–SM’06) was born in Iran in 1967. He received the BS, MSc and PhD degrees in electrical engineering in 1989, 1992 and 1995. He joined Tarbiat Modares University as assistant Prof. in 1965. He is now a Full Professor in Power Systems at the Tarbiat Modarres University (TMU), Tehran, Iran He is a Senior Member of the IEEE (and IEEE Iran Section Industrial relationship officer). Also he is a research Fellow of Alexander Von Humboldt in Germany. He has been awarded by DAAD and AvH in 2001, 2006 and 2009 for research stays in German universities. He was as visiting Prof. in university of Calgary, Canada in 2003. His main research interests are Power System Restructuring, Power System Reliability, and Electric Distribution System.
153
