SIMULATING ELECTRICAL PERFORMANCE OF STATIONARY FUEL CELLS FOR DISPERSED GENERATION

Transcription

1 SIMULATING ELECTRICAL PERFORMANCE OF STATIONARY FUEL CELLS FOR DISPERSED GENERATION Dirk Audring, Gerd Balzer Darmstadt University of Technology Department of Electrical Power Systems Darmstadt, Germany Abstract - Proliferation of dispersed generation makes high demands on development of power system. A dynamic simulation model using EMTDC program was developed in order to assess the impact of fuel cells on power system. The model of the electrical part of a fuel cell will be explained and the results of simulations on model networks will be discussed. Examinations include short-circuit performance, interconnection protection, grounding and isolated operation. Recommendations are proposed to integrate fuel cells into the power system. Keywords - Fuel cells, dispersed generation, interconnection, protection, simulation, isolated operation 1 INTRODUCTION Increasing amount of small generation units lead to dispersion of feeding points on distribution networks. Dispersed power generation requires new strategies for topology and control of the network, since existing distribution networks are built on the base of large generators and a specified power direction [1]. This paper considers fuel cells as dispersed generators, but a lot of results may be applicable to other distributed energy converters. Stationary fuel cell systems will be integrated in the existing infrastructure at market entry. Natural gas will be reformed to hydrogen on site. Generated electricity will be fed to power system or supply a micro-grid or isolated load. Distributed generation makes new demands on local distribution companies. Hence they have to assess the impact of distributed generation on power system planning, operation, protection and tariff. Liberalization of electrical energy market offers great opportunities for development and marketing of dispersed generation. Efficient and simple integration of fuel cell systems into the power system is a challenge in the near future. In case of occasional isolated operation of a microgrid, standardized equipment is installed for distribution and protection. Tripping range of protection relay s must guarantee the same selectivity in a micro-grid as in connection with power system. Operation modes of these fuel cells must incorporate both possible network topologies to ensure reliability and safety. Services of power system, like voltage and frequency stability, must be provided by a fuel cell or electrical energy storage during isolated operation. 2 APPLICATIONS OF STATIONARY FUEL CELLS FOR DISPERSED GENERATION A lot of manufactures develop fuel cells for a wide range of applications. One field of application is dispersed generation. Stationary fuel cell systems may be economically available in the near future. Advantages of fuel cells are low CO and NO X emission, a high ratio of electric power to thermal power of about 1 and a part-load capability at a nearly constant and high efficiency. Since fuel cells convert chemical energy directly to electrical energy, efficiency is not limited to the CARNOT efficiency [2]. Sophisticated fuel cells may be suitable for various applications, such as manufacturing plants, office and public buildings, department stores and residential buildings. Produced heat can be utilized depending on operating temperature of fuel cell. Cogeneration of heat and power (CHP) raises exploitation of fuel and hence reduces CO 2 emission. The high ratio of electric power to thermal power enables CHP for buildings and may achieve a high exploitation of fuel [3]. Rated power of dispersed generation units and operating modes depend on objectives of manager and energy demand of supplied facility. Rating and limitations of rated power regarding operating modes and supply objectives are not discussed in this paper. Some manufactures equip their fuel cell systems with communication interfaces in order to operate a large number of dispersed fuel cells from a central control as a big power station ( virtual power plant ) [4]. Local distribution companies have the opportunity to produce their system services, like reactive power generation and peak-load supply, with dispersed fuel cells. Additionally they can sell produced heat to customers without a long-distance heat transmission. Disadvantages are expenditures of measurement devices, control, data transfer and information processing. Stationary fuel cells may be used as an uninterruptible or back-up power supply. Since charge of an electrical energy storage, like battery or energy storage capacitor, is limited, a fuel cell can generate required energy of an isolated load or micro-grid and extend considerable operating time. A micro-grid with fuel cells may be also an alternative to supply remote and undeveloped areas with electrical energy. Fuel cells have to prove reliability, availability and economic efficiency, before they take on important tasks of electrical power supply.

2 3 SIMULATING MODEL OF ELECTRICAL PERFORMANCE 3.1 General energy conversion of fuel cells A single cell of a fuel cell has a non-linear currentvoltage characteristic with a non-load voltage of 1.2 V. Cells are stacked to get an appropriate voltage level. A step-up converter adjusts the voltage level of the fuel cell stack to feed an inverter. A self-commutated inverter is used to convert DC-voltage of fuel cell to a three-phase alternating current (figure 1) [5]. Feeding of single-phase alternating current to power system is limited to units smaller than 4.6 kva according to present guidelines [6]. Many installed single-phase inverters have to be symmetrical spread over all three phases. Figure 1: General energy conversion of fuel cells Power injection mainly depends on control of inverter. The rate of change of power output is limited by capability of gas reformer to produce hydrogen and peripheral devices for gas preconditioning. Fuel cells may increase power output faster, if hydrogen is supplied from a storage. Starting time depends on type and operating temperature of fuel cell and reformer. 3.2 Simulation model The electric power conversion of a fuel cell system is emulated in order to determine electrical performance at operating and fault conditions. Simulations are carried out with program EMTDC. A self-commutated inverter was developed to examine different methods of connection to network (figure 2) [5] [7]. includes a limitation of power gradient. Therefore timeconstants of fuel cell stack are not modelled in detail. Switching sequence of inverter is computed from reference voltage using pulse-width-modulation (PWM). PWM pattern is calculated with sinusoidal PWM. Alternatively a module was developed to compute PWM pattern based on space vector of reference voltage. The reference voltage is calculated by the control of inverter (see section s 4.1 and 5.2). Smoothing reactors and filters to reduce injection of switching frequency are installed at ACside of inverter. Inverter with or without a transformer at network side may be used with fuel cell systems. Hence simulations can be carried out with both configurations. If an inverter-transformer is used, the phase angle of transformer is taken into consideration at synchronization of inverter. A circuit breaker connects fuel cell system to network. Protection scheme acts on circuit breaker and disconnects fuel cell from network during loss of mains and fault conditions. 4 CONNECTION TO POWER SYSTEM 4.1 Control of inverter Fuel cells can only be connected to the grid, if inverter is synchronized to network voltage. Therefore network voltage is measured at connection point. The phase angle of one phase is calculated by means of a phase-locked loop. Network voltage is transformed to d-q-components (u d and u q ) using the PARK transformation. Setpoint of active (P) and reactive power (Q) or alternatively power and power factor are given and reference value of current is calculated as d-q-components (i d and i q ). Since EMTDC program does not compute the power invariant d- q-components, setpoints of active and reactive power has to be corrected by multiplying by 2 3 (equation (1) and (2)). i d = 2 3 P u d + Q u q u 2 d + u2 q (1) i q = 2 3 P u q Q u d u 2 d + u2 q This reference current in d-q-components is transformed to three-phase reference current (i ref ) by means of instantaneous value of phase angle of network voltage. Deviation between reference current and measured current injection is computed. New reference voltage is calculated in order to eliminate current deviation within next operating sequence (t seq ). Pulse-width-modulation pattern is computed from reference voltage (equation (3)). (2) Figure 2: Inverter model DC-side of inverter is fed by fuel cell stack and stepup converter. The step-up converter provides inverter with a nearly constant DC voltage. DC ripple depends on coupling capacitance [8]. Power gradient depends on gas supply as described in section 3.1. The control of inverter u ref = u FC,AC Limitation of power injection L t seq (i ref i FC ) (3) Dispersed generation units may be connected to all voltage levels of distribution network. Power injection

3 leads to a voltage variation at point of common coupling (PCC). Voltage variation due to functional switching (i.e. normal switching duty) of a single unit is limited to 3% at low-voltage networks and 2% at medium-voltage networks according to present specifications [6] [9] [10]. Voltage variation depends on injected power and power factor angle of fuel cell as well as short-circuit power and angle at PCC. Hence it depends on network topology and voltage level. Voltage variation u of a single unit can be calculated by equation (4) [1]. with: u [p.u.] = S FC S sc,pcc e j (ψsc,pcc+ϕ FC) S sc,pcc - short-circuit power at PCC ψ sc,pcc - short-circuit angle at PCC S FC - complex power of fuel cell - power-factor angle of fuel cell ϕ FC The resulting voltage at PCC with power injection of a fuel cell can be derived by vector summation of voltage variation and node voltage at PCC without the fuel cell. Absolute value of the resulting voltage variation is computed as represented in Equation (5). u pcc [p.u.] = S 1+2 FC S sc,pcc cos(ψ sc,pcc + ϕ FC )+ S2 FC S 1 (5) sc,pcc 2 (4) a power factor angle of 45 at PCC and a fuel cell system feeds 60 kva with a power factor angle of -15. This is a ratio of power injection of fuel cell to short-circuit power of 0.02 and leads to a voltage variation due to functional switching of 1.74%. If power factor angle of fuel cell is unknown or may change during operation, the graph with a difference of angles equals 0 have to be chosen to consider the worst case. Power injection of all dispersed generating units are limited to meet voltage range of networks. Voltage range at operating conditions is limited from 90% to 110% of nominal voltage at medium and low-voltage networks from the year 2003 [11]. 4.3 Short-circuit performance Fuel cell systems are connected via a tie circuitbreaker to power systems. The protection scheme of a fuel cell system consists of an undervoltage and overvoltage relay. Typical setpoint values are 1.1 p.u. for overvoltage protection relay and 0.8 p.u. for undervoltage protection relay [6]. In case of a short-circuit in the AC power system, the inverter contributes short-circuit currents, until undervoltage relay trips the tie circuit-breaker. Figure 4 describes a low-voltage model network with radial feeder from one secondary substation. Figure 4: Single-line diagram of investigated low-voltage model network corresponding to figure 5, figure 6 and figure 7 Figure 3: Voltage variation at PCC due to functional switching of a single fuel cell unit Figure 3 shows the resulting voltage variation at PCC ( u pcc ) as a function of ratio of power injection of fuel cell to short-circuit power at PCC dependening on difference between angle of short-circuit power and power factor angle. Considering a short-circuit power of 3 MVA and Figure 5 shows terminal voltage and current injection of one phase of a fuel cell at node B1 system during a remote three-phase short-circuit at node A2. At occurrence of short-circuit, voltage drops to 0.7 p.u. at PCC. Fuel cell feeds operating current, until undervoltage relay trips tie circuit breaker. Since inner control loop of inverter controls current, fuel cell acts as a current source. Higher level control is power control with a loop time constant of

4 a few cycles. Current control stops increase of current injection, if rated current is fed to network. Hence fuel cells feed maximally their rated current to the network. In this case, fault is cleared after 100 ms by fuse F1 and voltage restores on feeder B. Non-directional overcurrent protection relay s of feeders are not affected by fuel cells, since they maximally contribute their rated current to power system. Fuel cells may cause reverse currents through directional protection devices, like a meshed network master relay at distribution transformer. Hence this may lead to tripping of directional protection relay, if setpoint of tripping current is lower than operational current. Therefore if dispersed fuel cells are installed in a network, tripping range of directional protection relay s have to allow operation currents in reverse direction [12]. transformer at network side is direct grounded, zerosequence impedance is reduced and hence unsymmetrical short-circuit currents increase. Since zero-sequence currents distribute corresponding to impedance s, high currents may flow via neutral point connection through the inverter-transformer of fuel cell. Figure 6 shows currents through inverter-transformer of fuel cell at node A3 during a single-phase-to-earth fault at node A2 (see figure 4 for simulated model network). All fuel cell systems are direct grounded. Rated power of this fuel cell is 10 kva, which leads to a rated current of about 15 A. Transferred short-circuit currents multiple exceeds rated current of inverter-transformer. Especially neutral conductor has to carry about 1 ka peak current in this case. If a design with grounded inverter-transformer is selected, single-phase-to-earth short-circuits have to be considered. A star-point reactance may reduce zero-sequence current at inverter-transformer. Figure 5: Short-circuit performance of a fuel cell at node B1 during three-phase short-circuit at node A2 4.4 Neutral point connection Zero-sequence is energized during unsymmetrical faults to earth or unbalanced load conditions. Hence grounding of dispersed generation influences zerosequence currents. Ordinary grounding schemes are possible, like solid earthing, earthing by means of a neutral reactor or an isolated neutral point. Methods of neutral point connection of network have to be taken into account, in order to avoid interference with protection system. Ungrounded units do not influence zero-sequence of power system. Short-circuit current at fault location is only raised by percentage of voltage rise at fault location due to dispersed fuel cells. Hence no problems for operation and protection of network are expected. Direct grounded inverter without a transformer at network side inject some high-frequency currents to zerosequence. Control of inverter must measure and reduce zero-sequence current injection. If inverter with Figure 6: Currents through terminals of a direct grounded invertertransformer of fuel cell at node A3 during a single-phase-to-earth fault at node A2 Third harmonic current increases with dispersed grounded transformers, since inverter-transformers have different rating and saturation hysteresis loop. This leads to third harmonic current in neutral conductor. Figure 7 shows absolute values of frequency spectrum of operating current in neutral conductor to distribution transformer at secondary substation. Case a) displays a configuration, where star-points of dispersed inverter-transformers are connected to neutral conductor, and case b) shows the operation with isolated star-points of dispersed generators. Distribution transformer at secondary substation and loads on low-voltage network are grounded in both cases and contribute current through neutral conductor. Third harmonic current is larger in case a) due to saturation effects of inverter-transformers. Fundamental frequency component of current (here 50 Hz) is larger in case b), because zero-sequence currents can only flow through distribu-

5 tion transformer. A star-point reactance reduces third harmonic current. Figure 8 shows an examined medium-voltage model network. Dispersed fuel cells are connected at medium and low-voltage networks. A three-phase short-circuit occurs at the medium-voltage side of node A2. Voltages at all connected low-voltage networks drop below 0.8 p.u. (Figure 9). Hence all connected fuel cells are disconnected from network by their undervoltage relay after 50 ms. The overcurrent-time protection relay at faulted medium-voltage feeder trips here after 500 ms and disconnects faulted medium-voltage feeder A as well as all connected low-voltage networks to feeder A. This shortcircuit leads to a loss of dispersed generation. Loads of distribution networks must be fed from transmission network after voltage recovery. Hence load current supplied from transmission network increases. Disconnected fuel cells may synchronize and reconnect to power system a certain time after recognition of voltage recovery. Figure 7: Low-order harmonics of current in neutral conductor to distribution transformer, case a) direct grounded inverter-transformers and case b) isolated star-points of inverter-transformers 4.5 Undervoltage protection As described in section 4.3, fuel cells may disconnect during a fault by their undervoltage protection relay. Since voltage drop occurs at faulted feeder as well as at faultless feeders, fuel cells at faultless feeders also disconnect from power system. Hence selectivity is not given for dispersed fuel cells. Figure 8: Single-line diagram of medium-voltage model network corresponding to figure 9 Figure 9: RMS-Voltages at low-voltage side of distribution transformers The loss of dispersed generation can be avoided by delaying the tripping of undervoltage protection relay longer than tripping time of feeder protection. This requires fuel cell inverters, which are capable to feed current to reduced terminal voltage at PCC. The same problem occurs at other voltage levels as well, if tripping of faulted feeder takes longer than tripping of undervoltage relay s of fuel cells [3]. Since voltage drop may be detected at remote networks, instantaneous tripping of undervoltage protection relay of dispersed generation is not useful. 5 MICRO-GRID OR ISOLATED OPERATION 5.1 Conception of a micro-grid A micro-grid is a small isolated network with the purpose to supply loads at remote, undeveloped areas or during temporary disconnection from power system. A main characteristic of a micro-grid is low short-circuit power, since generating units are rated to peak-load. Load demand may fluctuate and load forecast may not be possible due to unpredictable behavior of energy user [13].

6 If just one fuel cell supplies a micro-grid or isolated load, the fuel cell must be capable to feed peak load and change of power output must be fast enough to meet load demand. Hence hydrogen have to be supplied from a storage and rating is based on peak demand of load. Fuel cell has to set voltage and frequency of micro-grid. Therefore control of a stand-alone fuel cell operates like the master control (see section 5.2). Depending on load characteristic, the fuel cell may operate mostly in part-load condition. Fuel cells have a part-load capability at nearly constant electrical efficiency from 100% of rated power to approximately 30% of rated power. Below 30% of rated power electrical efficiency drops towards zero. The fuel cell system can be supported by an energy storage, like batteries or capacitors. Energy storage supplies peak load demand and smooth load characteristic. Fuel cell supplies energy demand over a certain period of time. If power demand of micro-grid is lower than power injection of fuel cell, the energy storage will be charged. Otherwise fuel cell supplies base load to micro-grid. The control of the energy storage is master control and adjusts voltage and frequency of micro-grid. Master control computes required power demand of micro-grid and transmits the setpoints of power to subordinated fuel cells. Advantage of this modular configuration is, that one master control can manage several fuel cell systems. This leads to a higher reliability and easy extension of micro-grid. 5.2 Master control Master control has to adjust voltage and frequency of micro-grid. Hence setpoints of control are voltage and frequency. Reference voltage of inverter is calculated from these setpoints. The inverter with master control acts as a voltage source. Injected currents result from topical load demand and power injection of subordinated fuel cells. Load demand can be calculated by means of measured currents and voltages of other distributed generating sets. Setpoints of power of subordinated fuel cells are computed from topical load demand in order to reduce or eliminate power injection from electrical energy storage [14]. Rated power and rate of change of power output of the specific fuel cell have to taken into account for computation of setpoint of power output. If several dispersed generating sets are installed, different part-load capabilities and available power have to be considered in master control. The charge of electrical energy storage is to be considered in order to recharge at low load times. Control of inverter of subordinated fuel cells may be the same as for operation in parallel with an ordinary power system. Inverter synchronizes to voltage of microgrid and feeds requested power. Communication interfaces are necessary in order to transmit the setpoint of power to dispersed fuel cells. Operational status messages have to be transmitted to master control. Since shortcircuit power of a micro-grid is much less than in a large power system, additional filters may be necessary to reduce high frequency current injection due to switching of inverter. 5.4 Short-circuit performance Short-circuit performance is explained on a simple configuration with a stand-alone fuel cell feeding a threephase alternating current to a busbar with two bays (figure 10). Nominal phase-to-phase voltage of micro-grid is 400 V and nominal phase-to-earth voltage is 230 V. The microgrid is modelled as a TN-C system. Loads of 10 kva and a power factor of 0.9 are installed at each feeder. As described in section 5.2, an inverter with master control acts as a voltage source. Hence current increases at occurrence of a short-circuit. Since current carrying capacity of transistors is limited, the control restricts current flow to a maximum permissible current. Therefore shortcircuit current is limited by control of inverter. Figure 10: Examined basic micro-grid 5.3 Subordinated fuel cells Figure 11: Short-circuit currents at feeder A and at inverter side of transformer Figure 11 shows short-circuit currents through feeder A during a three-phase short-circuit at this feeder. Currents increase at short-circuit occurrence until inverter limits current through transistor valves. Each phase may disconnect separately from busbar. Hence three-phase shortcircuit changes to a two-phase short-circuit. If second

7 line protection opens, fuel cell feeds a single-phase short circuit. Short-circuit currents increase nearly to the double of maximum permissible current of inverter. Currents at inverter side of inverter-transformer are limited to a maximum permissible current of 50 A. Since modelled inverter-transformer is delta connected at inverter side, the current through a delta winding is the difference of two line currents. Transformation ratio of inverter-transformer is set to 1, hence currents at the side of micro-grid can reach maximally the double of maximum permissible current (here 100 A). A 16A - fuse for line protection at feeder A would need approximately 10 seconds to disconnect the faulted feeder from the busbar [15]. This melting time is to long to ensure safety. A solution may be a combination of short-time increase of maximum current of inverter and fast melting fuses at feeder protection as proposed in [16]. Selectivity and fast response time of protection in a micro-grid remain problems and further research is necessary on that topic. 6 CONCLUSIONS A self-commutated inverter model was simulated with EMTDC. Control of inverter was developed to meet the requirements of a fuel cell system and to simulate electrical performance of a fuel cell. Controls for network connection and to feed a micro-grid are emulated. Different designs can be modelled and compared. Voltage variation due to functional switching of a single fuel cell limits power rating and can be calculated by means of short-circuit power and angle at PCC as well as rated power and power factor angle of fuel cell. Since inverter is current controlled at inner control loop, fuel cell contributes maximally their rated current to network during a short-circuit. Voltage drop during a short-circuit may trip undervoltage protection relay s of fuel cells at faultless feeders, if tripping time of feeder protection is longer than tripping time of undervoltage protection relay. Hence selectivity is only achieved by delaying tripping of undervoltage protection relay s until faulted feeder is disconnected. Direct grounding of inverter-transformer reduces zerosequence impedance and increases third-harmonic currents at neutral conductor. During unsymmetrical earth faults high zero-sequence currents flow through direct grounded inverter-transformer. A star-point reactance reduces zero-sequence currents of inverter-transformer. Short-circuit performance at a basic micro-grid was explained. Problems with response time of conventional line protection in a micro-grid are pointed out. Technical problems with integration of fuel cells in power and micro-grids will be solvable. 7 ACKNOWLEDGEMENTS This research project is sponsored by the German Research Society (Deutsche Forschungsgemeinschaft). REFERENCES [1] Cigre Working Group 37.23, Impact of increasing contribution of dispersed generation on the power system, Cigre Report 137, [2] R. H. Wolk, Fuel cells for homes and hospitals, IEEE Spectrum, pages 45 52, May [3] D. Audring, G. Balzer, O. Schmitt A. Wildenhain, Impact on power system by fuel cells supplying residential buildings, IEEE Porto PowerTech 2001, IEEE, [4] J. Berg, Das Vaillant Brennstoffzellen-Heizgerät, Brennstoffzellen-Heizgeräte zur Stromerzeugung in Haushalt, HdT Essen, [5] J. Domergue, A. Rufer, N. Buchheit, Dynamic model of a solid oxid fuel cell stack and power converter, Third European Solid Oxid Fuel Cell Forum, Nantes, June [6] VDEW-Richtlinie, Parallelbetrieb mit dem Niederspannungsnetz, VWEW-Verlag, [7] Z. Chen, E. Spooner, Voltage source inverters for high-power, variable-voltage DC power sources, IEE Proceedings-Generation, Transmission and Distribution, Vol. 148, No. 5, IEE, September [8] L.E. Lesster, Fuel cell power electronics, Fuel Cells Bulletin, No. 25, Elsevier Science, [9] VDEW-Richtlinie, Parallelbetrieb von Eigenerzeugungsanlagen mit dem Mittelspannungsnetz des Elektrizitätsversorgungsunternehmen (EVU), VWEW-Verlag, [10] IEC 1000: Electromagnetic compatibility (EMC) - part 3: Limits, IEC, [11] EN 50160: Voltage characteristics of electricity supplied by public distribution systems, European Standard CENELEC, [12] A. R. Wallace, Protection of embedded generation schemes, Protection and Connection of Renewable Energy Systems (Ref. No. 1999/205), IEE Colloquium, [13] H.L. Willis, Energy storage opportunities related to distributed generation, IEEE PES Summer Meeting, IEEE, [14] J.F. Chen, C.L. Chu, Combination voltagecontrolled and current-controlled PWM inverters for UPS parallel operation, IEEE Transactions on Power Electronics, Vol. 10, No. 5, IEEE, September [15] IEC 60269: Low-voltage fuses, IEC, [16] B. Burger: Transformatorloses Schaltungskonzept fuer ein dreiphasiges Inselnetz mit Photovoltaikgenerator und Batteriespeicher, Dissertation Universität Karlsruhe, 1997.