1640
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 3, AUGUST 2005
Optimal Allocation of Embedded Generation on Distribution Networks Andrew Keane, Student Member, IEEE, and Mark O’Malley, Senior Member, IEEE
Abstract—As a result of the restructuring of electricity markets and the targets laid down for renewable energy, increasing amounts of embedded generation (EG) are being connected to distribution networks. To accommodate this new type of generation, the existing distribution network should be utilized and developed in an optimal manner. This paper explains the background to the technical constraints faced by EG projects, and a new methodology is developed using linear programming to determine the optimal allocation of EG with respect to these constraints. The methodology is implemented and tested on a section of the Irish distribution network. Results are presented, demonstrating that the proper placement and sizing of EG is crucial to the accommodation of increasing levels of EG on distribution networks. Index Terms—Dispersed storage and generation, linear programming, power distribution planning.
I. INTRODUCTION
E
MBEDDED generation (EG) can be defined as small-scale generation, which is not directly connected to the transmission system and is not centrally dispatched. Generation is now being connected at distribution level, which has led to the characteristics of the network being changed. If increasing levels of generation are to be accommodated, then there must be a change of thinking regarding the planning and design of the distribution network. There is increasing demand for the development of the distribution network from a passive network to an active one in order to facilitate increasing levels of embedded generation [1]. It is envisaged that there will be a large investment in the distribution network, transforming it from a passive to an active network. If this is to be the case, a number of steps should be followed. First, best use should be made of the existing distribution network by optimal allocation of EG. Second, the development from passive to active should be planned optimally with all relevant considerations taken account of. The placement of generation on a first-come first-serve basis invariably limits the overall capacity of EG [2]. Here, a new methodology is developed to determine the suitable locations and ratings for EG on distribution networks with respect to the technical constraints, which will enable a high penetration of generation on the network and avoid network sterilization. Network sterilization results when capacity is allocated to the bus/buses whose voltage Manuscript received September 30, 2004; revised January 28, 2005. This work was conducted in the Electricity Research Centre, University College Dublin, which is supported by ESB Networks, ESB Powergen, ESB National Grid, Cylon, the Commission for Energy Regulation, Sustainable Energy Ireland, and Enterprise Ireland. Paper no. TPWRS-00525-2004. The authors are with the Department of Electronic and Electrical Engineering, University College Dublin, Dublin 4, Ireland (e-mail:
[email protected];
[email protected]). Digital Object Identifier 10.1109/TPWRS.2005.852115
and/or short-circuit levels (SCLs) are most sensitive to power injections. Thus, no more generation can be connected as the buses are constrained. In [2], the authors employ an optimal power flow technique to maximize EG capacity with respect to voltage and thermal constraints. SCLs, SCRs, equipment ratings, and losses are not considered. The effect of network sterilization is clearly demonstrated by comparison between allocating generation to buses individually rather than as a group. In [3], genetic algorithms were used to place generation such that losses, costs, and network disruption were minimized and the rating of the generator maximized. The constraints considered were voltage, thermal, short circuit, and generator active and reactive power capabilities. Generation is placed in single units at individual buses, while ignoring the interdependence of the buses and the network sterilization that can result from improper EG placement. In [4], the authors use a heuristic approach to determine the optimal EG size and site from an investment point of view. Once again, short circuit constraints are not considered, and the focus of the objective function is on optimal investment rather than maximizing renewable energy. It uses a cost benefit analysis to evaluate various placements of EG. The basis for the new methodology is in exploiting the interdependence, if any, of the buses with regard to the constraints. The constraints all have either linear or approximately linear characteristics with respect to increasing power injections, or they place a fixed and independent limit on the power injection. The new methodology described here determines the optimal allocation with respect to all of the relevant technical constraints. Optimal allocation ensures best use of the existing assets and achieves a higher penetration of EG in a cost-effective manner. In Section II, the technical constraints on EG are outlined, and the implementation of the methodology is explained. A section of the Irish distribution network is modeled in DIgSILENT Powerfactory using network data obtained from the distribution network operator (DNO) and is described in Section III. The methodology is tested on this section, and results and discussion are shown in Section IV, which illustrate how network sterilization is avoided if EG is optimally allocated. Conclusions are given in Section V. II. OPTIMAL ALLOCATION METHODOLOGY The objective is to maximize generation subject to the constraints outlined. Under the European Union (EU) Directive 2001/77/EC, Ireland must provide 13.2% of its electricity generation from renewable sources [5]. The EU Directive for renewable energy penetration is part of a strategy to meet the Kyoto Protocol national targets for reducing greenhouse gas
0885-8950/$20.00 © 2005 IEEE
KEANE AND O’MALLEY: OPTIMAL ALLOCATION OF EG ON DISTRIBUTION NETWORKS
emissions. The vast majority of embedded generation is from renewable sources. This methodology ensures optimal use of the existing network assets, thus helping to meet these targets in a cost-effective manner. An optimal allocation of EG exists for each section of the network. Generation capacity should be allocated across the buses such that none of the technical constraints is breached and the capacity maximized. Therefore, the proposed objective function is (1) where is the EG capacity at the th bus, and is the number of buses. Without loss of generality, it is assumed that there is one generator connected at each bus. The objective function J (MW) (1) is maximized subject to the constraints, which are described below. A. Thermal Constraint This is a stand-alone constraint; simply put, the rated current of the lines must not be exceeded (2) where
is the current flowing from generator to bus , and is the maximum rated current for the line between each generator and its corresponding bus. Under standard voltage and power factor conditions, the rated current of the line can be translated directly into a rated active power for that line. B. Equipment Ratings 1) Transformer Capacity: The amount of generation connected minus the summer valley load must not exceed the rating of the transformer at the higher voltage. If there is some existing generation, then this must be subtracted from the total. The result is the remaining capacity available below that station. In the case of two parallel transformers, the capacity is taken as the rating of the smaller transformer plus the summer valley load (3) where refers to power flow through the transmission subrefers to the rating of that station transformer, and transformer. 2) SCL: A maximum short circuit rating for all equipment is laid down in the distribution code [6]. A short circuit calculation is carried out to ensure that this constraint is not exceeded as the level of installed capacity increases. The SCL is highest at the transmission system bus. Buses close to this bus may find their capacity limited as a result. The constraint is given by SCL
SCL
(4)
where SCL is the SCL at the transmission substation busbar, is the highest current that switchgear can safely and SCL break under fault conditions. The contribution of increasing levels of generation at each bus to SCL is determined by short circuit analysis. The SCL contributions of generation
1641
at each bus are combined and formalized into an algebraic equation, shown as follows: SCL
(5)
where is the dependency of the SCL at the transmission station to power injections at bus , i.e., the slope of the SCL versus power injection characteristic of the th bus, as shown in is the power injection at the th bus, and is the Fig. 5. initial SCL at the transmission bus with no generation present. Equation (5) is a very accurate model of the short circuit characteristics when an intelligent choice of the range for calculation of the slopes is made. The optimization is initially calculated with respect to the constraints as outlined. The initial range for calculation of the slopes is selected such that the fixed limit constraints are satisfied. The allocation calculated by the optimization algorithm for each bus is then used as the new range over which to recalculate the slopes for the SCL constraint, thus improving the accuracy of the dependencies. When the optimization is recalculated with the new dependencies, it has been found that it gives a very similar answer to the initial allocation. Thus, it is found that the linear approximation is adequate but that the iteration process is worthwhile, as it does not add considerably to the calculation time and ensures an accurate determination of the optimal allocation. The SCL used at the transmission station is the winter peak level given in the transmission system operators (TSOs) forecast statement [7]. C. SCR The SCR is the ratio of generator power MW) at each MVA). It gives an indication bus to the SCL at each bus SCL of the voltage dip experienced near the generation in the event of a feeder outage. The connection of induction generators to high impedance circuits may lead to voltage instability problems if the SCR is not kept within acceptable limits [8]. The dip in wind farm terminal voltage that results from a fault can result in an acceleration of the induction generator, leading to overspeed. If the speed is increased to a level above the critical value, the generator will accelerate out of control. This will lead to voltage collapse as the induction generator absorbs more reactive power [9]. If the SCR is small enough, the transient voltage dip will be limited, and the system will remain stable. The contribution of other generators to the SCL is considered, as it may be significant, depending on the proximity of the bus. The phase angle at the busbar is omitted as an extra margin of safety. It could be included and the allowable ratio set to a lower value, such as 6%. A value of 10% is largely in line with values used by other utilities and is in line with the value recommended in the European standard EN50160 [10] SCL
%
(6)
where SCL refers to the SCL at the th bus, and is the power factor at the generator. A base value for the SCL at the th bus is calculated with no generation present on the network . The contribution, if any, of generation at other buses to this level is then calculated, allowing the short circuit characteristic
1642
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 3, AUGUST 2005
of each bus to be formulated into an algebraic equation. The SCL at the th bus is given by SCL
(7)
tions at bus . This analysis is carried out under minimum load conditions, as this is the worst-case scenario for voltage rise. Both the standby and normal forward-feed conditions are considered. There is usually more than one possible standby feeding arrangement, but the most severe feeding condition is usually readily identifiable.
Subbing (7) into (6) E. Energy Resource and Customer Initiatives Constraint (8) For this constraint, the SCL used at the transmission station is the summer night valley level. D. Voltage Rise Effect From [11], the voltage at the generator is given as (9) where and are the active and reactive power of the generator, respectively, is the impedance of the line, and are active and reactive power at the bus, and and are the voltages at the generator and bus, respectively. Thus, it can be seen that the generator voltage will be the load/bus voltage plus some value related to the impedance of the line and the power flows along that line. It is evident that the larger the impedance and power flow, the larger the voltage rise. The increased active power flows on the distribution network have a large impact on the voltage level because the resistive element of the lines on distribution networks are higher than other lines. This leads to an X/R ratio of approximately 1 rather than a more typical value of 5 on transmission networks. The voltage must be kept within standard limits at each bus (10) where and refer to the minimum and maximum voltage limits at the th bus. The relationship between voltage and power injections at each bus is determined. As MWs are added at each bus, the voltage rises. Increasing levels of generation are added incrementally at each bus in turn, and load flow analysis is carried out to determine a voltage versus active power characteristic for each bus. Next, the interdependence of the bus voltage levels is examined. Once again, increasing levels of generation are added incrementally at each bus, but now the voltage level at every other bus is examined. Thus, characteristics are determined for voltage levels at each bus due to generation at all other buses. By combination of these characteristics, the voltage constraint may be formalized into algebraic equations for each bus, as shown in the following:
(11) where is the dependency of the voltage level at bus on power injections at bus , i.e., the slope of the voltage versus power injection characteristic of the th bus. refers to the inirefers tial voltage level at the th bus with no generation, and to the dependency of the voltage level at bus on power injec-
All of the network buses may not have a substantial energy resource available, if any. This can be factored in by adding an inequality constraint for each bus, limiting its allocation to the available energy resource at that bus. On the other hand, a customer may wish to install a generator to serve its load and export any surplus generation onto the network. This constraint allows the effect of proposed generation at individual buses on the overall EG capacity to be assessed. These constraints are formalized in (12), with the allocation to each bus being at least equal to any existing generation and at most equal to the available energy resource at each bus (12) where and are the available energy resource and any existing generation at the th bus, respectively. A suboptimal allocation results when generation is allocated on a first-come first-serve basis, resulting in network sterilization. The voltage levels and SCLs at the buses are interdependent; thus, some buses will have more influence on the voltage and SCLs than others, as illustrated in Figs. 3 and 5. The formalization of these interdependent constraints and the independent constraints along with the use of linear programming ensures that this interdependence is taken into account, and the capacity is maximized while satisfying all the constraints. A flowchart of the methodology is shown in Fig. 1. There is a varying level of interdependence between the buses, which is dependent on a number of factors: • proximity to transmission station; • relative distance between buses; • type of conductor. The proximity of the bus to the transmission station has a large bearing on the sensitivity of that bus to changes in voltage. As would be expected, the closer a bus is to the transmission infeed, the higher the voltage at that bus. Also, these buses tend to be less sensitive to generation at other buses and also less sensitive to generation at the bus itself. The converse is also true. The distance between the buses affects their level of interdependence, and perhaps more importantly from an allocation point of view, the relative distance will define each of their relationships. The type of conductor also has a bearing on these relationships. For example, a 630XLPE cable has a lower impedance than a 100SCA overhead line, and therefore, the voltage rise effect due to generation at other buses will be much reduced. In terms of SCLs, the closer a bus is to the transmission station, the higher the SCL is. Once again, the relative distance between the buses will determine their interdependence, as will the position of normally open points on the network. The impedance of the conductors will affect the SCL, and the typically high impedance distribution networks will result in a low SCL.
KEANE AND O’MALLEY: OPTIMAL ALLOCATION OF EG ON DISTRIBUTION NETWORKS
• • •
1643
direct feed to 110 kV/38 kV or 110 kV/MV station; direct feed to 38 kV/MV station; tee connection onto existing line.
If the capacity of a project is less than 5 MW, connection to the MV network may be possible and is typically cheaper. Tee connections are not generally permitted, as they can lead to a decrease in the security of the system. The structure of the network modeled is radial, with normally open points separating sections of the network. Most EG in Ireland is located in rural areas. In Ireland, the maximum number of buses that are fed along one line from the 110-kV station (i.e., interdependent) is typically four or five. There may, of course, be more stations fed from the same 110-kV station; however, these are separated into independent sections by normally open points. In such cases, they all have the common constraint of the SCL and transformer capacity at the 110-kV station. All generators are assumed to be fixed speed induction generators. They are connected in parallel at each bus with a power factor correction capacitor bank to maintain the power factor at 0.95 as required by the DNO. All necessary network data were obtained from ESB Networks, who is the DNO in Ireland [12].
IV. RESULTS AND DISCUSSION Fig. 1. Optimal allocation methodology.
III. NETWORK STRUCTURE AND PARAMETERS In Ireland, EG is typically connected at 38 kV or MV (10 or 20 kV). It is a mixture of renewable and conventional energy sources—wind, small scale hydro, combined heat and power (CHP), photovoltaic, and landfill gas. The vast majority of applications for connections coming into the DNO are wind farm projects. These wind farms are typically located in remote areas, where there is very little existing network and also very little demand for electricity. As of January 2004, there was approximately 500 MW of generation either connected to the Irish distribution network or with signed connection agreements, of which approximately 70% is wind [7]. In Ireland, voltage rise tends to be the dominant constraint and quite often the critical factor when offering capacity on the network. This is due to the existing network in Ireland, which outside Dublin is typically a weak rural network with a large amount of conductor. A weak network means a network with a low SCL or fault level. There is approximately 159 000 km of distribution network in Ireland. These “long stringy” networks are plainly not equipped to support a large amount of generation. Due to these high impedance lines, the sending voltage at the transmission station is usually set to a higher than nominal value (i.e., 41 kV at 38 kV) to ensure that the voltage is within standard at the end of the line when no generation is present, thus maximizing the load that can be supplied from these networks. Lowering this sending voltage would facilitate increased penetration of EG but at the expense of the DNO’s customers who would experience a decrease in the quality of their supply due to the increased occurrence of nonstandard voltage levels. Within each voltage level, there are a number of options for connecting generation into the system. These are as follows:
The methodology was tested on a number of sample sections from the Irish distribution network. Results are given here for a tail fed 38/110 kV station with five buses shown in Fig. 2. The methodology has been tested on larger networks. This section is chosen to illustrate the results, as it demonstrates the potential for network sterilization very well. The most severe case for voltage rise is the standby feeding condition shown in Fig. 2 with the out-of-service line in grey. The individual voltage sensitivity characteristics were calculated and are shown in Fig. 3. From Fig. 3, it can be seen that bus A is the least sensitive to power injections, and therefore, based on this constraint, only will have the largest allocation of generation. It is the least sensitive to power injections due to its proximity to the 110-kV station. It can also be seen that buses B and C have very similar profiles. They are separated from the rest of the network by line 5, which is 41 km long, which explains the significantly lower (kV/MW) in (11) are voltage at these buses. The values for determined from this graph. They are shown in Table I as the are also shown in this diagonal elements. The values for table. The interdependence of voltage was also calculated, resulting in a graph for each bus. The graph for bus C is shown in Fig. 4 It can be seen that buses C and B have a different relationship than the other buses; this is due, once again, to the line connecting buses A and B, which is 41 km long. The values for (kV/MW) in (11) are determined from these graphs and are shown as the off-diagonal elements in Table I. The individual sensitivity of the SCL at the 38/110 kV station to power injections at individual buses is also calculated, resulting in Fig. 5. As would be expected, the effect of the buses on the SCL at the 110-kV station is dependent on the distance from the station. Relative to the other buses, bus A is very close to the 110-kV station and, as such, has a much bigger impact on the SCL at that
1644
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 3, AUGUST 2005
Fig. 2. 38-kV five-bus radial distribution network diagram.
Fig. 3.
Individual bus voltage sensitivities to power injections at buses. Fig. 4.
station. The values for (kA/MW) used in (5) are shown in Table II. The SCL contributions of generation at other buses have a significant bearing on the SCR constraint. As can be seen in Fig. 6, some of the buses have a negligible impact on the SCL at bus B; this pattern is repeated for all the buses. In this case, only the contribution of bus C is significant. The values of (MVA/MW) in (8) are shown in Table III. The diagonal elements are blank, as the contribution of generation at any bus to the SCR at that bus should not be included in the calculation. Two scenarios are evaluated to demonstrate the performance of the methodology and the effect of network sterilization.
Dependence of buses on power injections at bus C. TABLE I VOLTAGE INTERDEPENDENCY
Scenario 2: 7.6 MW of Customer Generation at Bus E: In this scenario, 7.6 MW of customer generation is allocated to bus from (12) is set to 7.6 MW. By application E, i.e.,
KEANE AND O’MALLEY: OPTIMAL ALLOCATION OF EG ON DISTRIBUTION NETWORKS
1645
TABLE III SCL INTERDEPENDENCY
TABLE IV OPTIMAL ALLOCATION
Fig. 5. SCL at 38/110 kV station versus power injections at individual buses.
SCL
TABLE II DEPENDENCY
Fig. 6. SCL contributions of other buses to bus B.
of a linear programming algorithm using the calculated sensitivity values for the constraints formalized in (2), (3), (5), (8), (11), and (12), the maximum allocation is determined after two iterations to be 11.20 MW, as shown in Table IV. Scenario 2: No Generation Preallocated: In this scenario, no generation is allocated prior to optimization. The methodology is applied again, and the total allocation is determined after two iterations to be 22.74 MW, with the generation allocated as shown in Table IV. The 22.74 MW allocation represents a further 11.54 MW of generation capacity, which is approximately a 100% increase on scenario 1. This is an extreme situation but is as likely as any other allocation when applications are dealt with on a first-
come first-serve basis. Using the energy resource and customer initiative constraint, the sterilizing effect of generation at bus E has been shown. This constraint can be used to check the effect of existing or proposed generation at all the buses on the total EG capacity. The methodology works on all sections of network that are operated radially. The severity of network sterilization varies with the level of interdependence between the buses. The network section modeled in the paper has a high level of interdependence, since under standby feeding conditions, all five buses are connected to the transmission bus by a single line and is typical of Irish rural networks, thereby illustrating clearly the effect of network sterilization on potential EG capacity. With larger systems, there are typically more lines connecting the buses to the transmission bus, with normally open points separating sections. This reduces the overall interdependence of the voltage levels—but not of the SCLs. Thus, the potential for network sterilization is reduced. As a result, when tested on larger systems, network sterilization, while still significant, is not as severe as in the illustrative case shown in this paper. This methodology determines the optimal allocation from a capital investment point of view and does not take account of losses. Losses are an operational issue and, therefore, cannot be easily integrated into the methodology. Over a 10 to 15 year period, which is the typical lifetime of a wind farm project, some of the network conditions will change, possibly causing constraint breaches. First, it is assumed that the load will grow. This load growth will only serve to lessen the voltage rise effect. It will increase flows along the lines and, therefore, losses. In addition, the SCLs will possibly increase, which may cause problems with equipment ratings. The predicted SCLs in the TSO’s forecast statement enable an accurate determination of future network conditions. The mix of generator technologies may have a significant bearing on some of the constraints. It has been assumed that all the generators are fixed-speed induction machines. A different mix of technologies could result in a different overall short circuit contribution, thus altering some of the constraint characteristics [11].
1646
IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 3, AUGUST 2005
Existing distribution networks are passive, in that they were designed and built purely for the delivery of electricity to the customer. The introduction of EG is changing the characteristics of the distribution network. It has led to increased and bidirectional active and reactive power flows, along with wider variation in voltage levels, both of which affect the operation of equipment on the network and the level of losses. As such, EG can defer planned network reinforcement but may also require network reinforcement of both the distribution and transmission network. It has been assumed that the transmission system can accommodate the high levels of EG. An investigation into the validity of this assumption is also required. A significant penetration of EG, which in Ireland is almost exclusively from renewable energy sources, can be achieved with no investment in the distribution network other than the cost of installing the generator. V. CONCLUSION A new methodology has been proposed for the allocation of EG on distribution networks. It has been implemented using linear programming and tested on a sample section of the Irish distribution network. Distribution networks across the world generally have the same structure and operating principles, and thus, this methodology should be applicable to all such distribution networks. If renewable energy targets are to be met in a cost-effective manner, best use of the existing distribution network must be made. From the results presented here, it can be seen that the allocation of generation capacity on a first-come first-serve basis does not achieve this. The methodology described provides a tool that DNOs can use to aid in planning the distribution network. Severe instances of network sterilization will continue to occur if EG proposals are dealt with on a case-by-case basis. It is evident that while EG may pose significant technical problems, these can be overcome to the advantage of society.
[2] A. Wallace and G. Harrison, “Planning for optimal accommodation of dispersed generation in distribution networks,” in Proc. CIRED 17th Int. Conf. Elect. Distrib., Barcelona, Spain, May 2003. [3] B. Kuri, M. Redfern, and F. Li, “Optimization of rating and positioning of dispersed generation with minimum network disruption,” in Proc. IEEE Power Eng. Soc. Gen. Meeting, Denver, CO, Jun. 2004, pp. 2074–2078. [4] W. El-Khaltam, K. Bhattacharya, Y. Hegazy, and M. M. A. Salama, “Optimal investment planning for distributed generation in a competitive electricity market,” IEEE Trans. Power Syst., vol. 19, no. 3, pp. 1674–1684, Aug. 2004. [5] The promotion of electricity produced from renewable energy sources in the internal electricity market [Online]. Available: http://www.europa.eu.int [6] Irish Distribution Code. ESB Networks Standard vl.2, 2002. [7] ESB National Grid Forecast Statement 2004–2010 [Online]. Available: http://www.eirgrid.com/ [8] I. Holdsworth, N. Jenkins, and G. Strbac, “Electrical stability of large offshore wind farms,” in Proc. 7th Int. Conf. AC-DC Transm., London, U.K., Nov. 2001, pp. 156–161. [9] V. Akhmatov, H. Knudsen, M. Bruntt, A. H. Nielsen, J. K. Pedersen, and N. K. Poulsen, “A dynamic stability limit of grid connected induction generators,” in Proc. IASTED Int. Conf. Power Energy Syst., Marabella, Spain, 2000, pp. 235–244. [10] Voltage Characteristics of Electricity Supplied by Public Distribution Systems. CENELEC European Committee for Electrotechical Standardization Standard EN50 160, 1994. [11] N. Jenkins, R. Allan, P. Crossley, D. Kirschen, and G. Strbac, Embedded Generation, ser. Power and Energy. London, U.K.: IEE, 2000. [12] ESB Networks Infrastructure [Online]. Available: http://www.esb.ie/esbnetworks/
Andrew Keane (S’04) received the B.E. degree in electrical engineering from University College Dublin, Dublin, Ireland, in 2003. He is currently working toward the Ph.D. degree in the Electricity Research Centre, University College Dublin. His research interests are in power systems planning, embedded generation, and distribution networks.
ACKNOWLEDGMENT The authors would like to thank I. Codd, D. Phelan, and T. Walsh of ESB Networks and their colleagues in the ERC for their help with this work. REFERENCES [1] S. Liew and G. Strbac, “Maximizing penetration of wind generation on existing distribution networks,” Proc. Inst. Elect. Eng., Gener., Transm., Distrib., vol. 149, no. 3, pp. 256–262, May 2003.
Mark O’Malley (S’86–M’87–SM’96) received the B.E. and Ph.D. degrees from University College Dublin, Dublin, Ireland, in 1983 and 1987, respectively. He is currently a Professor in University College Dublin and the Director of the Electricity Research Centre, with research interests in power systems, control theory, and biomedical engineering.
