(488d) A Supervisory Control Framework for Fault-Tolerant Dispatch of Distributed Energy Resources

The convergence of competition in the power industry with the arrival of environmentally friendly micro-turbines, fuel cells, photovoltaics, small wind turbines, and other advanced distributed power technologies has sparked strong interest in distributed power generation [1]. This convergence of policy and technology promises to transform the electric power system from one relying primarily on central generation to one in which Distributed Energy Resources (DERs) provide most of the power needed. The resulting major improvement in power reliability, quality and efficiency; and the greater flexibility to respond to changing energy needs could save billions of dollars now lost each year because of power disruptions. These considerations have motivated a significant and growing body of research work on the modeling, analysis and control of DERs (e.g., see [2]-[6] for some results and references).

An important challenge with the evolution from central to distributed generation -- especially in light of the increasing number and diversity of DERs spread over the grid -- is the development of integrated monitoring, control and coordination strategies that would ensure optimal dispatch of the energy resources to provide highly reliable services under various disturbance and failure scenarios. This is significant given the fact that the distributed power market is driven by the need for reliable high-quality power, and the substantial impact that local disruptions in power flow can have when the DERs are integrated to support grid operations. In this context, process control and monitoring are critical tools for ensuring that the supply of DER networks matches demand whilst ensuring the stable and sustainable operation of each DER unit. However, the stability of DER networks and how they perform in the presence of local unit faults and failures has not been rigorously assessed. The need for these networks to function in the presence of faults is critical as to be a viable option for electricity generation, and these DER networks must be able to perform at the same standards of quality, stability, and robustness as current electrical generation. Meeting this challenge calls for a hierarchical monitoring and fault-tolerant control architecture where distributed monitoring and fault-tolerant control systems that perform timely fault identification and control system reconfiguration at the local level are integrated with a higher-level supervisor that optimally coordinates power generation in the event that local fault rectification is not possible. This hierarchy enables the timely mitigation of faults with supervisory oversight.

In prior work, fault-tolerant control of DERs has been studied in the context of a small-scale network of solid oxide fuel cells, where the focus has been on the detection and recovery from faults at the local level without supervisory oversight [7]. DER health monitoring was conducted through use of an alarm threshold on a properly designed observer-based output residual. Exploiting inherent actuator redundancy in the component subsystems, a number of stabilizing controller configurations were designed, and a methodology for active switching between the configurations in the event of a threshold breach was developed. The results were subsequently extended in [8] to include explicit fault estimation and performance-based fault accommodation capabilities in the local fault-tolerant control system design. In both studies, the system under consideration was only looked at from the local level with no supervisory communication. This negated the need for any logic or control at anything other than the local level which would be necessary in the event of a severe local unit fault in order to bring the local unit to an adequate safe parking state or initiate a safe unit shutdown. When all locally feasible corrective measures cannot rectify the local fault, an intervention by the supervisor becomes necessary. The local fault diagnosis information in this case must be communicated to the supervisor to be able to reconfigure the power load distribution for the entire network and compensate for the loss in power supply caused by the local failures.

The objective of this contribution is to develop an optimization-based supervisory control framework for fault-tolerant dispatch of DERs at the network level, which builds on the local fault identification and accommodation strategies developed earlier. The proposed framework extends the previous results in two important directions. One direction is incorporating logic on the local unit side to communicate unit fault scenarios at the local level with the supervisory controller. This would be in the case of a unit being hindered to the point where operation at a level deemed necessary by the supervisor is no longer a feasible option. The other direction is incorporating a supervisory controller to optimally allocate load to each unit on the network with logic in place to handle fault situations at the local level. The supervisory controller is designed using model predictive control (MPC) techniques which allow optimization of resource dispatch in the control action while incorporating various operational constraints on the local units. In the event of a local unit fault, the supervisor would communicate with the local unit to determine a sustainable operating condition, whether that be complete unit shut down or some diminished safe parking state. At that time, the supervisor would need to reallocate the load deficit caused by the faulted unit in an optimal way across the remaining units of the system. To this end, a finite-horizon constrained optimization problem is solved on-line to determine the optimal future power references for the functional DERs. Only the first part of the reference trajectories are transmitted to the local control systems, and the optimization is repeated when new measurements are sent to the supervisor. The receding horizon implementation strategy provides robustness against intrinsic uncertainties in the power generation capacity of certain DERs (e.g., renewables subject to environmental variations and intermittencies), as well as the uncertainty associated with long-term future power demand. The proposed framework is illustrated using a multi-source hybrid renewable DER system consisting of a photovoltaic energy conversion system, a wind energy conversion system, a fuel cell, an electrolyzer and a battery bank.

References:

[1] R. H. Lasseter, "Microgrids and distributed generation", Journal of Energy Engineering, 133, 144â??149, 2007.

[2] M. Deshmukh and S. Deshmukh, â??Modeling of hybrid renewable energy systemsâ?, Renewable and Sustainable Energy Reviews, 12, 235 â??249, 2008.

[3] S. Barsali, M. Ceraolo, P. Pelacchi, and D. Poli, â??Control techniques of dispersed generators to improve the continuity of electricity supply,â? Proceedings of IEEE Power Engineering Society Winter Meeting, pp. 789â??794, 2002.

[4] M. Marei, E. El-Saadany, and M. Salama, â??A novel control algorithm for the DG interface to mitigate power quality problems,â? IEEE Transactions on Power Delivery, 19, 1384-1392, 2004.

[5] Y. Sun, S. Ghantasala and N. H. El-Farra, ``Networked Control of Distributed Energy Resources: Application to Solid Oxide Fuel Cells," Industrial Engineering and Chemistry Research, 48, 9590-9602, 2009.

[6] W. Qi, J. Liu, X. Chen, and P. D. Christofides, â??Supervisory predictive control of stand-alone wind/solar energy generation systems,â? IEEE Transactions on Control Systems Technology, 19, 199 â?? 207, 2011.

[7] Y. Sun, S. Ghantasala and N. H. El-Farra, â??Monitoring and fault-tolerant control of distributed power generation: Application to solid oxide fuel cells,â? Proceedings of American Control Conference, pp.448-453, 2010.

[8] J. T. Allen and N. H. El-Farra, â??A model-based framework for fault estimation and accommodation applied to distributed energy resources,â? Renewable Energy, in press.