Control For Energy and Sustainability

EPSRC Programme Grant

Project PS-C: Risk Profile of Future Power Systems

Manager: Goran Strbac

Investigators: Goran Strbac

Research Staff: Rodrigo Moreno (Research Associate)

Collaborators: Richard Vinter

Sponsors: GE Energy, National Grid

Start date: 01/10/2010

Linked Projects: UT-A and UT-B

Summary. In order to deliver Government targets on renewables, it is expected that about 10GW of wind generation will be connected in Scotland and about 20GW of offshore wind generation in England. The projected investment in network reinforcement is at the level of the asset value of the entire National Grid. However, present deterministic network design standards developed in the 1950s prescribe the level of network redundancy, and do not explicitly consider operational measures that potentially could substitute for network reinforcement and facilitate a more cost effective connection of renewable generation. On the other hand, the application of new emerging control technologies, due to their reliance on communication systems and uncertainty associated with availability of responsive demand, may potentially pose additional operating risks, as these may fail to operate correctly due to inappropriate settings, faults, corrupted data, communication delay, etc. This risk needs to be understood, quantified, and managed to ensure that the integrity and security of the overall system is not compromised. Efficiently managing risks associated with the increased application of operational measures is critical for the future development and implementation of alternative network control and design philosophies, which is the key focus of this project.

Given that future networks will operate with significantly less redundancy, consideration of such High Impact Low Probability (HILP) events will need to become an essential component of network operation and design strategies and hence radically new risk assessment based techniques are required to facilitate development of cost effective resilient networks.

The aim of this project is to assess strategies for protecting large networks from catastrophic failures and to quantify their resilience. Here, optimization techniques from Project UT-A and analytical and accelerated Monte Carlo simulation tools, developed in the Project UT-B for investigating the propagation of faults in random networks, will have important roles.

Furthermore, it is important to re-examine the validity of conventional deterministic scheduling methodologies when applied to systems with significant contribution of wind generation, which was a major part of the investigations last year.

Current Status:

Risk modelling of transmission network operation and investment

We have carried out comprehensive analyses that demonstrate that any attempt to fix a single deterministic rule for operating the transmission networks, as in the present deterministic standards, will lead to potentially significant inefficiencies. We also demonstrated that various operational measures (such as generation and demand response) can be very effectively used to release additional network capacity. The results clearly demonstrate that the probabilistic approach provides the basis for the development of future network operation and design standards that would maximise the value of transmission network to users, enhance the utilisation of existing networks, foster the entrance of new technologies that may complement and provide alternative network reinforcement and hence facilitate efficient integration of renewable generation. Furthermore, our recent progresses show that the limited amount of capacity released to network users by the deterministic standards in operational timescales drives unnecessary costly network investment that do not necessarily provides higher levels of security. Moreover, our work demonstrates that deterministic-based investment can drive major risks in system operation, incorrect network designs and higher levels of wind curtailment. We showed that these problems can be corrected by using a probabilistic approach for network security and that the GB system will be benefited if probabilistic standards are adopted. We have expanded the above work to deal with the operation of various FACTS devices (like series compensators and phase-shifting transformers) under uncertain wind generation in order to: (i) release latent network capacity of the existing network assets while co-ordinating AC and DC power flow control, and (ii) determine robust set points of flexible devices against various potential wind generation outputs. Hence the proposed probabilistic network model optimally coordinates available preventive and corrective control actions from flexible network technologies, generation and demand, including generation dispatch and reserve commitments, post-contingency exercise of generation reserves, demand curtailment and generation intertrips, while explicitly considering the likelihood of post-contingency events and wind uncertainty. We aim to compare the efficiency and reliability performance of network operation under two different strategies to control flexible network devices: preventive-only mode with deterministic security (consideration of N-1 and N-2 events only), and combined preventive and corrective mode with probabilistic security. We demonstrate that the preventive control-only approach would limit the ability of the system to integrate renewable generation. In contrast, through the application of corrective control, an optimal trade-off between accommodating uncertainty in wind generation outputs and dealing with system outages can be determined. We have also demonstrated that the flexibility of a transmission network with HVDC and FACTS devices may be significantly restricted when applied in preventive mode, particularly when uncertainty in wind generation output is explicitly considered. Additionally, we demonstrate that an appropriate control strategy of flexible network devices could facilitate reduction in reserve holding levels and therefore improve the efficiency of the system operation and simultaneously enhance the ability of the transmission network to accommodate increased amounts of intermittent wind generation.

Risk modelling of extreme events

Given the above approaches, future networks might operate with significantly less redundancy due to the use of corrective control and thus consideration of malfunctions associated with communication and control infrastructure will need to become an essential component of network operation and design strategies. For instance, the increased use of corrective control to manage post-fault transmission overloads in the form of generation and demand curtailments, has raised concerns related to angular and voltage stability of transmission systems given the reliability level of the communication infrastructure on which the control relies. In this framework, we have demonstrated through several dynamic simulations that the risks of communication delays and the consequent dynamic instability risks can be limited by using complementary assistive measures. Indeed, in order to deal with angular and voltage stability problems that can be caused by malfunctions associated with communication and control infrastructure necessary to manage post-fault transmission overloads in the England-Scotland interconnector in Great Britain (GB), we have proposed three Transient Assistive Measures (TAM). They supplement the generation and demand curtailment (intertrip) actions, to ensure stability during the immediate post-fault period. The proposed methods have been assessed on a dynamic system model representing the GB transmission network. A 2 GW High-Voltage Direct-Current Link with Current Source Converters (CSC-HVDC) was modelled and incorporated into the full dynamic model of the GB system to represent the planned western bootstrap. The proposed TAM are (i) a HVDC Power Oscillation Damping (HVDC-POD) controller, (ii) an adaption of the HVDC power order, and (iii) switchable reactive support devices. We have demonstrated, on the basis of studies performed on the GB transmission network, that without the proposed transient assistive measures, the benefit of corrective control is compromised and the risk of transient instability due to possible communication delays (in inter-tripping generation/demand), is significantly higher.

Additionally, we are undertaking fundamental research on quantifying the risk associated with rare events in power systems. This involves the application of advanced Monte Carlo methods to increasingly complex problems in the field of power systems. In order of complexity, the systems under study range from stationary (e.g. generator availabilities) to non-stationary memoryless systems (e.g. generation adequacy with seasonal load variations) to systems with integral constraints (storage) or ‘smart’ agents that anticipate future states. When existing methods in the power system literature are not suitable for the complexity of the problem at hand, we attempt to extend the methods through original work (see also project UT-B) or by adapting methods that have been developed in different disciplines, notably within the field of physical chemistry. In tandem with the simulation approach, we aim to obtain analytical expressions for relevant risk indices. For all but the simplest problems these analytical results cannot be computed. However, they can provide interesting limit cases that may be used for the testing of simulation methods and to provide error bounds on results. Further work started involves detailed assessments of network integrity under various disturbances considering cascading outages and failures of secondary systems and inadequate demand response. This will be used to assess the risk profiles associated with particular operational measures, focusing on inter-tripping schemes of various complexities. We use algorithms and tools developed at Imperial in conjunctions with a range of commercial software tools for reliability and risks analysis. We investigate how alternative statistical estimators can be used to measure propagation of outages, assess the proximity of blackouts, and analyse, measure and manage the distributions of blackout size associated with alternative network operation and design philosophies. We shall apply the both new and previously developed system risk assessment methods, to quantify reliability performance of transmission networks that employ generation and demand inter-tripping schemes.

Risk modelling of system reserves

Another strand of work involved the analysis of operating reserve requirements and the need for flexibility at high levels of penetration of intermittent renewable generation, particularly considering the value of storage in supporting grid balancing. In traditional scheduling methodologies, deterministic wind and demand time series would be used as inputs to the cost minimisation algorithm, while reserve requirements would be specified as exogenous criteria. Such deterministic methods rely on defined exogenous reserve criteria that are necessarily static, whereas the economically efficient amount of reserve would vary according to the dynamic cost of providing it, and the risk and cost of unserved energy. Such pitfalls can be avoided by simulating system operation using stochastic scheduling that can account for the uncertainties explicitly, by providing the commitment algorithm with a range of possible outcomes (e.g. wind realisations) that are weighted according to their probability of occurrence. We have enhanced our methodology for stochastic unit commitment to include evaluation of storage technologies. We have demonstrated that the stochastic value of storage in system with wind generation can be 50% higher than what a deterministic approach would indicate. In a related subject, we have started research and modelling development to study the impacts of increased volumes of generation reserves on releasing network capacity and on network investment over a multi-busbar network. This will be particularly important to determine the efficient levels of network utilisation for scheduled energy transfers when also considering the redundancy needed to potentially increase these transfers and access (standing) reserves under several realisations of wind outputs. Our early results demonstrates that the need to access reserves may restrict energy transfers, increase constraints costs and drive further network investment. It is also shown that accessing remote standing reserve rather than local spinning reserve may allow SOs to minimise wind curtailments.

Risk modelling of co-optimised system reserves and transmission investment

Expanding on the above stochastic scheduling to optimise reserve services, we have developed a planning model that co-optimises such system reserves with transmission investment. We fundamentally balances costs of network reinforcement against the (expected) costs of network operation (i.e. constraints/congestion, generation reserves, and network losses) and unsupplied demand, minimising the overall cost of network expansion, operation and security of supply under uncertain forecasts of renewable generation. In operational timescales, we determine a generation dispatch that optimally deals with the expected scenario in a given hour and that is sufficiently robust to also deal with those unexpected (scenarios are derived from an array of potential outputs of different sources of renewable generation, e.g. wind, solar etc, foreseen in advance, usually 4 hours ahead). Within this framework, model may constrain network transfers to secure network operation by leaving network headroom unutilised to access remote generation reserves under unexpected events. For example, if a exporting area has important amounts of flexible (spinning and fast-start) and economic generation, and a importing area has high outputs of wind generation (which can be accompanied by the generation of conventional units), transfers between these areas can be constrained in the expected condition in order to leave network headroom unutilised and thus be able to balance unexpected large wind drops in the importing area with the support of reserves in the exporting area. Hence, transmission network capacity can be optimally allocated between energy transfers (i.e. those derived from scheduled generation outputs that supply demand under expected operating conditions) and reserve transfers (i.e. those derived from generation outputs that supply demand under an array of potential wind outputs, different than those expected) and this can affect network investment needed to transfer these two services. Although we have demonstrated that at a national GB level it may not be justified to build further transmission to access reserve (i.e. standing reserve in England is sufficient to deal with local wind uncertainties), this may be of importance at an international level and thus sharing of network capacity may play an important role in the utilisation and construction of new cross-border interconnectors.

The PS-C project has led to the establishment, from November 2011, of regular fortnightly meetings on risk assessment methods for power systems. These work discussions are open for all researchers from Imperial’s Control and Power Group and provide a platform for dissemination of knowledge between this project and other related activities in the group.


[CMS14] Y. Chen, R. Moreno and G. Strbac, OPtimal coordination of HVAC/DC control to enhance the ability of the system ti accommodate intermittent renewable generation, IEEE Transactions on Power Systems, 2014

[MPS13] R.Moreno, D.Pudjianto and G.Strbac, Transmission Network Investment with Probabilistic Security and Corrective Control, IEEE Transactions on Power Systems, Vol. 28, Issue 4, pp3935-3944, November 2013.

[MPS12] R. Moreno, D. Pudjianto and G. Strbac, Integrated reliability and cost-benefit based standards for transmission network operation, Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, Volume 226, Issue 1, pp75-87, 2012

[SS11b] A.Sturt and G.Strbac, Value of stochastic reserve policies in low-carbon power systems, Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, Volume 226, Issue 1, pp51-64, 2012

[SS11c] A.Sturt and G.Strbac, Time-series modelling of power output for large-scale wind fleets, Wind Energy, Volume 14, Issue 8, pp953-966, 2011

[SS11a] A.Sturt and G.Strbac, Efficient stochastic scheduling for simulation of wind-integrated power systems, IEEE Transactions on Power Systems, Volume 27, Issue 1, pp323-334, 2011

[MPS11] M. Castro, D.Pudjianto, P. Djapic and G.Strbac, Reliability-driven transmission investment in systems with wind generation, IET Generation, Transmission and Distribution, Volume 5, Issue 8, pp 850-859, August 2011

[SMPC11] G.Strbac, R.Moreno, D.Pudjianto and M.Castro, Towards a risk-based network operation and design standards, Proceedings of the IEEE PES 2011 General Meeting, Detroit, USA, July 2011

[SS11] A.Sturt and G.Strbac, A time series model for the aggregate GB wind output circa 2030, Proceedings of the IET Renewable Power Generation Conference, 2011

[MSPMB10] R.Moreno, G.Strbac, F.Porrua, S.Mocarquer and B.Bezerra, Making room for the boom - new regulatory avenues and network infrastructure changes for accomodating renewables, IEEE Power and Energy Magazine, Vol 8, issue 5, pp 36-46, Dec 2010