Framework for comparing power system reliability criteria Evelyn Heylen Prof. Geert Deconinck Prof. Dirk Van Hertem Durham Risk and Reliability modelling for Energy Systems day November 12th, 2014 Introduction Current situation in power system reliability management • Deterministic N-1 criterion with various shortcomings • Major evolutions in the power system • Increasing uncertainties Probabilistic reliability management • Take into account probabilities • Could tackle shortcomings of N-1 • Many academic references • Not fully used in practice Amongst others due to lack of quantified benefits Framework for Comparing power system reliability criteria 2 Outline • Framework for comparing power system reliability criteria o o o o o Overview General schematic of the implementation Implementation of deterministic reliability management module Assumptions Comparison of deterministic criteria • Conclusion 3 Outline • Framework for comparing power system reliability criteria o o o o o Overview General schematic of the implementation Implementation of deterministic reliability management module Assumptions Comparison of deterministic criteria • Conclusion 4 Framework for comparing reliability criteria Objectives of the framework: 1. Quantification of performance of various power system reliability criteria and their management 2. Comparison of performance 3. Identifying alternative reliability criteria 5 General schematic 6 Deterministic reliability management module 1. N-1 criterion: ‘System should be able to withstand at all times the loss of any one of its main elements (lines, transformers, generators, etc.) without significant degradation of service quality.’ 1 2 2. State enumeration: • Run power flow • Check for operational limit violations 3. Decision = balance reliability and cost 4. Reliability actions • Corrective actions • Preventive actions 7 4 3 Reliability management modules Reliability assessment methods Reliability criteria Reliability control Analytical contingency enumeration N-k Preventive/corrective control Event tree/fault tree analysis Optimization Random sampling (Monte Carlo) Markov analysis 8 Limits on reliability indicators Optimization and limits on reliability indicators Asset management System development Reliability management modules Reliability assessment methods Reliability criteria Reliability control Analytical contingency enumeration N-k (i.e. N-0 and N-1) Preventive/corrective control Event tree/fault tree analysis Optimization Random sampling (Monte Carlo) Markov analysis 9 Limits on reliability indicators Optimization and limits on reliability indicators Asset management System development General schematic 10 Framework for comparing reliability criteria 11 Data generation • Unit commitment model • Monte Carlo Data modules • Matlab m-file Events and triggers • Input reliability assessment 12 PF = Power flow SW = Social welfare OPF = Optimal power flow Test Setup (I) • Reliability criterion e.g. N-1, N-0 Extended problem formulation using islanded systems • Reliability control e.g. preventive, corrective Interlinking constraints between islanded systems 13 PF = Power flow SW = Social welfare OPF = Optimal power flow Test setup (II) • Objective function Minimal cost for society == maximal social welfare • Probabilistic approach in large system State selection 14 PF = Power flow SW = Social welfare OPF = Optimal power flow Simulation • Optimization, e.g. SCOPF, OPF Economic dispatch of generators satisfying operational limits, reliability criterion and control constraints 15 PF = Power flow SW = Social welfare OPF = Optimal power flow TSO actions • Short term: Preventive and corrective actions • Medium term: asset management & operational planning • Long term: system development 16 Reliability assessment • Check performance of reliability criterion and reliability control using PF and OPF for: • All contingency cases • Contingency cases of truncated state space • Specific scenarios (i.e. events) 17 Events • Evaluate performance of reliability management for specific cases • Time series including results of events due to: o o o Natural hazard Operational conditions Human behaviour • Which can lead to: o o o o 18 Discrepancy between generation and load Generator/branch outage at particular moment in time Failure caused by several simultaneous faults (failure of cable or power line in same trace etc.) … Economic evaluation • Based on market model • Social welfare evaluation • Total cost evaluation Could be substituted by more complex evaluation techniques 19 Comparison of reliability criteria • Methodology for comparing reliability criteria • Appropriate metric for comparing reliability criteria 20 Assumptions • • • • • • • • • No generator ramp rates or minimal on/off times Knowledge of Value of Lost Load at every node Linear cost curves constant marginal costs of different units Corrective actions o Generation redispatch o Load shedding Constant failure and repair rates exponential distribution Aggregated branch models No failure of corrective actions Reliability assessment considers only branch outages No forecasts errors included (wind, load…) No stochastic, multi-stage optimization • Single TSO, Single area • DC power flow 21 Comparison: Results • Three node test system • Comparison of N-0 corrective o N-1 preventive o N-1 corrective Varying value of lost load (VoLL) o • Performance of reliability criteria and their management dependent on VoLL 22 Outline • Framework for comparing power system reliability criteria o o o o o Overview General schematic of the implementation Implementation of deterministic reliability management module Assumptions Comparison of deterministic criteria • Conclusion 23 Conclusion • Comparing power system reliability criteria is important • Framework for comparing power system reliability criteria and reliability management o Objectives of the framework: 1. Quantification of performance of various reliability criteria and their management 2. Comparison of the performance 3. Identifying alternative reliability criteria o Quite complex, even with many assumptions included • Preliminary result: Performance of reliability criteria and their management dependent on VoLL 24 Thank you! Questions? Evelyn.Heylen@esat.kuleuven.be Geert.Deconinck@esat.kuleuven.be Dirk.Vanhertem@esat.kuleuven.be The work of Evelyn Heylen is funded by: Research in the framework of the Garpur project