Operational Risk Assessment – Time for a Smarter Look at Reliability for Power Transmission Systems
Written by Math Bollen and Zunaira Nazir
The number of supply interruptions originating in the power transmission system is very small in most industrialized countries. This high reliability is achieved by combining a long-term and a short-term approach: ensuring sufficient components like transmission lines and substations during the (long-term) planning stage; maintaining sufficient reserves during (short-term) operation. The latter is achieved through a number of criteria that have to be fulfilled, somewhat simplified, referred to as “(N-1)-criterion.” There are some limitations with this approach: it is occasionally insufficient, it can create unnecessary barriers against new production and consumption, and it does not allow any trade-off between operational risks and other risks.
Using stochastic methods offers opportunities to overcome these limitations. Stochastic methods are used for long-term planning of transmission systems, typically referred to as “reliability assessment.” Such studies quantify the grid performance over time windows of several years or longer.
On the other side of the scale, for short lead times like 24 hours or less, stochastic methods are rarely used. The basic principle for this so-called “operational risk assessment” was introduced in the 1960s, it has been further developed in a range of studies, and several attempts have been made to get system operators to use it.
The operational risk of a transmission system is defined as the product of the probability p(c) of a contingency occurring within the lead-time and a severity factor F(c) for that contingency, added over all contingencies, c=1⋯Nc.
This can also be interpreted as the expected value of the severity factor. This equation rules the steps and calculations in operational risk analysis.
A contingency is the situation where one or more elements in the transmission system are unavailable at the end of the lead-time, whereas they were available at the start of the lead-time. Calculating the probability of this occurring requires an unavailability model for the components involved as well as failure rate and repair rate data. This includes data on common-mode, hidden and dependent failures. Obtaining such data is a serious challenge, but not the only one. On the modelling side, challenges include smart-grid elements like HVDC, customer involvement, and dynamic line rating. The number of possible contingencies is very large: for a system consisting of 30 simple elements, there are one billion contingencies. Thus, already for a rather small system, it is not possible to obtain the above sum over all possible contingencies. A selection has to be made; this is referred to as “contingency filtering” and it is one of the main challenges addressed in the research on operational risk assessment.
Another remaining challenge is obtaining suitable definitions for the severity factor. The severity factor is a measure for the impact of a contingency; this could impact on the system like overload or undervoltage; it could be impacts on the network users, like the number of customers not supplied, or it could be expressed purely in economic terms. There is a large number of possible severity factor definitions, but the scientific literature does not include any work towards finding the most suitable definitions; many studies do not even explicitly define a severity factor. A power-system study is part of almost any study on operational risk assessment. Most commonly, a load-flow is used to determine how the loss of one or more elements (a contingency) affects the network and/or its customers. The main limitation in the number of contingencies that can be included is in practice the amount of time it takes to calculate a severity factor. The definition of the severity factor requires a trade-off between accuracy of results and computation speed. A specific challenge is to define severity factors that are relatively easy to calculate but allow a sufficiently accurate measure for the cascading outages risk.
The concept of operational risk has been around for several decades, but it has not made practical applications. The two main barriers against its practical use have been the success of the (N-1) criterion and the limitations in models, data and computational power. However, computational speed and power has increased such that it should no longer be a serious challenge. There is also more experience with the use of stochastic methods than several decades ago. At the same time, the need for stochastic methods has increased as wind and solar power introduce additional and new types of uncertainty. Different types of social development also make it more important to balance different risks. Operational risk for the transmission system shall be balanced with economic, societal, environmental, and political risks. All of this makes that the (N-1) criterion may no longer be the most suitable tool for the future transmission system.
There is a range of possible applications of operational risk assessment.
As an example:
- Analysis of major contingencies. The operational risk can be calculated as part of post-mortal analysis for the system state just before or after a major event. This application would allow transmission-system operators to obtain experience with operational risk assessment before using it in actual operational planning or operation.
- Risk-based day-ahead planning. The operational risk can be calculated based on the hourly generation, consumption and load-flow resulting from the day-ahead market settlements. The calculated risk can be used to decide if intervention in the market is needed.
- Risk-based operation. The operational risk can be calculated again when the conditions change significantly compared to the day-ahead planning. This can be because of the loss of a large production unit or a major transmission line; but it could also be consumption or wind-power production deviating a lot from its prediction. When the risk is deemed too high, measures can be taken.
The main remaining challenge, before coming to broad practical applications of operational risk assessment, is, according to the authors, the definition, presentation, and interpretation of the assessment results. The authors consider as an essential first step the establishment of a common set of severity factor definitions; such a set would allow benchmarking and exchange of experience between transmission-system operators. Such experience, in parallel with further research and development, will result in a smarter way of obtaining high reliability for power-transmission systems.
This article was edited by Vigna Kumaran.
Math Bollen is a professor in Electric Power Engineering at the Luleå University of Technology, Skellefteå, Sweden. He discovered operational risk assessment while giving a lecture on the subject as part of Prof. Ron Allan’s course on power-system reliability in 1993 and he has been attracted to the subject ever since. firstname.lastname@example.org
Zunaira Nazir is a Ph.D. student in Electric Power Engineering at the Luleå University of Technology, Skellefteå, Sweden. She is working towards bridging the gap between academic research and practical applications for operational risk assessment of power transmission systems.