Management of electrical distribution assets on both public and private networks is an increasingly complex issue. The strengthening of health and safety legislation had a significant impact on how equipment is designed, installed and operated. This has now been reinforced by increasing demands on the electrical network by customers, regulators, owners and other sectors of the business.
The impact of failure of identical assets at different points in the network can be significantly different. Performance of equipment itself can vary depending upon its environment, maintenance regime and operational exposure. The network itself is subject to considerable variability with influences such as weather and third party damage.
The load itself can also have a major influence. A network may perform adequately under normal load, but much worse under high load, even to the point of a mini-collapse of the network (as experienced in recent years in Auckland and Chicago).
For an industrial network the impact of poor reliability on the core business can also be severe. Nevertheless, it is difficult to make a case for expenditure on electrical infrastructure unless it is possible to quantify the financial risk created by incorrect or under investment.
Targeting of investment to improve short and medium term network performance demands a quantifiable engineering and financial justification to help achieve capital efficiency and customer service improvements.
The techniques described here have been successfully used with owners and operators of utility and industrial networks throughout the world. The techniques are equally applicable to the operation of existing networks, the design of new networks or a combination of the two. The analysis will also allow the impact of increased or reduced expenditure to be assessed, ie budget constrained planning.
The techniques also lend themselves to the development of performance based contracts or the securing of external finance.
Decision-making support
With such a complex multi-dimensional problem and with the ‘big wins’ often having been already addressed simplistic approaches are unlikely to deliver solutions that balance all of these risks and protect the bottom line of the business.
What is necessary is a system wide approach. This does not mean throwing away all of the traditional tools of volts and amps and good operational practice but does mean supplementing them with new techniques such as sensitivity, root cause and risk analysis.
Advanced network modelling techniques are now being implemented world-wide by asset managers to prioritise network investment based on performance, cost and risk.
A key feature of such an approach is the ability to model the electrical reliability and operational characteristics of the network on a systems basis as opposed to simpler feeder level analysis. The analysis takes into account the complete network topology, consequences of failures, capacity constraints, multiple options for re-supply etc ensuring that the analysis mirrors the way in which the actual network operates.
The model is based on the connectivity of the network and the way in which the network is operated. Fault rates, repair times and restoration switching are included for each component making the model highly representative of the network, electrically, geographically and operationally.
A considerable amount of information is contained within this model, including line types and fault rates for each section, along with customer numbers and loading information. The increasing use of asset registers, databases and GIS systems allows such data to be compiled quickly. Data can then be manipulated electronically and imported to assemble a network model. This allows the effort to be focussed upon analysis and output rather than model assembly.
Analytical techniques
Simulation
Expected levels of the network performance are calculated using an analytical simulation. An analytical approach simulates a contingency on the network, determines the impact of this contingency on system reliability, and weights the impact of the contingency by its probability of occurrence. This process is repeated for all possible contingencies. Typical results of an analytical simulation would include figures for:
• momentary interruptions (per year)
• sustained interruptions (per year)
• interrupted hours (per year)
• switching operations (per year)
A contingency occurring on the network model is followed by a complicated sequence of events. Because of this, each contingency may impact many different customers or loads in many different ways. The key to the analytical simulation is to accurately model the sequence of events after a contingency to capture the different consequences for different load points.
Root cause and sensitivity analysis
Root cause analysis allows rapid identification of plant items which contribute significantly to the poor performance of the network being modelled. This quickly allows investment options to be refined and targeted at the appropriate parts of the network.
Sensitivity analysis further identifies the effectiveness of proposed investment options at improving the performance index or indices being considered.
The combination of a root cause and sensitivity analysis, identifies the critical components that have the greatest effect on the performance indices being considered and those aspects of the plant item such as switching time or failure rate that if improved would deliver the most significant network performance benefit.
Network performance variation risk assessment
Risk assessment (Figure 1) can be carried out using a methodology called Analytical/Monte Carlo Hybrid Simulation. In this technique components are assumed to have a constant failure rate over the course of one year.
If 100 identical power distribution systems are built, the calculated expected reliability of each will be identical. In a given year, however the performance of each network may vary and be far worse or better than expected reliability. This variation is natural, and it is vital to understand it when determining the likely range of the network performance as well as the probability of exceeding pre-set thresholds of reliability.
This allows the probability of the component failing a specific number of times in a year to be computed, and the analytical simulation described in the previous section can be modified to simulate a random year rather than an expected year. This is done by determining the number of times each component will fail, a priori.
An analytical simulation can then be performed that substitutes component failure rates with the number of times that they will fail in the random year being simulated. Using this process, many years can be simulated, a list of outcomes can be recorded, and distribution statistics can be computed.
The Analytical/Monte Carlo Hybrid Simulation is computationally intensive, but provides statistical results not obtainable by using purely analytical methods. This statistical information is vital when assessing technical and performance risks associated with network investments, providing network performance indices at stated levels of confidence and accepting that the performance of power distribution networks may vary considerably from year to year.
Not only can the average or expected performance of different options be quantified using these techniques but also an assessment be made of the impact on the possible spread of performance, ie standard deviation.
Northern Electric – a case study
Using the techniques outlined above a computerised model has been designed and built by ABB Utilities to help Northern Electric’s distribution business (NEDL) with its plans to improve the reliability of the electricity supply network.
The aim in creating the model was to provide NEDL with an independent, rigorous and credible assessment of how its existing network performs, using advanced techniques to identify causes of poor reliability. In addition, the model is being used to evaluate how the network will perform under various improvement schemes so that NEDL can target future investment programmes to increase network reliability.
ABB built the network model by importing details of 568 km of overhead cable and 650 substations directly from NEDL’s database. By building in all the relevant information concerning cable lengths, switching times and repair times ABB constructed what is effectively a working scale model of every switch and connection within the network. Figure 2 illustrates the same process used in a similar analysis in the USA.
ABB took this model mirroring the electrical, operational and performance characteristics and constraints of the network and carried out advanced statistical analysis to profile how the existing network should be expected to perform over a period of time – within the limits of confidence. This allowed a risk assessment of the potential variability in performance to be undertaken. Figure 3 shows a hypothetical case of this process.
Potential options for improvement included extending the remote control facilities within the network to enable switches to be reset remotely, without an engineer having to visit the site, which would dramatically cut the switching times; as well as network reinforcement, such as the construction of new substations.
The ABB analysis confirmed that such measures could produce significant improvements in network performance and quantified the degree of improvement expected. Most importantly, the analysis indicated that as well as improving the average performance the range of performance could be greatly reduced by reducing network susceptibility to the extremes associated with major storms.
Commonwealth Edison
Commonwealth Edison (ComEd) is a unit of Chicago-based Exelon Corporation, one of the nation’s largest electric utilities with nearly $15 billion in revenues, a customer base of 5 million, and a summer peak load of 22 GW.
In the summer of 1999, the company’s customers experienced serious power outages. David Helwig, ComEd’s executive vice president of energy operations commented “In September 1999 we set out on a comprehensive turnaround strategy to make sure that within two years our customers would experience a noticeable improvement. We invited ABB to assist in identifying what remedial action was required, and to implement the necessary work”.
Having established where capacity and redundancy were most critically needed, five priority turnkey projects were undertaken. These projects were completed within six months, rather than the two years that would normally be required for such projects.
What started as one project evolved into a partnership between ComEd and ABB, with ABB having a formal role in ComEd’s continuing asset management work. This covers for example, reliability assessment, upgrades and refurbishment of critical substations, development of formal contingency plans and maintenance practices, as well as business processes and business management.
Results
In a detailed report ComEd noted that it had reduced the average number of outages per customer by 38 per cent system-wide since December 1998, and decreased the average length of outages by 46 per cent over the same period.
The specifications highlighted in the report were:
• Increased capacity to make the system stronger including the completion of 23 major projects to increase capacity
• Rigorous maintenance to reduce the frequency of breakdowns
• Improved equipment monitoring to catch problems before they happen
• Re-organised operations to improve the management of the system
• Better planning to meet future needs more effectively
• Enhanced communications to provide customers and public officials with more and faster information, especially during outages.
