 | Dr Grant Read Energy Modelling Research Group |
Introduction
Traditionally, the energy sector, and particularly the electricity sector, has been dominated by monopoly or near monopoly enterprises, typically either owned or regulated by government. The sector has also been a very significant focus of traditional Operations Research (OR) activity of the type documented by Read et al [1992]. Indeed the success of electricity sector applications in the Edelman Prize competition for applied OR, the proliferation of specialist conferences and publications on various aspects of "Energy Modelling", and the emphasis on OR techniques in non-OR publications such as IEEE Transactions on Power Apparatus and Systems, suggests that the electricity sector may well provide the single largest demonstration of successful OR in action.
The nature of the electricity industry is changing very rapidly, though, with a widespread international movement toward competitive electricity markets. Some countries, such as Norway, Chile, Japan and the United States have traditionally been supplied by a large number of different regional utilities, and had developed a variety of mechanisms to allow some form of trade between them. In most cases, though, each power company had a near monopoly on supplying customers in their own region and often controlled at least part of the transmission system which any competitor would have to use to compete for those customers. Thus, despite some important precedents in those countries, the 1990 breakup of the old Central Electricity Generation Board in the UK into a state owned transmission company plus several competing privatised generators, heralded a major paradigm shift with regard to electricity sector organisation worldwide. Culy et al [1996] show that, while privatisation has proved politically difficult, New Zealand has been near the forefront of this reform movement since the assets of the former Electricity Division were formed into the Electricity Corporation of New Zealand (ECNZ) in 1987. Australia started this process later but, although it is still in the process of implementing a unified national market, it has already moved much further in terms of extensive asset privatisation.
Given the sweeping nature of these changes, it is natural that we should ask what implications they may have for the future of OR in the electricity sector. Read [1996] documents the profound effect which changes in New Zealand have had on OR activity there. Fortunately, current indications are positive, in two respects:First, it is evident that OR is making a very valuable, and highly valued, contribution to developments in the sector. In fact, it would be difficult to ignore the OR contribution, given that the central feature of the market of the New Zealand market is a Linear Programming (LP) model which determines the dispatch of all generators, on the basis of offers and bids received from participants, and implicitly, via its dual variables, the price of electricity for each location in the network, along with a reserve price in each island.Second, as a result, OR activity has expanded, rather than contracting, under the new regime. The style and content of OR activity has definitely changed, though, toward what might loosely be called a "dual" approach, in which the emphasis is on setting, and responding to, price signals, in a competitive environment, rather than on top-down "primal" optimisation.
Given the small size of the New Zealand electricity sector, these observations may not be thought very significant in the international scene,. But developments in New Zealand follow those in the UK, where Bunn [1994] and Littlechild [1996] report that the electricity market has also been a significant focus of OR activity, and very similar developments are now being pursued elsewhere. Thus a modified version of the same software is currently being introduced in a national electricity market for Australia, and similar approaches will soon be implemented in major US electricity markets. These developments are being closely watched, and emulated, in electricity markets around the world, and may well be paralleled in other sectors, with a Victorian state gas market now being developed along similar lines. Our experience suggests that this international movement constitutes a major opportunity for the OR community to make a unique, and ongoing, contribution to the development of each and every country in the Asia-Pacific region, and beyond.
The lessons learned in this sector may also acquire greater significance when we reflect on the fact that electricity sector changes have not occurred in a vacuum, but in the context of a wide-ranging economic change process. Duncan and Bollard [1992] document the New Zealand experience, in which government assets were formed into "State Owned Enterprises" operating on a commercial basis, and then generally privatised, with the implicit regulation involved in public ownership thus being replaced by reliance on market forces operating within a neutral competitive framework. Many Asia-Pacific countries have already, or will soon, experience similar processes. Leaving aside the issue of privatisation, though, most businesses are increasingly relying on contractual/pricing mechanisms to coordinate the activities of smaller, more flexible, production units, rather than attempting to optimise the behaviour of monolithic organisational structures. Recent surveys, by Fildes and Raynard [1997] for example, suggest that traditional OR groups, and activities may not be faring well in this new dynamic environment. On the other hand, our experience suggests that it is possible for OR to adapt, grow, and contribute positively to the new world. Thus the lessons emerging from our experience may be of value to the wider international OR community.
The Role of OR in the New Zealand Electricity Sector
Read [1996] discusses the contribution of OR at all levels in the sector, ranging from high level policy modelling, down to short term cost minimisation. Rather than repeat that discussion, or attempt a thorough survey of more recent work, some of which is covered in papers presented to this Conference, we briefly expand on three themes raised in the earlier paper.
First, the previous paper suggested that commercialisation had increased pressures for efficiency, and hence for traditional OR modelling focussed on cost minimisation, while OR had a continuing role in the policy analysis area. It was also suggested that development of the market had led to more OR people being employed in the sector than before, from a wider variety of backgrounds, using more diverse and innovative techniques for a wider variety of employers. These trends have continued.
The earlier paper highlighted examples drawn from our own experience at the Energy Modelling Research Group at the University of Canterbury, and CORE Management Systems Ltd, on projects for ECNZ, Trans Power, and other clients. To some extent, that work builds on foundations laid under the old organisational structure by models such as PRISM/SPECTRA (Read et al [1987]. But the sector is no longer dominated by a single OR group, model, or modelling style, and the last decade has seen the development of innovative models from OR departments at other Universities; from local academics, consultants or software developers not formerly involved in OR; and from international software developers. Thus, for example, although many of the ideas in the market clearing process are based on research carried out at EMRG, and reported by Hogan et al [1996], for example, the actual market software uses a different formulation developed by Trans Power, and developed by ESCA Corporation of the USA (Alvey et al [1997]). While ECNZ's mid-term reservoir management is actually still determined by the SPECTRA model, alternative models, using newer stochastic techniques, have been developed by EMRG, Auckland University, and a Brasilian software vendor. Although the short term river chain optimisation LP described in our earlier paper is used to determine offers to the market, the hydro groups within ECNZ have each decided that different models, from different sources, provide a better match to their needs. Naturally, non- ECNZ generators employ their own approaches.
Second, the earlier paper also noted that, now that many more decision-makers are involved in the sector, there is a greatly increased demand for "market intelligence", with decision-making units now having to determine how to interact with the market in ways which maximise their own profits. Again, the examples cited were CORE/EMRG projects, but OR people are employed in this kind of role by several market participants, while a market simulation model is now sold by an independent software developer.Finally, it was suggested that new market developments may involve new types of OR application. Typically, the sectors which are now being opened up to competition are those in which it has not, historically, proved easy to find adequate market clearing mechanisms. Electricity, in particular, combines technical complexity with a requirement to provide real-time coordination, including virtually instantaneous reaction to contingencies. An experimental market clearing model, and the Dispatch Based Pricing model described by Hogan et al, were cited as illustrating the role which OR models can play in supporting such "delicate" markets by determining "optimal" price signals which ensure that all parties are not only informed of "the market solution", but have appropriate incentives to willingly make their capacity available to the market, or limit their demands on the market, in such a way as to maximise their contribution to total economic welfare. Such applications were seen as providing a clear illustration of the "dual" nature of much OR work in the new environment, a theme examined in more detail below.
To those examples we should now add the software described by Alvey et al, which is actually used to clear the market, and software developed by CORE for ECNZ to prepare offers for the market. We should also note two other aspects of this "dual" OR role. First, this kind of market structure requires the active participation of OR people in establishing a consistent legally watertight set of "Market Rules", which provide pragmatic and robust dispatch and pricing procedures in a form which can be implemented via LP, say. The market also requires that the corresponding formulation, and software, be audited. Second, it is inevitable that there will be legal disputes arising from the nature of the Market arrangements, and/or relating to market outcomes. Such proceedings, while painful, form a necessary part of the overall social system which progressively re-defines the environment in which decision-making occurs, and OR people will play a crucial role as expert witnesses in that process.
Implications
This brief sketch has hardly touched the surface of the wide range of OR activity being undertaken in relation to electricity market developments, worldwide. Many of the New Zealand examples cited here, could also have been paralleled with examples drawn from Australia, or elsewhere. More importantly, though, these examples may be seen as illustrating a much broader set of themes, relating to the changing role of OR in a decision-making environment which is increasingly organised around a "competitive market" paradigm. On this topic we can do no better than quote from the previous paper, from which the remainder of this paper is drawn.
Broad Changes in Decision-Making Philosophy
The competitive market is not without its critics as a way of organising the electricity sector, and some of these criticisms are doubtless valid. Here we outline what that regime is trying to achieve and emphasise some of its strengths, in order to explain the motivation for the modelling approaches adopted, and to highlight the positive opportunities which may open up for OR applications in such an environment. In OR terms, the old approach can be described as having a "primal" orientation, whereas the new approach focuses on what might loosely be described as the "dual".
Under the old approach, Government, or head office, decision-makers attempted direct control of decisions made at lower levels in the hierarchy. Thus it was natural to formulate their problems in terms of primal optimisation, implicitly assuming that lower levels would implement "the solution" efficiently, irrespective of organisational structure. The new approach attempts to create an environment in which individuals, or firms, have the opportunity, and incentives, to act in ways which make a positive contribution to the economy or enterprise as a whole. Such positive "incentivisation" is provided in an environment where individuals or firms reap at least some of the increased productivity flowing from their actions, and face at least some of the consequences of poor decision-making, or effort. We refer to this as a dual approach because "prices" of various kinds, including interest rates and remuneration packages, play a key role, both in "signalling" the value which the wider system is prepared to put on the resources consumed and output produced, and in providing incentives.
In order to make this dual approach work, contracts are typically required to define, measure, and ensure accountability for, the goods and services delivered from one decision-making unit to another. Regulation, if it occurs, focuses on "structure" and "conduct", rather than on "outcomes", the idea being to provide a "level playing field", on which all "players" compete, with emphasis on regulating "the rules of the game" to ensure fair competition. Thus, rather than regulate electricity price or rate of return, the New Zealand electricity sector has been restructured by breaking up integrated generation/transmission monopolies, with the transmission grid, which is arguably a natural monopoly, owned and operated by a single SOE so as to ensure equitable access for all parties, and the state's generation assets formed into competing firms.
This involves a potential loss of coordination efficiency (ie primal optimality) in short term operations, and requires duplication of many functions in different organisations, and in the market structures themselves. On the other hand, it allows each operator to concentrate on optimising their own operations. This can lead to significant gains because they are more familiar with their own circumstances, have greater freedom to innovate and to deal with issues in a way which suits local conditions, and have clearer goals and stronger incentives to improve performance of their own operations. Critics of such re-structuring can always point to the fact that whatever improvements are implemented by the new management "could have been" implemented under a centralised structure, too. This is correct, in the sense that the same technological options are open to both, but management in a centralised structure may not have the freedom, or the incentives, to actually implement many of those options. Further, although a large organisation should be able to attract "the best" talent, in optimisation as in any other area, no one individual knows as much, or has as many ideas, or as much opportunity to observe and experiment, as the collectivity of all those involved in a more decentralised decision-making process. Such factors are becoming increasingly important as technological change places an ever greater premium on the ability to innovate and change.
The Role of OR in the New Environment
Opinions differ as to whether a decentralised and competitive approach to economic management will ultimately prove to be effective, or socially acceptable. Some would argue that current trends are merely a brief digression on the path toward a broader, more holistic, and more cooperative approach to decision-making, and that decentralisation leads to unacceptable losses from "sub-optimisation", as decision-makers concentrate on pursuing their own objectives without considering wider implications for the system as a whole. Within the OR community, some seem hostile to the new environment. In part this attitude may relate to an underlying conviction that there must be a single "optimal" plan, that a rational "objective" analysis will reveal that plan, and that a single "decision-maker" must be responsible to implement it. From this perspective, the duplication of effort involved in having a variety of parties perform their own analyses seems inherently wasteful, and the fact that they cannot afford the complex OR models we would like to build may also be disappointing. Others, perhaps, believe that the OR profession should now be preparing for the inevitable swing back toward more centralised, and broad based, decision-making and some sections of the OR community are broadening the scope of OR toward addressing this kind of "messy" problem in a holistic way.
On the other hand, the current consensus within the government and business communities who form our principal client base is that the "central planning" paradigm has now been so thoroughly discredited, by its performance in practice, that it will never again become popular. They would see development of the kind of national and international planning framework within which a more holistic decision-making approach might operate as both improbable and counter-productive, because such institutions blunt incentives, inhibit initiative, and distort objectives. From this perspective the very existence of an organisational framework within which global optimisation might be pursued would represent "sub-optimisation" within the wider socio-political system.
The probability and desirability of these opposing visions of the future may be debated, but the OR profession must face the practical issue of how to react to current trends. For the medium term, it seems clear that societies around the world will be relying far more on market mechanisms, and correspondingly less on centralised planning processes, than has been the case in the period since the Second World War, within which OR has grown to maturity. This constitutes a significant change in the environment in which OR activities are undertaken, and suggests that the role of OR may also need to change significantly. Traditional OR activity has focussed on "optimisation" of (primal) problems over which a decision-maker has direct control. As technology has developed, there has been a clear trend toward including more areas, more factors, more detail, more complexity and more objectives within the scope of our analyses. But decentralisation of decision-making implies a demand for simpler analyses, focussed on the key concerns and responsibilities of particular decision-makers. Conversely, a whole new range of OR activity may now open up in relation to the design and operation of market or organisational structures, so as to provide incentives for individuals to optimise their own contributions. This implies a greater role for dual analyses, and for the integration of the insights of modern economics into OR practice.
Acknowledgment
This work has been carried out for, and with the support of, a number of clients, including the Electricity Corporation of New Zealand, Trans Power (NZ) Ltd, and the Electricity Market Company of New Zealand, under the auspices of the Energy Modelling Research Group at Canterbury University, CORE Management Systems Ltd and Putnam Hayes and Bartlett Asia-Pacific Ltd. However the author is solely responsible for the views expressed here, and for any errors in fact or judgement.
References