Incorporating greatly increased levels of distributed generation into Ontario’s electric system will not be easy. Accelerating DG connections across the province raises a multitude of concerns including operational management, physical upgrades, connection procedures, the balance between provincial and local incentive structures, and cost-sharing, to name just a few. Although each issue is manageable in itself, it will be difficult to ensure they are all addressed on a timeline that meets the government’s objectives, and the public’s apparent appetite, for new capacity from distributed generation. The following talk given by Dr. Jan Carr, the previous CEO of the Ontario Power Authority, was given at a conference on distributed generation, October 26 and 27, at the University of Waterloo. This is an edited transcript of his comments.
There are three broad areas I would like to cover in outlining what will shape tomorrow’s grid.
First, it’s important to understand the starting point, and the forces that made the current system what it is, because not only is it the starting point but future developments will depend on changes that we make to the decision making process that got us to where we are.
Secondly, since the issue of interest is distributed generation (DG) and how it interacts with the grid, we need to look closely at what’s needed to accommodate DG. Then finally we can look into what changes might be necessary to the status quo, and how we get there.
I’ll start with a bit of history. Today’s energy systems evolved into their present form in response to economic and reliability constraints – that is, the system had to be reliable, operable, and provide service to customers at a reasonable cost. This led to an approach that relied on economies of scale. Larger facilities turned out to produce lower unit costs, and allowed grid operators to optimize usage based on the life-cycle costs for various kinds of plants. That is, some fuels are expensive, but the plant to use them is relatively cheap; other plants are relatively expensive to build, but the fuelling costs are low, like hydroelectricity for example. So since we have to produce electricity as and when customers’ need it, it made sense to deploy a portfolio of different types of technologies – those with low operating costs were put in place to provide baseload, and those with high fuelling costs were used sparingly, to meet peak demand. That worked out well because the high capital cost of baseload plants could be amortized across many hours of operation while the relatively lower capital costs of peaking plants could be affordably amortized over relatively fewer operating hours. So the reason we have large plants and a portfolio of technologies is to optimize the cost of electricity supply to customers.
As a result of the attempt to optimize reliability, we wound up with a centralized control system. The reason we did that is, fundamentally, the cost of storing electricity exceeds the cost of generating it. To make an analogy, you could say that for electricity the cost of a warehouse exceeds the cost of a factory. In such a case, you obviously manufacture just in time rather than manufacturing a surplus at some times and warehousing it for use later. So we wound up needing centralized control because the whole system is most economically organized for just-in-time operation. We had to make sure that everything was very precisely coordinated on a real-time basis.
What we have, then, is the lowest-cost supply, because we’re using large-scale generators, the highest-density energy sources available, and we have reliability from just-in-time production.
On the subject of fuel energy density, here again costs have been minimized. Exploiting energy sources that have the highest density requires smaller and hence cheaper conversion machinery and so high energy density sources become the first choice when the economics are worked out. Thus we use fossil fuels instead of biomass since, for example, coal produces more heat than wood per unit weight. Similarly falling water produces more energy per unit volume than does blowing wind. A one megawatt hydroelectric turbine could be less than a metre in diameter whereas an equivalent wind turbine could be a hundred metres. Smaller machinery costs less than larger similar machinery so we get the lowest-cost generation technology by using the highest-density energy sources.
The difference between transmission and distribution
The next point in understanding our present electricity supply is the hierarchical system architecture that we wound up with, and the natural break point between the transmission and distribution systems. The transmission system is purpose-designed, and it has very sophisticated monitoring and control systems. It’s designed to connect major supply sources with major load centres. It has relatively few elements to it, compared with the distribution system, and in comparison it is very expensive and sophisticated.
The distribution system is almost the antithesis of that. It is designed on a handbook basis using standardized elements. It has relatively simple monitoring and control systems. It’s basically designed to connect the customers to the transmission system, so its design is driven largely by geography – where are the customers, and where do the lines need to go to connect them?
Thus we have two very different systems in terms of their technical performance, their sophistication, and their unit costs. And ultimately, they differ in their reliability.
To summarize, the starting point is one of a small number of large generators, high-density energy sources, a hierarchical transmission and distribution system, and centralized control. The implication is that any change in approach will generally increase the costs. I say that because the present approach has resulted from a search for the lowest costs – so if we aim anywhere else, we will wind up, for one reason or another, with higher costs.
Challenges of connecting DG
Another thing to note is that the capability of the existing system is lowest at the level where distributed generation needs to connect. This means that some enhancement of the distribution system will be necessary to accommodate DG on a sustainable basis.
ne: Turning to what’s needed to accommodate distributed generation: first of all, DG is small scale. That means it must connect to the distribution system. The cost of connection goes up as the voltage goes up, and the transmission system operates at a much higher voltage. So, to avoid having connection costs dominate, small-scale projects typically need to connect to the distribution system.
Second, we need to accommodate uncoordinated development. We’re talking about a large number of small-scale plants rather than a small number of large-scale ones. There will be multiple owners, each of whom has its own business case, limitations, capabilities, resources and ideas. Development cannot be coordinated centrally as is the case when only one or a few projects are involved.
Third, the generation is non-dispatchable – that is, it runs when it wants to. Renewable DG like wind or small-scale hydro will be intermittent and non-dispatchable. Fuelled generation will probably be cogeneration, probably a small-scale plant whose operation is dictated by the needs of the heat host, not the electricity system. If it’s a biomass operation, there could be fuel storage constraints, and so on.
So the challenge I have outlined can be stated as: How do we evolve the least sophisticated and lowest value part of the grid to accommodate uncoordinated development of a large number of small scale supplies with narrow financial margins and random operation while continuing to ensure electricity customers receive supply that is as reliable and affordable as has been provided through coordinated planning and centralized operational control?
More complex control systems
Where does meeting this challenge lead us? Let’s look first at some of the technical factors.
First, we need some sort of distributed control arrangement to accommodate a large number of small operators. We clearly need a different architecture than the existing centralized control structure but the new structure would also have to be compatible with the centralized one. In designing the new structure we can take advantage of the fact that individual small-scale plants will not need to be controlled with the same degree of refinement as each of the traditional larger ones. As well, we have to avoid a labour-intensive approach that involves control rooms staffed 24 hours a day, as is traditional. One approach might be to have small generators automatically negotiating by computer with a local hub to determine when they’re allowed to run. The hub in turn would know whether to let them run based on instructions from the conventional centralized system operator. The system operator would see the DG in clusters, each of which he could control while leaving the members of each cluster to sort out among themselves how they met the obligations and constraints required by the system controller.
Second, we need to convert the distribution systems into a hybrid distribution and gathering system that can also collect power from generators. Presently the distribution system is designed to maintain correct voltage at all points along a feeder in the context of power flowing in one direction only – from transmission to individual customers. That will obviously change as power flows both ways to accommodate DG.
We also need to change the physical geography of the distribution system, since small-scale generators may be located in a different pattern to the customers that the system is presently designed to serve. An obvious example of this is windfarms where the best sites are typically in relatively sparsely populated rural areas that are consequently served by relatively low capacity distribution systems.
Dealing with uncoordinated development
The need to allow uncoordinated development will require an increased reliance on experience-based probabilistic methods for planning and operation. Planning a system based on a small number of large elements can be fairly deterministic. Reserve margins can be assigned on the basis of the resources available, and their known reliability. As soon as we go to a decentralized, distributed system it becomes impracticable to predict things like maintenance schedules. But if there are enough generators it becomes possible to rely on some percentage that will be available at any given time. It is also possible to expect that all generators will be affected in the same way by a storm moving through the area, for instance.
So we will wind up with a system where small-scale generators will not have unrestricted operation as they do at present. They will have some form of automatically negotiated right to operate. They may also incur costs related to controlling voltage profiles, and they may have to work together to insure that probabilities are within some accepted margin. These obligations and constraints on individual suppliers are what is required to deliver the quality of electricity service that customers have traditionally received.
Looking now at the commercial factors, the first requirement we encounter is the need for some kind of standing set of contract conditions, along with a standard set of connection rules. This will give generation developers a known basis for getting into business, and – most importantly – the ability to choose their own timing. Such standardization reduces transaction costs, and is consistent with the handbook design for distribution systems. This, of course, is exactly how the OPA’s Standard Offer Program works.
Distributor revenue
But one thing we need to do is make sure that the distribution system owner receives some revenue. In effect, there must be a fee for the use of the system by generators so that serving DG becomes a bona fide business for the distribution system owner. This allows a case to be made for the investment needed to expand the system to accommodate the new distributed assets. Such a revenue stream will justify and rationalize capital spending on the distribution system that is necessary to both adapt it to its new role as a gathering system and also to expand it to reach new locations where generators are located. The OPA’s existing Standard Offer Program is based on using existing system capacity and it does not generate any revenue for the distribution system owner who therefore cannot justify any capital expenditures.
This “Distributed Generation Tariff” would most logically be administered by the distribution system owners in much the same way as they presently administer the existing distribution tariffs under OEB regulation. This is because the most complex aspect of putting a DG system in place is the local connection, and that is something the distribution owner has to deal with. To provide one-stop shopping for DG developers, it would make sense to also have at least the administrative responsibility for the commodity contract rest with the local distribution system owner.
Additional rules
There are some very important details that need to be dealt with before rushing off and implementing all this. Equitable queuing rules are needed to ensure maximum use of system capacity. System capacity is, economically speaking, a scarce resource and that means there will always be queuing. What might some of the necessary queuing rules be? It seems to me the right to connect should be non-transferable to prevent hoarding and scalping. Much like mining claims or the old homesteading policies each licensee must “use it or lose it.” Also the grace period between the time of signing the contract and the start of operations of the generator needs to be shortened from its present 3 years under the OPA’s Standard Offer Program. A good arrangement might be to have projects under development meet specified milestones along the way toward start of commercial operations as a condition of continuing to hold a contract and the right to connect.
The balance between regional and provincial benefits and costs also needs to be considered in designing a Distributed Generation Tariff. This is because some of the benefits of DG are local, and some are province-wide. The local benefits include making better use of the local distribution system, and so reducing the unit costs of its use for all local users including existing customers. In contrast, energy costs are presently pooled province-wide and the commodity aspects of a Distribution Generation Tariff would need to be established on a provincial basis.
Another consideration is the transmission constraints that make generation, whether large or small, uneconomic in some regions. With the exception of the southern and central parts of Ontario, most areas of the province are generation-rich; that is, local generation capacity exceeds local demand, so that power flows out. Putting more generation in most locations therefore increases the outflow and hence increases transmission loading. Ultimately we will need some form of locational pricing for generation, applied both to large-scale as well as distributed generation.
Simplicity vs. thoroughness
Let me also say a few words about simple versus comprehensive. Simple arrangements can be put in place quickly, but cannot capture all the opportunities. The existing Standard Offer Program is relatively simple, in that it uses existing capacity on the distribution system, while it does not provide for any expansion of the system. It has had the advantage of letting us get out the gate quickly, but for the same reason the disadvantage of not being sustainable over the long run. We have to balance the desire to get off the ground quickly with the need to allow the system to evolve for long-term sustainability. Too often, it seems, we are tempted to try to be all things to all people on day one. Comprehensive arrangements take time to establish, with benefits possibly not showing up for years. I think the watchword here is “don’t let perfection stand in the way of the good.”
I firmly believe that market-based pricing is required to meet customers’ needs in a flexible and sustainable fashion and so another complexity to resolve is dovetailing a tariff with standard pricing in a system where electricity prices are determined by the balance of supply and demand. It would seem that the 20-year fixed price arrangement used to-date is not sustainable and a better arrangement that involves some form of indexed pricing during the contract term needs to be implemented. There is considerable room for innovation here.
Again, to summarize: The issue is one of evolving the least sophisticated and lowest value part of the grid to accommodate uncoordinated development of a large number of small scale supplies, with narrow financial margins and random operation, while continuing to ensure electricity customers receive supply that is as reliable and affordable as has been provided through coordinated planning and centralized operational control.
Questions put to Dr. Carr by IPPSO FACTO:
Q: You’ve described DG as a pair of problems – it’s both more expensive and less reliable. What then is driving it? Either it’s becoming less expensive or more reliable, or there is some other advantage that you haven’t touched on.
A: It’s mainly the drive toward renewables, and high-efficiency cogen. Because both energy sources use less dense energy sources, we wind up requiring small-scale generating plants that are geographically spread out. So if you want to make greater use of renewables, for reasons such as climate change, which trump the traditional economics inherent in the electricity system, then you wind up with DG.
Q: Aren’t some of the larger windfarms more like large generators than DG, as far as the system is concerned?
A: Yes. But the read I’m getting from the press is that there’s an increasing backlash against these large windfarms, so that we may be seeing more installations with one or two turbines. I’m sure we’ll continue to see both large and small developments and DG is really about the small ones.
Q: Can we learn from Germany, which has these small cooperatives scattered all over the country?
A: I don’t know the details of the business model, but I should point out that electricity costs twice there what it does in Ontario, so the economic margins there are way bigger, and some things that wouldn’t make economic sense here do in Germany. As well, European electricity distribution systems are designed quite differently to those used in North America, which is the result of a number of factors, such as our lower population density, lower utilization voltage (120 volts instead of 230 volts) and inherently lower energy costs due to plentiful hydro electric and fossil fuels.
Q: Is it still the case that large units are always going to be lower cost? Perhaps there are technological changes that could reverse that?
A: I think for a given technology there’s an optimum size, and it won’t be the same for every technology. But for every technology, the economic pressures tend to drive the size up, not down, because every technology has some costs that are insensitive to size.
Q: Will we still be relying on large centralized plants and planning for baseload?
A: Yes, I think we are going to continue relying on centralized planning. But it will include a probabilistic component to deal with distributed resources, and that will allow us to avoid construction of some central resources to the degree that some percentage of all installed distributed resources can be counted on to be available and in operation. A lot of it will be driven by experience, and I don’t doubt there will be some unpleasant surprises – in both directions – but over time it should be possible to have a mixture of central planning at the high level and the probabilistic model accommodating DG.
In the end, it’s about the customers, and they will not put up with unreliability, regardless of what other benefits there may be. I think as well, there is a limit to the premium that customers will pay to be virtuous. So, all in all, I put a lot of stock in allowing things to evolve along conventional economic lines within a broader umbrella policy direction that prescribes such things as renewables set-asides or portfolio standards.
Q: Does the distributor at present get some of the tariff?
A: No. it’s just a commodity contract and the generators are making use of surplus capacity on the grid, which is already being paid for by customers.
Q: Has there been resistance on the part of any distributors, because of that?
A: The only distributor that’s got into this in any substantial way is Hydro One, because most of the resources being developed are renewable, and hence rurally located. However, the cogen will be almost entirely urban. So ultimately, all the distributors will have to get involved. But for the moment Hydro One has born the brunt of the enthusiastic response to the Standard Offer Program. I am not aware of any reluctance on their part to connect generators but I don’t see how either they or the OEB can justify expanding the distribution system at local customer expense when the beneficiaries will be generation developers selling into the provincial market.
Q: Did the designers of the Standard Offer program, of whom you were one, anticipate that it wouldn’t be sustainable over the long run?
A: Remember that the original Ministerial Directive concerning the SOP had two addressees: myself and Howard Wetston at the Ontario Energy Board. And that recognized the regulatory implications and potential capital spending on the grid that needed to be worked into the program. Did we know that it was not sustainable? The short answer is yes. Did we know how fast the uptake would be? Absolutely not. We thought we were talking about a few hundred MW, and maybe a hundred contracts, not thousands.
Q: I guess, if the purpose was to attract proponents, then you got the pricing right and did what you were supposed to do.
A: I take the perspective that no system, including the electricity system, can survive if it doesn’t reflect the interests of the customers as distinct from the interests of the suppliers. APPrO members will only have a sustainable business if what they’re doing is sustainable from the perspective of the people who buy and use electricity. So maybe we didn’t get the pricing right, because it was too popular. Maybe we’re paying too much. This is why I suggested indexing the price. In the present Standard Offer Program, we’re effectively setting price administratively, and that is always a danger. It’s easier to set the price wrong than it is to set it right. Indexing reduces everyone’s risk in the end because it avoids overcharging customers – and hence having them revolt – and avoids underpaying suppliers – and hence having them go bankrupt. In effect, indexing is a concept that sets price based on a long term rolling average of the competitively determined market price
The conference at which this paper was originally presented was titled “Distributed Generation and the Future of Ontario’s Electricity Grid”. It was presented by the Queen’s University Institute for Energy and Environmental Policy in Waterloo on October 26 and 27, 2008. For more information on the conference, see the presentations online at this location: http://www.queensu.ca/qieep/events/distributedGeneration.html