By Stephen Kishewitsch
The Independent Electricity System Operator, together with stakeholders, has been developing the rules and procedures for dealing with variable forms of energy, notably wind and solar, so that they can be properly integrated into the grid and market operation. IPPSO FACTO asked the IESO’s Darren Finkbeiner, Manager - Market Development, about some of the wider implications of the development. The following has been edited and condensed slightly.
The IESO has framed the subject as three basic issues: visibility into variable generation on the distribution grid, forecasting and dispatch. Once those are fully implemented, is integration of renewables accomplished?
No. Renewables represent one significant piece of a broader changing supply mix. At the IESO we are looking at that changing supply mix and trying to ensure we are getting the maximum benefit from all resources and resource types. That includes maximizing the flexibility that is inherent in all the technologies. Wind and solar have tremendous flexibility, and Ontario has not fully tapped into all of that with these rules.
At the outset of SE-91 [the stakeholder consultation process] we put together a number of design principles, and one very important principle was not addressed by these rules: the future consideration for wind and solar’s participation in ancillary services. As an example, wind and solar can respond very quickly – in minutes or seconds– and can be used to provide operating reserve to respond to sudden system contingency.
We also have a requirement to continuously balance supply and demand and use regulation service to help ensure the second-to-second balance between load and generation. This service requires the use of signals sent from the IESO on a second-by-second basis, typically to the Beck hydro facility at Niagara Falls. But using hydroelectric resources for regulation can be less efficient and more complex, as it can affect the water management requirements of the river system. If we could use wind or solar to do that, maximizing a technical flexibility, it may lead to an overall more efficient market, creating additional benefits from the renewable investment.
In addition to that, the nuclear fleet is becoming more flexible, the NUGs are being re-contracted and maximizing their flexibility and operational benefits. Where we have untapped potential we want to explore how to use it. We will start taking on that after September 11 [when the dispatch rules go live].
There is already substantial wind and solar generation connected on the distribution grid, where the IESO exercises no dispatch control, except for the few facilities registered as market participants, and more is expected. As a result, as I take it, these facilities are free to produce all the power the wind will allow, while transmission-connected facilities may have to be dispatched down. There would seem to be quite an incentive to develop windfarms on the distribution grid, in preference to connection on the transmission grid, simply to avoid the possibility of being dispatched off. Is there any need anticipated, from the IESO’s perspective, to develop the ability to dispatch distribution-connected variable generation?
Theoretically there would be some incentive to connect on the distribution network, but at the same time there are technical limitations. We looked at the OPA forecast for transmission and distribution connections, and the number of MW on each. If that comes out as anticipated, there will be no need to dispatch the distribution assets because we feel we have enough operational control with the transmission assets, so long as we know what the distribution assets are doing, through the forecasts. We’ve chosen not to do so in part, because of the complexity involved – on the transmission system, we calculate limits and see and understand flows and many other inputs like the outage program. We do not monitor or control to that level of detail when it comes to the distribution system of all the utilities. There might be a situation when we want to reduce a wind farm’s output, but Hydro One as the distributor might say they need that output because of some condition they’re dealing with on their distribution at the same time. Any such dispatch instructions would involve a more complicated, three-party discussion. The IESO does not pretend to understand the nuances of the various distribution systems, with the transferring of loads from one feeder to another and so on. It’s never been our mandate and we’ve never built the capacity.
Readiness
The IESO is of course preparing training programs. At this point, how ready does it seem the various sizes of generators are to work with the rules and procedures, once they’re finalized? E.g. spend on control programs, install hardware, train operators? Do some perhaps have more to do than others?
We are just embarking on the outreach program for all the renewable generators being connected. All the stakeholders have been very supportive of the direction we’re headed. They have had their contract issues, but everyone is very keen to participate. Some will be more sophisticated than others, because they have other types of assets that are already being dispatched, in Ontario and elsewhere – gas plants, hydro, and so on. A solar-only generator that hasn’t had that experience will be less sophisticated. But we’re working with all of them to equalize that learning curve, and keep the implementation and the technology as low-cost and low-complexity as possible.
For the telemetry data, we’ve worked with the distributors and transmitters to piggyback off the mechanisms they already have in place. Hydro One, for example, requires all the generators on its grid to provide output data, so we see all that same data. There’s a solar farm on Sarnia’s distribution network, which sends its output data to the utility there, and instead of putting in new communication links directly to the IESO, they could just pass the information to us over the existing systems we currently use to communicate with the LDC. We are also looking at internet-based dispatch for its lower cost and complexity. We will be using our IT people to make sure everybody’s on the same page with respect to technology. We’ll be holding training programs so they understand how the market will be working. We’re starting that outreach in a few of weeks.
Why do the distribution-connected generators need training, if they’re not being dispatched?
Training for the distribution-connected resources will be solely about forecasting and the data-sharing. Let’s say their telemetry fails, or they have an outage on their system. They have market rule obligations governing how fast they have to report and fix that. We have to make sure they understand those rules and responsibilities.
For the transmission-connected folks there’s a whole additional level. It’s about how you submit offers, what a five-minute dispatch looks like, and how you’re to respond to that. Here’s what your settlement statements and your bills are going to articulate. Here’s how we’re going to calculate your compensation. Here’s how we’re going to be passing on your information to the OPA so they can settle your contract. And much, much more.
Most of these dispatch orders are going to be sent by computers following algorithms, and received by computers running algorithms to control blade pitch and so on, isn’t it? What kind of training is involved?
There will be IT training, about just what kind of internet connection is needed, the latency in response time and so on. At the same time, generators can choose to respond to dispatch instructions in separate ways. Some generators read our dispatch instructions and then issue their own instruction to their facility.
Even on a five-minute basis.
Yes. Some let the dispatch instructions go through automatically, others don’t. And there are times when the automated systems are not sufficient. For example, when we calculate a five-minute dispatch instruction, we are not doing it for this instant or for the five minutes starting now. Let’s say it’s quarter to the hour. We’ll be issuing an instruction we want a facility to follow from ten to, to five to. The algorithm that’s doing that uses a snapshot of the system taken at quarter to. But a contingency can happen at any second. We can lose a wire or a generator, suddenly making that instruction wrong. So we do have to intervene from time to time. We can stop the signal and send a new one, with the participant never noticing the difference, or we can pick up the phone. They still need somebody who can respond to that phone call saying we need you to do x instead of y. That’s part of the training – also here’s when you call the IESO; here’s when the IESO will call you. Here’s how fast you have to respond. If we see a windfarm going out of service, we don’t know – is it faulty telemetry, is it coming right back on, or in an hour? We need someone to call us and explain, so that we can then make the appropriate adjustments to the system.
Dispatch over the internet
How satisfied is everyone about cyber security in having variable generators manage dispatch instructions over the internet, instead of fixed wires as initially conceived? Hackers based in China (according to analysis of the cyber attacks) recently were able to gain the control codes for substations along gas and oil pipelines in Canada and the US.
We have a very extensive IT group here, and they’ve looked at the various protocols and they’re very confident. We have to meet NERC’s [the North American Electricity Reliability Corporation] cyber security standards, and what we’re proposing is well within those standards. We have a section in our IT group responsible for internet security, and in fact every IESO employee must take cyber security training annually, so we are not as vulnerable and they know what to look for, e.g., worms and trojan horses and things like that in their emails.
Recognizing the impact of contract terms on grid operation
The industry has expressed concern that market rule changes could render some contract provisions inapplicable. How does the IESO see its role in accommodating contract provisions that are related to these rule changes and procedures? Was this taken into account in developing the dispatch rules, is there still work to do on the subject?
When we look at the needs of the changing supply mix, we understand that there are contracts. We worked from the beginning with the OPA and the Ministry on the GEA to try and get contracts that were suitable to integrate with the market. And as a result, we saw with the FIT contract design that it went a long way towards that. That work however initially didn’t deal with the earlier contracts. It’s our understanding that now all the parties have come to some resolution of the issues created with the introduction of the rules. [See “What the generators think,” page 28.]
We looked at the contracts to understand what the behaviours would be without these rules, given the contracts. If you look at the RES rules, those generators had no incentive to move under any condition whatsoever. Those contracts were designed such that you get paid only if your turbine spins, and it doesn’t matter what the electricity price is anywhere in Ontario – the OPA was going to pay the contract rate. So there’s 1500 MW or so that had no incentive to respond to the market’s needs, whether surplus or locational contingency. We could not manage that situation on an ongoing basis. We looked at the FIT contract, and saw that they were incented to move for global issues, where price is important, but locational issues don’t show up in price, so they had no incentive to move for a problem in a local area such as southern Ontario. We needed to also put rules in place for operability needs. The fact that there’s an economic benefit is important, but the operability requirements meant we had to put these rules in place regardless of the contracts. We worked with the generators and the OPA so they understood what types of changes were being considered so they could resolve their contract issues. They’ve come to that resolution, as we understand. They’ll get compensated appropriately, based on the terms of the new contracts. The application filed by the RES group of wind generators with the OEB was withdrawn and with that these rules were passed. None of the other wind / solar generators applied.
Locational oversupply and choosing amongst generators
Previous to having these new procedures in place, you could order a facility off for stability reasons, but in the case of surplus generation, with windfarms producing excess power, you would have had to turn off one of the conventional generators, like hydro or gas?
I’ll give an extreme example. If we had a lot of surplus and had to shut down four nuclear units, they would stay shut down for three days. We may have enough gas plants to replace them, but at a higher cost and with a larger effect on the environment. These rules allow us to move the wind with nuclear, in that situation reducing the need for the gas operation.
If we had so much wind we had to shut down five nuclear units, and the fifth one meant we didn’t have enough gas plants to keep the lights on for the next three days, then that is a reliability problem, and [in the absence of the new rules, only for that reason] we could shut off the wind. Surplus starts as an economic issue and turns into a reliability issue at its extremes.
But sometimes it’s not just about this surplus. Sometimes, once there’s more capacity on the system, the wind may want to go from 2000 to 3000 or 4000 MW in a very short time. We may not have enough resources on the system at any given point in time to respond to that increase. We will need to control how fast those wind assets ramp. The new rules allow for the dispatch process to manage these types of operability concerns. [See also the overview story, page 22]
We know the OPA is investing aggressively in supply – we cannot assume those renewable assets will operate at only their average output and no more. If, as an example, there was a wire with 100 MW of spare capacity, you could put 100 MW of wind capacity on it. But that 100 MW of wind on average may only operate at 30%, leaving 70% of spare capacity on that line unutilized. Why not, to get more value from that line, install 200 MW of wind. On average it’s still sending only 60 MW over the line. Now you’re using the line more efficiently, it’s the smart thing to do from an investment perspective. But what happens when from time to time we get 60% total output from that 200 MW wind farm? During those instances we would have a locational surplus that exceeds the spare capacity of the line and that needs to be managed through dispatch.
So we have three types of dispatch: the global surplus, the ramping case, and the locational oversupply issue. The local and ramping issues are operational and not always reflected in the Ontario price.
The growing challenge of managing diversity on the grid
Just as a historical point of interest, the new dispatch rules and procedures represent an ongoing movement from verbal instructions over the phone, with a small number of central generators, to automatic, computer-controlled, algorithm-based dispatch with many more generators scattered across the grid. When did that process start and how close is the trend to running its course?
In the old Ontario Hydro days we had one main generator and one control point, and we could pick up the phone and have a single conversation (300 up here or 200 down there, etc).
Almost overnight, when we launched the market, we had generators like the NUGs that did not have that type of relationship with the IESO system operator. Early on we also had early movers like Brighton Beach, that we spoke with directly. Brookfield Power acquired ~300 MW of OPG’s hydro assets in the north and became another communication point. Bruce Power (initially British Energy) leased the Bruce station, and all of a sudden we started seeing more and more generators who were submitting offers and bids and we were dispatching a greater number of organizations. And over time, with every RFP the OPA has issued, with every merchant decision to connect another generator, transmitter, dispatchable load becoming part of the equation, the number of participants we sent dispatch instructions grew and continues to grow. Every year we get new generators that we have to dispatch. The GEA is another example where we’re going to see a large number of participants in short order enter into this five-minute dispatch realm. We have morphed almost overnight from that market opening day in 2002, to steady growth in participants.
Once you have this algorithm-based automatic control system in place, I guess you can just keep adding generators?
You can, but it becomes increasingly complicated as you add resources. You have the potential for more contingency events. Computers have breakdowns too, and upgrades, maintenance, different levels of sophistication amongst participants, more players to contact and communicate with during contingency events. We’re always balancing complexity with operations. It’s a challenge.
Is there an irreducible tension between reliability, which means system redundancy, and economic efficiency, which often means less investment in system resources?
There are very few scenarios when we don’t have a number of options for solving a problem. Take the case of a global surplus. Sometimes we have exports available, and we don’t have to dispatch anything off. We have industrial load incentive programs, with pricing options, where we can incent them to consume more during surplus periods. What we look to do is pick the option that respects reliability, economic efficiency, and is the most environmentally appropriate.
In the US, it seems they’re relying mostly on transmission being able to shift wind generation out of an area of surplus to some other area with more demand, and so they haven’t developed the kind of dispatch rules.
NYISO has been dispatching wind since 2008. They haven’t as much of a problem, with only about 1600 MW of wind on the system now, and only about 100 of that on the distribution system. They have a completely different set of issues. In the next 2 1/2 years we’re going to have more than three times as much on our system as them, and their system is bigger by about 5000 MW. Our lowest system demand, say an Easter Sunday evening with springtime temperatures, can be a little over ten thousand megawatts. You put, say 4000 MW of wind on the system, on top of a 9500 MW nuclear fleet, plus must-run hydro during the spring thaw, and we have a completely different issue from New York.
PJM dispatches as well using nodal pricing. All those generators are exposed to those prices in a way they’re not in Ontario. As an example, that wind generator in PJM might have a contract that gives it $50 in tax credits. So when the price goes to -$51 they’re losing money, and they’ll go off, because it’s costing them to stay on. We do not subject generators with our contract designs to a situation where they are losing money. They don’t have to pay to stay on. [Ed. Note: this is a separate matter from the negative floor price for wind, discussed beginning on page 23.] Even the contracts that expose wind and solar to price never expose them to a negative settlement outcome. We have to be careful how we compare those other jurisdictions, because they may all have dispatch, but they have different contract and pricing structures from Ontario. If, for example, the FIT contracts had fully exposed generators to a nodal price where they could end up paying, we wouldn’t necessarily need the dispatch rules for the locational issues. [See also “Renewables integration south of the border.”]
What we do with dispatch, they do with locational pricing.
Yes. In fact, our dispatch logic is based on a locational marginal pricing model. That algorithm for coming up with the dispatch instructions is 100% nodal. We don’t settle on the nodal but we dispatch on the nodal. The difference between the two is managed with constrained-on and -off payments.