Testimony of Kellan Fluckiger

Comments on Elements of the ISO MD02 Market Design Initiative

 

Submitted by Kellan L. Fluckiger

I will organize the comments as follows: First, I will make general comments regarding the scope and efficacy of the proposed effort. These should come first, since the correct scope is paramount in achieving the correct result. Then I will comment on the specific elements presented by the ISO at the Focus groups held during the week of January 14, in Folsom.

I believe that this design effort at the ISO is absolutely critical. We have one chance to get this right or it will be done for us. We must no longer make assumptions that we do not know will work. We must slightly overreach and then back off of particular elements if they are not necessary as opposed to undderreach and hope it works out. We have taken that second approach far too long and with disastrous results. At the same time, we must also be aware of the costs imposed on the consumers of decisions that we make. We cannot leave that responsibility to others.

 

General Comments

 

The ISO MD02 proposal remains fundamentally flawed in several respects. It does not address the fundamental causes of market dysfunction that we have seen over the last several years. It continues to rely only on uncertain incentive mechanisms that have proven to be completely misunderstood by the current designers. It is impossible to imagine that we would continue to trust the performance of the electric system to such uncertain mechanisms when the results have been blackouts, runaway prices, market manipulation and staggering long-term economic impact.

We must correct the fundamental flaws, and then examine the results so we can gradually depend more and more on the benefits that competition can offer, and yet not expect good results in areas where competition does not make sense. Electricity is a fundamentally different commodity, and is viewed in our current political and social paradigm in a different way than other commodities. We must recognize these differences. Failure to do so will only result in a continuation of the current dysfunction and a repeat of the disaster in greater or lesser degrees of the last 24 months. I will list fundamental missing elements and why I believe they are critical.

 

1. Accurate Load Scheduling

The ISO must require explicitly that load be accurately scheduled. This requirement should be equipped with enforcement mechanisms. This means words in the tariff like "accurate and complete load scheduling within acceptable bandwidths is a requirement in the ISO." For example, it should be required that 90% of all load be scheduled before day-ahead processes are run and 95% is scheduled before hour ahead processes are run. These expectations and numbers are reasonable and have the force of history and logic behind them. No utility serving load would have failed to own or contract for needed energy far in advance of the operating day. Historic errors in forecast are small, on the order of 1-2% and a 95% requirement leaves the load-serving entity (LSE) the flexibility to purchase in the spot market.

No utility historically would have simply assumed that an unlimited quantity of energy would be available on the spot market to meet its load. We have learned through sad experience the hazards of overreliance on the spot market. To continue in such a framework at this time is irresponsible. In addition, the ISO as control area operator must know where load is going to appear and where the resources are going to come from to cover that load.

No utility operator would have imagined managing transmission constraints effectively without knowing where and when the load would likely materialize. ISO running congestion management without accurate day-ahead schedules, both with respect to location and quantity is ridiculous.

Concern about penalties that arise from excess forecast error can easily be addressed by tying the bandwidth for tolerance to the ISO own overall load forecast error. The more ISO forecast error, the wider the tolerance for LSEs. This will account for sudden weather changes and other unexpected occurrences.

 

2. Generator Performance Mandate

A similar requirement must be on generators and other sources of supply. The requirement to perform according to schedule must be explicitly spelled out. Simple imbalance penalties, even augmented with loss of capacity payments are insufficient. Tariff words like "Generators and other sources of supply are obligated to perform in order to supply power into the ISO grid." This should also come complete with enforcement mechanisms.

No historic utility operator would enter a day or a week not knowing whether or not a plant would follow its schedule or instructions given to it by the system operator. Absent physical failure, compliance with system operators forecast plans and real-time instructions is essential. The transaction side of the equation where LSEs and Generators work out deals is a completely different issue, and is handled in the financial arena, but after schedules are given to the ISO, compliance should be required.

Congestion management and transmission allocation is also a sham without accurate generation schedules and the assurance of performance of those units. Attempting to meet this requirement with only day-ahead residual commitment and potentially inaccurate load schedules is risky, and open to dispute. If accurate load schedules and generator performance are required, then there is no issue.

Concern about penalties is less troublesome since units can be instructed up or down when load forecast error occurs and those instructions are part of the schedule against which compliance is measured. An appropriate bandwidth for unit performance is also expected. With all these features, there is no reason not to require generator performance.

 

3. FTR Design

FTRs are a significant problem as well. Historically, when utilities purchased power, they also bought delivery capability or transmission rights as well. FERC actions separating transmission from generation did nothing to change that paradigm, it simply made utilities treat their own load purchases consistently and equitable with other requests for transmission use. The ISO is committed to equitable treatment, but fails completely to offer transmission service that makes sense.

If we intend to have generators commit to build capacity, then we need LSEs to be able to sign long term contracts with them to offer financial stability and get projects financed. If we want those contracts signed, then either the generators must be built at the load, or transmission rights must be acquired. The ISO must offer transmission rights for a 5, 10 and 20-year variety to allow for this sort of contracting. This reform is supposed to be fundamental, and get to the core of problems. If we do not address this problem then not only will adequate capacity be jeopardized, then the munis and other vertically integrated groups will not be able to join the ISO, because they require such transmission service.

This approach is entirely consistent with FERC directives, even to the ISO, where they have talked about needing longer FTRs in previous orders. Concerns about hoarding can be addressed with strict enforcement. 20-year FTRs must be scheduled perhaps one year in advance, since someone actually serving load would only buy a 20-year FTR to accommodate a particular supply contract. Shorter term FTRs could have shorter release times to accommodate their intended usage. Failure to comply should result in uncompensated loss of the FTR Loss of unit, loss of load and other extraordinary circumstances should all be allowed in. The concern for enforcement and fear of gaming should not preclude the creation of the products needed by legitimate LSEs.

 

4. ACAP Design and Implementation

ACAP – This is a proposal by the ISO that has significant merit. It is essential that LSEs be put under obligation to procure supply. This mirrors responsibilities held by those who have an obligation to serve. However, the current proposal to have 115% of the highest hourly load each month available for the month will be incredibly expensive, and will produce far more capacity than historically required. It actually has the potential of producing a requirement that LSEs have nearly 200% of needs tied up in firm commitments during off-peak hours. This is unacceptable. I understand that subsequent modifications have been made, but they are nowhere near sufficient as of yet.

When resource planning was done historically, capacity available daily or seasonally was adjusted to harmonize the twin variables of cost and reliability. This approach must continue. The ISO role should be to verify that appropriate capacity (115% or some other number) is available each hour of the day. This must be accompanied by stiff enforcement obligations in addition to the cost of ISO procurement, since the ISO will keep the lights on. This kind of obligation and enforcement must accompany and mirror generator obligation to perform so that no penalty is simply an adder to cost in contract negotiations as seen by the other side.

An additional critical component of ACAP is that it must take into account the pre-existing contracts. Just as at start-up when the ISO was required to accommodate existing transmission contracts, (ETCs) the ISO must now accommodate supply contracts that already exist as of the date of implementation of this concept. To fail to do this would place an unfair burden on those who negotiated contracts not knowing of the future world in which those contracts would operate.

A final component of ACAP is that it will be critical to deal with the interties. California has historically been, and remains today, dependant on imports. This is not a bad thing. Seasonal diversity exchanges with winter peaking areas makes sense and saves money. It will be critical to properly address transmission capacity requirement with respect to ACAP delivered on interties. It would not make sense to have a requirement on intertie ACAP that it be accompanied with pre-purchased firm transmission, particularly since, at 115%, there is a good chance that it will go unused. It will only add unnecessary costs.

Having said all this, I must reiterate, ACAP is a good idea. It should be tailored to each hour of the day, with the intent kept in mind. The intent is to assure sufficient capacity to meet load and ensure competitive markets. A daily verification, with an hourly requirement is a start. The following items should be added.

  1. A tolerance for unit outages below a threshold, say 5% that need not be replaced if the unit or source becomes unavailable after the day-ahead verification. Obviously, this should be subject to appropriate verification. And appropriate market power and manipulation protection.
  2. Verification should be daily, with hourly amounts adjusted for actual projected load. This goes hand in hand with accurate load scheduling.
  3. During off-peak hour when excess capacity is more li8kely available, the requirement should decrease. For example, during peak hours, it should be 115% subject to a) above, and during shoulder hours it should decrease to 112% and gradually decrease so that during 10 or 12 off-peak hours, it is at 107%, or only enough to cover load and a/s. The premise for this is that during off-peak hours, energy from partially loaded units is more readily available. I realize that units come off at night, and this can be adjusted as needed.

 

4. Market Power and Compliance Enforcement

Market Power and compliance. The DMA at the ISO is understaffed and underpowered. They need to have the requirement to report and act instantly on suspected manipulation and enforcement activities. The ISO has far too long waited and waited and gathered data, while hoping that errors correct themselves. The DMA should be empowered to file problems instantly, with FERC taking quick action. If triggers are not perfect, then they can be adjusted as we go along, but we have so far erred so badly on the side of "let ‘er rip," that billions of unnecessary wealth transfer has occurred.

These are the big areas of market design that must be addressed that are either missing or completely inadequate in the current proposal. I believe that without these being addressed, then the rest of the design changes don’t mean much and as I have said before WE WILL REPEAT THE EXERCISE OF THE LAST 24 MONTHS AGAIN. I can already hear the supply side participants salivating at the thought of an incomplete or late reform that allows a second bite at an already decimated apple.

There are other critical areas of needed reform such as settlements and transparency, but they are beyond the scope of this paper. They actually should be addressed at the same time, but this may be beyond the ability of the ISO to do at this present time. I am quite distressed to see in the latest from the ISO that the ACAP is already getting the possible hatchet for near-term implementation because it seems to be too hard to do. This is typical of the last several reform efforts, with "delay" being the most often implemented feature.

I will now get to specific elements of the design proposal. I will list the element and then comment on support as is or list needed changes as I see them.

ACAP Obligation
  • It should be created
     
  • It should be placed on LSEs
     
  • It should be phased in over 2-3 years
     
  • It should be measured and verified daily
     
  • It should be only a fixed percentage over each hourly load, and this should be adjusted for on and off-peak needs
     
  • ACAP can be provided from owned capacity, contracts, Interties (Firm transmission should only be required day-ahead, since prior commitment is often more expensive.. If there seems to be a lack of performance, the ISO should have the flexibility to apply a derate factor to any supply proffered by and LSE if that resource fails to perform. Such a derate might say in essence "that tie does not perform always when you offer it, it only counts for 75% of offered value" This would require the LSE to acquire more capacity to meet whatever the total obligation since some portion was derated because of lack of performance. The ISO could have a schedule of de-rates, non-performance each time causes a source to be derated by 5%, etc.)
     
  • There should be a penalty for ACAP insufficiencies. It should be significant, and should be phased in over time, since in a situation such as we have now, where there is insufficient capacity to ensure either a competitive spot market, or a competitive negotiating market in which to form contracts, just means that market power is either exercised in real-time, or in the negotiations and captured in a contract. We witnessed such situations when DWR did contracts. I do not believe the crisis has yet passed, and indeed I do not believe that it will be over until these reforms are in place!!!
     
  • Generators must perform as scheduled, and if they do not, there is an imbalance penalty (at 15-%?), an ACAP charge and perhaps some other penalty. This also must obviously be bracketed to allow for normal machine operation, etc.
     
  • ACAP availability should be submitted daily, verified as needed by the ISO and penalized for non-performance with spot checks.
     
  • The ISO is already the supplier of last resort. It already does residual unit commitment and residual ACAP supply today in an ad hoc disorganized way to ensure that the lights stay on. Appropriate formalization of this process is necessary, and it may perhaps be a place where people can intentionally come to purchase ACAP shortfalls, but we must deal with timelines and synchronization of this process with residual unit commitment. In any case, to fit with the day-ahead unit commitment, if the ISO does an ACAP market, perhaps it should be 2 days ahead, or in advance of the unit commitment process to ensure dispatchability and mesh with those other processes.
     
  • All generators must bid or have default availability in day-ahead. These must obviously be mitigated bids is market power conditions exist, and the ISO should not be shy about putting them in print. Mitigated bids should include start-ups and cost plus 10 or 15% so no one can complain of lost money. There are simply too many transmission constraints and system conditions and places where certain units or groups of units are needed to pretend that we don’t need a must offer requirement. It is incumbent on the ISO to craft a must offer mitigated bid proposal that deals with market power, not only for steady state conditions, but that can be invoked for transmission conditions involving either scheduled or forced outages.
     
  • The ISO claimed that ACAP would solve locational congestion problems. This is only true if loads are accurately scheduled, and units offered must run. This is fundamental. It is not sufficient to depend on residual unit commitment. This would incent load to ignore local reliability requirements and let the ISO handle them all with mitigated bids in the residual unit commitment portion of its procurement. This will never move toward load taking care of itself.
     
  • The ISO must have all units offered at mitigated prices in constrained situation, since there is a condition that can be created for nearly unit in the system where it is in a "must-run" posture, and it would be better to create a process and rule, than managing by exception.

Forward Congestion Management

  • Optimal Power Flow, Full Network Model should be used
     
  • Schedules must be explicitly required to be accurate, 90% Day-Ahead, and 95% hour ahead. Without this, forward cong will be useless as it has been since start-up.
     
  • Generators must perform to schedules or forward cong will be meaningless and resulting transmission allocation and pricing signals insignificant.
     
  • WSCC reduced network should be used
     
  • Trading hubs and Demand Zones are fine
     
  • Market Separation should be forever abandoned. It was a bad idea to start with and it would be insane to continue its use
     
  • Locational pricing for generators is good. Zonal or aggregated pricing for load zones may be necessary, although it may interfere with the effects of real-time Pricing or other attempts to incent demand response. These have to be coordinated with PUC tariffs and actions, since the extent to which price signals mean anything to loads, it is through a PUC process. Slamming LSEs with a pricing structure they cannot accommodate is meaningless and a waste of money.
     
  • Recallable transmission is a good idea that should have been done long ago. In addition, the hour-ahead process should be completely changed. We should not try to run cong, but instead have an oasis system that allows first come first served transmission allocation for residual scheduling that is compatible with WSCC timelines. RTS is also necessary even if this is done since ETC transmission is not available to the ISO in the conventional way. However it might be worth exploring having a consolidated web site with the Munis if the timelines are the same and the customers who use Muni transmission are willing to pay.
     
  • The proposed Forward congestion Model DOES NOT ensure that final schedules are feasible. It does not ensure accurate load schedules, it does not ensure generator performance, and it does not ensure that the result is dispatchable, since the values of the two critical parameters, generation and load, are not accurate. Garbage In Garbage Out. Simple residual Unit commitment may put the right units on line, but waiting till real-time to dispatch those units to the right level means nothing. If other units that are scheduled actually don’t come on line, or if everyone does whatever they want you still have the mess you have today. The ISO needs to get real with this and require performance. If things work out later, these rules can be relaxed, but the nightmares of the last years should be enough lesson.
     
  • The proposed Cong does eliminate inter/intra distinction, but this is meaningless without a requirement for performance on both loads and generation, so as proposed it does not ensure accuracy, transparency, or proper allocation of transmission.

FTRs

  • Point-to-point FTRs and point-to-hub FTRs should be created as neede4d to assist business.
     
  • Point-to-point FTRs should receive scheduling priority.
     
  • Nodal prices should be used to derive congestions costs.
     
  • ETCs should convert to FTRs
     
  • FTRs must be created to support 20-year and shorter transactions, or the above comment will never happen.
     
  • FTRs must be use it or lose it, but with different provisions for different length of FTRs
     
  • FTRs should only be valuable to those who schedule on the transmission system, and should lose all value if left unscheduled. We do not have a sufficiently robust transmission system to allow the other type of FTR. There are too many that just need the FTR for scheduling and cannot complete in the sophisticated trading world of derivatives that will emerge if the other type are created. That may be a later development, but not now.
     
  • Obligations not options

Forward Energy Spot Market

  • No separate PX energy market is needed. If the ISO eliminated market separation and balanced schedules, then the Cong process will perform the residual energy matching functions and created the needed SC-SC trades.

Residual Day-Ahead Unit Commitment

  • The ISO must have authority to commit needed units Day-Ahead that are not scheduled. This must include start-up payments. This must occur at both the system and local levels. This would be rendered unnecessary if the ISO required accurate scheduling, and made the ACAP requirement Locational. Cong could handle this.
     
  • Yes, ACAP resources should be required to bid in.
     
  • All resources need to be available at appropriately mitigated prices if this system is to work. Even Non-ACAP resources. If a needed resource is not ACAP and a particular transmission configuration occurs, then how do you get that unit to run? OOM?? Aren’t we trying to get rid of that stuff?
     
  • A unit should lose start-ups and no-load payments if it is already on line for any other reason, a/s, uninstructed deviations, etc., but there must be a bandwidth so that the unit is not penalized for coming on line at the projected load and then slightly deviating from it.
  • - Yes market payments should be netted from start-ups. A legitimate issue is the costs of running a plant that may have been on reduced staffing if it is called to run for several days. In addition, minimum run times of longer than 48 hours may need to be considered for some older units.

Ancillary Services

  • Ancillary services should be don simultaneously with Cong, and not sequentially
     
  • Regulation energy should be paid. Perhaps the money for this service could come from loads and generators in proportion to their use of the Imbalance market, or uninstructed deviations, since in addition to natural variations, use of the imbalance market has been the source of the huge quantities of regulation required.
     
  • Sequential markets benefits I believe are outweighed by the gaming of separate bidding. When energy bids are submitted which bear actual relationships to units "costs," something that is more likely to happen when market separation is relaxed and unbalanced schedules are permitted with the resulting inter-SC trades, then optimal a/s procurement is more likely. Rational buyer may bring some benefit in a sequential market, but if the imbalance market is kept small by a 95% hour ahead scheduling requirement, a load accuracy scheduling requirement and a generation performance requirement, plus paying regulation and charging it to those use the Imbalance energy market, I think regulation costs will go way, way down, and the rational buyer missed opportunities will be negligible. Single pass procurement is the way to go. 

Hour-Ahead Market

  • The timing of the hour-ahead market must line up with the WSCC
     
  • ETC and hour ahead ISO scheduling must look identical
     
  • This may mean ditching hour-Ahead Cong, and doing OASIS procurement on a first-come, first-served basis.
     
  • Those who have short term FTRs may be able to schedule them in the HA, and thereby have capacity already. If differential FTRs are created with differing lengths of use it or lose it, then some may be permitted to expire only at HA. The only thing that would require is that FTR holders who still have valid HA FTRs would have to claim there use 2 hours ahead so that unused FTRs could be available for the intervening hour for OASIS-type procurement.

Real-Time Changes

  • Real Time Full network model must be used with Locational prices.
     
  • Real Time bid mitigation is needed, but only if required by Locational problems, as evidenced by the need to go out of sequence, or some other measure.
     
  • Uninstructed deviations outside the bandwidth (above) should not be paid. Below the bandwidth should be charged heavily (150%?, I think that is the current proposal?)
     
  • Load following of one’s own load must be accommodated, at least in MSS areas, and perhaps in load aggregated pricing hubs. Residual neutrality problems due to differing Locational generation prices could be paid/charged to that same load.
     
  • Real time clearing of Beep Stack should continue (Target Price change)

Damage Control Price Cap

  • This is certainly needed. We have had enough bad experience so that this should be kept so that things don’t get out of control. New capacity price signals can clearly be sent under such a cap. The real question is the relationship between the level of this cap and any demand response/Real-Time Pricing signals being sent.

Additional Comments

  • Coordination with the PUC procurement process is essential. The ISO should be involved since they overlap, and the IOUs may be told to do something they can’t accommodate at the ISO. Renewable procurement and scheduling comes to mind
     
  • Intermittent/Renewable/Demand sources all tend to have differing characteristics. While it may be nice in theory to treat them all the same, and pretend they are all gas fired generators, it is not practical and it is easy to argue that each type is "not similarly situated"