DAWG Subgroup A

minutes of meeting - 1 July 1996, LAX

Announcement by Dave Kaplan. DAWG meetings July 9-11 will be at PG&E auditorium in San Fran, at 245 Market Street (at California Street). May enter at 77 Beale or on Market St. Tuesday July 9, 2-5 PM, will be full DAWG "checkpoint meeting" with CPUC. Commissioners will get presentations on DAWG progress. There may be a full DAWG meeting in the morning, but this has not been decided yet. Wednesday July 10 are Teams A & C meetings, and Thursday July 11 are Teams B & D.

Correction to minutes. Nancy Day notes that on p. 2 she and Dave Kaplan are tagged to cover costs of implementation. Nancy should not be on this. Dave is still on.

New Items

Bill Booth (CLECA) - See CLECA handout. Follow-up on load profiiling. Use of upstream & test meters in aggregated groups can minimize cost shifting associated with aggregation. If we can aggregate Res class, the 800 MW limit is not terribly relevant. The number of transactions is the crucial issue. Propose first year exempt Res from 800 MW limit & divide it among other classes. Allocation should reflect availability of meters as of start date. If participation does not fairly represent all classes, admission should be based on historical usage (say, 1995-97).

Mac McCay - 2 items. 1st, the 800 MW limit irrelevant, and therefore pro rata shares of DA are not meaningful. Basic limit should be transactions. 2nd, universal metering should be the goal of state policy for technological reasons. If anyone believes policy should be not this, they need to pur forward their reasons.

Carolyn Kehrein - Load profiling. Upstream metering is not load profiling. Concern is rate freeze or cap. If there is cost shifting between DA-Res and non-DA-Res, then the rate cap may impede fair allocation of costs, so there are problems with widespread Res load prof; i.e., cost-shifting could hurt other classes. We should have an open season to assess interest. If interest in DA is greater than feasible limit, then pro ration shares and allow customers to opt out if they get too small a share. If they get a small share, they could still get it all from their DA provider, but provider must buy the non-DA portion from the PX. This results in fewer interactions for the customer, and since DA provider must rely on the PX for imbalances anyway it just increases the amount the provider must buy from PX.

Lorenzo Kristov (see handout) - Two issues group should recognize as distinct scenarios if we cannot reach agreement on how to proceed. Issue 1 is phasing, whether 1998 must be a limited DA year or can be universal DA. CPUC decision orders limited first year, but some parties argue full implementation is feasible. Issue 2 is universal metering. We have two different scenarios of the future depending on whether we recommend universal RT meters, or we allow customers to choose to not have RT meters. Are there any reasons, other thatn uncertainty about costs, why parties would want to preserve the option not to have a RT meter?

Mac - Can have universal RT meters & still have monthly average pricing optioin. Meter does not foreclose that possibility. Universal RT meter idea would include meter reading and data processing system, for maximum efficiency.

Sean Casey to Bill Booth. Can Small Com be excluded from 800 MW or not? Response - PG&E data indicates that a relatively uniform load curve for Res applies throughout the service territory. May not be so for Com. Sean guesses that load profiling could be equally successful for Small Com as for Res.

Kaplan - PG&E has better data for Res than Small Com. In theory load prof might be viable, but need more data to confirm that. Different Com entities have very different load profiles. PG&E has less reliable data at present time.

Sean - If all groups to have access in 1998, then Small Com must be included somehow, which will require load prof.

Nancy - Does anybody want to argue for MW rather than transactions limitation? John Fielder - SCE has already taken this position. Nancy - Suppose we determine factual basis for limits in year 1. Once we embrace those limits, perhaps there is space unsubscribed. Would that mean no further phasing needed? Suppose transactions are the limit, say 500,000 statewide allocated among IOUs. First year only 250,000 subscribe. If every year the number of subscribers is below transactions limit, then no phase in is needed. Suggests one scenario: If first year unfilled, then don't limit further subscriptions, just allow market to decide.

Dan Carroll - CPUC says there will be 1st-year phase in. Suggest we go with that & require proponents of alternative to develop proposals outside DAWG.

Ed Yates - Wants to advocate full DA year 1, but we need more info to identify technical barriers and see whether they can be overcome. Good to ask what does demand look like?

John Fielder - SCE technical viewpoint is it's foolhardy to go 100% on day 1. Transactions today: 200 wholesale per day in SCE control area. Then a transaction is defined as one customer DA contract, i.e., one large customer, or one aggregation. At present you can think of SCE doing one large transaction for entire UDC. This notion of transactions refers to schedules at wholesale level, sometimes combined energy and capacity.

Mac - proposes alternative transaction definition: Hourly interval meter (i.e., 24 hourly data points) is read once per day.

Roger Johnson - think more in terms of schedules for ISO. John's example of a single aggregate ignores different ties on the transmission system.

George Samaniego - Our limitation is really processing data as needed for settlement, based on control areas. Main thing is number of meters being read, not number of customers per se. The ISO coordinates transactions between grid & UDC. Limits have to do with ability of the UDC to process. The ISO has ultimate settlement responsibility.

John Fielder - If all food processors are aggregated, the UDC will still read meters to render bills for T&D + CTC + imbalances. The UDC will pick up the bill for spot purchases (this point was contested later), so the UDC will have to continue to read meters, although aggregators can still aggregate across UDC boundaries.

Mike Jaske - Continuing role of UDC is an issue. 1- Many parties suggest pro rata splitting of customers. If you do, all these customers still buy some of their load from the UDC, meaning the UDC must be involved in data. Clear designatino of customers as DA or non-DA without pro rata apportioning makes meter reading simpler. 2- Have no idea yet what D tariff will involve. The "ratesetting working group" has not yet talked about it and the IOUs not made 7/15 filing. There may be computation required for D tariff, so the UDC will need meter data. If want to minimize transactions for UDCs, get UDC out of energy responsibility for individual customers. As an alternative, let the energy supplier do metering, billing, etc., so the UDC need not do this. John F - SCE expects to read meters & issue bills for D service at least.

Anthony Mazy, DRA - In future will see many more aggregators. Now seeing prototypes of independent energy retailers, & UDCs will be expected to partition themselves between disco & retailer. Idea of UDCs having to retain meter reading function is holding onto past way of doing things.

Nancy - For SCE to provide transport service it does not need hourly reads. If SCE system cannot process hourly / daily, that's OK because UDC does not need it. George - UDC must also submit forecasts & do settlements, so needs to know. Carolyn - UDC can forecast for the whole system using econometrics, then all DA scheduling information gets aggregated, so UDC load is just the difference. Can have summation of all DA loads by geographic area, but don't need real time info. John Fielder - Until all customers have RT meters, then system George mentioned has to be continued. Must know net loads on system for balancing. Once every cust has meter then may revise this system, but until then it must continue. Roger - DISI team activities indicate that if there are real time imbalances attributable to SCs serving DA customers, those imbalances are charged through to those DA customers, and not a burden to the UDC.

The rest of this discussion was suspended until after feasibility presentations, next on agenda.

Randy - Prepared a packet of draft material on Schedule Coordinator (SC) after consultation with WEPEX DISI team. Gary Matteson has contribution a draft also. Please read & comment.

Dimitra is on WEPEX team that will deal with SC, but this team is temporarily on hold while it resolves funding. Should be picking up this task soon.

Randy noted that the model of SC specifies that the SC must represent the total DA load of one T-D interface and/or the entire output of one generator. Dimitra said there is a significant reason why this is done, but would take too long to explain. End users can do business with as many suppliers as they like, but must designate one party to schedule load with the ISO. Some parties at WEPEX are not completely satisfied that this is necessary, but have adopted it as a working assumption. Still a proposal, not a decision.

Randy now needs input on his outline item #6 - hardward & software requirements for SC to communicate with ISO.

Roger - WEPEX-DISI are focusing on infrastructure requirements, but not software. The goal is to allow for flexibility in software choice, but there must be protocols so communication with the ISO works.

Gary Matteson - Working through the exercise of developing the SC's job description, found two theses: 1- There are some mandatory tasks for SC, and his paper gives a short list, and there are elective tasks, and there are some SC should not do. 2- A main problem - who should be responsible for oversight, control, licensing, etc. His article argues that the market should dictate controlling features, & have given brief examples of how it might work.

BREAK

Gary Schoonyan presentation. See handout. Material represents work in progress.

Objective is to commence transition 1/1/98. Greater complexity could jeopardize success.

WEPEX complexities. California often compared with other countries, is perhaps most like UK but much more complex because of reasons listed in handout. E.g., ancillary services will be a competitive market. Also must be integrated through WSCC with other states, for frequency control etc., which requires coordination. Direct scheduling by participants, day & hour ahead with demand bidding. Congestion management, zonal procing, managementt of transmission losses on zonal or facilities-based approach.

Metering standards - this refers to grid level, not yet down to customers. Refers to SC interface with ISO, UDC, etc., which will need sophisticated systems. Ultimate settlement responsibility is with the ISO, so it needs accurate & timely info. Significant inaccuracies or missing data lead to settlement delays, which in turn result in cash flow delays. Who will be responsible to cover losses in such cases?

Diagrams. Diagram for the PX would also apply to SC, because it would have to do all the same functions. Credit subsystem is essential. When balances are not right & timely, which entity is responsible to cover?

Phase-in considerations. Concern about failures on 1/1/98 - must be able to do settlements even if new systems have bugs. Operating systems now about 99% available, redundant systems. When they started out they were 90-95% available. The ISO is key to future phase in. SCE suggests to observe for 3 years, then let ISO decide. Maybe different ideas of what's possible, but in the end ISO should be the one to decide. "Big Bang" installations have not been done. Think about the problems of getting DAWG papers on Internet. Doing this all at once not practical. Will be both cooperative & competitive players. When problems arise, competitors have incentives to exploit problems rather than solve them. All vendors expect a phase in.

Dimitra - WEPEX has not yet determined exactly which functions must happen on 1/1/98, & which others may be gradually added.

Numbers for test case (sample case; we don't know enough yet to say whether these are limits): 500 schedules per hour or 12,000 per day, from 300 generators + 200 meters. How derived? Started with example, tried to see where system would not support those numbers. Backup for these transactions, in case of ISO system failure, would have to be in place at ISO level. Cannot do at IOU control area level because the required backup systems do not exist today. PG&E & SCE control rooms as they exist today will be phased out, replaced with primary & secondary ISO control rooms.

All DA customers should have hourly interval meters. Can take global or piecemeal approach. Global approach will require a 4-5 year project. Commercial implications refer to billing & settlements, and what happens when meter goes bad and can't be read. Today this is handled in bundled package by utility. How to handle in commercial setting?

Mark (PG&E). Update on Group C, these are the issues under discussion.

- Data requirements

- What are options available for delivering DA (load profiles, meters, other technologies)

- How much do these cost & who pays?

- Who owns meter - who controls it, ie, provides, installs, accesses and reads, maintains?

- What does it take to install meters + required infrastructure? Some believe in a plug'n'play concept, others see 4-5 year timeline.

- Metering standards (lack of, both from WEPEX and end user viewpoint). There are as yet no standards for automated or real-time meter reading. There are lots of vendors, but no incentives for working toward standards because all vendors would like to have their own adopted.

Dilemma: If someone could define DA data requirements, how many customers want DA, where they are located, then we could answer with the best technology, how much it would cost, and how long it would take to deploy. At workshop in Sacramento, the technology comes down to 2 basic types, with lots of versions: automated meter reading network (AMR), or smart meter with phone line interface.

Smart meter + phone similar to what exists today for large indust utility cust. If there are only a few thousand DA customers spread out, then prefer this. Other, fixed network AMR using radio frequency, would fulfill universal meter requirement because we install statewide infrastructure. Could be new meter or retrofit, allowing Res & Small Com customers to participate because 100 percent saturation means low per unit cost.

Progress: Now near concluding data requirements, moving into ownership issues & approaching consensus. Roger - likely both technologies will be installed in different areas, depending on the density of geographic area.

Gary - billing systems. Was concern billing systems would be technical constraint. Now they are not expected to be, but are still issues. Key variables must be determined: Will there be 1 or 2 bills per customer, who does billing, and how much unbundling on the bill? Need to be decided early to be ready for 1/1/98. If UDC collects, then a portion gets forwarded to the third party supplier. Must be commercial contracts covering risk for nonpayment, etc.

George Samaniego - What are limiting factors? See diagram of the Revenue Collection Cycle. Compare present system with Transition Plan. Table shows how the number of data points is multiplied with DA, from 64 m to 20700 m per year. Meter data flow shows two limitations: data transfer & data processing. ISO would like to get daily data for settlements. SDGE is now upgrading, with new system the data processing limit is not hit until 100,000 customers. Other constraint is more uncertain because don't do today. Existing MV-90 system allows 3-5000 transactions (meters) per day, based on hourly data downloaded daily, using new modems for data transfer. Do this today for some remote locations. Can be expanded modularly, so can multiply in 3-5000 units, but soon becomes inefficient compared to a new universal system.

Nancy - SDGE needs to measure & settle energy delivery for DA customers. Every SC must report to ISO, who does settlement daily for imbalances. If no dispute, then goes into data base for billing. But daily ISO must verify that two parties agree on data, with SC responsible for this. For non DA customers, the PX will be the SC.

Dimitra - SDGE forecasts for tomorrow, sends to ISO as do DA customers via SC. At end of day there is difference between scheduled & actual. Within the SDGE zone, the calculated gross deliverable from grid to SDGE = sum of all G + net imports, adjusted for losses, which must = all load delivered at all grid interconnections within SDGE service area. At the same time, the ISO receives actual metered loads from SC for DA customers. The ISO subtracts DA loads from total SDGE load. The ISO must depend on SC metered data to determine imbalances for all UDCs, until all customers have interval meters. Nancy - Thus SDGE does not need to maintain hourly data itself for this purpose. John F - UDC either gets data from customer or from ISO because it needs to bill for D service, depends on loads.

Sean - If UDC does not read the meter, if data goes directly to ISO, does limit still apply?

PG&E - Still run into numbers problem, and it will take time for all meters to be installed.

Will pick up discussion right here at 0900 on July 10. Dave Kaplan's cost of implementation item will be held until July 10 meeting.