Exhibit 99.420
Distribution: Dariush Shirmohammadi, Paul Gribik, Al Suding, Jeff Amlin
- --------------------------------------------------------------------------------
Subject: SCE project
Author: gbackus@boulder.earthnet.net%smtp at x400po Date: 5/7/97 12:07 PM
Paul,
William Heller has asked me to put together a smart gaming system that looks at
the present and past conditions, find patterns, and determine moves (day, hour
and spot) that maximize profits for generation and minimize costs for the UDC.
One component of the system needs to analyze the (probabilistic spectrum of)
impacts of a move before hand. Is your L-P model (as in your paper) THE model
that the ISO will use. Is the model (in theory) fully available to all those who
participate in the California system? To SCE?
Have any information on the run time for you model? (Is each run an hour or a
day of hours?) I would need to make 40-100 runs for each hour to determine the
possible outcomes and the correct portfolio of bids.
Thanks
George
- --------------------------------------------------------------------------------
From: Paul.Gribik@pscmail.ps.net
To: gbackus@boulder.earthnet.net Subject: Re: SCE project
Date: Wednesday, May 07, 1997 2:08 PM
George:
The paper describes the model that the ISO will use to allocate and price
interzonal transmission capacity to the various Scheduling Coordinators for
their energy transactions. There are some as yet incompletely defined protocols
that will be used to allocate and price interzonal transmission capacity for
ancillary service transactions. The paper does not address these. The paper also
does not describe the procedures that will be used to adjust energy schedules to
relieve intrazonal congestion and the settlements for this.
The software that solves the interzonal congestion management LP is being
developed by PCA in Arizona. This will not be public domain software. I also
don't think that SCE will be just given a copy; however, SCE does own a license
for PCA's OPF. This is the package from which the congestion management software
is being developed with many significant changes.
I am sure that PCA will sell to others, but they won't give it away. In any
case, for your probabilistic analysis, I think that a much smaller LP would give
acceptable results (many fewer nodes in the transmission network model). This
could be developed using a commercial LP package. The ISO will solve each hour
in isolation. Base software to solve the intemporal problem was not available
and modeling 24 interrelated hours was considered too risky.
Paul
- --------------------------------------------------------------------------------
Subject: Re: SCE project
Author: gbackus@boulder.earthnet.net%smtp at x400po Date: 5/7/97 5:54 PM
Thanks for getting back.
How do the utilities check to make sure the PX or ISO is not screwing them if
they don't have the model themselves to at least run tests on? We want to check
what did happen (we need to see what data is going to
be public off the PX and ISO) with what could happen. The "aggregate" model
would need to be close or have a well defined bias. If we can't have the "real"
model would you be available to "make" the "gaming" version? This implies you
would have the data set from the real model.
G
- --------------------------------------------------------------------------------
From: Paul.Gribik@pscmail.ps.net
To: gbackus@boulder.earthnet.net Cc: Alan.Suding@pscmail.ps.net
Subject: Re[2]: SCE project
Date: Thursday, May 08, 1997 8:09 AM
George:
I don't think that ABB, E&Y, etc. agreed to give licenses to use the ISO's
software to anyone other than the ISO. If someone wants to use the software to
check the ISO operations, I believe that they must make arrangements with the
vendors to obtain licenses. The vendors are under no obligation to provide
access to their intellectual property to anyone for free.
Let's look at the aggregate model. The interzonal congestion management model
will include a DC OPF model that can handle several thousand buses in the WSCC
system model (approx. 6000 buses I think). I am not convinced that the
probabilistic analyses you may want to run will need this level of detail. We
may be able to get by with a model that uses only around 100 buses. This will of
course require some analysis of the transmission network to see what we can do.
Another part of the problem is just dealing with the volume of data to model the
6000 bus network. Shrinking the network will reduce this problem.
The interzonal congestion management model will also include a market separation
constraint for each scheduling coordinator. In phase 1, the ISO want to be able
to treat 1000 SCs. In phase 2, 4000 SCs. For your gaming analyses, I don't think
that we will need to treat this many SCs. We may be able to get by with treating
approx. 100 SCs. There won't be anywhere near this many major players in the
energy trading arena.
Because of this, I think that writing a small interzonal congestion management
model in MINOS, CPLEX or other LP solver should be a fairly cheap and relatively
quick. Definitely cheaper and quicker than waiting for PCA to develop their
software and buying a license.
The 6000 bus model will probably be needed for intrazonal congestion management.
For this model, we will use a full AC OPF without market separation constraints.
This will be solved using a minor extension of PCA's off the shelf OPF. Running
full AC OPFs can be very tricky. Just defining a reasonable case can take weeks
of effort. I don't know if the effect of intrazonal congestion will be worth the
effort to analyze. I am afraid that it will offer gaming opportunities so it
probably will be an area to analyze. The gaming opportunities will arise since
the number of resources that can relieve an "intrazonal" constraint violation
will be very limited in may cases. This will give rise to locational market
power.
At present, I don't know what my availability will be. My time may be fully
booked on the ISO Alliance systems build. Personally, I think that your project
is more interesting. It could also be more in line with my abilities and
background. However, PSC has a lot riding on the ISO systems development and if
they want my full or nearly full time, I'm sure that they would get it. I wish
that I could give you a better estimate of my availability, but this is as
clearly as I can see into the future. I personally think that building the
gaming version of the interzonal congestion management software would be doable
and interesting. It could also give a valuable back-up to the full model being
developed by PCA.
Paul
- --------------------------------------------------------------------------------
Subject: Re: Re[2]: SCE project
Author: gbackus@boulder.earthnet.net%smtp at x400po Date: 5/8/97 12:19 PM
Paul,
Thanks much for this one. You gave me great information. I agree with your
assessment and understand it. I would like to find a way to make it worth PSCs
while to have you help on this project. From an intellectual point of view I
like to think about the AC gaming just because it is such a tough problem. As a
follow on to what I hope we can do DC in phase 1, AC could be looked at in phase
2. In phase 1, I am actually trying to get the CPUC to recognize the mess they
are causing with their pricing and marketing rules and relieve some of the
restrictions so that the market can actually behave like a market (as in supply
and demand cause the price the customers see). It should also help SCE survive
until the CPUC changes its perspectives. In phase 2, it would seem to be worth
money for SCE to look at AC issues and I just want to satisfy my own curiosity
on the dynamics. (In that I see SCE probably as a big loser before deregulation
is done, helping them game the system doesn't seem all that predatory -- maybe
it's just my rationalization.)
George
- --------------------------------------------------------------------------------
Subject: Re[4]: SCE project
Author: Paul Gribik at PSC-LADWP
Date: 5/8/97 3:10 PM
George:
The project definitely is interesting. It should also have good follow-on
potential.
I think that several areas of the protocols have large potential for gaming. I
don't know if we want to try to get the CPUC, FERC, ISO and PX to try to plug
the holes. I am afraid that it may be too late. It may be best to help SCE guard
against attacks and develop profitable strategies under the existing protocols.
I hope that we can get together on this. It should be fun and profitable. I look
forward to talking about this more.
Paul
- --------------------------------------------------------------------------------
Author: George Backus:
Glad to hear of your interest. I didn't mean affecting the CPUC directly. It's
just that once the CPUC sees the gaming they may rethink things and in a few
years get the system cleaned up to work better.
G
- --------------------------------------------------------------------------------
Subject: Re[5]: SCE project
Author: Alan Suding at PSC-LADWP
Date: 5/8/97 5:25 PM
First order of business is the project proposal then. In the meantime e-mail
discussion can continue as it has been regarding the work. How about a
conference call on Monday with the intent of getting the proposal out?
If everyone would toss out suitable times and phone numbers, I'll coordinate the
call. I'm open all day Monday.
Another option is to use Kinko's video conferencing facilities (if one is
convenient to everyone). PSC will cover $. Not so much to see each other but to
see documents/drawings at the same while talking goes on.
PSC has a proposal outline that I can send to all as a starting point.
In addition, I'll sort thru the e-mail todate send a summary of the conversation
and suggestions in the form of meeting minutes (kinda) such as; planning for
subsequent marketing to other companies, ISO model availability and options,
computational needs, pending coordination, priority, and approval questions
within PSC and such.
George, who's jamlin@earthlink.net?
- --------------------------------------------------------------------------------
Subject: Re: Re[5]: SCE project
Author: gbackus@boulder.earthnet.net%smtp at x400po Date: 5/9/97 12:53 PM
Jeff is my business partner and the head of Systematic Solutions, Inc. They
primarily do the model development and client support/project work. Jeff already
has many of the bases covered for model design and functionality. We do need
Paul's smarts on the ISO submodel. We have a good PX submodel but would surely
take the "real" California model. We also need the good technical analysis
"wall-street" PSC consultant to make sure that we are not deceiving ourselves on
what to look for and what it means. The "consulting" here is mostly cross-check
and fatal flaw stuff because, unlike the stock market analysis tools, we will be
having an affect on the out come (we cannot assume our acts are independent of
the market). Nonetheless, if we go "co" on a product then the "stock
market/commodity" folks of PSC will be needed in marketing and training.
I need to talk to Jeff on our thoughts of going in with you to sell the product
elsewhere jointly. I am positively inclined but we need to think clearly here.
(Jeff and I have already touched bases with other clients on the product and are
a few steps ahead of your here.) Also the gaming AI stuff we bring along with
the systems dynamics techniques raises many long term business issues with us
should they become "yours." I think we can easily work this all out but I need
to think of what it means.
Paul is far superior to us in the models of the ISO and PX, and crucial to the
California work. But SCE will make sure we can only work for them in California.
PERIOD, PERIOD!!!! This may bring some "exclusivity money" but we would want to
market elsewhere for better $. We can do "elsewhere" ourselves (Jeff and I)
including the ISO model, but having you guys on board would certainly increase
what we could do (given our size) and take advantage of Paul's expertise and
productivity (as compared to ours).
I have a conference call with the SCE lawyers at 10:00 AM your time on Monday to
determine what we can put on paper and how hard nose SCE will be about residual
rights. We need to have some contact among us before then to see what our mutual
needs are -- before I inadvertently give the "farm" to SCE.
Can we talk early Monday AM your time and then get into the proposal thing later
Monday? I can do Kinko's video conferencing if you think it would help. At the
moment I am not so sure we have much to "see."
The proposal has to phased I think because we need to see along each step what
high-cost tasks may have been overlooked and whether we are smart enough to be
sure we can successfully complete the next step (as in producing money making
decisions/gaming for SCE). We will need to study the CPUC rules to determine all
significant gaming strategies/loop-holes, we will need to have some early play
with humans to see if the system works and whether "real" people do things we
have overlooked. We also need to test out the data crunching and run time of the
software so that it gives an answer before it's needed. We will then have to
test the AI/statistical algorithms to see if it finds the gaming information and
plays well against humans. Then we have the real January-plus tests where we
need to be on-site closely watching and fine-tuning the model -- jumping in if
we see the model do something stupid and stop Art Channing from doing what the
model said. If we find overbooking useful, the model may be betting on over $50
million per day. A few days of big losses may be frowned upon by SCE management
- - and their contract lawyers.
George
- --------------------------------------------------------------------------------
(Intermediate discussion regarding conference times and flights deleted).
- --------------------------------------------------------------------------------
Author: George Backus
If I flew in, I would not be able to get to SCE until 10:00 for their "meeting."
We would not have time to talk before hand. I could "call" them from your
offices for the conference call but still we would only have a few minutes to
get our act together. I hope Hemant can be on the Monday morning "chat" as well.
Sounds like he has some good experience at making SCE responsive. As such, the
telephone may give us the most time. Should we start early as possible on the
telephone Monday morning with others coming in as they arrive?
I am open to the flight (I think) but it's not clear it helps the Monday
problem. I am gone the next Thursday through the next Thursday I am in San
Francisco on Thursday but the day is full. I can make some contact over those
days but not much. That does mean that next Monday through Wednesday are
proposal go-nuts days. I think I really do need to get something to Heller next
week. I would guess we need to strike while the iron is hot.
Depending on how we agree on Monday AM, I am open to PSC being the lead. I can
then load up the laptop and head your way for Tuesday to work on the proposal or
I can emial my parts on Monday and Tuesday. If we can somehow assume we WILL
make this work, I could come out on Monday to work on the proposal and have the
SCE legal stuff as just a side show. Given everything else I am doing and needs
to be done yesterday, I can only be in LA for one day, however. I'm now
confused. What's your perspective.
George