electric utility market deregulation

["William Steinhurst"]

Check out "The Dynamics of U.S. Electric Utility Deregulation,"
George Backus and Susan Baylis, US DOE, Office of Utility Technology.

Also, Systematic Solutions Inc., of Ohio has an electric industry
deregulation simulation game thats a lot of fun.

Bill Steinhurst
wsteinhu@psd.state.vt.us

["Shayne Gary"]

Responding to:
> Subject: QUERY electric utility market deregulation (SD0558)

> Has any modeling been done on the electric utility market that is
> similar to the postings Ive seen about the telecommunications industry?

There has been an enormous amount of system dynamics modelling work
done within the electric utility industry. In the U.S., Andy Ford
has a long list of energy journal publications stemming from his
modelling of the industry. One cite is given below.
A. Ford, M. Bull and R. Nail (1989) Bonnevilles conservation
policy analysis models. Energy Policy 15, 109-124.

Also in the U.S., Geraghty and Lyneis utilised system dynamics to examine
utility performance in:
D. M. Geraghty and J. M. Lyneis (1985) Feedback loops: the effect
of external agents on utility performance. In Strategic
Management and Planning for Electric Utilities. (J. Plummer, E.
Oatman and P. Gupta, Eds) pp 249-267 Prentice-Hall, New Jersey.

I suspect there are more recent examples, so perhaps others from the
list can provide other citations.

In the U.K., there has been quite a lot of system dynamics modelling work
done over the last 5 years to examine the effects of privatisation on the
electric utility industry. The industry was privatised in 1991 from
a single, integrated pubic utility into two generation companies, one
transmission company and 12 distribution companies. Indepedent
generation companies were encouraged to build new capacity and
compete with the existing generators to sell electricity into a
market pool.

At London Business School, Bunn and Larsen have published several
articles which examine the market mechanism established in the U.K.
D. W. Bunn and E. Larsen (1992) Sensitivity of reserve margin to
factors influencing investment in the electricity market of England
and Wales. Energy Policy 20, 420-429.

D. W. Bunn, E. Larsen and K. Vlahos (1993) Complementary
modelling approaches for analysing several effects of
privatisation on electricity investment. Journal of Operational
Research Society 10, 957-971.

Also at London Business School, Isaac Dyner recently completed his
Ph.D. thesis using system dynamics to examine exporting
components of the U.K. market mechanism to Colombia.

That should give you some articles to start with...


Michael Shayne Gary
London Business School
44-171-262-5050 ext 3486
sgary@lbs.lon.ac.uk

["Suding, Alan D"]

Has any modeling been done on the electric utility market that is
similar to the postings Ive seen about the telecommunications industry?

Im with a California electric utility that wants to explore what system
dynamics can do to explore strategic questions such as the impact of new
players into a market that has been closed for 80 years. Questions about
pricing, service, marketing, etc.

Im new to system dynamics modeling having just completed a short course
earlier this year. My background is information technology and business.

Alan Suding
818-302-3173
Southern California Edison
sudingad@sce.com

["George Backus"]

In response to the interests of Alan Suding:

An invited system dynamics (words not to be said out loud) analysis of
deregulation dynamics will be presented as the warm up speech preceding
Sect. of Energy Hazel OLearys keynote speech at the annual Western
Systems Coordinating Council (WSCC) in San Diego Nov 21. 9996. WSCC is the
NERC region that brought us the major blackouts this year in the Pacific
Northwest.

The analysis is based on the CIGMOD (Competitive Industry Gaming Model)
originally developed in 1986 explicitly to analyze electric utility
deregulation (a few too many years ahead of its time). The model was used
in the UK for deregulation analysis and is now used heavily by US
utilities. We played the game (it allows any combination of computer or
human players) for the Western Interstate Energy Board (the 11 states on
three provinces of the West) several months ago. Among many commission
and utilities participating, was Enron and Portland General. Enron won
often as it forced the "rules" to change. It won against Portland General
and then against others. Four month later, it played the game for real
when it acquired Portland General. We also have this story for other
utility to utility mergers when we presented CIGMOD analyses in-house.
When the model is played with all computer players, it becomes a nasty
analysis tool that uses gaming and our confidence analysis tool HYPERSENS
to find the strategies that can stop all others despite uncertainty in the
business environment and regulation.

A really neat outcome so far is that the US rules for deregulation are
hopelessly inconsistent from a dynamics perspective. The rules have to be
circumvented or changed (as they are in the model). This generates gaming
behaviors that look and feels like chaotic behavior (also seen in the UK).
What is really wild is that there appear to be only 4 attractors: two weak,
two strong. (The repetitive prisoners dilemma may act as a good metaphors
to explain these -- we say it is so but our BELIEF varies with each new
game we watch.) An experienced user (2 days experience) can quickly see
the "pattern" in the chaos and determine the attractor most likely to
dominate the game and thus act accordingly to win a lot of (real?) money.
The larger the number of "creative" human players that there are (who
generate their own noise), the easier it is for the pattern to be
recognized and reinforced. When the computer plays itself we have NOT
found a logic to tell it what to expect amid the early chaos. (The
computer AI can easily wipe out novices or those who cant "learn" to see
the pattern.) Even those of us who can see the pattern can not explain
specifically what we see because we are assigning a probability to what we
see as the correct interpretation and then acting on it as if it were
definite. We win probabilistically not deterministically. The "knowledge"
seems to come more from our "other experiences" rather than the experience
with playing the model. Thus, the boundaries of the model preclude the
"information" that the model would need to be good at seeing the pattern---
maybe. The wonders of human pattern recognition in survival settings!
There may be a new field of science here in chaotic pattern recognition
with limited data.

Further, the gaming seems to have infinite combinations and possibilities
with dramatic impacts, yet there are only a few finite solution "areas"
that appear to me to defy von-Neumann or Nash interpretations..

BTW, we have seen very sr. mgmt get so involved in the game they truly
sweat out the iteration between moves -- almost panicked. The great
skeptics are our ultimate asset. We initially make it so that the computer
only does tautological accounting such that the model "behavior" is ONLY
from human to human interaction. After a few games where every body
changes the rules, and still the same results occur, the skeptics turn
believers - big time--, and the we kick in the computer players to
enlighten them as to what a really nasty competitor could do. Humans
eventually, however, out do the computer for "legal" nasties. Note that
changing the regulatory rules DO NOT change the "trajectory" toward a
specific attractor. The system evolves from and locks into its
self-created history of environmental (economic and physical) pressures.
Trifling regulatory efforts are no match for a pent-up competitive market.
Note also that the transition always lasts 5-7 years before stabilization
to a "mature" market. This number seems related to the number of trials
needed to determine the preferred trajectory to the attractor. Thus, I am
not really going to say it is chaotic, it may be just an interesting
start-up search noise that needs its own name.

I have a "final draft" report that was prepared for the US DOE on the
implications of these dynamics. When it was presented 18 months ago, it was
considered "totally wacko" by utilities -- a direct quote. Now that 2/3rds
of the prognostications have happened on schedule, the finale parts are
being taken seriously. The report is 65 pages or a 7.5 Meg. Word document.
CIGMOD is provided to the utilities after we teach them to use it. When
running CIGMOD, we have roughly 15 teams of humans playing the computer.
Each team is a company, and the computer can play as many teams as the
human mind can tolerate. When the "real utility data" are used the model
is historically validated to capture each companys "historical inertia"
and "culture" from a decision-making perspective. At this point the game
ceases to be "just a game" for all who play it.

We have applied the model to the UK, South America, Australia, Canada and
the US. So far it is batting a thousand. It does not say exactly what will
happen but it accurately hits the envelope of the trajectory every time.
Its almost scary.

I will send documents and literature (propaganda) email or snail mail to
who ever is interested. CIGMOD itself requires a corporate checkbook for
access. (Note that because of the data and computer gaming, this 100% SD
model requires multi-dimensional array capabilities not available in any of
the "conventional" SD simulation packages.)

Finally, Southern California Edison (Alans company) used SD for modeling
(ENERGY 2020) for many years via us and PACE Consulting. The conditions
associated with Nat Mass leaving PACE caused PACE to scramble in ways not
overly useful to SCE. I dont think the model has been used in the last few
years. The person knowledgeable in SCE is Carl Silsbee.

Because there are long delays in changing the physical electric utility
system but no storage of electricity (immediate physical responses
"controlled" by short-time constant information), and because deregulation,
by definition, changes the system from one operating regime to another,
the "chaotic" transient of the change is much different than I ever would
have expected -- yet it appears to match real world behavior.


George
George Backus
gbackus@boulder.earthnet.net