New S.M. Model from Tokyo Electric
"Type-58 S.M. model is now available in ATP" was the title of
a formal announcement that was made April 2nd by Dr. Hiroshi Okamoto
of the Power Engineering R&D Center of TEPCO in Japan. His E-mail of
the Fargo list server declared: "We are very pleased to inform you
that a new type-58 S.M. model is now available in ATP-EMTP. Tokyo
Electric Power Co. (TEPCO) and Toden Software Inc.
(TSI, a subsidiary of TEPCO) have been working on a phase-domain
synchronous machine model in order to improve numerical stability
of EMTP simulation. Mr. Xiang-lin Cao of TSI is the author of the
new type-58 S.M. His new model has the following features:
1) Numerical stability. The type-58 S.M. is based on phase-domain
machine modeling, whereas the type-59 is based on dq0-domain modeling.
The conventional dq0-domain modeling requires prediction of
electrical variables as well as machine variables in order to
interface a synchronous machine circuit with the transmission
network. We found that the prediction of electrical variables
may cause numerical instability in some systems. Phase-domain
modeling of the S.M. can get rid of prediction of electrical variables,
and is expected to improve numerical stability. You can see this
advantage by testing a new standard benchmark data case DCN20.DAT
2) Input data compatibility with type-59 S.M. Data cards for the
type-58 S.M. are fully compatible with data cards for the type-59
S.M. All you need to do, to convert to the new phase-domain S.M.
model, is replace 59 by 58 in columns 1-2 (the source type code).
In addition, you can use both type-58 and type-59 machines in
the same data case."
This story began in the October, 1996, newsletter. Any reader
wanting background is encouraged to consult this initial mention of
the project. Included are illustrations of simulation speed.
Referring to these, Dr. Okamoto*s announcement explained:
"The disadvantage of the type-58 model is that simulation generally
will be slower, since phase-domain modeling requires
re-triangularization at each time step. However, we think that
the difference usually is acceptable."
The Type-58 source code was created and implemented entirely
within Japan, it should be emphasized. Program developers in
Portland simply supplied Salford translator output
(non-commented source code), and it was returned with the new
model installed. Although your Editor volunteered to answer
questions about program structure, none ever was asked.
Upon first seeing the modified source code, your Editor*s
reaction was as follows (from E-mail to Dr. Okamoto):
"First, my observation about structure and programming is this :
there is no observation. Usually when we receive
coding from others, there are many things that require changing.
Most commonly, there are fixed-dimension arrays and
COMMON blocks, or explicitly-typed variables other than
COMPLEX. I was truly surprised to discover that there was
no general need to change anything, in the case of your
Type-58 S.M. code. Your implementation was as professional
and as sophisticated as if Dr. Brandwajn (author of the Type-59 code)
himself had done the implementation. You clearly do understand
the details of Type-59 storage, and have merged your model with it
in expert fashion. You are to be congratulated for a job very
well done *"