IEEE Power & Energy Magazine - July/August 2017 - 100

frequency at 93 Hz on the dc transmission line reacting with a positive sequence 153-Hz ac system resonance
generated between the rectifier ac filters
and the Long Spruce generating station.
The initial failure of the emTDC simulation based on its slow-responding dc
transducer suggested the problem's fix.
The addition of a capacitor to the input
of the analog current control amplifier
in the bipole 2 dc controls effectively
slowed the current feedback signal to
about a 10-ms response time. This solution stayed in effect for many years,
and the 93-Hz oscillation was never observed again.
Today, using modern graphical
emT tools, the design and analysis
of high-speed controls in new devices
such as renewable generation, SVCs,
and HVdc ties is much easier, and in
many cases the actual firmware code
in the controllers may be embedded in
emT offline simulation models. The
traveling-wave effect of high-speed
transients on transmission lines, previously invisible in the simpler phasorbased models, is now recognized as
a significant factor in transient behavior. a very large industry push for
better models and academic research
(hundreds of papers) resulted in the
frequency-dependent (phase-domain)
models that most modern emT programs use today. machine models are
now extremely accurate. Lightning
performance evaluation, harmonic
studies, and many other types of analysis are relatively easy to perform.
However, the many advantages provided by detailed emT simulations are
limited by significant constraints. The
first constraint is the acquisition and
management of the detailed data required
to obtain accurate simulations, and the
second, once again, is the limit imposed
by available computing capacity.

Real-Time Simulation
Currently, most emT simulations are
performed offline using pCs. The realtime computation of emT algorithms
is also possible using extremely fast
purpose-built parallel processing computers and is now commonly used to
100

ieee power & energy magazine

perform real-time tests of complex controllers before they are put into service.
In 1978, Woodford, then heading
the newly formed HVdc Research Center in manitoba, visited Roger Duggan.
Duggan was working on a real-time
simulator using simple hardware processors at Carnegie melon university.
The desire for real-time computing
was driven by engineers who wished
to test actual hardware (such as protective relays or HVdc controls) within
simulated power systems to ensure that
controls and circuitry would function
as expected when commissioned in the
unforgiving real world.
Seeing potential extensions of the
emTDC algorithm, which was becoming more mature in its offline form,
Woodford secured a Canadian federal
research grant through the national Research Council (nRC) to build a new
real-time digital simulator. However,
because floating-point arithmetic digital
signal processors (DSps) were not widely available at that time, that development effort was unsuccessful until 1986.
It was then that the Dommel algorithm was first programmed into a neC
DSp by a small research team working under Woodford. Later, the nRC
approved more federal money to continue research. The first commercial
real-time digital simulator (RTDS) was
sold to Ontario Hydro for relay testing,
paving the way for its further development. The RTDS was subsequently
manufactured commercially by a new
company in Winnipeg (named simply
RTDS) that undertook further research
and development as competitive products also began to appear.
Today, hardware-in-the-loop, realtime simulation is widely used in HVdc
and flexible ac transmission system
testing applications as well as protection equipment testing. It permits simulation in a variety of system contexts
rather than in a single representative application. The shift to real-time digital
simulation marked a major benchmark
in modeling power systems. until this
point, all simulation was performed
either by mathematical algorithms of
growing sophistication or by physical

modeling, often made more convenient
by the use of similitude rules. The use
of real-time technologies allows the
system context to be arithmetically
simulated, while the prototype being
analyzed is represented by an actual
physical device under test.

The Future of Simulation
although the capabilities of digital offline and real-time simulation have increased drastically since the early days
of emT, power-f low, and dynamic
analysis tools, there remains an ongoing
need to improve models to design better, more economical, and more reliable
power systems. modern power systems
continue to increase in complexity with
the advent of renewable technologies
and their sophisticated power electronic
interfaces to the grid as well as the increased use of smart technologies.
The limits of today's power-flow and
transient stability tools already frustrate
engineers wishing to represent all relevant characteristics of complex networks.
given sufficient data and accurate component models, emT tools get the details
right but run slowly and cannot model a
complete network to ensure all effects of
one area upon another are represented.
power-flow and transient stability tools
cannot always model details with the
precision needed, while certain network
elements such as offshore HVdc grids
cannot be modeled at all. Weak systems
can also cause numerical convergence
trouble for phasor-based tools, and these
tools may introduce approximations into
the control models which have important
performance implications.

Massive Parallelized
Computing
Simulation run times depend on both
the complexity of models and the size
of the network. Since ordinary pCs
begin to struggle with increases in the
demands of emT simulation programs,
the parallelization of processors has
become more attractive. Real-time
tools have long used parallelization to
reach real-time computing speed requirements but require expensive customized hardware.
july/august 2017

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - July/August 2017