REAL-TIME simulators do exist and have been widely
used for the analysis of electromagnetic transients
(EMTs) and testing physical control/protection platforms
[1]–[4]. Existing real-time simulators are based on parallel
processing, where multiprocessors, either general-purpose
processors (GPPs) or digital signal processors (DSPs), or
computer clusters are utilized [5]–[11]. The existing real-time
simulators have limitations on the minimum time-step size, the
frequency bandwidth of the simulation results [12], and the
accuracy of the adopted models. Therefore, their application
are limited, particularly when power electronic-based apparatus
with high switching frequencies are of concern. These
limitations are the main motivation behind the development ofa new field-programmable gate-array (FPGA)-based real-time
simulator [13] which addresses such technical limits/challenges
of real-time simulators based on a new methodology for implementation
of the system equations in a FPGA environment.
The salient feature of the implementation methodology of [13]
is that it maintains the calculation time, within each simulation
time-step, nearly fixed irrespective of the size of the system.
This paper extends the parallel implementation methodology
of [13] to simulate ac machines with a nanosecond range simulation
time-step in an FPGA environment. The proposed implementation
methodology for ac machine models enables the use
of a small simulation time-step, in the range of tens to few hundred
nanoseconds. Based on exploiting the nanosecond range
simulation time-step and the large response time of ac machines,
the proposed method: 1) eliminates the need for predictive-corrective
action for machine electrical and mechanical variables,
2) allows parallel simulation of both the electrical and mechanical
subsystems of the machine, and 3) allows the use of the terminal
voltages calculated at a given time-step to solve the machine’s
variables in the subsequent time-step, without compromising
accuracy or numerical stability of the simulation. Based
on the proposed parallel implementation methodology, a massively
parallel (i.e., having a large number of independent arithmetic
units that operate simultaneously) customized hardware
architecture is designed and implemented on a FPGA.