IEEE Electrification Magazine - June 2015 - 30

Te∗
Propulsion Motor Plant
TL

ωrm

Vship

∗
ωrm

Propeller Torque (%)

Ship Speed/Thrust Controller

Motor Speed Controller

Te∗

Robinson Curve (Load Machine)
300
200
Increasing Vship
100
0
-100
-200
-300
-100 -80 -60 -40 -20 0 20 40 60

80 100

Propeller Speed (%)
Ship Hydrodynamic Model

Thship

Figure 10. A propulsion system diagram for the PHILS.

microsecond time steps. Hence, the average voltage model of
the PWM converter is commonly used. This model assumes a
constant output voltage during the switching time so that the
time step can be the same as the hundreds of microsecond
sampling (switching) period. This simulation model can be
useful for relatively slow dynamic system levels such as ship
maneuverability, excluding crash stop.
The aforementioned simulation model verification does
not involve practical issues such as switching harmonics and
ripples and the nonlinear property of the BESS. Modeling
these stiff or nonlinear properties is not easy. Hence, the
power-hardware-in-the-loop simulation (PHILS), or downscaled PHILS, is performed. The PHILS setup, shown in
Figure 11, consists of software models and down-scaled
hardware implementations of a ship IPS. The components
that need a small step time or have a higher nonlinear property should be implemented in the hardware, such as the
AFE converter, the BESS, the PWM converter, and the propulsion motor. The components that lack accessibility or safety
should be modeled as software. This includes engine generators and their associated controllers, pulse loads, propellers,
Robinson curve, and ship hydrodynamics.
There are several considerations for the connection of the
software and hardware. One consideration is the synchronization between the simulation model and the actual hardware
component. The simulation should be run in real time just as
with the hardware. As a low-cost solution, a multicore processor can be used to simulate partitioned software models based
on a well-known ideal transformer model. In addition, the data
transfer between software and hardware must be considered.
The one-way transfer from hardware to software can be done
easily with sensors and communication, while the returning
transfer from software to hardware needs additional high-performance voltage amplifiers to closely imitate the PCC. An
example of the amplifier is a high-frequency switching multi-

I E E E E l e c t r i f i c ati o n M agaz ine / j un e 2015

level inverter with an inductor, capacitor, and inductor (LCL) filter, which can cover the fault situation transient due to its
extended voltage regulation bandwidth. Furthermore, the electrical and mechanical dynamics between full- and downscaled systems must be preserved. The power, torque, speed,
inertia, voltage, current, impedances, and other values should
remain the same per unit. Some of the parameters can be kept,
while others need virtual parameters after proper conversion.
This PHILS verification validates most control system levels
such as the current and voltage harmonics as well as fast
dynamic situations including the crash stop, pulse, and faults.

ensuring IPS Grid Power Quality
with Onboard BeSS
The onboard grid power quality is currently an important
issue, especially in naval ship IPSs. Ship crews suffer from various power quality problems such as voltage dips, blackouts,
abnormal changes in the voltage and frequency, electrostatic
discharges, electromagnetic fields, flicker, and harmonics. The
AFE convert and BESS of naval ships enhances both the grid
voltage and frequency quality simultaneously while suppressing the grid current harmonics. Their operation can be effective at steady load conditions as well as at transient load conditions. This article verifies some ship load conditions through
the experimental results of the PHILS. The transient load
example involves an active power load that causes sudden
and severe grid-frequency changes, such as a generator trip,
and an active and reactive power load that causes sudden frequency and voltage changes such as a pulse load.
For example, SDG1 in Figure 9 is tripped during the MGO
mode when LDG1, SDG1, and SDG2 are in line with an
almost 90% load factor in an outage. The total capacity of all
online generators is 4.5 MW, and the total loads demand is
4 MW. Thus, SDG1 trip overloads the system. As shown in
Figure 12(a), the PCC frequency drastically drops within a

Table of Contents for the Digital Edition of IEEE Electrification Magazine - June 2015