Professional Documents
Culture Documents
Designing Energy Efficient Traction Machines For Electric and Hybrid Vehicles
Designing Energy Efficient Traction Machines For Electric and Hybrid Vehicles
P. Wendling
(Altair Engineering, USA);
Abstract
The design of powertrain motors for electric or hybrid vehicles presents many
challenges. The designer faces multiple constraints on weight, compactness,
cost, efficiency and temperature rise. The efficiency of the motor impacts
directly the autonomy of the vehicle. In a race to sustainability, designers need
innovative methodologies and fast tools to meet the many requirements of
electric mobility.
Rotating machines are traditionally optimized at their rated operating point. For
powertrain applications, motors operate over a wide range of speeds and
torques. The efficiency of the motors must be optimized considering the full
driving cycle of the vehicle. Today, such optimization can be achieved using
new dedicated and fast analysis tools. These design tools deliver reliable
evaluation of the motor efficiency throughout the driving cycles. These tools
help the designer select the most efficient machine topology early at the
predesign stage.
The impact of the losses on the temperature rise can be evaluated. An adjusted
design can mitigate the losses and better meet the thermal constraints of the
machine. A coupled magnetic/thermal simulation is needed. In such a
computation, electromagnetic losses are the sources for the thermal problem.
1. Introduction
The design of the traction machine is split into two phases. In the first phase, to
perform the optimization over a full driving cycle, assumptions are made in the
calculations, enabling higher solving speed and keeping reasonable fidelity of
the results. In the second phase, more accurate models are used, leading to a
better evaluation of the losses in the machine. However, only one
representative load case is considered. For this load case, the impact on the
losses of the PWM power supply is evaluated.
Using this two-step approach, the selection of an optimal structure for the
traction application can be achieved. This will be illustrated through an
example.
known to deliver high torque and high power as well as high efficiency. The
machine outer dimensions are set. It is a 3-phase machine. A 54 slots stator has
been chosen. During an initial pre-design study, 4 different rotor topologies are
evaluated. Table 2 shows the performances for these topologies at the base
work point. Rotor A and B offer the best performances. However, Rotor C is
chosen for cost and manufacturability reasons.
Once the rotor topology has been selected, the winding scheme must be
chosen. The design of the winding can also be easily performed using the same
tool. Table 3 shows key parameters for 4 different configurations of the
winding.
The stator and the rotor are made of laminated sheets (M330-35A) and the
permanent magnets are made of NdFeB (Br=1,05T).
Figure 1: Efficiency map and locations of the work points considered to evaluate
the efficiency over the WLTP duty cycle.
DESIGNING EFFICIENT TRACTION MACHINES FOR EVs and HEVs
Fig. 1 shows the efficiency map. It displays the efficiency for each load point
on the driving cycle. The global efficiency of the powertrain is the sum of the
efficiencies over driving cycle.
The computations are based on the Park model [2], using the direct and
quadrature characteristics of the PM motor. The table of the flux linkages, Φd
(direct axis) and Φq (quadrature axis) are calculated using a static finite
elements solution. The linkage inductances are evaluated in function of the
currents id (direct axis) and iq (quadrature axis). The iron losses in the stator
are evaluated based on local values of the magnetic flux density. These losses
are adjusted with the speed. These tables are used to evaluate the transient
behavior of the motor and the losses, for each load point. The value of the
currents (direct and quadrature) for each load point are quickly obtained using
an optimization procedure.
This method gives a very good compromise between accuracy and solving
speed: the saturation effects are fully considered, but no finite element transient
solving is required. At this stage, iron losses in the rotor and eddy current
losses in the permanent magnets are neglected.
The tool used for the optimization is Altair HyperStudy. It directly couples to
FluxMotor. It is based on a global response surface optimization method
(GRSM). The rotor resulting from the optimization is shown on Fig. 2. This
new topology leads to a global efficiency of 73.7% compared to the initial
efficiency of 70.9% for the initial topology. The base torque is 181.5 N.m and
the base speed is 6226 rpm.
Magnetostatic computations are quick and efficient. Their use is key during the
pre-design stage of a machine where speed is needed to evaluate multiple
topologies. However, the losses in the rotor, especially the eddy current losses
in the permanent magnets, must be considered for more accurate values of the
motor efficiency. To evaluate the losses, a transient simulation with motion of
the rotor and circuit coupling is performed using Altair Flux software. Only
one work point is considered. The speed is imposed in the model.
The work point is chosen to represent the typical load the motor will have to
sustain during long periods of time: motor running at base speed (6226 rpm)
for half base torque (90.75 N.m). This work point is also interesting from the
thermal stress point of view.
and quadrature inductances (Ld and Lq) [3]. The current fed to the motor is
computed from this simulation and exported into tables of current versus time.
These tables are used as sources to a transient finite element model where the
losses will be evaluated.
In this study, the finite element tool directly integrates an electric circuit. The
equations describing the circuit are solved simultaneously with the magnetic
equations in the finite element matrix system. The PWM signal is introduced
in the definition of the voltage sources connected to the motor by comparing a
sine wave to a triangular signal (intersective PWM). The chosen switching
frequency is 5,6 kHz. A parametric study is performed to get the values of the
voltage and control angle to be used to reach the considered work point. The
current is displayed in Fig. 3.
The losses are evaluated on a full cycle at steady-state. The time sampling,
needed to capture the effects of high frequencies in the current, leads to the use
of a small time step. Because of the time constant of the machine, several
cycles are needed to reach the steady state operations. To reduce the number of
time samples needed to steady state, the transient computation is initialized
with a static computation. The static computation establishes the magnetic flux
produced by the permanent magnets. Also, to damp the numerical transient and
shorten the solving time, the resistivity of the winding is increased during the
initial time steps. Permanent magnets are considered as solid conductors in
which eddy currents and losses can develop. Iron losses are calculated using a
Bertotti type model [4]. The mesh for this computation is shown on Fig. 4.
Close to 4500 time steps are necessary to obtain the results.
Figure 4: 2D mesh of the machine using periodicities and using about 6800 nodes
Table 6 compares the results obtained with and without PWM and the impact
can clearly be identified on the iron and permanent magnet losses.
A critical constraint for the machine is the temperature the insulating parts and
the permanent magnets can support. Using the joule losses averaged over a
steady-state period as heat sources, a thermal computation is performed. It is
based on a 2D model in which conduction is solved using the finite element
method. Heat exchanges at the boundaries of the domain are evaluated using
simple convection coefficients. The machine is considered to be water cooled
from the outside. A transient simulation is performed to evaluate the
temperature distribution in the motor after running for 3 hours at base speed
and half base torque. Fig. 5 shows the temperature distribution for two
simulations, the one on the left without PWM and the one of the right with
PWM. Because of the higher harmonics in the current, the model with the
DESIGNING EFFICIENT TRACTION MACHINES FOR EVs and HEVs
PWM included shows higher temperatures in the rotor than the model without
PWM. The temperature of the magnet is key to the survival of the magnets.
6. Conclusion
7. References