Marius Rosu, Ph.D.
Koichi Shigematsu, Ph.D.
Edited by Leland Teschler
The handwriting is on the wall. The next generation of vehicles will largely carry electric traction and hybrid-electric powertrains. These powertrains are inherently more intricate than those built solely around internal combustion engines.
Complicating matters is a trend toward decentralized design processes: Large subsystems that must work together in the finished vehicle may be developed on two different continents.
Under this scenario, it is logistically difficult and time consuming to put hardware subsystems together and see how they behave. The only practical way to explore design variations is via simulations in software. No surprise, then, that the high degree of complexity and wide variety of domains involved in this kind of design has fostered use of simulation and analysis tools.
In particular, there has been much attention paid to the modeling of automotive drivetrains or powertrains. In hybrids, the drivetrain increasingly looks like an electrical system. For example, a typical hybrid drivetrain consists of all the components that generate power and deliver it to the road surface. This includes the engine, transmission, energy storage device, motors and motor controller, driveshafts, differentials, and the final drive (drive wheels).
Simulations of these subsystems can take place today at several different levels of detail. For example, there are simulations of electromechanical components that model the behavior of fields in electric motors. But other simulations represent motors at a higher level of abstraction. Here the motor is just a black box that generates specific outputs in response to inputs.
The same can be said for widely used vehicular subsystems such as electronic inverters, regenerative brakes, and various kinds of actuators. High-level simulations can now predict how the overall system will behave. The high-level simulation gets inputs and works with detailed electrical models of component parts.
Use of these simulation packages lets automotive engineers work in one integrated design environment. Software can help predict efficiency and fuel economy of vehicle designs, power management strategies, and guide the electrical design. Engineers can simulate complete drivetrains by combining fuel cells, batteries, power electronics, and electric motors in one model. Many of these components have themselves been modeled and are available for use in multiple designs.
DEVIL IN THE DETAILS
Today's electrical equipment for vehicles can be extremely sophisticated, with a lot of design features to optimize efficiency and minimize costs. Similarly, it is important for drivetrain simulations to accurately model these features to the level of electric field properties, torque qualities, and so forth. These details provide a basis upon which higher-level simulations can accurately simulate the behavior of bigger subsystems.
An example of electric-drive-system architecture might consist of a power supply, inverter, actuator, mechanical load, and appropriate controllers. Simulation software can model this topology both at the system level and with more details that take into consideration different behavioral aspects of each component.
In particular, detailed simulations for hybrid drivetrains often involve predicting the performance of electric motors. Electric-motor technology is taking over from such mechanically actuated automotive systems as power steering, fuel and water pumps, and even the traction system. Brushless motors increasingly handle variable-speed applications where space is tight, as in fuel pumps and electronic power steering.
Brushless motors are widely used in automotive actuators because they are fast, quiet, efficient, and last a long time. Their compact size, controllability, high efficiency, low EMI (electromagnetic interference), and high-reliability are also pluses.
Brushless motors increasingly are deployed in sensorless commutation schemes. This eliminates the Hall-effect or optical sensors normally used to detect rotor position. Sensorless operation is also preferred for rotors immersed in fluid such as fuel, oil, or water. In sensorless control, back-EMF zero-crossing serves as a signal for commutating the motor windings.
Another type of electromechanical system now found in hybrids is the ac variable-speed drive. Since they started to become widely used in the 1970s, their basic mode of operation has hardly changed. A diode bridge rectifies the incoming ac to produce dc. The dc then feeds an inverter bridge that, in turn, produces an ac output to drive the motor. A control loop continuously adjusts the frequency and voltage of the drive's ac output so the motor delivers the right speed and torque.
Regardless of how its components are arranged, the electric drive system has a number of shortcomings. For example, rectifier systems are notorious for feeding high levels of harmonics back into the supply, and they also have a poor power factor. In addition, the so-called dc link between the rectifier and the inverter bridge requires the use of large electrolytic capacitors. These take up a lot of space and are also the components most likely to limit the useful life of the drive: They lose capacity over time, especially when operated at high ambient temperatures.
Finally, the link between the incoming supply and the motor in this conventional design works in only one direction. This is a problem if the motor acts as a generator as, for example, when a car coasts downhill or when the driver applies the brakes. The only way to feed energy back to the supply in these cases is through use of a special converter, which is expensive. In general, braking not only wastes energy but also demands use of a bulky, heat-producing braking resistor as an energy sink.
For these reasons power electronics in modern hybrids often incorporate a modular matrix converter topology. Here the three phases of the drive's incoming supply go directly to the motor through a matrix of bidirectional semiconductor switches. By correctly sequencing the operation of these switches, it's possible to precisely control the voltage and frequency of the output to the motor.
The benefits of the technology include minimal generation of harmonics, regenerative braking without the need for additional resistors, automatic return of energy to the supply during braking, and longer operating life. Matrix-converter drives operate efficiently in all four quadrants. This method of operation reduces harmonic generation to approximately 8% of the level associated with conventional drives, while also offering almost unity power factor.
The voltage converters are based on pulse-width modulation (PWM) of their ac output voltages, which takes place through countless switching operations running at high frequencies. Modern semiconductor power devices (GTO thyristors, MC thyristors, and IGBT transistors) are nowadays used for this purpose. The switching operations are controlled through sophisticated algorithms, which now are based almost exclusively on microprocessors.
Hybrid electric drives must be accurate, stable, and efficient. Their operation demands that designers optimize switching strategies and switching elements alike. This is a complicated multidimensional problem.
A number of simulation packages can work together to give designers an idea of how motors, drives, and other electromechanical systems will interact. For example, RMxprt is specialized software for the design of rotating electric machines. It calculates key performance metrics of electric motors so engineers can evaluate design trade-offs early in the design process. It computes performance metrics for specific designs through a combination of classical analytical motor theory and equivalent magnetic circuit methods. RMxprt also can create statespace system models incorporating the physical dimensions of the machine being designed, winding topologies and nonlinear material properties. These factors are then typically used by another simulation package called Maxwell 3D.
Maxwell3D uses finite-element methods to solve electromagnetic-field problems in the time and frequency domains. Its typical use is to analyze the fields associated with electromagnetic actuators and components. With Maxwell 3D, users solve for such electromagnetic-field quantities as force, torque, capacitance, inductance, resistance, and impedance. The package can also generate-state-space models to help optimize design performance.
Engineers can use the Rmxprt state-space model to explore electronic control topologies, loads, and interactions with drive-system components in another simulation package called Simplorer. Simplorer is a simulator specifically for system simulation including power electronics. It works by modeling power semiconductors with different degrees of complexity.
In Simplorer virtual prototyping takes place using VHDL-AMS language. The VHDL-AMS is a superset of IEEE Std. 1076-1993, a language standard for the description and simulation of multidisciplinary mixed-signal systems.
Either RMxprt design or Maxwell can work in a Simplorer environment to define the behavior of rotational actuators in the hybrid drivetrain. Models developed this way can be configured in different design domains at different levels of abstraction. This sort of description also lets the model be cosimulated with specialized simulation packages and proprietary code, linking to Maxwell for high-fidelity modeling of electromechanical components.