The push to produce hybrid electric vehicles (HEVs) is stretching the imagination of automotive engineers around the world. It is also proving to be fertile ground for the latest computer modeling and simulation tools.
Two major players in the national effort to build HEVs have created simulators based on products from The MathWorks that have become mainstays in HEV design. The Advanced Vehicle Simulator (Advisor), developed at the National Renewable Energy Laboratory (NREL; Golden, Colo.), and the PNGV Analysis Toolkit from Southwest Research Institute (SwRI; San Antonio, Texas) both take advantage of Simulink's user-friendly, object-oriented environment for developing models, performing dynamic system simulations and designing and testing new ideas. The tools have set the course for HEV development at national labs, universities and among the Big Three automakers. Researchers and automotive system designers believe that these tools provide the means to reach the next level of advanced vehicle design.
The nation's efforts to build practical and efficient hybrid electric cars by 2004 began in earnest with software developers involved in the Partnership for a New Generation of Vehicles (PNGV), a government initiative announced by President Clinton in 1993. With increasing U.S. dependence on foreign oil imports-most of which is used to fuel America's cars-and increased awareness of the perils of air pollution, the subject of alternative transportation had risen to national attention. HEVs seemed to offer the best hope of a near-term solution. Computer models and early prototypes indicated they could boost fuel economy and cut emissions while still providing the safety, performance and range that American drivers demand.
Under the PNGV program, the NREL signed contracts with Chrysler Corp. (now DaimlerChrysler), Ford Motor Co. and General Motors Corp. to produce hybrid propulsion systems by 1998, first-generation prototypes by 2000 and market-ready HEVs by 2003. Engineers at the Big Three are working to meet the PNGV's lofty goal of tripling conventional gas mileage to 80 miles per gallon. Ford and GM, which have been in the program since 1993, are now testing their concepts while Chrysler, which joined the initiative in 1996, is still in the modeling phase. In addition to the industrial partnerships, the NREL has sponsored much HEV research at the nation's leading engineering schools. More than 20 organizations now use Advisor. The tool is continually fed up-to-date component test data through user and university validation efforts.
To truly understand the value of simulation tools like Advisor or the PNGV Analysis Toolkit to HEV design, it helps to look under the hood of a hybrid vehicle. All use an electric motor and a bank of high-voltage batteries in conjunction with a "heat" engine, usually a conventional internal-combustion engine. The components are arranged either in parallel or series configurations. Computer modeling with Advisor shows basic characteristics for each.
In a parallel hybrid vehicle, both the electric motor and the conventional engine are connected mechanically to the drive wheels. The conventional engine provides supplementary power when the vehicle is accelerating or climbing steep grades. Parallel HEVs do not need generators and have lower mass and better highway efficiency. In addition, they have redundant drive power, which means the car can still be driven if one of the power systems fails. However, parallel vehicles require a complicated mechanical coupling and a multispeed transmission. Also, since the internal-combustion engine must perform over a range of speeds, tuning can be difficult and efficiency is diminished.
In a series hybrid, only the electric motor drives the wheels. The conventional engine runs an alternator that provides power to the electric motor and the battery bank. Series HEVs have simpler mechanics, allowing more freedom for component placement. Because the electric motor provides enough torque throughout the range of required speeds, a single-speed gear box can take the place of the transmission. Full power is available at all times in a series HEV, and because the internal-combustion engine runs only when the batteries need charging, it can be tuned to run efficiently across a well-defined and controllable speed range. On the downside, series HEVs lack the redundant drive systems of their parallel counterparts and they are extremely sensitive to electric motor and battery efficiency. Their ability to negotiate long uphill grades is also questionable.
For the two basic HEV configurations, various control strategies not found in conventional vehicles must be considered. While the efficiencies of each component are important, it is the ways in which they are interconnected that determine the value of an HEV design. Enter Advisor, with its ability to model all of the HEV components, monitor their interaction and response and produce performance data. Advisor is termed "steady state" and "backward looking" because its simulations begin with desired outputs and work back toward component choices and configurations. With computer modeling and simulation, designers can evaluate any number of possible component and design configurations quickly and cost effectively. The component and configuration combinations that will meet performance and fuel economy goals readily reveal themselves and can move into the prototyping phase. Those that will not meet the goals, no matter how ideal the conditions, are easily discarded.
With all of the components modeled, Advisor allows HEV designers to use a variety of standard and custom driving cycles like the Federal Urban Driving Schedule and the Federal Highway Driving Schedule. Drawing on the power of Simulink, Advisor can calculate battery state-of-charge corrections. The simulator also provides designers with quick and accurate assessments of key vehicle performance data, such as fuel economy, emissions, acceleration and grade sustainability. Advisor results can then be analyzed with Matlab to find equilibrium points or ideal operating conditions.
Advisor's GUI made it easy for users to enter and modify elements, including drive train configuration, transmission, vehicle type, energy storage system, generator, driving cycle, control strategy and scaling. The GUI corresponds to a top-level Simulink block diagram that shows the model's data flow from trip, through road load, motor and energy storage and on to numeric outputs for fuels used and tailpipe emissions. Looking down two layers into the road load subsystem shows the interaction of next required speed and previous speed through other subsystems representing roll, climb, aerodynamics and acceleration, to arrive at figures for torque and speed required at the virtual vehicle's wheels.
In 1996, the NREL tested five vehicle configurations with Advisor-three lightweight (series, parallel and internal combustion) and two conventional weight (parallel and internal combustion)-in pursuit of the PNGV's goal of 80 mpg. Simulink's model-based graphical simulation environment allowed designers to manipulate each component and determine the strengths and weaknesses of each design. Using Advisor, NREL engineers found that parallel and series vehicles showed the same fuel economy sensitivity to most parameters, but series vehicles were three times more sensitive to battery and electric motor efficiency. The results also revealed that to reach the 80-mpg goal, internal-combustion-engine efficiency would have to be raised to nearly 35 percent and vehicle weight would have to drop to 2,200 pounds (1,000 kilograms), about half of today's conventional vehicles.
Although the conventional-weight vehicles modeled did not make the fuel efficiency goal, they did manage a 20 percent improvement in fuel economy over today's cars. The NREL went on to specify 176 distinct vehicles for evaluation, verification and data storage. The findings will likely fuel research into battery and electric motor technology and will accelerate the NREL's stated goal of having most metal automotive components made of aluminum by 2015.
Last year, the NREL used Advisor in a new test that allowed for different control strategies in the series and parallel HEVs, an analysis made feasible only with computer simulation. These valuable tests showed that the best-designed parallel vehicles achieved 24 percent better gas mileage than conventional cars, surpassing their series counterparts by 4 percent. In addition, Advisor has been well validated at the NREL and at a number of universities that use the simulator in their HEV design programs. Advisor's computer-generated findings have generally fallen within 2 to 4 percent of actual measurements on HEV prototypes, a figure of merit that shows excellent alignment between design models and real components. A validation study at Virginia Tech in 1997 found that "the validation process shows that Advisor has extensive values as a simulation tool for HEVs."
Meanwhile, software engineers at SwRI began to look for a way to take HEV development beyond steady-state, backward-looking models like Advisor and offer a dynamic environment for designing, testing and proving their ideas without hardware prototypes. Like the NREL's HEV designers, SwRI engineers adopted Simulink so that they could compile modular component libraries of engines, vehicles, mechanical couplings, driveline components, energy storage devices and a number of other essential HEV subsystems.
At the direction of the United States Council for Automotive Research (Uscar), a consortium of DaimlerChrysler, Ford and GM, SwRI used those libraries as the foundation for its PNGV Analysis Toolkit. Participants in the PNGV since its inception, engineers in SwRI's Engine and Vehicle Research Division were familiar with most of the widely used HEV simulators, including Advisor and Simple-V. SwRI chose Matlab and Simulink to develop its PNGV Analysis Toolkit because of their programming efficiency, Simu-link's self-documenting feature and customer familiarity with the tools.
Both the PNGV Analysis Toolkit's GUIs and component libraries offer a layer of design beyond simply choosing components, providing vehicle models with the required detailed descriptions of the components' performance values. In addition to testing the performance of the overall vehicle, this model-based design paradigm enables designers to validate an HEV's subsystems. Despite the dense models, even novices can use Matlab GUIs in the Toolkit.
The PNGV Analysis Toolkit has been used to evaluate new drivetrain options, to test and debug control strategies, to size and specify subcomponents and to evaluate software and hardware modifications. Uscar now owns the software. Its members-engineers at the Big Three-will use the Toolkit to develop HEVs that meet the PNGV's goals. Many involved in the PNGV program agree that dynamic and forward-looking simulators offer the best chance for fine-tuning HEV components, designing and testing advanced control strategies and optimizing HEV performance. The Toolkit may ultimately find wide use in conventional vehicle development as auto engineers move toward increased virtual design and testing.
Scott McBroom, Senior Research Engineer, Southwest Research Institute, San Antonio, Texas, David Helinek, Automotive Segment Manager, The MathWorks Inc., Natick, Mass.