Simulate a Vehicle’s Components

Automotive engineers make extensive use of computer simulation when they develop and test new vehicles and their components. But testing goes beyond computer models. You need to add real-world data to your model to better simulate a vehicle’s components and how they perform. That data comes from lab and road tests through sensors and data-acquisition instruments. Going further, you can provide real-world conditions in the lab by using microcontrollers and DACs to stimulate the model with realistic operating conditions.

You can create a model that simulates the characteristics of a dynamic system such as a vehicle’s motion, an ABS braking system, an engine, or a transmission. Simulation of a complex model helps you understand how a system will work under controlled conditions as well as under asynchronous or random changes to the input.

One example might be the instantaneous application of the brakes at any point in the simulation—a random yet imaginable condition in the real world. You can also change any variable, such as the shift speeds of a transmission, and see how that will affect other parts of the system’s performance, such as engine rpm.

Figure 1. Mathematical models simulate a vehicle and its components. (Courtesy of The MathWorks.)

Figure 1 shows a model that includes an automatic transmission. The top diagram shows the transmission (center block) as part of a model that includes an engine and a vehicle. The bottom diagram gives details of the automatic transmission’s shift logic.

Each of the blocks in the upper window contains its own models and functional parameters. For example, the transmission model (Fig. 2) consists of a torque converter model and a transmission-ratio model. The model of a torque converter is a three-dimensional array: speed ratio, torque ratio, and K factor.

The transmission-ratio model (Fig. 3) consists of a look-up table that represents the gear ratio. Mathematically, a transmission multiplies the gear ratio by the torque converter’s output. The model also multiplies transmission speed by gear ratio, which loops back to the torque converter.

System models often require loops that feed the results of a computation back into a block in the model. The model simulates a dynamic system. Figure 1 has several loops, while the transmission model (Fig. 2) has its own local loop. Not all loops provide feedback to the system. Some loops can simulate mathematical expressions. For example, you can evaluate a differential equation using integrals and loops.

The bottom diagram in Figure 1 is a state diagram with feedback. It simulates the logic in a transmission that decides when the transmission should shift gears. The shift logic tries to reach a steady state based on external inputs of throttle position and vehicle speed. In the steady state, the transmission is in the proper gear. Otherwise, the shift logic tells the transmission to shift up or shift down, depending on external factors.

Figure 2. A transmission model consists of a torque converter model and a transmission-ratio model. (Courtesy of The MathWorks.)

Figure 3. A transmission-ratio model consists of a look-up table and two multiplication functions. (Courtesy of The MathWorks.)

Where do you get the data to build a model? How, for example, do you determine the inertia of a vehicle, which affects how a transmission will perform? You can guess, you can rely on a previous simulation of the vehicle, or you can collect real-world data.

Collecting real-world data gives you the best model. Assume you want to simulate a new transmission in an existing vehicle. If you don’t already have real-world data, you can perform a road test on an existing vehicle using sensors to measure engine rpm, torque, fuel flow, and temperature. You’ll need a multichannel data-acquisition system, a computer, and signal-conditioning circuits in your vehicle.

Figure 1 is an example of a pure simulation. The simulation runs entirely in the computer, even if the model uses real-world data. Here, the simulation proceeds as fast as the computer can perform the computations. A pure simulation may run faster or slower than the real system.

You can, however, set up a simulation to interact with hardware. According to Martin Schrage, president of Xanalog (North Reading, MA), hardware-in-the-loop testing is used more for testing and verification, as compared to a pure simulation, which has more of a design role. To test a system using hardware-in-the-loop, you replace all or part of the mathematical model with real devices.

Hardware-in-the-loop testing started in the aerospace business where testing a prototype was dangerous and costly. Later, it expanded into the automotive field to speed products to market by reducing the amount of testing needed on the test track.

In a hardware-in-the-loop test, you might simulate a component’s signals using a microcontroller (for digital signals) or you can use a computer-controlled or microcontroller-controlled DAC to generate analog signals. For example, you might use a DAC to generate a simulated throttle signal so you can test a real transmission in a vehicle on a test bed. During the test, the computer uses digitizers and signal conditioners to monitor parameters such as vehicle speed, engine rpm, and temperature. To test the model in Figure 1, you might replace the transmission’s gear-shifting state diagram with a microcontroller. The DAC might simulate the throttle position. For more real-time simulations, you might simulate a vehicle component with a DSP instead of relying on a microcontroller or a PC to control the loop hardware.

Another option is to use the vehicle’s actual hardware in the simulation to accurately evaluate the control-strategy implementation under “live” conditions. In this way, you can evaluate modifications to the design of the control system in real time. You also can assess vehicle performance using operating hardware. T&MW

 You can contact Martin Rowe at m.rowe@tmworld.com