Virtual Modeling Helps Build Better Cars

Diesel engines in the past tended to emit large amounts of smog-producing fumes because of their high-temperature compression ratios. But, just as in gasoline-powered cars, a technology called variable-valve lift (VVL) helps reduce emissions by optimizing gas use in the cylinder. Until recently, VVL technology for diesels had lengthy development times and thus targeted premium vehicles where the added costs could be more readily absorbed into the sticker price.

Luckily, simulation software such as from LMS International in France stepped in, helping engineers develop a less-expensive and fuel-efficient diesel engine. Such software also helps engineers perform such automotive-design tasks as eliminating excess cabin noise and developing functional, aesthetic seating.

Virtual variable-valve lift
VVL does the same job as the conventional mechanical cams it replaces. Basically, the technology controls the valves’ sequential opening and closing (lift), depending on engine-load requirements and other factors. Unlike cams, which provide a “one-size-fits-all lift” for all valves in the entire engine-operation range, VVL uses electrohydraulic actuators linked to the engine-control unit (ECU) that individually control the timing of each valve, thereby providing a cleaner-burning engine.

In a recent project, Renault in France created a VVL-development strategy that takes a fraction of the time of past approaches, making the technology affordable for mass-market diesel cars. The project focused on creating a detailed computer-simulation model of the entire diesel engine, including the complex VVL controls using software such as our LMS.Imagine.Lab AMESim.

The virtual engine captures the design process and can readily be reused to develop numerous future vehicles. Engineers would simply enter the different parameters for the particular car model and engine-type instead of creating the entire VVL and engine configuration from scratch.

The virtual engine model of the recent Renault M9Rb engine was created in simulation software selected for its capability to represent hydraulic, pneumatic, electrical, thermal, and mechanical behavior as well as control logic in a single, unified physics-based model. Engineers created the model by dragging, dropping, and interconnecting simple icons in a building-block fashion to create what looks like a simple sketch that shows the relationship of the various elements. In fact, underlying the sketch is an advanced representation of how the physical system will operate in the real world.

Engineers started by using application-specific modules of the simulation software to create detailed representations of engine control, fuel injection, and compression-ignition combustion. Also modeled were the valves, intake and exhaust ports, as well as the turbocharger, low-pressure exhaust-gas recirculation (EGR) loop with exhaust back-pressure control flap, and a particulate filter. Simulation output from the model under various operating conditions accurately predicted engine performance including gas pressure and temperature.

Next, engineers modeled different VVL systems and evaluated them for emission level, fuel efficiency, and engine performance. They also built a representation of the special hydraulic system to integrate the VVL with the engine. A VVL model was selected based on its capacity to provide full coverage of all airflow possibilities from zero to maximum valve lift for optimal engine breathing, according to engine speed and load.

Engineers then developed a simulation method for comparing intake and exhaust states to estimate the amount of residual gas left in the combustion chamber. This is the critical burned gas ratio (BGR) that determines NOx (smog-producing gas) levels. The data was used to establish control rules for the EGR loop and VVL actuator settings as well as for the turbocharger and other elements. Cosimulations were also performed between the simulation software and Simulink to check the control logic in the ECU and VVL system.

Renault engineers fine-tuned the model, using it to investigate different exhaust-gas recirculation configurations, calibrate engine-control codes, and evaluate engine performance in detail. For example, the model let engineers see combustion-chamber behavior including temperature distributions, valve action, and residual gases, in minute steps throughout the entire engine cycle. Thanks to the use of the software, Renault engineers say the M9Rb development stayed on track, timewise.

Transfer-path analysis
Special analysis software provides the best of both worlds when it comes to conventional transfer-path analysis (TPA) and an alternative, response-only TPA approach known as operational-path analysis (OPA). TPA is a common method for studying noise and vibration in automotive applications. In this approach, the way vibrations and radiated noise travel from source to receiver — through structural connections such as engine mounts, for example, or airborne sound reaching passenger ears — is represented with noise-transfer functions (NTFs), a precise ratio of output at the receiver to input loads and forces for each vibration path.

A major drawback of conventional TPA is that considerable time is needed to quantify NTFs for all transmission paths. For example, it takes time to remove the vehicle engine and mounts and install the extensive instrumentation. Technicians position loudspeakers to replicate airborne engine noise. Structural vibrations are induced by placing vibrating shakers where the engine attaches to the frame. Vibration and noise levels are then measured with microphones in the vehicle interior.

To avoid such delays, some companies turned to OPA. Instead of NTFs, this method defines the transfer of vibration and noise energy as transmissibility — the ratio of acceleration or sound pressure at the noise source to that of the receiver. Structural measurements are made using accelerometers at attachment points between the engine and the vehicle body. Underhood acoustic measurements are taken with microphones mounted next to the engine. Other microphones measure the response noise of the combined structural and airborne sounds.

The advantage of this approach: It’s fast. All measurements are taken with the engine under load in one or more sequential test runs where no mount stiffness data is needed. An entire testing cycle is usually done in a day. But OPA is risky and unreliable. Cross-couplings between accelerations can happen, skewing some attachment-point measurements. And transmissibility varies significantly according to where the physical loads are acting. As a result, testing may falsely identify vibration paths or miss them entirely.

Special analysis software such as LMS Test.Lab OPAX combines the best of both approaches. For example, it lets users determine NTFs without taking out the engine. And once engineers input the noise data captured by vibration sensors and microphones, they can study potential remedies by modifying component attributes.

Kia Motors, Seoul, South Korea, for example, first studied a 280-Hz booming noise in a vehicle cabin that happened when the car’s six-cylinder, four-liter engine ran at full throttle. After three refinement cycles over several months, Kia concluded it must redesign the engine — an unacceptable option.

After positioning two loudspeakers in the cabin interior and measuring the underhood response with eight vibration sensors and six microphones, engineers entered the data in the software. Color maps of results clearly showed the culprit noise source to be an acoustic resonance not identified by previous tests. Indeed, the booming noise was entirely eliminated by a simple adjustment to the prototype vehicle, a relatively inexpensive remedy.

Comfortable and eye-appealing seating
Obviously, car seats are more than purely functional. They also affect a driver’s overall feel for a vehicle and sway consumer perception of brand value. So, automotive manufacturers usually go all-out to include a wide range of seat controls for changing the driver-seat height, cushion angles, back tilt, and body contour. To handle such complex mechanical designs comprising hundreds of contact and friction points, companies typically make use of manual calculations, experience with past designs, and countless build-and-test prototypes. Italian automotive-seating manufacturer Faurecia says motion-analysis software helps it simplify this arduous process and develop workable designs more quickly.

One recent project involved kinematic and load calculations on a seat-height adjuster. It was controlled by the driver or passenger via a pump lever with a low operating torque for easy actuation. Pressing the lever rotates a gear set that moves the seat up and down. A rotating ring with fingers that engage slots on the perimeter of the assembly serves as a clutch to hold the gear in place after adjustment. Inside the ring, a rotating cam moves ball bearings that compress a set of springs, creating a return torque that retracts the pump lever into its original position.

“The design was a trade-off among factors such as the fail-safe capability of the mechanism to stay locked into position during a crash, the pump force needed to actuate the mechanism, and the return torque of the ball-and-spring assembly in the control,” says R&D engineer Tanguy Moro. “In particular, the role of friction between the bearings, cam, ring, and springs was critical in providing sufficient return torque for the lever.”

To check these factors out virtually, engineers created a LMS Virtual.Lab model of the seat-height adjuster in motion-analysis software by importing data from Catia v5. Because the two programs work well together, part geometries, assembly joints, and kinematic constraints transferred directly.

Important for success was the motion-analysis software’s capability to represent nonpermanent contact between parts. This is a critical feature because contact plays an important role in determining friction between many of the moving parts. In this case, the height-adjuster model had 18 contact points between the cam, ring, and bearings, letting seat engineers predict normal and tangential forces on these components, and from that, friction.

The software then executed a series of design-of-experiment simulations on the mechanism model based on boundary conditions and constraints engineers typed-in. A comparison of the results for numerous combinations of parameters, particularly friction coefficients between the various parts, identified the variables with the greatest impact on mechanism performance. Results were mapped on a response-surface plot so engineers could more easily see the influence of different design variables.