Good vibrations: Using a driving simulator for NVH development
In article 5 of 8 of our Motion Series of articles, Cruden’s Dennis Marcus and Bastiaan Graafland explore the challenge of deploying a driving simulator to handle high-frequency inputs demanded by NVH testing
A decade ago, OEM vehicle development was often conducted in silos. Communication between specialised departments such as vehicle dynamics, body and structure and powertrain development was sparse and it was rare for multiple departments to share the same test tools.
That’s all changed now. At many car makers, specialists from different functions are increasingly working together, sharing knowledge and trying to integrate their research setups into centralised systems. One example is engine test benches and brake test systems being coupled with driving simulators to get human input on design decisions. Greater collaboration during vehicle development is leading to shorter time-to-market and a better product for the end customer.
Driver-in-the-Loop (DIL) simulators are an important point of integration for different engineering departments. One of the most recent departments to join the DIL party is noise, vibration and harshness (NVH), whose engineers are seeking ways to conduct subjective and more immersive experiments.
For many years, NVH research has been conducted by doing unit tests on, for instance, hydraulic, multi-axial simulation tables (MAST) as supplied by MTS, which also makes the actuators for Cruden’s DIL simulators. As you can see from the images below, both setups consist of a hexapod, but the two systems are very different.
- Hydraulic, multi-axial simulation tables (MAST) are commonly used for NVH testing. MTS system pictured.
- Hexapod motion platform with electromechanical actuators.
The MAST system carries nothing but a seat. In this well proven unit test approach, the seat is the test subject. Using accelerometers on the seat, it is exposed to an input signal that excites the seat rail. The response of the seat can be measured and evaluated; this generates reliable and objective data for the new seat design.
Vehicle engineering however is not only based on objective data, but also to a large extent on subjective aspects. These subjective aspects can be evaluated by a person in a DIL simulator where instead of just the seat, an entire vehicle body might be mounted to the motion platform.
- To collect subjective NVH feedback from humans, an immersive DIL simulator environment is necessary. Pictured: DIL simulator at the Toronto
- Rehabilitation Institute.
For the management and expert driver evaluations of, for example, two possible new engine mount designs, it is valuable to experience the NVH aspects of the two designs in an immersive DIL simulator environment instead of, or in addition to, comparing test result data sheets.
It is for these reasons that Cruden’s customers are considering combining, if they have not done so already, the more traditional DIL simulator use cases like ride and handling and ADAS testing, with NVH research. Depending on, for instance, the bandwidth requirements for the experiments, it is however not always completely straightforward to specify a DIL simulator that provides the best of both worlds.
We saw in our previous article how a motion system should be tailored to its application. If the goal of using a motion system is purely to provide immersion for ADAS testing for example, then a basic, more affordable motion system may be specified. By contrast, if the NVH department is also interested in using the simulator with accurate high frequency excitations, then a powerful and high bandwidth motion system will be required. This is challenging to combine with large workspaces and will drive up system costs.
So far in our Motion Series of articles, we’ve predominantly discussed vehicle dynamics simulation. The input frequencies from a real-time vehicle dynamics model are usually no higher than 15 Hz and therefore the motion systems traditionally used for DIL simulators can accurately replicate inputs at frequencies up to 15 Hz. Higher frequency components may still be part of the input signal, however their output will be subject to larger phase shifts and magnitude fluctuations – not only because of motion base performance limitations, but also because of compliance in the moving structures.
In addition to the 6-DOF motion system, Cruden also uses shakers to create engine and powertrain vibrations at higher frequencies, based on vehicle model states like vehicle speed, engine RPM and engine load. The lower accuracy in replicating high frequency input has only limited negative effect for most of the traditional DIL simulator use cases, since they are mainly used to increase the immersion and not for the evaluation of vibrations experienced by the driver. However, for NVH experiments, accuracy at higher frequencies is essential, which requires another approach.
The frequency response characterises the dynamics of a motion platform. It is a measure of attenuation and phase shift of the platform accelerations as a function of input frequency in all degrees of freedom including crosstalk attenuation. A better frequency response can be achieved by applying the measures listed below:
- Reduction of the total moving mass (actuators, platform and mock-up)
- Increase in the stiffness of the payload (platform and mock-up)
- Increase in the stiffness of the motion system (e.g. by reducing the actuator stroke)
- Using more powerful actuators
- Lower effective mass by reducing rotating inertia and use of direct drive (non-geared) actuators
- Introducing dedicated HF controls
Actions 1 and 2 will compromise immersion because they tend to make the mock-up look and feel less like an actual car. Unless significantly reducing the actuator stroke and with that the workspace of the motion system, items 3 and 4 will add to the costs significantly.
Finding the right compromise is an interesting but challenging process, which typically results in a different outcome for every group of users. If the NVH uses cases are focused on transfer path analysis in the higher audible frequency domain, there are no immersion-related requirements for the motion platform. If, however, the NVH studies include subjective validation of the comfort level of, for example, the vibrations felt through the structure of an autonomous vehicle, an immersive simulator with a motion system capable of accurately reproducing high bandwidth motion is needed.
Simulation software: real time vs replay
When using a driving simulator for vehicle dynamics studies, the goal is to allow the driver to drive the car and to assess how it handles. This can only be achieved with real-time simulation where the virtual car responds to input from the driver. In this simulation the characteristics of the car, engine, road surface and the inputs of the driver are used to calculate the input signal for the motion system. This real-time requirement will often not work for NVH studies focusing on the higher frequency domain. The calculations to generate the signals are too complex for real-time simulation, so for NVH studies it is more common to generate the input signals off-line. The result of an off-line simulation is an input signal, also called a drive file, for an experiment. Next to simulation-based drive files it is also common to record drive files whilst using real (reference or competitor) vehicles.
Using off-line simulated or recorded drive files allows for an iterative approach that yields high frequency flat magnitude and near zero phase shift that goes far beyond the specified bandwidth of the motion system. With this approach several reference points on the motion system are equipped with accelerometers. By comparing the input signal with that of the signals measured on the reference points, the iterative optimisation software will determine how to change the input signal such that in the next run the measured output signal will better match the original drive file. In the following iterations the result will be evaluated again and the input will be adjusted again to more closely match the original drive file with the signal measured on the reference point.
The accuracy of the match will depend on the structural properties of the simulator, the motion system performance, the length of the drive file and the number of iterations. The stiffer the motion system and moving structure, the easier (i.e. quicker) it will be to iterate to a drive file that will create accurate motion. However, stiffness typically comes with mass (m) which means more force (F) is needed to achieve the required acceleration (a). Yet another compromise to be made.
When using an off-line iterated input signal with the simulator, the driver or passenger will experience a replay of a scenario instead of a real-time and interactive driving experience. However, in comparison to the real time simulation, the accuracy of the motion in the higher frequency domain, increases significantly due to the iterative process. As the accurate reproduction of higher frequencies is essential for evaluating the comfort aspects of a vehicle, it is acceptable to work with a replay without driver interaction. With an iterative process, accurate reproduction of frequencies up to 60 Hz can be achieved using an electromechanical hexapod motion system.
At Cruden, our engineers are always happy to help and are very good at answering the list of questions you’ll have about your simulator project. Please don’t hesitate to consult us.
For more information, please contact Dennis Marcus via email@example.com or on +31 20 707 4646.
If you think a colleague would be interested in receiving the articles in this series, you can sign them up to our newsletters here.
Links to subsequent articles will be added below as they are published.
View all articles in our Motion Series of articles: here.
Article 1: Driving Simulator Motion Systems 101
Article 2: Space: The Final Frontier! Why bigger is not always better when it comes to driving simulator workspace
Article 3: Cascading motion systems: How to balance the need for workspace with simulator complexity, agility and costs
Article 4: Motion for immersion VS motion for vehicle dynamics and motorsport
Article 6: Acceleration is not enough: there’s more to accurate motion cueing than meets the eye
Article 7: A real car in a virtual world: the pros and cons of chassis mockups in driving simulators