Alessandro Toso

Performance Comparison of Real-Time and General-Purpose Operating Systems in Parallel Physical Simulation with High Computational Cost

Real-time simulation is a valuable tool in the design and test of vehicles and vehicle parts, mainly when interfacing with hardware modules working at a given rate, as in hardware-inthe-loop testing. Real-time operating-systems (RTOS) are designed for minimizing the latency of critical operations such as interrupt dispatch, task switch or inter-process communication (IPC). General-purpose operating-systems (GPOS), instead, are designed for maximizing throughput in heavy-load systems. In complex simulations where the amount of work to do in one step is high, achieving real-time depends not only in the latency of the event starting the step, but also on the capacity of the system for computing one step in the available time. While it is demonstrated that RTOS present lower latencies than GPOS, the choice is not clear when maximizing throughput is also critical.

Real-Time and Real-Fast Performance of General-Purpose and Real-Time Operating Systems in Multithreaded Physical Simulation of Complex Mechanical Systems

Physical simulation is a valuable tool in many fields of engineering for the tasks of design, prototyping, and testing. General-purpose operating systems (GPOS) are designed for real-fast tasks, such as offline simulation of complex physical models that should finish as soon as possible. Interfacing hardware at a given rate (as in a hardware-in-the-loop test) requires instead maximizing time determinism, for which real-time operating systems (RTOS) are designed. In this paper, real-fast and real-time performance of RTOS and GPOS are compared when simulating models of high complexity with large time steps. This type of applications is usually present in the automotive industry and requires a good trade-off between real-fast and real-time performance. The performance of an RTOS and a GPOS is compared by running a tire model scalable on the number of degrees-of-freedom and parallel threads. The benchmark shows that the GPOS present better performance in real-fast runs but worse in real-time due to nonexplicit task switches and to the latency associated with interprocess communication (IPC) and task switch.

Multibody System Modeling of Flexible Twist Beam Axles in Car Suspension Systems

Dynamic simulation techniques that are based on Multibody system approaches have become an important topic in studying the performance of various mechanical components that comprise an automotive system. One of the challenging issues in such studies is the ability to properly account for the flexibility of certain parts in the system. One example where this is important is the design of twist beam axles in car suspension systems where twisting deformations are present. These deformations are geometrically nonlinear and require a special handling. In this paper two multibody system approaches that are commonly used in overcoming such problem are examined. The first method is a sub-structuring technique that is based on the popular method of component mode synthesis. This method is based on dividing the flexible component into sub-structures, in which, the linear elastic structural theory is sufficient to describe the deformation of each sub-structure. Using this method the deformation of the beam is described using the mode shapes of vibration of each sub-structure. The equations of motion, in this case, are written in terms of the system’s generalized coordinates and modal participation factors. In the second method a Multibody System (MBS) solver and an external nonlinear Finite Element Analysis (FEA) solver are coupled together in a co-simulation manner. The nonlinear FEA solver, in this case, is used in modeling the deformation of the twist beam. The forces due to the nonlinear deformations of the flexible beam are communicated to the MBS solver at certain attachment points where the flexible body is attached to the rest of the multibody system. The displacements and velocities of these attachment points are calculated by the MBS solver and are communicated back to the nonlinear FEA solver to advance the simulation. The two methods described above will be reviewed in this paper and an example of a twist beam axle in a car suspension system model will be examined twice, once using the sub-structuring method, and once using the co-simulation method. The numerical results obtained using both methods will be analyzed and compared.Copyright

Integration of Time Waveform Replication Process in a Multibody Software for Reverse Load Identification

In the automotive industry, the need to meet the durability requirements in a very early stage of the development of a new vehicle model is becoming more and more crucial. This is a key factor that can reduce the time to market and avoids modifying substantially the design if a component fails earlier than expected. This is also a challenging task for several reasons; in the early phase the primary design suffers from a lack of knowledge about the loads that the new vehicle will experience in its life. In literature ([1][2][5][6][7]) several methods have been proposed; for instance the so-called digital test track approach ([1]) is a CAE-based tool in which the vehicle and the road are modeled in a multibody environment together with a detailed representation of the tire and the driver in order to perform a replication of a test drive. This predictive method is very valuable but still requires a lot of information about the vehicle’s components that is usually not available at this stage of the vehicle development. On the other hand a pure test-based procedure suffers from other problems such as the need of a mule vehicle and long and costly test campaigns that need to be repeated at each component’s modification. A hybrid approach has then been proposed and implemented successfully by LMS on industrial size cases. This approach known as Time Waveform Replication (TWR) ([2]) relies on a set of test data and multibody model available from test drives carried out on a predecessor or a vehicle similar to the one that is being currently designed. The data collected on a road test is used to back-calculate the equivalent spindle displacements that will cause the same forces on the multibody model that are experienced in the test sessions. This approach has several beneficial aspects with respect to the two mentioned before. The tire model does not need to be accurate since the displacement are applied directly to the spindles (but the application can be easily extended to “road profile identification” if a detailed tire model is available). Moreover it is well known that if the forces measured at the spindles are applied directly to the unconstrained multibody model, it will result in an undesired drift of the model due to a mismatch in the mass and inertial properties between the real vehicle and its model. This is even more important when measured forces are applied to a new vehicle model that is only similar to the tested one. The TWR approach relies on a linearized model of the vehicle that is derived directly from its multibody representation. Then the spindle displacements are back-calculated by pseudo-inversion of the Frequency Response Function of the system and the application of the desired target signals. This method gives a direct result only if the system is linear; this is typically not the case in the field of vehicle dynamics where the geometry of the suspension, the non-linear properties of the dampers and bushings together with the intrinsic non-linear nature of the constrained equation of motion implies that the linearized model used by TWR is valid only for small changes to the configuration at the instant of linearization. To cope with this problem, the TWR sets up an iterative process that uses the output error to update the input. In case of high non-linearities or large changes in the configuration the linearized model can be also updated. In this paper the integration of the TWR process in a multibody code such as LMS Virtual.Lab Motion is described. In particular a new tool named LMS Motion-TWR has been developed. The application guides the engineer in setting up the models inputs and outputs, allows to drive the multibody code to compute the linearized model and the association between the test data and the numerical responses of the model. The computation of the driving signals is performed by TWR core solver as a background process allowing the user to focus on the analysis of the results rather than spending time in dealing with file conversion and transfer from one software to another as was done in the past. Moreover several post-processing tools are available such as time and frequency domain plots, RMS error and X-Y plots. Finally this paper describes the application of the tool in an industrial case scenario using a model of a quad. A quad was equipped with several sensors and driven on a test track. The collected data is then used in the Motion TWR software to compute the equivalent spindle displacements. Since some of the front suspension parts are modeled as flexible bodies the reverse load identification analysis is completed by a durability calculation.Copyright

Handling and Primary Ride Comfort Development in Early Design Stage by Means of 1D Modeling Approach and Multi Attribute Optimization Process

Today the automotive market is ever more competitive and vehicles must satisfy the requirements of the customers in all respects: handling, comfort, acoustics, fuel economy, etc. Therefore OEMs have to launch innovative products in a short development timeline: the time to market (TTM) of new vehicles has continually decreased and nowadays the developing process of a new car is completed in less years than in the past. This scenario emphasizes the role of CAE in the vehicle design engineering design and the necessity of exploiting its potentialities, in order to shorten the TTM and to reduce the impact of experimental tests on it. In this context a step-by-step approach with multi-physics 1D environment such as LMS Imagine. Lab AMESim is proposed in order to monitor vehicle performances in all the design stages, thanks to the employment of models with increasing complexity. In addition the ultimate step can be employed for performing a multi attribute optimization on vehicle performance metrics in order to find the best attributes balancing and to pass the preliminary recommendations to the design with a considerable time-saving respect to 3D MBS models. This paper briefly describes the process for building 1D models with LMS Imagine.Lab AMESim and moreover it shows the definition of a multi attribute optimization algorithm in terms of handling performances with the most complex model.Copyright

A Comparison of Different Multibody System Approaches in the Modeling of Flexible Twist Beam Axles

One of the challenging issues in the area of flexible multibody systems is the ability to properly account for the geometric nonlinear effects that are present in many applications. One common application where these effects play an important role is the dynamic modeling of twist beam axles in car suspensions. The purpose of this paper is to examine the accuracy of the results obtained using four common modeling methods used in such applications. The first method is based on a spline beam approach in which a long beam is represented using piecewise rigid bodies interconnected by beam force elements along a spline curve. The beam force elements use a simple linear beam theory in approximating the forces and torques along the beam central axis. The second approach uses the well known method of component mode synthesis that is based on the linear elastic theory. Using this method the deformation of the beam, which is modeled as one flexible body, is defined using its own vibration and static correction mode shapes. The equations of motion are, in this case, written in terms of the system’s generalized coordinates and modal participation factors. The linear elastic theory is used again in the third approach using a slightly different technique called the sub-structuring synthesis method. This method is based on dividing the flexible component into sub-structures, in which, the method of component mode synthesis is used to describe the deformation of each substructure. The fourth approach is based on a co-simulation technique that uses a Multibody System (MBS) solver and an external nonlinear Finite Element Analysis (FEA) solver. The flexibility of any body in the multibody system is, in this case, modeled in the external nonlinear FEA solver. The latter calculates the forces due to the nonlinear deformations of the flexible body in question and communicates that to the MBS solver at certain attachment points where the flexible body is attached to the rest of the multibody system. The displacements and velocities of these attachment points are calculated by the MBS solver and are communicated back to the nonlinear FEA solver to advance the simulation. The four approaches described are reviewed in this paper and a multibody system model of a car suspension system that includes a twist beam axle is presented. The model is examined four times, once using each approach. The numerical results obtained using the different methods are analyzed and compared.Copyright