Picture Exp_7 Picture Exp_15 Picture Exp_16 Picture Exp_17 Picture Exp_18 Picture Exp_19 Picture Exp_20

Cloud Implementation For Suspension Optimal Design

Executive Summary

Suspension systems are present in many products in daily life: cars, motorbikes, etc. The automotive market for suspensions and its components (such as shock absorbers) is awfully significant for European industry. Automotive OEMs value vehicle safety and comfort, and so the technical specifications for components are high. This demand for safety and comfort is an outstanding characteristic in the EU5 market (France, Germany, Italy, Spain and United Kingdom). Global market of shock absorbers was estimated in $220 billion in 2016. Companies like KYB and Showa reported in 2016 annual turnovers of $3.2 billion and $2.5 billion, respectively. The EU5 is a net exporter of shock absorbers with figures close to €2.1 billion in 2016. The International Monetary Fund (IMF) predicts a considerable rise in the Gross Domestic Product (GDP) of all of the EU5 countries between 2016 and 2021. Estimated GDPs range from 4.0 in Spain and Italy to an impressive 5.8 percent in UK.

Outside of the passenger car market, there are significant opportunities with respect to shock absorbers, such as the OEMs that manufacture commercial vehicles, agricultural equipment and trailers. Special attention must be paid to off-road and competition vehicles. Suspensions are key elements in this type of vehicles and have a big influence on handling and performance. For this reason there is an increasing interest on the R&D sector on vehicle competition championships to provide the best quality of suspension adapted for each type of vehicle and each racing category.

In this scenario, the goal of this experiment is to provide cloud-based tools that allow for finding an optimum design for a given use of suspension systems considering the application–specific requirements and constraints. The solution proposed is based on the use of a general-purpose software tool for multi-body simulation. Therefore the applicability of the experiment results is not limited to just suspension systems but after slight adaptation can be used in many practical problems arising in industry.

Furthermore, this experiment is not just looking into the prediction of the behaviour of a mechanical system based on a set of design variables; it aims at finding the optimum values of the design variables in order to make the mechanical system achieve the desired behaviour. The challenge lies in finding the design variables efficiently using a “smart” optimization approach.

The current process of finding an optimal design is an iterative procedure requiring several physical prototypes, testing them, varying design parameters, building and testing again and thus subsequently approaching an optimum – or at least a design, which fulfils the requirements. Maybe better designs exist but cannot be elaborated due to time and cost constraints. The current process strongly depends on designers’ experience and is slow and expensive. In the market of vehicles for competition, designing and testing a single modification of an existing shock absorber may take 3 to 4 weeks and the associated costs are from € 3,000 to € 5,000. In the case that modifications had to be performed in the vehicle and suspension geometry costs may rise up to € 50k. Costs and time for a single iteration are spent without certitude of getting the expected results. Rarely the optimum solution is found in the first attempt; therefore this process has to be repeated 3 or 4 times which in practice makes the process unfeasible due to budget and the competition calendar limitations. A complete new design may take 2 to 3 months and the total cost can amount to as much as € 500k. In this context, simulation promises to explore a wider design space and find better solutions in the same or shorter time with a fraction of the expenses associated to the current process.

The challenges of the approach conducted in this experiment is to efficiently use HPC resources for the compute- demanding multi-body-systems simulation and to implement an efficient automated optimization procedure which deducts design variables from many simulations’ results. Thus, the use of this software tool will reduce the need and number of physical prototypes as well as will find better design solutions. This goal is achieved by a design of experiment (DoE) approach including sensitivity analysis of the goal function and constraints with respect to the design variables.

TECHNICAL IMPACT

The end user, Donerre is able: a) to change and enhance the design process by using simulation tools, b) to reduce the time required for the design of new products, c) to allow the R&D team to test design alternatives that would be unfeasible in a typical methodology based on real ground test and d) to improve product quality. The experiment results prove that the user is able to quickly introduce changes on the design parameters and rapidly analyse the effects in the vehicle behaviour with a very short learning curve. In the experiment the user devoted less than one training day to learn the use of the cloudified solution. These features lead to outstanding reductions (5 to 10 percent) of the lapse time devoted by the engineer in the design of a new product as well as to better understanding of vehicle dynamics. The new approach contributes to improve the quality of the final product. As far as simulation results are accurate (i.e., differences between simulation results and experimental measurements in a vehicle are lower than 10 percent), the use of simulation will reduce the need for building physical prototypes and will allow exploring a large number of design options, including modifications in the shock absorbers and the vehicle suspension.

STT contributes to the experiment with CMechStudio: a general-purpose multi-body simulation software. STT as ISV expected the following technical impact: a) reduction of execution time required by the parametric analysis between 10 to 50 percent, b) reduction of execution time required by an optimization problem by 10 percent to 20 percent, c) new possibilities to run parallel simulations on Cloud/HPC infrastructures, thus allowing solving more complex problems/models in less time. Within the experiment run tests were performed with three models: a simplified two-dof suspension model, the rear suspension system of a racing motorbike and the full model of a car boogie. The results obtained from these test cases proved that the technical objectives have been fully achieved.

ECONOMIC IMPACT

The simulation tool allows analysing a single modification in the design of an existing shock absorber in less than one week leading to savings in man labour of about 1 to 2 weeks. This time reduction is important especially if several iterations had to be performed. The number of physical prototypes will also be reduced. Expected overall cost savings will be higher than 50 percent.

Cost of simulation tools is also an important factor. In most cases SMEs cannot afford the use of simulation tools due to the price of the licenses. For example, the price of buying a license of MSC Software Adams Car is € 50k plus 20 percent of mandatory maintenance for 1 year or € 10k/year per licence. This factor limits the access of SMEs to simulation software. The pay-per-use approach proposed in this experiment will make simulation tools affordable to SMEs.

Using simulation tools Donerre will solve some crucial issues of racing vehicles. This competence could motivate customers to select Donerre instead of another brand. Donerre expects to create a new job for a design engineer in 2018. Total income expected from the execution of this experiment in the next three years is € 70k approximately.

STT, the ISV, estimates to increase in 2018 the turnover generated by the simulation tools in 2016 by a factor of 5 percent and 10 percent in the next three years to a total amount of € 145k. It is expected to create a new job for a mechanical engineer in 2018.

From the perspective of the HPC provider, Bifi, this experiment has contributed to improve BIFI‘s HPC-Cloud infrastructure. The expertise acquired will help BIFI in obtaining new projects and contracts. Bifi expects increasing sales of HPC resources to users of simulation tools; expected income accounts for € 30k in the next three years. It is expected creating 3 new jobs.

Suspension Optimal Design Suspension Optimal Design