Simulation shifts gear
The pressure on vehicle manufacturers to deliver cheaper and more efficient cars intensifies year on year. Concerns about global pollution have led to ever-changing rules and regulations to reduce vehicle emissions and improve fuel efficiency. Manufacturers are having to respond rapidly and constantly – and often there is significant variation between geographical territories. One significant government move has occurred in traffic-ridden Brazil, which has recently introduced an incentive scheme to encourage innovation in vehicle technology. The Inovar-Auto scheme offers tax incentives to manufacturers who can demonstrate significantly reduced fuel consumption for vehicles manufactured between 2013 and 2017.
‘As regulations change, the burden on engine designers is increased,’ says Frederick Ross, director of ground transportation at CD-adapco, whose Star-CCM+ software is used by automotive suppliers worldwide. ‘Example of these changes can be reduced weight, downsizing of engines, reducing frictional losses in the engine. With all these changes, there is also demand for higher output. Designing vehicles with both higher output and lower weight engines is a difficult engineering challenge.’
A key tool in meeting these challenges is simulation software, and many different packages are used in powertrain design. Alongside finite element analysis for structural testing, an engineer might also turn to fluid dynamics to check the flow of oil, air or coolant around an engine, or combustion cycle analysis to assess the behaviour of fuel.
The way in which vehicle manufacturers use simulation software has evolved over time. ‘Initially, we dealt mainly with stress analysis, for example, the stress on the intake/exhaust valves during operation,’ says Ross. ‘Computational fluid dynamics (CFD) for air flow and coolant flow was added to the mix around 20 years ago, looking at flow into the combustion chamber, plus the path of the coolant flow around the cylinders. Today, we are looking at simulating complete systems of components, including stress, thermal and flow in a single simulation. Identification of thermal stresses is very important, and simulation helps to identify where potential cracks may occur within the engine structure, and guide the modification of the design accordingly. Often the worst thermal stresses occur under non-standard operating conditions, so our simulation tools need to be able to account for the full range of possible use scenarios.’
The reasons why simulation software is used has not changed, however – it saves vast amounts of time and money.
‘Our customers, such as GM, Jaguar and Volkswagen,’ continues Ross, ‘are moving towards virtual prototypes, replacing expensive physical prototypes with numerical simulation, which is advantageous both in terms of cost and in that fact that virtual prototypes can be deployed much earlier in the design process, helping OEMs to identify and correct potential problems before they even occur. For example, simulation helps them to predict the performance of an engine over a wide range of operating conditions, such as, for example, driving up a hill or towing a trailer. Using simulation they can also model the performance of the engine over a complex drive cycle, such as stop-and-go city driving.
‘This approach also allows the engine system to be optimised to achieve maximum performance. Once a physical prototype is built, it’s very difficult – and expensive – to make changes, so simulation saves both time and cost.’ Even though there are savings against those incurred by generating physical prototypes, simulation software itself has had to become more efficient, as the packages themselves still need to be configured and executed. This is an area that CD-adapco has been addressing.
‘There are also some standard processes that we have managed to almost completely automate,’ says Ross, ‘such as the coolant flow through an engine, or the tumble and swirl of fuel and air in the intake ports. Rather than worrying about the details of simulation, these automated workflows allow engineers to concentrate on refining the design. This type of automation can also be used in conjunction with optimisation software to identify the best design configuration automatically. Today’s optimisation algorithms, such as Red Cedar’s Sherpa, efficiently help engineers to come up with the optimised designs quickly, involving a large range of design parameters.’
Managing multiple simulations
Esteco’s modeFrontier software is used extensively in the automotive industry by customers such as BorgWarner Turbo Systems, which recently employed it for the optimisation of a turbo charge compressor, and by Cummins to integrate GT-Valve train and GT-Power models for valve event optimisation.
Esteco’s president, Carlo Poloni, says: ‘There are many types of software simulation used in powertrain development, from finite element analysis for structural development, to fluid dynamics and combustion cycle analysis. All of these tools are very specialised, but our software helps link all these tools together. We can iterate a series of simulations, according to the needs of the designer, in the background. To ensure any process or part is optimised, it is necessary to carry out multiple calculations using multiple tools in order to generate a complete performance map. modeFrontier acts as a software robot that executes a number of simulations according to these optimisation goals. Our software not only runs all these simulations, but passes data between them constantly. The post-processing functionality it offers means users can quickly work out what is best and for what reason. It provides business intelligence to the engineer.’
A similar product is on offer from MSC Software, as Peter Dodd, VP for system dynamics, explains: ‘A software package such as Adams is particularly suited to analytical development because it enables information from different types of physics – such as structural, fluid flow, hydraulics and gas dynamics to be coupled and solved simultaneously. Typically, these would each need a different code. We’ve satisfied a user desire to be able to connect these different codes together. Adams enables information from the different codes, which each may have been taken at a different point in time, to be extrapolated and interpolated accurately.
‘In order for these different codes to work together, a standard interface (or protocol) is required. Adams supports FMI (Functional Mock-Up Interface) which is an industry-standard tool for coupling disparate software codes.’
Dodd also believes that speed of analysis is an increasing factor for today’s engineers. ‘With more of the design development done on the computer, the analyst wants an accurate result in a shorter period of time,’ he says. ‘Coupled with that, dynamics solutions involve solving for equilibrium many times during the same analysis – the time between each solution being dictated by the frequency content of interest in the model. So, an analyst using Adams to predict the ride comfort of a passenger car may only be interested in frequencies up to, say, 70Hz whereas the analyst modelling the engine of that car would likely be interested in frequencies of 1kHz or more.’
Control design challenges
Alongside the mechanical design aspects of engine and transmission manufacture, there are also considerations on control design. Not that long ago, the presence of a single ECU (electronic control unit) within an engine was considered the very height of sophistication. Today’s production vehicles have anything between 15 and 50 such controllers. Maplesoft’s Maplesim is one of the tools used in the design of the controllers and how they interact with each other, as Paul Goossens, VP for application engineering, explains: ‘The worlds of mechanical and control design are starting to merge quite significantly. Traditional model-based design (MBD) tools can certainly handle the control design, but in order to develop the controller, you need to have a pretty good understanding of the dynamics of the system that you’re going to control. This has been awkward and time-consuming to do in these tools, so that’s where we come in – allowing engineers to develop the plant model through a multi-domain tool that can understand electronics, hydraulics, mechanics and thermal properties.
‘Often, this process will occur before the mechanical design is complete. Engineers will have an idea of what the mechanical design parameters will be, allowing them to sketch out the control design using our plant modelling tool, and put it through some duty cycles to see how it behaves. The model that we create can be used for testing – first using software-in-the-loop that effectively exercises the code with a virtual engine through many drive cycles. The next stage is hardware-in-the-loop, which would usually be a prototype embedded controller, and is then run in real time. All of this can be done before a prototype engine needs to be built.
‘The functional specifications that come out of our part of the process are then used in the final mechanical design process.
‘In our world, the engineer can play around with things like numbers of cylinders, bores, strokes and so on. He can get it to a point where it is working the way it should, then pass it over to the CAD guys to do their mechanical design.
‘Customers are demanding tighter integration between what we do and what CAD does – these have been two largely separate entities until now, but they are starting to merge.’
Simulation software is now an essential part of the powertrain design process, and as the quest for more efficient vehicles continues, engineers are using these packages with ever more intensity to deliver the results the market needs.
GKN Driveline is a leading supplier of automotive driveline systems and solutions, working with the majority of vehicle manufacturers worldwide, providing products such as constant velocity sideshafts, propshafts, differentials, all-wheel drive systems and eDrive systems.
‘As a company, GKN Driveline wants to be in a position where we get our product right first time,’ says Steve Minter, global engineering/ITS director at GKN Driveline. ‘We intend to advance our simulation capabilities to a point where we’ll only ever have to build one prototype, and be absolutely sure that it will work. This will make us much more like an aerospace company than a tier-one automotive supplier.
‘Lead times are much shorter than ever before. Moreover, both rig and vehicle testing are very expensive – and slow. The same is true for physical prototypes, so we want to reduce our physical testing to an absolute minimum. Getting our simulations as close to reality as possible is a key part of our strategy.
‘In addition, where there is a need for repetitive simulation, we look to automate it. We recognise that our experienced simulation staff are a valuable asset. We should be using them on innovating the next generation of products, rather than on repetitive tasks. So, over the past few years, we have developed a knowledge-based engineering system that has helped to automate many such tasks.
‘The way we work with customers is that, usually, we will be given vehicle performance data and the packaging in which the product will have to live. We’ll also be given details about any interfaces – i.e. how and where our product will attach to the rest of the vehicle. After that, it’s over to us – the customer entrusts the design details to GKN.
‘Customer requirements have increased steadily in recent years. The performance and reliability expectations for modern vehicles are immensely high, the vehicles themselves are becoming more and more complex, and there are more variants. A lot of products that were once purely mechanical are now intelligent, electronically controlled and using embedded software. That takes us into whole new areas of simulation.
‘20 years ago, we might have been satisfied with linear static analysis, which would help prove that the component wouldn’t break the first time you moved the car. The durability and fatigue testing would have been done physically, either on a test rig or via an accelerated vehicle test.
‘Nowadays, we won’t just look at the static strength. There are many non-linear calculations, along with models for crash, durability analysis, NVH (noise, vibration, harshness) analysis and so on. There is also great pressure on weight reduction, which takes you into areas such as shape optimisation, where specialist software can help you do the best you can with the material available.
‘We use a range of simulation software at various stages of our processes, but during the early design stages, our preferred tool is PTC’s CreoSimulate. It uses a very clever p-type solver, and allows people with relatively modest simulation skills to produce useful results very quickly. This speeds up initial design loops considerably.
‘A big question that has emerged in recent years is “What do you do with the data?” We are generating data at a pace undreamed of 10 years ago. So, having some sort of simulation process and data management system in place is now essential. We believe that the ability to effectively deploy simulation and manage the resulting data is a key differentiator for our business.’
PSA Peugeot Citroën
Altair’s HyperWorks – and in particular its OptiStruct finite elements solver product – is used by many major manufacturers, including PSA Peugeot Citroën. Julien Guyé, project manager at Altair, says: ‘The accuracy of the simulations has increased in all fields, as well as size of the finite elements models. Simulation-driven design – where you start with simulation and optimisation before undertaking any CAD – has become a mainstream way of thinking, including intensive usage of optimisation techniques to leverage mass reduction and faster design.’
As a result of the changing demands of the industry, Altair is always revising and updating its tools. ‘A recent example is two years of hard work to implement a powertrain-dedicated solution,’ says Guyé, ‘from 3D component meshing (for which we provide a meshing tool called Simlab) to dedicated features in OptiStruct.’
OptiStruct is used for topology optimisation – that is, starting with the available volume and allowing the software to optimise the best use of that volume for the engine part required. After this design is interpreted using CAD and validated using CAE, OptiStruct with its embedded morphing technology, can then be used to optimise the shape and size of the design, to ensure the most efficient use of materials. ‘The entire process can leave you with a shape you weren’t necessarily expecting,’ says Guyé.
Bernard Charlet, engine design justification specialist at PSA Peugeot Citroën, has been using Altair’s tools for several years. ‘At the time we started using OptiStruct, its features were quite unique,’ he says. ‘When Altair later acquired the provider of crash simulation software that we had already been using, it smoothed the process of being able to use the two in conjunction with each other.
In a recent project on oilpan design for PSA Peugeot Citroën, the optimisation process began with crash analysis, followed by linearisation of crash loading. This led to a sub-model and subcases for optimisation, followed by topology optimisation. Finally, the optimisation results were integrated into a crash analysis to validate the concept obtained. This process took much less time than the classical cycle of designs and computations.