In the high-octane world of motorbike racing, a balance between simulation and physical testing can be crucial to improved performance, writes Robert Roe
The development of motorbikes requires both simulation and physical testing – but it is the knowledge generated through these processes that is of the biggest benefit to users. It can be used to understand complex unsolved challenges, as well as helping users to develop experience that can be shared across an organisation, thereby driving more efficient development of future projects.
The development of new vehicles or other automotive products requires a careful balance between simulation and physical testing, at the very least, to validate the model used for the simulations. While analytical methods for simulation may seem to compete with physical testing, in reality the two practices work to complement the overall strategy and to help generate knowledge about the specific product.
Pete Dodd, vice president of system dynamics at MSC Software, explained that his work focusing both on physical testing and simulation has helped him to understand the balance that must be struck between the various disciplines that go into automotive development.
He said: ‘In general when it comes to balancing design, simulation and testing, you want all of your components that go on to make the final product to integrate well. That means that you want every component to perform its intended function. You don’t want any redundancy; you do not want anything to be over-engineered or under-engineered – and of course, at a component level, you need to meet the design objectives.’
‘You want to learn as much as you possibly can to improve your design process and that is why this balance of simulation and testing is important,’ concluded Dodd.
This knowledge generation can come in two distinct flavours. On one hand it can be used to develop an understanding of a particular component or design and this knowledge can be shared throughout an organisation. But simulation can also be used to understand complex phenomena – which, in the case of racing motorcycles, can be extremely difficult to test with physical prototyping, because of the conditions that need to be met in order to repeat the test accurately with a satisfactory precision.
An example of this is the research carried out by Yukio Watanabe, senior development and consulting engineer at Mechanical Simulation, and Robin Sharp, visiting professor at the university of Surrey, a motorcycle researcher who first defined the vibration modes that have been attributed to weave and wobble in the 1970s.
The two researchers used a simulation tool called BikeSim – of which Dr Watanabe is the chief developer – to predict and identify a vibration mode called ‘chatter’. This affects MotoGP racing teams when operating at high-speed and with high cornering loads – defined as more than 50 degrees of lean, with lateral acceleration of about 1G. Chatter was found through studying the combined vibrations of motions in steer axis rotation, fork/chassis twisting rotation and longitudinal fork bending.
Bike Sim is used to provide accurate forecasts for motorcycle dynamic behaviour. The latest edition, offered as a stand-alone product or through Altair’s Partner Alliance, also includes upgraded features such as a comprehensive powertrain model, a more detailed riding position and refined tyre characteristics to improve the accuracy of simulation results.
Watanabe commented: ‘These new features will further assist design engineers by providing an accurate and repeatable alternative to track testing in motorcycle design.’
A paper published by the two researchers in 2012: ‘R S Sharp & Y Watanabe (2013) Chatter vibrations of high-performance motorcycles, Vehicle System Dynamics, 51:3, 393-404, DOI: 10.1080/00423114.2012.727440’ explains that the term ‘chatter’ was used by Cossalter et al in a previous study, but some uncertainty remains about its meaning.
High speeds, high pressure
While chatter had previously been used to describe a vibration in the front wheel under hard braking, this did not accurately describe a different phenomenon that was caused during cornering at high speeds, normally only found by MotoGP teams under race conditions.
The researchers created a model describing a Suzuki GSX-R1000 motorbike, and then tested this under various conditions using BikeSim. While the motorcycle model makes up the core of the software, the suite also includes a database of machine, tyre and rider parameters, a graphical user interface, road surfaces and profiles, a plotter and an animator.
At the high speeds and with a very high angle of lean during cornering, as described in the paper, Watanabe explained that this is effectively a different system from that of a motorbike under normal conditions. This is largely because the suspension effect is reduced as they are not directly over the wheel due because of the very steep cornering angle.
Dr Michael Sayers, CEO and founder of Mechanical Simulation, stated that understanding these complex problems would not be possible without simulation tools such as BikeSim. Sayers said: ‘Without simulation this kind of testing is just not feasible because of the difficulties in recreating these conditions with physical testing.
While normal physical testing could not provide a comprehensive answer to describe this phenomenon, the simulation environment provided by Mechanical Simulation – through its BikeSim software – was used to identify the existence of a vibration mode that aligns with the experience of the riders using these bikes under these cornering conditions.
Watanabe explained that further research using BikeSim tools for linear analysis showed that all motorcycles have the potential for chatter vibration as well as weave and wobble motions. However, the chatter is usually stable and doesn’t resonate under normal riding conditions; it mainly appears under race conditions when the chassis structure is not stiff enough.
The use of simulation, in this case, enabled the researchers to identify and understand this vibration mode which was very hard to test accurately without simulation because of the very steep angles that these racing motorcycles would achieve while cornering at the very edge of the limits of the tyres’ grip.
Limits to physical testing
Another aspect to consider is that while the physical testing could be used to identify the problem, it cannot be used to elucidate the root causes fully, and thus cannot always be used as a tool to give the researchers and engineers knowlege about how to fix the problem.
This is an area where simulation can perform much more efficiently than physical testing alone, as a sensor does not need to be attached to every critical part as physical testing would require. In a simulation environment, you can collect much more information, which can then be used to influence future design changes.
‘The physical testing tells you the “what” but not necessarily the “why”,’ said Dodd. ‘Understanding the “what” and the “why”, means you can learn about your design and you can make progress. You don’t have to do the same thing next time you design it; you add to your knowledge and make things better the next time around.’
One way to increase the knowledge generation process is to couple different simulation disciplines into a single environment. This can be used to provide a more comprehensive solution to a particular problem, as changing one system or component can have a knock-on effect on another.
Dodd explained that recognising that this ability to generate more knowledge from simulation is seen as critical to driving automotive design for MSC users – not only because it provides more information to users but also because it can increase the fidelity of the simulation response. Dodd said: ‘Recognising that, and working with our customers in the past, we have provided the capability to couple these tools. We can run a co-simulation, which is where two of these tools talk to each other as they are doing the analysis, and they can swap information.’
One example that MSC software supplied to illustrate the use of MSC software in the motorcycle industry, was a project that they worked on in partnership with Mahindra Motorcycles, an India-based motorcycle manufacturer that wanted to improve the design.
In the past, the design of new two-wheeler models at Mahindra was based on building prototypes and driving them on a test track. The obvious limitations of this approach were that prototypes took an average of five weeks to build, and had to be run for about two weeks to evaluate component durability – leading to a long development cycle and the extra costs associated with the construction and testing of these prototypes.
A major improvement to this system came from the development of test rigs, which were introduced to recreate the conditions of the test track, using automated equipment that eliminated the need for a driver and could be operated 24 hours a day. Although this approach saved time, costly vehicle prototypes were still required for each major design change.
The final solution was to use MSC software – specifically ADAMS for its multi-body dynamics capabilities – to simulate this vehicle and the test track. In this way, the process can be largely automated and used to optimise the process, as many different design iterations could be tested in a much shorter period of time and at a significantly reduced cost.
ADAMS is an acronym that stands for Automated Dynamic Analysis of Mechanical Systems. ‘Multi-body simulation, as the name suggests is taking multiple parts together’ stated Dodd. He went on to clarify that an example of this is a motorcycle chain, which is made up of many links, all joined together by a pin. ‘Now that is a very simple example, but it is a good case of a system with multiple bodies that these individual links interact with to describe the behaviour of the overall system, in this instance, the chain’ said Dodd.
This process can be extended to all kinds of components such as gearboxes, engines, the crankshaft, the connecting rod or the pistons and how they interact with the stiffness of the cylinder block. Dodd said: ‘All of these components are moving relative to one another; all of these components are generating forces, and the kinds of motions can be rotational, translational. Sometimes this can be at relatively high frequencies – a car engine, for example, can typically spin at 6,000 to 8,000, RPM but a motorbike can be up to 16,000 or 17,000 RPM.’
Multi-body dynamics enables users to break down the complex systems made up of multiple moving bodies so that the interactions can be fully understood. This gives engineers a much better understanding of the forces involved, and this can be translated into stresses and strain on components, which can be used to improve design performance or to look at a product’s fatigue in a certain set of conditions.
While this process did not involve any simulation – merely building a test rig that enabled the researchers to move testing from the road into the lab – the second step involved creating an analytical model that could replicate the actions of the test rig, so that the process could be moved into a simulation environment, reducing the need for physical testing other than to validate the analytical model itself.
Dodd said: ‘These tests allow the team to build up confidence in the predictive nature of ADAMS and FEA, so that they can then use this virtual rig for testing out possible designs.’
Simulation on the rise
This process highlights a trend in the automotive industry where many companies want to move as much of the design and engineering to computer-based simulation.
This was a sentiment shared by Altair, as Michael Johnson, senior application engineer at Altair explains: ‘A trend in motorcycles, and for the wider automotive community, is moving validation of components from the road to the lab, and now to the screen – which provides significant cost and time savings.’
Another trend highlighted earlier is the desire of many automotive companies to build up as much knowledge as possible. This can be passed through the company to other teams and individuals – so that engineers do not have to re-invent the wheel every time they start a new design.
By coupling simulation tools, and by exploring and solving challenges previously unsolved through purely physical testing, simulation engineers generate a huge amount of data. The next challenge is to understand how to utilise this data fully to generate knowledge for the organisation that puts all the time and effort into these simulations.
Altair provides a comprehensive software portfolio – Altair Hyperworks – which enables users to take a product from the initial design stages through simulation to the final product, including structural analysis, durability, safety, NVH, CFD, aerodynamics, multibody simulation (MBS) , vehicle dynamics, optimisation, materials analysis, engine/powertrain, and control systems and model-based development.
By offering all of these solutions, and those of the Altair Partner Alliance – of which Mechanical Simulation is a member – means that users can couple simulations as well as helping users to move data quickly from one stage of the vehicle design process as it is needed.
This is also true of MSc software – as Dodd explains – by using a software portfolio allowing users to concentrate on the engineering rather than moving data and selecting the appropriate file formats: ‘Using an analytical tool like ADAMS gives you the ability to post-process the results. You can start looking at loads, stresses and strains, different characteristics around the whole vehicle to try and understand why something happened. It is very difficult in a physical test, unless you happen to have put testing instruments in the right area and you can’t instrument them all.’
‘In theory, you can retrieve information from any part of the model when it is on the computer – but you can only understand how the part performs under physical testing if you have got some kind of instrumentation in the right spot.’
Using simulation allows engineers to switch their interrogation as soon as some new piece of information indicates a problem could be solved elsewhere – rather than having to re-tool, create new prototypes and design new experiments, as is the case with physical testing.
Dodd said: ‘It gives the ability to learn because you can interrogate the model to find out why something happened. If you came up with a poor design for whatever reason, then you can understand why it did not work and then you can make sure you don’t do that again.’
‘It’s not enough to say that as an individual will not make that mistake again – we need to pass that information on as a company to make sure that the enterprise does not make that same mistake again,’ concluded Dodd.