Simulation yields better bearings

The conventional calculation method for determining bearing life is often referred to as the 'catalogue method'. The bearing industry has agreed to these particular calculation methods found in the specification ISO 281. The basic life calculation considers bearing load, speed, load rating and bearing type. However, bearing life calculations need to be much more complex if they are to yield the accurate predictions of bearing life and reliable operation that many of today's application demand. Increasing use of alterative materials, lubrication considerations, life expectancy, performance; engineers are continuously pushing the boundaries. This means that the understanding of 'in-service' bearing behaviour needs to be much more accurate. The variety of bearings available as commercial-off-the-shelf has steadily been increasing and this means that it is possible to get a bearing that is very closely optimised to particular operational conditions. Perhaps the most powerful tool available to engineers to help in the process is virtual simulation. Early virtual simulation is getting more ubiquitous and evermore accurate. Many of the major bearing manufacturers have been developing their own software to make sure that design engineers have the tools they need. Getting it right at the early stages has profound knock on benefit. One area in particular where bearing selection has become particularly crucial is in the renewable energy industry. Turbines used to capture energy from the winds, tides and waves need to be hardy to the external environment and are expected to work for 20+ years. However, the practicality is that bearings are not lasting this long and in many cases are prematurely failing. Bearing manufacturer SKF has made significant effort to develop its understanding of what happens to bearings during this type of operation. It has found that gearboxes of wind turbines are suffering from persistent and unusual failure. The gearbox is one of the most expensive parts of a wind turbine system and premature failure adds to cost through turbine downtime as well as unplanned and expensive maintenance and repairs. The failures that were occurring were abnormal and though SKF had encountered similar failures in other industries they were not at the rate encountered in wind turbines. "We don't think the failures are due to classical fatigue," says Paul Meaney, manager of the Renewable Energy Application Development Centre for SKF. "There are other mechanisms that are driving the failures." As the typical calculation to select bearings is based on stress and fatigue it had to question the tools and calculations used. The failures were more due to dynamic effects so the calculation had to include the dynamic effects and time domain so to better capture, at the design phase, events which could lead to this sort of failure. "It is difficult because you traditionally look at the loads, lubrication and cleanliness and you can make a good prediction about the life of the bearing," says Meaney. "Now we are saying there are dynamic events which will occur that will initiate some damage. It is not that the older methods are not valid, it is just that they are no longer sufficient." The company has been using and developing a multi-body simulation tool called Beast to better understand the behaviour of bearings in application. It is used during the product development phase. However, as it is a computer intensive tool it is not a tool you will apply for every bearing position for every design you do. Rather it is used where it is critical. "It is used to optimise the bearing for particular applications and will be used in critical positional," Meaney adds. "Beast is one tool that we have; it is not the only tool. It looks at a single bearing or a very small system. We have other tools that look on a system level like a complete gearbox." NSK too has been exploring virtual opportunities for optimising bearings for wind turbine applications. Its technical experts have developed sophisticated programs to better understand bearing dynamics and static load conditions and increase the accuracy of estimating bearing life. Application data, environmental conditions, structural stiffness as well as many other influencing factors can be simulated by NSK to better predict the life in a given application. For wind turbines it is the excessive wind velocities and resulting dynamic loads that dramatically impact the turbine and subassembly. Just calculating the lifetime of this type of bearing requires a large degree of expertise. This is why selecting bearings for wind turbines is more complex than in other areas. Numerous parameters must be taken into account. Besides the bearing loads and the rotational speeds in the application context, design engineers must consider the construction in which the bearing will be placed, i.e. the configuration of the shaft and housing, their materials, and their tolerances. In ISO 281, annex 4, the calculation of modified service life rating is based on simplified rolling bearing geometry. In order to increase the accuracy of the results of these calculations, NSK has developed its STIFF simulation software application that takes the above mentioned parameters into account along with the exact interior geometry, internal operating clearance, pre-load, deformation of the shaft bearing system, lubrication conditions, load area and the load distribution between rolling elements and raceway. This model divides the rolling elements into lamina sections. A modified service life rating is determined for each lamina section of the rolling element and bearing raceway. This data is then integrated using the application spectrum of the bearing application. For bearing arrangements in the wind turbine gearbox, the modified service life must be 175,000 hours, i.e. 20 years. The scope of the NSK's STIFF calculation software delivers results that enable rapid parametric analysis to assure the system design meets the design life criteria. Another example of calculation methods employed by NSK is FEA work which examines the distribution of stress within the bearing and its supporting components. It can also run frequency analysis to examine vibration generation of the rolling bearing within the application. Peter Kohl, a wind engineer at NSK says: "By the use of sophisticated numerical calculation tools which are able to simulate circumstances of the application and material properties such as temperature, lubrication, deformation of bearing components and contamination, it is possible to give a more reliable prediction about the real lifetime of the bearing." The problem has also been recognised by Schaeffler, which is working closely with a gearbox manufacturer, a wind turbine manufacturer and a software developer. It is developing new simulation software that is better able to calculate the dynamic operating loads acting on wind turbine powertrains. "Rolling bearing calculation software, Finite Element Analysis (FEA) and dynamic simulation tools will play a critical part in developing next-generation renewable energy systems," says Dr Steve Lacey, Engineering Manager, Schaeffler UK. "Used in combination with FEA tools, this complex, multi-body simulation model will enable design engineers to optimise the design of individual powertrain components and to establish how these interact with other systems across the entire powertrain." Simulation tools such as these, which can be used at the initial wind turbine design stage, will prove invaluable in helping to make future wind turbine designs more reliable and cost effective under a wide variety of load conditions. New bearing designs can be developed and tailored to specific wind turbine requirements, including turbines for both onshore and offshore wind farms.