One of the persistent bugbears of the automotive manufacturing industry, most obviously in actual vehicle assembly operations, is the incidence of musculo-skeletal disorders (MSDs) – in layman’s terms, pulled muscles and twisted joints – among assembly workers.
Someone who confirms it as a real problem, and one which both vehicle assemblers and relevant equipment suppliers are well aware of, is Professor William Marras, director of the Center for Occupational Health in Automotive Manufacturing (COHAM) at Ohio State University in Columbus, right in the traditional heartland of the US automotive industry. Set up in 2007 specifically to provide a centre of excellence to research such issues, COHAM does so by combining the application of highly sophisticated monitoring technology with the recreation of real vehicle assembly activities at full scale.
Full scale tests
Professor Marras explains that COHAM has three ‘pitches’ providing space to recreate three assembly stations, each with the capacity to contain a complete vehicle body. There is also enough room around those areas to allow for the co-location of some highly sophisticated monitoring equipment that can be wired up to people carrying out relevant tasks on the vehicle assemblies. He mentions, for instance, near-infra-red spectroscopy that can measure the oxygenation level of particular muscles.
The basic consequence is that it is possible to measure rather than merely infer such factors as “muscle force, ligament force and disc forces in people’s backs”. In a real production environment achieving this degree of precision and quantitative measurement would be effectively impossible, states Professor Marras, because of the sheer impracticality of installing the monitoring equipment and connecting people to it. Instead, where studies do take place in such locations they tend to rely on information gathering techniques such as video recording, which he describes as being of relatively “low fidelity”.
This combination of capabilities makes COHAM highly distinctive – possibly even unique. Professor Marras says he does not know of any other facility in the world that can offer the same level of competencies. Unsurprisingly, it has also attracted the interest and support of a number of automotive manufacturing companies. One of them is Honda, which sponsors Professor Marras’ professorship, though he is reticent about confirming any other names. COHAM has recently released details of a project that epitomises the approach it is pioneering. In this case it investigated the use of a tilting conveyor to mitigate the risks of MSDs for assembly workers and to enhance the overall efficiency of the assembly process.
The core piece of equipment, reports Professor Marras, was supplied by German-owned conveyor manufacturer ROFA. COHAM carried out a comparative survey in which a test group of six experienced automotive workers were matched against a test sample of six students in carrying out a series of assembly operations on a real vehicle body at various angles – horizontal and then tilted at 15° increments right through to 90° to that initial upright position (though only the horizontal and then 15, 45, 60 and 90° tilts were subject to detailed analysis).
Among the nine vehicle areas involved in the study were interior, underbody and engine compartment. As for the subjects carrying out the assembly procedures, the sheer range of equipment to which they were attached amply underscores Professor Marras’ contention about the level of such instrumentation that COHAM can achieve. In this case it comprised a lumbar motion monitor to measure low back posture; an electromyography system to record muscle activity; nine different magnetic/gravitational sensors to track general body posture; and goniometers to measure elbow and wrist motion.
At the end of the project, reports Professor Marras, the effectiveness of the tilting conveyor as a means of reducing the risk of MSDs was fully endorsed. Moreover, though the conveyor was infinitely adjustable, the research found that there was no need for a multiplicity of positions to achieve significant ergonomic benefits. In fact, despite the number of vehicle regions investigated, only two tilt angles were necessary to achieve optimal results in all cases. In two out of the nine instances a tilt angle of 90° was found to produce the best results, whilst 45° had the same effect in all other cases.
Looking to the future, Professor Marras says he has no particular expectations of the type of issue that COHAM may find itself investigating since he has not noticed any overall pattern so far. But he concurs with two notions that indicate that the organisation is unlikely to find itself without things to do.
The first is a continuation of the trend towards increased product differentiation through mass customisation that has been evident in the automotive sector for at least the last couple of decades. This is bound to mean a continuing high reliance on the use of people to carry out many assembly procedures, the variability of which will render them unsuitable for automation. The second, interestingly, is his implicit endorsement of the term ‘ergonomics pitfall’ which is becoming increasingly current among professionals in the field. It essentially means that the solution to one ergonomics problem may well form the seedbed for the growth of another.
On the first of those counts, Professor Marras believes, the challenge as likely to be psychological as purely physiological. “Customisation changes the nature of the risk,” he says, explaining that it presents assembly workers with the need not just to fix parts in place but to choose them correctly from a wider selection for the particular vehicle assembly involved while maintaining the speed of the line. The approach – known as ‘kitting’ – while practical and effective has the effect of creating “a mix of both physical and cognitive work,” he states.
Meanwhile, on the second count Professor Marras says that one possible source of future problems that needs to be recognised and acted on might, slightly perversely, be the increasing lightweighting of components – very likely due to the spread of composite materials down from the top-end of the market to more volume production vehicles. The ‘pitfall’, he says, is that managers may assume that lighter workpieces will facilitate more extended movements on the part of assembly workers and that MSDs caused by excess weights may be replaced by others resulting from personnel trying to reach too far into vehicle interiors to install them.
Interestingly, one of the subjects that Professor Marras cites as having the potential to be an increasingly prevalent cause of ergonomic stress to shopfloor workers in the automotive industry – the presentation of parts to them on assembly lines producing multiple product variants – has already been the subject of detailed study on the other side of the Atlantic, in Sweden to be precise*.
A team of researchers set out to compare the relative efficiencies of presenting components to workers through two quite distinct methodologies. The first was the common European practice of placing full homogenous pallet-loads of components adjacent to the line. The second involved placing components into much smaller containers that were loaded onto lineside racking that was constantly restocked from the back, a technique more commonly associated with Japanese-influenced manufacturing practices. The location was the plant of a major Swedish producer of automotive diesel engines. The research team was also multi-national with three members coming from the Chalmers University of Technology in Gothenburg and the other from Ryerson University in Toronto, Canada.
The latter was Patrick Neumann, a researcher at the Human Factors Engineering Laboratory at Ryerson, who has made the study of materials presentation techniques something of a personal speciality. He describes the project as quite small but still highly indicative of the benefits that can be reaped from appropriate workstation design. In this case just three assembly stations on the line were reconfigured from their pallet-based materials presentation approach to one involving smaller plastic containers on ‘gravity-feed’ racking – in other words, shelving tilted slightly towards the front so that not only did operators get a better view of the contents but also fresh, full containers automatically slid into place when empty containers in front of them were removed. The work required to complete the change was surprisingly easy. Neumann confirms that the changeover took just six hours to effect.
The line itself did not utilise a conveyor but instead an automated guided vehicle (AGV) carried assemblies between the different workstations. The use of pallets for lineside materials storage made considerable space demands and also, because of the widely variable rates at which different components were used, meant that different pallets were exhausted and replenished at widely different intervals. Some were changed several times a day, others every few days and some only once every several weeks.
Reducing non-value-adding time
The improvements that resulted from the reconfiguration, however, were considerable. For instance ‘non-valueadding’ time in the overall cycle from the completion of one assembly operation to the completion of the next was reduced by just over 20%. A major contribution to this was the reduction of the ‘walking time’, the period in which the operator quite literally walks away from the assembly area to retrieve parts and then returns to it. This went down by some 33%, largely because the actual physical distance covered was reduced by half. The floorspace required for lineside parts storage was also slashed by over three fifths. There were also direct ergonomic benefits for the assembly workers involved. Neumann explains that the ease of access to components by assembly workers was assessed under a colour coding scheme related to the height at which the components were stored. According to this scheme ‘red’ is the most ergonomically unacceptable – with parts stored either so high that workers have to reach up or so low they must bend down to retrieve them – ‘yellow’ an intermediate category and ‘green’ the optimal height. Before the reconfiguration, the numbers of parts allocated to these categories were, respectively, 45, 42 and four. Afterwards these figures were dramatically altered to four, 29 and 58. The consequences in terms of actual postures that the workers had to adopt were similarly directly quantifiable. Before the reconfiguration, Neumann confirms, the researchers found 85 instances in every hour when workers had to bend forward more than 45° from the waist. This figure was reduced to nine. There was also a complete abolition of instances in which workers had to raise their arms by 90° from the horizontal. The relevant figure went from 230 occurrences per hour to zero.
Supply chain focus
On the surface this looks like a win-win situation – a whole series of measurable, quantifiable improvements in process times, space utilisation and assembly worker ergonomics achieved through nothing more than the deployment of some simple racking. So why haven’t these procedures become standard practice?
Here Neumann has some interesting and, at times pithy, observations to make. Perhaps the most fundamental is his assertion that supply chains in the automotive industry – at least those in North America and Europe and by extension offshoots elsewhere with their roots in those regions – tend to be designed macroscopically. “Logistics tend to be designed by people primarily concerned with the supply chain,” as he puts it. In other words, eternal supply chain considerations, such as how and when parts will be delivered to the assembly location, are given an overwhelming priority. Detailed planning for what will happen when they reach the assembly plants is factored in, if not quite as an afterthought, then at least as a subsidiary consideration.
The reasons for this, Neumann continues, are deeply embedded in the culture of the industry, which means that supply chain and production issues are not planned in an integrated manner. In addition, production engineers “simply aren’t trained to think about people.” His answer to the suggestion that surely, in an industry as complex and mature as automotive manufacturing, things would be different, is simply to agree: “You would think so.” However, despite the positive results of the project Neumann cautions against the idea that new lineside procedures can be ‘bolted-on’ to existing logistics systems. He explains that they can have immediate implications for other procedures and systems inside the assembly plant, quite apart from the wider supply chain. The ‘ergonomics pitfall’ can make itself apparent anywhere assembly and materials handling procedures are modified for what might seem straightforward reasons of efficiency and productivity. For a start, Neumann points out, modifications of this sort will alter the overall timings for assembly procedures at different workstations so a line-balancing exercise will need to follow. But compressing the time for a particular assembly operation also raises the question of what should be done with the extra time which then becomes available. Simply increasing production rates is not necessarily the answer. For one thing there may not be the demand. For another, accelerating production rates, even with ergonomically enhanced procedures, may introduce other ergonomic issues related to the increased frequency of body movements. That type of circumstance, says Neumann, is a classic example of the ‘ergonomics pitfall’. “If assembly efficiency goes up you have to work out how to deal with the consequences,” he states.
Human modelling module
Finding ways of working such things out is, in fact, now a preoccupation for Neumann. He says he is currently focussed on the development of computerised simulation technologies that could assess the ergonomic implications of assembly procedures.
But such systems are already in use. Volvo Cars in Sweden, for instance, uses a system from Siemens PLM that includes a human modelling module called Jack. Moreover, according to Dan Lämkull, developer with the company’s methods and IT tools division in Gothenburg, the company has subjected its use of such tools to detailed analysis and continues to refine their application.
Late in the last decade, for example, Lämkull was part of a research team that sought to compare with reality the assembly operations for several Volvo cars predicted by the Siemens Ramsis programme the company was using at the time. The vehicles involved were Volvo’s S80, V70, XC70 and XC60 models, the first three of which were being made at Torslanda in Sweden and the last at Ghent in Belgium. Over 150 ergonomics test cases were evaluated and in general simulation and reality were found to accord reasonably well with each other, with some caveats. Lämkull reports that there was a good correlation where standing and unconstrained working postures were concerned but that results involving more detailed or complex behaviours – including hand access, push and pull forces and routines involving leaning and balancing – were less satisfactory. A continuing constraint on the use of such systems, adds Lämkull, is their highly specialist nature and need for very experienced operators. Users within the company now, he reports, are typically ‘simulation engineers’ – individuals with at least twelve years experience in simulation work and very likely also a previous background in shopfloor assembly work. He describes simulation work as “rather tedious” and also stresses that the system involved is a “complementary tool” that can in no way act as a substitute for the skills and experience of a trained ergonomist. “It is always the ergonomist who makes the decisions,” he states emphatically.
In fact Volvo is now intent on developing significantly enhanced simulation capabilities. Lämkull reports that the company is involved in a research project with the aim of developing a much more sophisticated type of virtual manikin, one with the ‘intelligence’ to move itself and to allow for the analysis of time and motion in relevant analyses. At present, he says, such analysis tools are static and permit only the creation of ‘semi-dynamic’ analyses through the sequential consideration of a series of static models. “It is a tough goal, though it is possible,” he states, adding that it is likely to take at least another three years to achieve.