Automation is making a difference at BMW Dingolfing but humans are still at the heart of production

P90249837-highResOn any given working day, ideas around information and automation present themselves to Reinhard Weindl. In his role as quality manager for technical planning at BMW Dingolfing he has witnessed an ever-growing range of digital solutions for vehicle production. However, Weindl retains a level-headed view and sticks to his priorities.

“It’s the spirit of the plant to look at the process – we want to have the best one. We don’t find a ‘cool’ bit of automation, put it in the line and see what happens. If we have good things to put into the process, like automation which makes sense, or help from a robot in the right place, then we do it.”

A total plant headcount of 17,500 employees contribute to 1,600 vehicles a day emerging from BMW Dingolfing (output reached 340,000 in 2016). The challenge for Weindl and his colleagues at the plant is the mix of product: “We have so many variants within two assembly lines. It means a high level of complexity for us and we always need solutions to be able to handle it and have as many derivatives as possible on the same line.”

Weindl acknowledges that gap and flush is one of the first aspects that a customer notices: “It’s how they recognise quality so it’s very important for our premium cars. It’s important to us to get it right first time and that we move fast into high volume.”

Reinhard_Weindl_1

Reinhard Weindl, quality manager for technical planning at BMW Dingolfing

For Weindl, the objective is not to react but to recognise early when things are not how his team believe they should be. In measurement, what’s needed is a kind of complete picture that reveals trends, portrays delivery quality and, as Weindl stresses, provides enough data to have a statistical relevance.

“We want to be consistent, not just measuring in one place, but to have a process chain – bodyshop, paintshop and assembly – with data from all three places. If we change something we can see what happens across the whole process. It’s one thing to measure, it’s another to have the right interpretation so you can really understand any problems and deviations.”

Statistical relevanceWeindl explains: “If we measure three times a day, the process charts show upper and lower tolerance limits. You are happy if you are in between the two. If you measure a lot of cars, then you can see if you are comfortably inside your limits. It’s important to know the deviation between the mean value and the nominal value for the process steps that follow. We may always be 0.3mm wider or more narrow, then from that trend an adjustment can be made.”

In keeping with the information agenda, the past two years has seen a flurry of new digital activity at the plant. BMW is placing ever greater emphasis on state-of-the-art measurement technology and analysis methods at its plants to ensure high quality and precision for new models from the outset. AMS was able to see several of these innovations in operation, including an optical measuring cell, a self-learning door-mounting system and a compact gap-and-flush quality station integrated into the line itself.

3D area scan measuringPiloted at FIZ, BMW’s R&D centre in Munich, an optical measuring cell has played a pre-production role since 2016. Sensors mounted on free-moving robotic arms scan the whole vehicle to create a 3D data model. The FIZ system is now available to the measurement technology team at the Dingolfing bodyshop with the latest BMW 5 Series being its first subject.

Beside the change from CMM probe methods, a key difference is preparation time. Stickered points and other geometrical references do not need to be attached to the body. It takes less than three hours to cover the entire BIW, with an accuracy to less than 100 micrometres. The results are displayed in colour for easy, error-free interpretation. As it prepares for the series ramp-up of the 5 Series Sedan, BMW says the system has already provided valuable insights for fine adjustment of processes. With series production of the 5 Series successfully launched and full capacity production reached, the measurement technology team are now turning to further launches.

_DSC2759 Automation of the door mounting is a self-learning system, using data from 50 most recent fittings, adjusting its first position and minimising correction stages

Best fit The door mounting line for the 5 Series (featuring both Sedan and Touring) operates with a fully-automated process in what BMW called its “best fit” principle. Optical measuring technology is used to assess the side frame of the vehicle where the system determines the best possible mounting location from the actual dimensions of the attaching parts, compensating for any deviations.

The doors undergo pre-assembly, with the hinge screwed onto the door by an associate adjacent to the line. On the other side of the guard fence, the robot prepares to lift and place the door. It first fires its scan for an initial reading and recognition of the door. It then takes the door, sets it on the car, and makes a positioning measurement. It will then adjust for ‘best fit’ based on all the measurement values it has collected. As it places the door, a partner robot couples to loosen the screws, press on the hinge and retighten the screws. With the hinge then taking the strain a quality measurement is taken.

Weindl points out: “This is the important one for us, because it is our delivery quality measurement.”

The ‘Best fit’ idea steers the robot to the door aperture not by a fully preprogrammed setting but from information the system has collected from a series of previous instances. It’s self-learning – retaining the information of ‘best fit’ from the most recent 50 cars that passed. Knowing the average position of these most recent BIWs means it can adjust to take a new first position setting and avoid further correction stages.

Weindl explains: “It finds the best fit between this [particular] door and this [particular] car. They don’t have a lot to correct. They are quite close to the ‘best fit’ position in the first movement. Normally we have two fitting steps, sometimes just one.”

The flow of information not only makes for better results in each fitting but also supplies comprehensive measured results without the need to transfer a single vehicle from the production line into a test room.

Parallel quality assuranceAt the end of the 7 Series doors and closures line, an inhouse innovation has provided a clever automation solution to measure gap and flush dimensions. As well as a quality assurance installation, the set up help Weindl’s team gather information about what is being delivered to assembly.

Previously, a visual check by associates was employed but it didn’t provide enough useful data. The associate undoubtedly had reliable judgement but they couldn’t collect, record and display precise deviations.

Weindl adds: “Also, only a few cars per day could be measured so not much of the statistical relevance that’s proving so important.”

“The challenge was that we had little space to install something, one takt. We couldn’t make a full station, with security fences, so instead we used the lightweight cobots.”

No large redesign of the line was necessary. A simple arrangement of four cobots floor-mounted at the end station provide a gap-and-flush check. Equipped with measuring heads, the system collects more than 50 measurement values with absolute consistency of point, angle and condition. With the increased volume of data, Weindl’s team can detect deviations early and take appropriate action. Crucially, associates can walk safely in and out of the station, applying or removing the conditioning brackets which placed inside ahead of the scanning. The associates can also see the results on the screen mounted above the station — if everything’s green then it’s fine.

It was a solution devised in-house by BMW. Weindl explains: “We could buy a cabin full of automated measuring but it would take up more space and be more expensive. We couldn’t extend the line but the important this was to get the quality data, to get it here [at the line] to give the team fast feedback and to have information about what is being delivered to assembly. We can tell [assembly] if we have a change in any averages and ask them what it means for their process – is it ok or do we have to react here?”

“It’s part of my function to be the interface between bodyshop, printshop and assembly. In the end if we are standing in front of the car and there’s a question, we need information - why is it happening – we don’t ask who is responsible or guilty, that doesn’t help us.”

By the end of 2017, BMW plans for comparable systems to be employed in the paintshop and assembly phases.

At the heart of productionA further BMW example of collaboration between man and a classic large, industrial robot - fitting the gearbox to rail-suspended engine. The associate takes the robot ‘by the hand’ bringing it in line with the screw fittings are to be placed, laser sensors activate to help the associate determine the optimum position. The robot rotates a plate six times, one for each screw, again taking on the strength-based task. It’s a good example of how to mix automation and manual approaches. Handling large components requires mechanical assistance but to full automate the process would require a change from the simple rail-suspended engine conveyor and a switch to a sophisticated, extremely precise (and no doubt expensive) mechanism that still wouldn’t match the nuanced and flexible judgement of a human associate.

A time- and trouble-saving idea in operation is the work glove fitted with barcode scanners. It speeds up a process by avoiding the cumbersome steps of picking up and placing down the scanner. The scan is now triggered by the worker pushing a button on the index finger with their thumb. It’s a way to improve and accelerate individual production processes, pushing operational quality and occupational ergonomics. There are 230 of these innovative work gloves in use at BMW.

Another example of digital assistance is the use of augmented reality applications on smart devices such as tablets. An image of a component can be overlaid with virtual specifications. The tablet then displays comparisons and evaluates the actual and target states, so the worker can determine if the part matches the target requirements. Potential issues can be identified and resolved early on.

BMW sees more and more scenarios where the old days of clear distinctions between what humans and machines can do are shifting. Machines are used to support people in their work. Where there are higher degrees of automation, humans can still be at the heart of production but supported more than ever by given devices.

The wide variety of examples of practice improvement, step by step, will modernise the working world, BMW says. That is its Industry 4.0 approach.

Across BMW production globally, there are around 45 cobots installed. This will increase to 65 in the coming months but the OEM doesn’t see this as a wave of automation – it’s simply a response to concrete use cases where automation makes most sense. 

_DSC2934 At the window line, man and machine combine with an automated adhesive application but a manual fitting to the car

Side by sideBesides the manual fitting of windows for the i3 and i8 production in Leipzig, a similar example operates at Dingolfing. A lightweight robot assists the assembly associate with installation of the triangular side window in the BMW 3 Series Gran Turismo, BMW 4 Series Gran Coupé and BMW 5 Series Touring models. At a workside alongside the line, the associate places the window into a table-top frame. The cobot sweeps its nozzle along window edge, applying a uniform bead of adhesive. Automating the process maintains the optimum line and height of the adhesive, ready to be collected up with the associate and inserted onto the car body.

The previous manual system required more space as, to apply the glue by hand, the associate needed to walk 360º around a plinth upon which the pane was placed. Conversely, a fully-automated system would no doubt require stations either side of the line, leading to far greater demand on space and equipment. On Dingolfing’s window line, the associate and the cobot combine to make the most of their respective strengths – flexibility and consistency.