Virtual testing, real progress
Simulation models enable programmers to test machine software before the actual machine even exists. And when the machine is finally delivered, the customer benefits from more reliable software support. The R&D team at Optima Nonwovens has demonstrated this in the case of a packaging machine for personal care products.
If the machine's control system does not detect that it is only controlling a simulation model, then Christina Weitze and Jan Buck have done their job correctly. The former, a mechanical engineer, was the one who “built” the virtual machine. The latter, an electrical engineer, can put “his” software through its paces, even though the individual components of the actual machine are still only just being manufactured. Weitze and Buck are part of the R&D team at Optima Nonwovens. They have been exploring the potential of virtual machines for several years, and using them in customer projects. One such was a special packaging machine for sanitary pads, which is already in operation at the customer's company.
At Optima Nonwovens, no two projects or machines are the same. However, in this case it was clear as soon as the order was placed that the team would have to introduce a lot of innovative features. For example, the individual modules needed to be flexibly adjustable, and the machine had to use a special kind of product transportation system. All of this presented new challenges for the machine software. Developing a simulation model, on which the software could be tested was definitely going to be worthwhile. Christina Weitze was on the case. She took all the dimensional drawings that the mechanical engineers had produced for the process, i.e. all the dimensions relevant to processing, and incorporated them in the simulation model. She also reproduced the sensor system, such as the light barriers, and the actuators, such as the motors and pneumatic cylinders. "The virtual machine has to include all the components from which the software would expect to receive a response in reality," explains Weitze.
Despite the effort involved – which must not be underestimated – a virtual machine is, of course, up and running far sooner than the real one, for which the individual components have to be manufactured. So it is worth Jan Buck trying to program the machine software as quickly as he can. He can then test it thoroughly on the simulation model and optimize it if necessary.
Christina Weitze and Jan Buck work on producing the simulation model and testing the virtual machine in parallel.
The virtual product feed is better than in the FAT
“Of course, the machine is reproduced in a simplified form in the simulation,” explains Buck. However, the logical processes are the same. If, for example, product pusher A goes off course, it must not collide with product pusher B. That is just as true for the virtual machine as it is for the real one, and can be ensured in the course of the simulation testing. Virtual products are even produced, transported and discharged from the machine at the end. In this respect, the virtual commissioning is closer to how the machine will be operated for real at the customer's premises than the commissioning of the actual machine in the FAT (Factory Acceptance Test). This is because, on the virtual machine, the products are fed in continuously, just like they will be later when it is used in practice. In the FAT process in the machine room, on the other hand, often only small quantities of the product are manually placed in the machine for packaging.
Buck is very pleased with all the new possibilities. He recalls, “previously, I was really only able to start the programming work when the machine was virtually fully assembled. Then there was not much time left before it was delivered.” He adds that doing the software engineering further upstream made the test phase longer, and that “we can now use that time to adjust the settings using the actual product.” Buck can also test out scenarios in advance which could put the plant into critical mode, such as an unexpected intervention by an operator. After all, “In the simulation, nothing can get broken,” says the software engineer, smiling. This means that the customer receives a machine that has been tested particularly thoroughly. The result? More reliable quality with no delays in the delivery time, especially in the case of unusual machines with special specifications for the machine software.
Software support instead of traveling the worldThe virtual machine certainly still served a useful purpose even after the actual machine had been delivered. Buck has already used it several times since – whenever the customer has a special request or needs to eliminate an unusual, unpredictable fault. Buck emphasizes, “We can initially test out improvements and upgrades here at the office on the 1:1 simulation model. Only when the software is working perfectly here do we send it to the customer.” If, as in this case, the customer is operating the machine in a distant country, a great deal of time is saved, and the engineer can concentrate on his software support work instead of spending time traveling.
If Buck had not already been convinced of the benefits, the reactions of his customers would have persuaded him. “They are absolutely delighted with the simulation process.” It is quite something when they can see on a video that their request has been implemented on the virtual machine and then the software can be installed directly. It is also significant that a follow-on project is already getting under way. Despite the fairly major mechanical changes, the enhanced machine will soon be running perfectly, first virtually, then for real.
Important for you
- Virtual machine models combine basic mechanical data with information about sensor technology and actuators.
- The simulation enables logical processes to be tested before the actual machine even exists.
- This means that faults in the machine software can be quickly spotted and rectified. The quality of the software improves, while the delivery time remains the same.
- The simulation model can be used throughout the life cycle of the machine, for example if it is ever modified.