Polishing of an airplane wing

 

Students: Pepijn van EckYuk Ho Chung, Schelte van der Horst, Jorge Luis Díaz Onofre, Florens Vernooij

FOKKER logo HHS logo TU Delft logoSAMXL logo
Images 1 through 4: Logos of Fokker Technologies, The Hague University of Applied Sciences, University of Technology Delft and Smart Advanced Manufacturing XL

Introduction

For a combined project, Fokker and the University of Technology Delft requested The Hague University of Applied Sciences to automate the prepolishing process of the leading edges of a G650 Gulfstream aircraft wing. Currently these leading edges are manually prepolished by skilled employees, even though the process of prepolishing is not very complex and can be easily automated. The automation of this process will leave the employee with more time to focus on other tasks like the more difficult finishing polish of the assembled wings.

Gulfstream G650
Image 5: Gulfstream G650

Process analysis

The process to be automated is the polishing of the leading edge with the necessary steps for the desired result. For a leading edge to be polished a polishing disk must be attached to the polishing tool and polishing paste must be applied to the disk. The employee checks whether the leading edge has been polished completely and corrects it, if it is not. The polishing disk gets dirty over time and will be cleaned and at some point be replaced by the operator. As this project was conducted mostly for exploratory reasons, the replacing and cleaning of the polishing disk has been left outside of the scope.

Creating a setup that can automatically polish a leading edge requires several different challenges to be taken on. Replacing an employee with a robot for the initial polishing is a fairly easy task, as it is a task that has to be identically executed repetitively. Of course it takes some time and effort to set everything up for it to work properly, but with the right tools and information it is rather straightforward. After this initial round of polishing, the quality might not be exactly as required, therefore the visual inspection by the operator has to be exchanged for a technological alternative. The real challenge however, comes with integrating and controlling all of the subsystems through a PLC program. This program controls the flow of the whole system by considering the information available in the system, which is gathered from sensors and internal feedback. Depending on the current state of the system, the PLC will tell the actuators to execute the required actions. While integrating the separate parts, a calibration has to be done to make sure everything is aligned correctly and the sensors return the required information.

Project breakdown

Image 6: UR10

The robot used in this project is a Universal Robot 10, which six degrees of freedom allow the end-of-arm-tool to reach every location in every position. As this robot actually a cobot, a cooperative robot, employees and operators are able to work within the same area without a safety fence being required. The aforementioned end-of-arm-tool used consists of a mount for both the polishing tool and the glare sensor. The orientation of this mount can be changed to allow for the use of tool or sensor.

End effector: polishing tool and sensor
Image 7: End effector used for the polishing and checking

Along with the right tool orientation, the correct amount of pressure must be applied for a good polish to emerge. The pathing of the robot to polish every bit of material is adjusted to always push on the wing with the same amount of pressure. This adjustment is made by using a force sensor to test the amount of pressure and correct accordingly.

Image 8: Leading edge mounted for polishing

Quality check and adjustment

Even with the right orientation and pressure, certain areas might not be polished well enough. The quality check and following adjustments by an employee are recreated by sensory checking and automatic correction. The glare sensor is used to detect the level of glare of the polished surface, so the quality of the polishing can be checked. With the knowledge of the location of the sensor combined with the data from the sensor, repolishing can be done to achieve the requested result.

The tool used in this project is a handheld polishing tool, similar to the one currently used at Fokker’s. Handheld tools are designed to be ergonomically correct and since ergonomics are not required for a robot, the design is not ideal. The reason to work with this kind of tool over a tool that’s meant to be mounted on a robot is that this project is all about the automation of the process. While picking a better suited tool would add value to the project, it requires research to base the decisions on. This research would have taken a lot of time which was not available in the few weeks this project was conducted. For the same reason a glare sensor was included instead of other kinds of machine vision, as the quality check was a requirement while the localisation or orientation was not.

Conclusion

The proof of concept built in this project shows great promise towards automating and optimizing this polishing process. While there are plenty of ways this setup can be improved, those possible improvements show what the potential of this project is. This project was conducted in collaboration with SAM|XLFokker Technologies, the University of Technology Delft, The Hague University of Applied SciencesMetaaltechniek B.V. and SICK B.V., so a great thanks to them for the support of information, materials and opportunities!