Werkse! is a company that offers supported employment to people with physical and/or mental learning difficulties, to enhance their future job prospects. Werkse offer a number of services, positions on assembly lines and packing and sending various components to external clients, where they try to work as lean as possible to reach their goal by limiting unused stock in between phases of an assembly process.
Original situation:
We were focussed on one of the assembly lines of Werkse. There are four employees assembling a electrical plug, where all of them do a part of it. The tasks are the following:
- 1st employee:
Placing four nuts in an empty connector
- 2nd employee:
Placing left pin and screw it tight (one screw)
- 3rd employee:
Placing left pin and screw it tight (one screw)
- 4th employee:
Placing lid and screw it tight (two screws)
The Problem:
The problem of this assembly line was a bottleneck at the 4th station. This task took longer than all the others, which resulted in a big pile of half-assembled connectors between station 3 and 4. Since they pursue a lean work environment, they wanted to eliminate the bottleneck. The goal was to find a solution without taking people out of the process with the use of a robotic arm (Universal Robot 5).
The problem for the employees is currently, the screws are very small and some of these employees have difficulty screwing manually, over a long period of time. For this reason, we concluded that automating the screwing would satisfy the conditions given to us by Werkse and SMR. The group will combine function of the robot arm with a vision system, for the safety of the Werkse employees – which is fundamental in this project.
Our solution:
After we got our assignment, we went to the company to do the entire process. We wanted to see for ourselves how fast we could do it, to get a good overview of the assembly line. After doing it ourselves, we found out that the last part was indeed a bottleneck. We brainstormed about possible solutions and came to the conclusion that we should change the current assembly line by redefining tasks. This will mean that the people will have something else to do, but it made the workload of the tasks more evened out.
The figure above shows the material flow for our new assembly process, the new tasks are divided as follows:
- 1st employee:
Fits 4 bolts into connector (Same as before).
- 2nd employee:
Slides both pins into connector, loads carrier of connectors, and places into work area 1.
- 3rd employee:
Fills screw cartridge with small screws, collect carrier from work area 1, pass to employee 4 with lids.
- 4th employee:
Fills screw cartridge with big screws, collect final connectors and place in output location
The robot will screw 4 (2 big and 2 small) screws into each connector.
We received a spare drill from the company, the same they are using now for the manual assembly. The drill is mounted inside an end-of-arm-tool, which is attached to a robot arm. The robot arm will pick the screws from a cartridge and place them in the connectors, which are mounted in a carrier.
The carriers and cartridges are designed ourselves. We 3D-printed multiple concepts to check what the best designs would be. The difficulty in this process was the accuracy of the 3D printer, since it comes down to millimetres.
It turned out to be difficult to create a sufficient solution within four weeks, but with more time and the right equipment, this problem is definitely solvable.
Major decisions during project:
The drill we received has two possible connectors, where one of them facilitates the function of turning the drill on and off remotely. Since that connector was unavailable, we soldered two cables inside the drill to fix this ourselves.
A big problem we encountered was the centering of the drill. It turned out that the axis of the drill is not centered accurately, which gave a variation of around 4 millimetres. This variation was almost as big as the screw. We tried different approaches to solve the problem including:
- a bearing around a bit holder (where it lost its magnetic force, required to pick up the screws)
- conic shapes in the carriers and cartridges
- varying the speed of the drill (to reduce the inaccuracy.
In the end we fixed the position with electrical cables tied around the screwing bit, but this is not very accurate. Since there was no time left, this was the best solution possible in this amount of time.
The cartridge was created to simplify picking up the screws, but it turned out this was more difficult than initially thought. The robot has to remember a lot of positions, a different one for each screw. The company that sells the drill, also has a automatic screw supplier. This machine supplies a screw in a predestined position where the robot can pick it up. When it is picked up, automatically it will put a new screw in place to be picked.
The drill we received is made for manual use. The same company sells drills made for automation. The drill is more robust, has a higher accuracy and has an aluminum casing, made for mounting.