Automated Weld Grinder


Students: Wesley Damen, Martin van Langevelde, Wouter Lettinga & Youri Verschut (2020)


Pipes with a diameter from 16 inches (approx. 40 cm) and larger are made by bending a steel sheet into a cylinder and welding the edges together. The orientation of this weld can be parallel to the axis of the pipe or the welds can be spiral around the pipe.

The weld is an imperfection on the smooth surface of the pipe. This imperfection can cause problems during the connecting of the pipe pieces. Also, it presents a risk during the coating process. To eliminate these risks, the welds at the ends (150cm) of a pipe need to be grinded smooth.

Selmers, a company specialized in equipment for blasting, coating and handling of pipes, are seeing an opportunity to automate this process. Automating the process of removing the weld will eliminate the need for people in a dangerous workspace, where big pipes and equipment are moving around. Selmers asked us to build a prototype which solves the problem.


The prototype consists of a Universal Robot 10 robotic arm, an end of arm tool and a Python program for control, with GUI (graphical user interface).

End of arm tool

The end of arm tool is compactly designed, containing a grinder to remove the weld, two induction sensors, ‘guiding wheels’, an end switch to detect the pipe and the weld and lastly several springs and guides to make the end of arm tool compliant. Compliancy is needed for flexibility, since pipes are almost never perfectly circular. Also different sizes of pipes could be processed and each size needs different settings of the end of arm tool. 

The ‘guiding wheels’ are used as a mechanical limit for the grinder, so it can never grind into the pipe. This is because grinding material away from the wall thickness of the pipe would make the pipe unusable according to the standards for pipes. 

The grinder disk is set on the exact same height as the ‘guiding wheels’ before every grind movement. This is achieved by pushing the disk on the pipe until the induction sensor, which is installed near the ‘guiding wheels’ notices the pipe. At that moment, two electromagnets are activated which, using a construction, limit the grinder from pressing dropping below the height of the ‘training wheels’. A spring makes sure the grinder is still pressed on the weld.

Python program

All the hardware is controlled by a Python program. First the program determines the location and orientation of the weld. Based on the orientation, it generates a path for the end of arm tool to follow during grinding. After the height of the grinding disk is set by turning on the magnets, the program starts the grinding tool and path.

The back end part of the program can be controlled by an easy GUI. The only input for the GUI is the diameter of the pipe, after which the program can be started. During the whole process, the GUI gives feedback to the user.


We can conclude that the proof of principle for a automating process of weld removal can work within an industrial environment, because we can grind different types of welds (spiral and linear) and different sizes of pipes with only the input of the pipe diameter.

After testing a lot of different sensors the induction sensor is the most precise sensor to find the weld. The only downside is that the pipe isn’t completely round, so without some kind of compliant system the induction sensor will run into the pipe or will give a invalid output because the pipe gets too close to it.

Second of all it’s very import to make the EOAT compliant, because grinding is a delicate operation with a small margin of error. By making the EOAT compliant and adjustable with a lot of springs it make the EOAT modular. Therefore we can grind different sizes of pipes and it makes it possible to grind the weld from the inside and outside.

The result that we could get with limited budget and materials is very promising for further development of the final solution for Selmers.