We have been making the BA 386 axle-box housing for years. Valid TSI certification, hundreds of thousands of units delivered.
Logic says: don’t touch it!
I see it differently. A proven product means proven safety. It does not mean proven production efficiency. Those are two different things, and almost every cost calculation I look at reminds me of it.
That is why we tested another robotic welding option.
What we wanted to find out
- Repeatability. The robot puts the same heat into the same spot in every part.
- Traceability. Weld parameters are recorded, not reconstructed afterwards.
- Predictable distortion, meaning less straightening and smaller machining allowances.
What was harder than we expected
- Casting tolerances. A robot cannot even out what a welder corrects by feel.
- Clamping. Most of the work was not in the robot program but in the fixtures.
- Access to welds inside a closed pocket.
The sentence I say most often in this field: safety is not a variable in the calculation.
On the railway, a welded joint is part of an approved manufacturing procedure. Changing the welding technology is therefore not a purchasing decision but a qualification process: a new procedure, new samples, new documentation, new validation.
So we look for savings elsewhere: in part preparation, clamping, logistics between operations, and in how much manual work straightening and re-grinding swallow. That is where I see room for about 10 %, without touching anything certified.
Automation is not a goal for us. It is a way to stay competitive in a country where we can no longer win on labour cost.
I would like to know how colleagues in other regulated industries handle the qualification of a new welding procedure. How many months did it really take you?