Design of a prototype device for filament winding of pressure vessels

The CADWIND software was supposed to supply the motor motions. Several simulations were performed to understand its limits. During the simulations it was also recognized to suggest solutions about both the windable shapes and the machine layout. It is because the motions of both the mandrel and the trolley are calculated. In this way the information about the step to be run by the motors could be drawn.

Moreover, knowing the motions to run it is possible to check whether the linear guide of the machine can run the distance. Acting in this way it is possible to test the machine about the requested shapes and to try different parameters to make the machine to wind the mandrels. Unfortunately CW returns the axis motions but do not return the motor motions.

Although these motions are strictly related, the software does not care of the large variations between two subsequent instructions. During the simulation it is also possible to check for the tow overlapping, giving some useful inputs back to the vessel modelling as well. Unfortunately it is not possible to evaluate the local thickness all over the mandrel.

The simulations were set feeding the software with the informations about:
– the machine dimensions
– the size of the motor step (according to CW: the resolution)
– the acceleration and the speed along the axes
– the processing time
– the shape to be wound
– the tow material

During the mandrel modelling and the winding path definition some suggestions about the mandrel shape were obtained. Some of the pressure vessels to be wound are 1 m long, suiting open ends and a 200 mm diameter. This vessels look like a pipe and it was not possible to find the tow path to wind such vessel length with the requested winding angle. Also some problems arise for winding above the turning frames, because it was not possible to totally wind the vessels. This problem was not related to the friction coefficient but to the trajectory necessary to turn the fibre back.

Being demonstrated that the cut of the helicals is not a feature limiting the pressure performances, the solution was found in simulating the pressure vessels with shaped (conical) ends, to be cut later on. In this way it is possible to wind the 1 m long vessels with winding angle equals to 25°, while before it was not possible to wind below a winding angle equal to 70°. This result relates to the winding path, thus it needs to tested over the machine layout.

The simulations over the machine layout underlined the relationship between the feed eye position (minimum distance from the mandrel surface) and the winding possibilities. Keeping the feed eye far from the mandrel surface it is not possible to wind long mandrels. The reason for this limit is the extra-stroke needed to complete the winding. This problem imposed the necessity to move the feed eye closer to the mandrel surface. In the case relating the 1 m long vessel, it was found that reducing the minimal distance down to 20 mm it was not possible to wind the mandrel.

For safeness reasons it is preferred to avoid winding below such distance. The same limit does not show to wind a 900 mm mandrel. In this case it is possible to wind with a winding angle of 45°. Such limit was considered acceptable.