PySLM: Geometric Hatch Overlap Check/Visualisation

Designing scan strategies for PBF techniques, we are not entirely aware of situations that arise where the powder-bed is not fully exposed due to a mismatch when scan vectors are not sufficiently overlapped. Typically, unoptimised placement of hatch vectors lead to the creation of irregular porosity or voids in L-PBF parts.

This can arise along the intersections between the contour and interior hatches, especially a long concave regions such as sharp corner features with acute angles.

The approach is not an efficient way to examine the presence of , but provides a representative view for checking this geometrically.

Method

The approach takes advantage of the relatively new Iterator classes available within the analysis module, which vastly simplifies the generation procedure for manipulating and examining existing scan vector geometry. Firstly, generate or alternatively import the Layer and its LayerGeometry groups to examine. The group of layers are passed to the ScanVectorIterator class, which will iterate across every scan vector from both ContourGeometry and HatchGeometry objects within a Layer. Single point exposures are not considered.

import pyslm.analysis
from shapely.geometry import LineString, Polygon, MultiPolygon
from shapely.ops import cascaded_union

scanIterator =  pyslm.analysis.ScanVectorIterator([layer])

After the creation of the ScanVectorIterator, this can be readily expedited to process across all scan vectors across the Layer. The basic process relies on converting each scan vector to a Shapely polygon objects and then processing them using the geometry tools available.

For this case we use Pythonic notation to compactly operate across each scan vector and collect them. We convert each scan vector to a Shapely LineString, which has a method to then offset or buffer.

# Laser Spot Radius
laserSpotRadius = 0.04

# Iterate across each scan vector and buffer than geometry
lines = [LineString(line).buffer(laserSpotRadius) for line in scanIterator]

After offseting all the lines, this can be easily visualised by conversion to a Shapely MultiPolygon.

# Merged the offset lines into a Shapely Multi-Polygon Collection
multiPoly = MultiPolygon(lines)
pyslm.visualise.plotPolygon(multiPoly)

The geometrical result is shown below. It is relatively quick to generate and plot individual scan vectors as shown below:

Geometrical overlap of scan vectors in SLM processed by PySLM
Overlap of hatch vectors represented by geometrically offsetting the individual scan vectors within a Layer.

Each scan vector that is offset is represented by a Shapely.Polygon. It is trivial to perform a boolean operation with the Shapely library. Although, it is recommended to use the more efficient, albeit still relatively slow, shapely.ops.cascaded_union function to merge multiple geometries together:

# Cascaded union is a more efficient boolean merge for multiple polygon entities
multiPolyMerged = cascaded_union(multiPoly)

pyslm.visualise.plotPolygon(multiPolyMerged)

The combined result is shown here with a slightly smaller hatch distance to exaggerate the effect and highlight regions where the laser beam may not sufficiently provide exposure to the powder bed:

Illustration of regions that may
Regions that have insufficient coverage observed after performing a boolean merge after the scan vectors have been offset

This post shares a relatively simple example exploration using geometrical operations and the iterator class to understand potential issues related to hatch overlaps.

Final Conclusions

Potentially, this check could be extended into 3D using morphological operations. This would provide a more qualitative examination of porosity generation as a result of Furthermore, the combined use of Neural Networks or reduce order models could provide a representative exposure area to provide the geometrical offset in 2D and provide a prediction for coverage spatially in 3D.

Leave a Reply

Your email address will not be published. Required fields are marked *