some slicing issues in rapid prototyping · 2016-06-16 · in the past, the issue of slicing has...

91

Journal of Mechanical Engineering Vol. 8, No. 1, 91-101, 2011

ISSN 1823-5514© 2011 Faculty of Mechanical Engineering, Universiti Teknologi MARA (UiTM), Malaysia.

Some Slicing Issues in Rapid Prototyping

Boppana V. Chowdary Department of Mechanical and Manufacturing Engineering

The University of the West IndiesSt Augustine Campus, Trinidad

Email: [email protected]

Divesh R. SahatooIn-Corr-Tech Ltd., San Fernando, Trinidad

Email: [email protected]

ABSTRACT

Rapid prototyping (RP) processes such as fused deposition modeling (FDM) and stereolithography (SL) are being widely used in the industry with great potential for direct digital manufacturing of functional parts. In the RP process, surface finish and build time are important and are affected due to improper selection of slice thickness. It is evident that there is a strong relation between the slicing and tessellation process for which more scientific efforts are necessary to develop methods for creating more efficient STL files. The aim of this study is to show the effect of slicing on the surface finish, layering error and build time of a prototype. Furthermore, this study shows how tolerance and slice thickness are related in creating more efficient STL files. In this regard, three objects were modeled for generation of STL files. Each object’s STL file was sliced with different slice thickness values. Screenshots approach was used to show the slicing effect on layering error. The build times were calculated and documented. The study results show that slice thickness has a great impact on several factors. The results will serve the industry in understanding how the STL generation and slicing process can be carried out effectively for efficient utilization of various resources. Keywords: Rapid Prototyping, STL, Tolerance, Tessellation, Surface Finish, Triangular Facet, Slicing, Slice Thickness, Layering Error, Build Time

Artkl 6.indd 91 10/21/2011 11:00:07 AM

92

Journal of Mechanical Engineering

Introduction

In the rapid prototyping (RP) process, the standard triangular language (STL) model is “sliced” into layers, each of which is added physically by the RP machine in making the prototype. In the slicing process, the thickness of the slice has an effect on the eventual surface finish as well as the build time of the completed prototype. The thickness of the slice used to manufacture the prototype would bring about an effect known as a layering error [1] or staircase effect [2].

The RP process involves the breaking up of a 3-D CAD model into 2-D slices [1]. The accuracy of each slice is dependent on the position and number of triangular facets used. There are cases where too many triangular facets may be used to represent the object, which leads to an increase in the file size of the STL and hence the usage of more memory than is necessary when processing. In an effort to increase productivity of the RP process, this study places an emphasis on investigating the effect of slice thickness on the surface finish, layering error and build time of prototypes.

Background

The issue of slicing plays an important role in the RP process as it may affect the quality of surface finish and the build time of an object depending on slice thickness [3] As such, there are several publications which have dealt with this particular issue. The layering error issue has been dealt sufficiently by [1] and [2]. [3] also addressed the issue. However, no emphasis has been placed on how slicing issue is directly related to object tessellation [4].

[1] discussed an adaptive slicing algorithm that can vary the slice thickness depending on the geometry of the object to be prototyped in an attempt to minimize layering error and improve surface finish. [3] presented an adaptive slicing approach which significantly reduced fabrication time by 17-37 percent compared to conventional methods and improved surface finish by eliminating unnecessary slices. [5] looked at a way to directly slice CAD models using contours instead of going to STL hence, overcoming some problems which would have been encountered during the RP process. [6] outlined improving geometric accuracy using adaptive slicing with sloping boundary surfaces while [7] dealt with a tolerant slicing algorithm to overcome memory constraints and computational problems. [2]’s study shows the different issues involved in the tessellated process which include investigations on increased build time with small slice thickness and poor surface finish with a large slice thickness.

When an object is sliced, horizontal slice planes are intersected with the sides of the triangular facets. Those collective intersecting points are then adjacently joined by straight lines to form a contour at that height [1]-[2]. It is apparent that

Artkl 6.indd 92 10/21/2011 11:00:07 AM

93


if there are more intersected triangles, there would be more points available and thus, the contour would be more accurate. The slicing process is repeated until contours are created for the entire object from top to bottom. In this study, the number of triangular facets used in slicing was looked at with respect to slice thickness. The slicing process illustrated by [8] shows that the solid model is either tessellated or its surface divided into triangles. The triangles are intersected by slice planes which produce intersection points which are used to form slice contours. The space between adjacent contours is termed a slice, each of which is built upon each other to create the physical prototype. The thickness of the slice would also bring about a layering error, which would affect the surface finish of the prototype.

In the process of slicing, the triangular facets are intersected by slice planes for creation of slice contours [1]-[2]. As a result, there is a possibility of over-tessellation with respect to slicing [4]. This occurs due to the presence of triangles that are not intersected by slice planes and which do not contribute to the formation of slice contours. Thus, more scientific research is necessary for elimination of the redundant data that would minimize system hardware and processing time requirements.

In the past, the issue of slicing has been met with various levels of success. Restricted applicability to the study of the tessellation-slicing relation is one of the important limitations. In this regard, this paper tries to investigate various issues such as the effect of slice thickness on surface finish, layering error, and build time of prototypes. Additionally, the relation between tessellation and slicing would be studied in this paper.

Research Approach

In order to investigate the research issues, three objects as shown in Figure 1 were selected and modeled using SolidWorks package. Several screenshots were generated for purpose of the analysis. To show the effect of slice thickness on build time, each object was sliced and the build time was calculated for various slice thicknesses. For different slice thickness values, the relation between triangular facets and slice planes was established graphically. In the past, the screenshot approach has proven effective to overcome computer memory constraints and computational problems [7] and in analysing human anatomy models [9].

Three STL files of different tolerances were generated for each object to investigate the impact of the layering error and tessellation-slicing effect on surface finish and build time of the prototype. One STL file was chosen for each object to show the effect of layering error on surface finish. Graphs were plotted to show the effect of slice thickness on build time.

Artkl 6.indd 93 10/21/2011 11:00:07 AM

94


Results

Initially, one of the three generated STL files for each object was sliced and the resulting images are depicted in Figures 2, 3 and 4 which indicate the effect of slice thickness on the layering error and the surface finish of the object. For a better understanding of the findings, greater focus was paid to the edges of each object where the slicing effects were more evident.

The second generated STL files was sliced for different slice thicknesses and the resulting build times were calculated using Z Corp’s Spectrum Z510 3D printer at a speed of two slices per minute. The established graphical relationships for various objects are shown in Figure 5.

(a) Sphere

Figure 1: The Selected Objects of the Study

(c) Battery Carrier(b) Alblock

Figure 2: Screenshots of Sliced Sphere STL Files

(a) Thickness = 0.2 mm (b) Thickness = 0.4 mm

(c) Thickness = 0.6 mm (d) Thickness = 0.8 mm

Artkl 6.indd 94 10/21/2011 11:00:08 AM

95


Figure 3: Screenshots of Sliced Alblock STL Files



Figure 4: Screenshots of Sliced Battery Carrier STL Files



Artkl 6.indd 95 10/21/2011 11:00:08 AM

96


To show the relation between tessellation and slicing, the screenshots shown in Figures 6, 7 and 8 were used to represent the slice planes and triangular facets for each object. Figure 6 shows the front view of the tessellated Sphere for varying tolerances and slice thickness values. The views shown in Figure 7 are of the Alblock viewed from the top, featuring part of two bevels and a flat face.

The views given in Figure 8 are of the Battery Carrier (top view) showing a bevel and a flat face. It is noted that the Battery Carrier has mostly straight faces with some small features such as holes and bevels.

Figure 9(a) and Figure 9(b) shows the impact of build direction and slice orientation on the non-spherical objects. The non-intersected triangular facets which are perpendicular to the build direction are represented in green colour.

Figure 5: Build Time Versus Slice Thickness

Figure 6: Screenshots of Tessellated Sphere STL Files Against Slices

(a) Tolerance = 0.005 mm; Slice Thickness = 0.8 mm

(b) Tolerance = 0.01 mm; Slice Thickness = 0.2 mm

(c) Tolerance = 0.10 mm; Slice Thickness = 0.2 mm

Bu

ild

Tim

e (

min

s)

Artkl 6.indd 96 10/21/2011 11:00:08 AM

97


Figure 7: Screenshots of Tessellated Alblock STL Files Against Slices

(a) Tolerance = 0.01 mm; Slice Thickness = 0.8 mm

(b) Tolerance = 0.01 mm; Slice Thickness = 0.2 mm

(c) Tolerance = 0.20 mm; Slice Thickness = 0.2 mm

Figure 8: Screenshots of Tessellated Battery Carrier STL Files Against Slices

(a) Tolerance = 0.004 mm;Slice Thickness = 0.8 mm

(b) Tolerance = 0.004 mm;Slice Thickness = 0.2 mm

(c) Tolerance = 0.03 mm;Slice Thickness = 0.2 mm

Artkl 6.indd 97 10/21/2011 11:00:09 AM

98


Discussion

Figures 2, 3 and 4 show an increase in layering error and a decrease in the quality of surface finish with increasing slice thickness. Therefore, it is desirable to have a slice thickness which would provide the best surface finish possible with minimum build time. It was also noted that the layering error has no effect on flat faces of an object which are horizontal or vertical as shown in Figures 3 and 4 and therefore the surface finish is not of importance in this case. Further, it can be seen that curved and slant surfaces were affected by the layering error and hence, deterioration of surface finish was noticed. To minimize the effect of layering error and to improve the surface finish, the slice thickness should be reduced even though this would increase the build time of the object.

Figure 5 shows a decrease in build time of the prototype with increasing slice thickness for all three objects. Since build time is an important factor, it is desirable to choose the largest possible slice thickness available such that the prototype would be completed in the shortest possible time. This however needs to be balanced with an aim to improve the surface finish and for the optimal usage of resources. It was also noted that the build height of the object has a direct impact on the build time (refer Figure 5). The Sphere and Battery Carrier have almost the same build height and hence the results are following a similar trend.

In creating slices, slice contours must be created first. In creating these contours, triangular facets are intersected by slice planes. In order to find the most efficient contour setting, the following factors must be taken into consideration:

● Tolerancevalue● Slicethickness● Numberofnon-intersectedfacets

Figure 9: Impact of Build Direction and Slice Orientation

(a) Alblock (b) Battery Carrier

Artkl 6.indd 98 10/21/2011 11:00:09 AM

99


As can be seen in Figures 6, 7 and 8, there are cases where some facets which are not intersected by any slice planes. This situation brings about wastage of resources where there is a need for more memory and time needed to process the redundant facets.

Figure 6(a) shows facets which are not intersected by slice planes, whereas Figure 6(b) shows opposite in which all of the facets are being intersected while Figure 6(c) shows multiple intersections per facet. Furthermore, the surface finish as shown in Figure 6(c) is negatively affected due to over tessellation and thus, may not be able to produce the best quality prototypes. Therefore, the configuration as shown in Figure 6(b) seems to be the best choice of all the models since it provides a good balance between tolerance (0.01 mm) and slice thickness (0.2 mm).

For the Alblock (refer Figure 7(a)), there are several non-intersecting facets. Figure 7(b) shows all facets being intersected while Figure 7(c) shows multiple intersections per facet. Since the surface finish of Figure 7(c) is worse than Figure 7(b) because of a higher tolerance value, thus, the model shown in Figure 7(b) would represent the best settings for this object with a tolerance of 0.01 mm and a slice thickness of 0.2 mm. It would also be the best since the features shown are small bevels which require a small tolerance and slice thickness which facilitates more points in forming slice contours and may provide a more accurate prototype. Prototyping intricate details, however, would more likely lead to increased build time. It was also observed that non-intersecting facets are present on curved surfaces but are seldom seen on flat surfaces.

For the Battery Carrier, there are several non-intersecting facets as presented in Figure 8(a). Figure 8(b) shows facets with all being intersected while Figure 8(c) shows multiple intersections per facet. Since this is a bevel as well, more points are better in prototyping this type of feature and hence, Figure 8(b) would be the best model setting. It is also possible to use the model (at tolerance = 0.004 mm and slice thickness = 0.2 mm) as shown in Figure 8(c) but its surface finish was worst. Again, for smaller slice thicknesses, build times would be larger but are necessary if intricate details are to be prototyped accurately.

When slicing is done, each slice was oriented horizontally and built in a vertical direction. As mentioned previously, each horizontal slice plane intersects triangular facets to form contours which make up each slice. Also, as stated by [10] and [4], the vertex of each triangle is stored along with a normal pointing outward. As seen in Figure 9, there are triangular facets present which are perpendicular to the build direction, that is, parallel to the slice planes (horizontal). These are the differently coloured (in green) for each object. The triangular facets on these faces are not intersected by slice planes and hence they are redundant. It can be deduced that these need not be taken into consideration in the process of prototyping. Thus, a proper scientific approach is necessary to decide the build direction or orientation of the object while prototyping. From this research it is clear that the horizontal surfaces should not be tessellated. In

Artkl 6.indd 99 10/21/2011 11:00:09 AM

100


addition to savings in computer hardware, avoiding the tessellation of horizontal surfaces will further reduce the occurrence of STL errors and build time.

Conclusions

This research involved in determining efficient tolerance and slice thickness values for prototyping. To find the best values, the tolerance and/or slice thickness would have to be adjusted until an appropriate number of intersections are present. It can be seen that the layering error increases and surface finish decreases with increasing slice thickness. It can also be mentioned that the build time was decreased with increasing slice thickness.

Screenshots approach used in the research gave a clear view of how tessellation and slicing are related and also to find a more efficient means of choosing tolerance and slice thickness values. This could be done by minimizing non-intersecting triangular facets and slice planes and by ensuring that enough intersections are present to produce fairly detailed slice contours.

Also, it is evident from the research results that objects with more curved and slant surfaces have an increased amount of layering error than the objects with horizontal and vertical surfaces. It is also apparent that complex objects show greater signs of over-tessellation and hence, would require more processing time to make more efficient STL files. It was also shown in the study that horizontal surfaces of the prototype need not be tessellated since slice contours do not intersect with the triangular facets. Developing an algorithm in this direction would lead to a more efficient representation of STLs.

References

[1] Tata, K., Fadel, G., Bagchi, A. and Aziz, N. (1998). “Efficient slicing for layered manufacturing”, Rapid Prototyping Journal, 4(4), 151-167.

[2] Pandey, P.M., Venkata, N., Dhande, R. and Dhande, S.G. (2003). “Slicing procedures in layered manufacturing: a review”, Rapid Prototyping Journal, 9(5), 274-288.

[3] Tyberg, J. and Bohn, J.H. (1998). “Local adaptive slicing”, Rapid Prototyping Journal, 4(3), 118-127.

[4] Fadel, G.M. and Kirschman, C. (1996). “Accuracy issues in CAD to RP translations”, Rapid Prototyping Journal, 2(2), 4-17.

Artkl 6.indd 100 10/21/2011 11:00:09 AM

101


[5] Jamieson, R. and Hacker, H. (1995). “Direct slicing of CAD models for rapid prototyping”, Rapid Prototyping Journal, 1(2), 4-12.

[6] Hope, R.L., Roth, R.N. and Jacobs, P.A. (1997). “Adaptive slicing with sloping layer surfaces”, Rapid Prototyping Journal, 3(3), 89-98.

[7] Choi, S.H. and Kwok, K.T. (2002). “A tolerant slicing algorithm for layered manufacturing”, Rapid Prototyping Journal, 8(3), 161-179.

[8] Kirschman, C. and Jara-Almonte, C.C. (1992). “A parallel slicing algorithm for solid freeform fabrication processes”, Proceedings of the 1992 Solid Freeform Fabrication Symposium, August 3-5, Austin, TX, 26-33.

[9] Swann, S. (1996). “Integration of MRI and stereolithography to build medical models: a case study”, Rapid Prototyping Journal, 2(4), 41-46.

[10] Noorani, R. (2006). Rapid Prototyping: Principles and Applications, (Wiley, U.S.A.)

Artkl 6.indd 101 10/21/2011 11:00:09 AM

some slicing issues in rapid prototyping · 2016-06-16 · in the past, the issue of slicing has...

Documents