Elsevier

Computer-Aided Design

Volume 40, Issue 2, February 2008, Pages 235-243
Computer-Aided Design

Extruder path generation for Curved Layer Fused Deposition Modeling

https://doi.org/10.1016/j.cad.2007.10.014Get rights and content

Abstract

Extruder path generation for a new rapid prototyping technique named “Curved Layer Fused Deposition Modeling” (CLFDM) has been presented. The prototyping technique employs deposition of material in curved layers in contrast to flat layers as in Fused Deposition Modeling (FDM). The proposed method would be particularly advantageous over FDM in the manufacturing of thin, curved parts (shells) by reduction of stair-step effect, increase in strength and reduction in the number of layers. The criteria for the generation of tool paths for CLFDM are proper orientation of filaments and appropriate bonding between adjacent filaments in same layer and in successive layers.

Introduction

Rapid prototyping (RP) refers to a group of solid freeform fabrication (SFF) processes that are capable of developing complex shapes without part-specific tooling in a short span of time. Newer RP processes are being developed and commercialized every year [1]. Layered Manufacturing (LM) technology is employed for most of the RP processes wherein a part is produced by employing layer-by-layer deposition of material. Fused Deposition Modeling (FDM) is one of the commercially exploited LM processes where a filament of heated (fused/semi-solid) thermoplastic material is extruded through a deposition nozzle (which would henceforth be referred to as the ‘extruder’ in this article) and applied over a flat surface to form a layer (Fig. 1). The main advantage of LM over conventional manufacturing is that complex shapes can be physically realized without elaborate tooling. However, there are some specific part shapes like thin, slightly curved shell-type structures (skull bones, turbine blades etc.) where the application of LM is poorly suited and may result in lack of strength, stair-step effect (poor surface finish) or large number of layers (higher build time) [2]. The reason behind such low part quality is the discontinuous nature of the filaments in building up the part by LM (Fig. 2).

Strength of parts made by FDM suffers from anisotropy [3] and adhesive strength between layers (or across filaments) is appreciably less than the strength of continuous filaments (longitudinal strength). Zhong et al. [4] studied the mechanical properties of short fiber-reinforced ABS polymers for use as a FDM feedstock material. On comparison of the longitudinal strength with the adhesive strength, it was observed that the former was substantially higher than the latter. Hence, discontinuity of filaments on the part shown in Fig. 2 produced by FDM would tend to reduce its strength. Apart from this, there is pronounced stair-step effect in the sample of Fig. 2 and it is obvious that layer thickness would have to be appreciably reduced in order to achieve better surface finish. This in turn would increase the build time, as mentioned earlier.

However, in a number of cases, proper choice of orientation of the part (build direction [5], [6], [7], [8]) in the FDM chamber may eliminate some of the above-mentioned drawbacks. For example, in the case of a typical curved thin part under consideration, continuity of filaments can be obtained in the concerned section of the part if it is held upright (Fig. 3) in the deposition chamber and the deposition carried out as shown (contour fill). However, discontinuity of filaments would still exist across these sections shown by dotted lines. In fact, there would not be any continuous fibers across these sections. Further, if the part has bi-directional curvature (Fig. 4) the selected build orientation would not serve the purpose.

Adaptive slicing is perhaps the main strategic response of FDM to solve these problems. Thinner layers in regions of low (near-horizontal) surface slope and high curvature would certainly reduce the surface roughness. At the same time, build time is not expected to increase substantially as higher layer thickness would be retained in places of high (vertical or near-vertical) surface slope and low curvature. However, the improvement in strength of the part due to thinner layers would be marginal. Further, if thinner layers are applied only at selected regions (the very idea of adaptive slicing), the part would remain as weak as the thick-layered regions. In the available literature, one may find a number of attempts for adaptive slicing of parts [9], [10], [11], [12], [13], [14], [15], which include both direct slicing and slicing of faceted surfaces in the form of STL files.

“Curved layer FDM” or CLFDM — as proposed in this work, may offer solutions to most of these issues for thin curved shell-type parts as discussed above. In this process, which proposes an entirely new building paradigm for FDM, the filaments would be deposited along curved (essentially non-horizontal) paths instead of planar (horizontal) paths.

If the literature is considered on RPT (Rapid Prototyping Technology) in general, the idea of curvilinear (non-horizontal) material deposition is not entirely new in other spheres of additive manufacturing. Klosterman et al. [16] have developed curved layer LOM (Laminated Object Manufacturing) process for monolithic ceramics and ceramic matrix composites (CMC). The advantages of this curved layer process are elimination of stair-step effect and improved surface quality, increased build speed, reduced waste, and easier decubing. Researchers from CREDO Laboratory, Clemson University [17] had started an endeavor on variable slice orientation in SLA process. Significant improvement in surface quality due to deposition at variable slice orientation had been envisaged. Kerschbaumer et al. [18] presented an algorithm for generating tool path for 5-axis laser cladding using adaptive slicing technology. It approximates the tessellated CAD model with a parametric surface and generates the tool paths by varying one of its parameters (Isoparametric paths). Although the generation of tool path is straightforward, isoparametric paths suffer from a major drawback of obtaining denser tool paths in some surface regions than others due to non-uniform transformation between the parametric and the Euclidean spaces [19].

The objective of the present work is the investigation for the manufacturing of curved thin parts by depositing material in curved non-horizontal layers using FDM. It is envisaged that there would be substantial improvement in the mechanical properties of thin-section curved shell-type parts made by CLFDM in comparison to FDM. It is also aimed to develop and implement an algorithm for generating 3D curved paths of the extruder head for filament deposition to achieve successful reproduction of part shape and proper inter-filament bonding.

With the development of bio-friendly materials, RP has been exploited in the field of biomedical engineering [20], which requires precision and flexibility. Biocompatible PMMA-resin was used for developing the part of a skull for replacement in the case of an accident victim [21]. The proposed method of CLFDM would be very appropriate for the manufacturing of functional prototypes of skull bones and other thin shell-type parts. Other potential applications are in the manufacturing of intricate and small sized turbine blades or objects of thin cross-section, produced for actual use or for design, verification and testing. The advantages of using CLFDM are — lesser number of layers for identical part, higher continuity of filament resulting in more strength and more bonding between consecutive layers (due to larger area of inter-layer bonding).

Section snippets

Process description for CLFDM

In FDM, 3-axis CNC is sufficient for the deposition of filaments in a flat layer. In CLFDM, the ideal choice would be a 5-axis CNC machine, such that the extruder axis would always coincide with the normal to the layer at the point of deposition. However, a 3-axis machine would also suffice when this surface normal (n) does not appreciably deviate from the vertical (Fig. 5). This condition is valid for slightly curved parts.

FDM uses 2C, P (x- and y-axes with contouring control and z-axis with

Results and discussion

The skull shown in the Fig. 9(a) as obtained from [2] is produced by SLA (Stereolithography) which is a LM process. It is included as an example to show an extreme case of stair-step effect — especially at the top dome region. Fig. 9(b) shows the same part (the top dome, a typical thin curved part) if it be made by CLFDM. This graphical simulation of the part Fig. 9(b) is obtained by using the approach outlined for the extruder trajectory in CLFDM. Only two deposition layers are shown for the

Conclusion

A new method called “Curved layer Fused Deposition Modeling” has been formulated and tested on parametric surfaces. The advantage of this method is in creating thin-section, slightly curved (shell-type) parts where the flat-layered FDM might fail to meet the strength requirements. Higher strength is obtainable by employing longer length filaments or roads and obtaining curved inter-layers of larger area per layer. The proposed method has the potential to increase strength of parts and to reduce

Acknowledgements

We sincerely thank ‘The British Journal of Oral & Maxillofacial Surgery’ for granting us the permission for using Figs. 9(a) and 11. We also would like to extend our thanks to Dr. John Winder for providing the digital copies of these pictures.

References (30)

  • H.-Y. Feng et al.

    Constant scallop-height tool path generation for three-axis sculptured surface machining

    Computer-Aided Design

    (2002)
  • G.V.V. Ravi Kumar et al.

    Computing constant offsets of a NURBS B-Rep

    Computer-Aided Design

    (2003)
  • G.V.V. Ravi Kumar et al.

    Computing offsets of trimmed NURBS surfaces

    Computer-Aided Design

    (2003)
  • L.A. Piegl et al.

    Computing offsets of NURBS curves and surfaces

    Computer-Aided Design

    (1999)
  • R.T. Farouki

    Exact offset procedures for simple solids

    Computer Aided Geometric Design

    (1985)
  • Cited by (0)

    View full text