Abstract
Textile-type structures realised by Additive Manufacturing (AM) techniques, also known as 3D Printing (3DP), have been the subject of growing international research for the previous decade. This is due to their significant design potential and their capability to produce net-shape textile artefacts for the development of personalised, high performance textiles applications.
The paper discusses the on-going research in this novel area of AM, concentrating on their design, modelling and the efficient generation of the 3D data required for manufacture. The paper also discusses the current applications of AM textiles and
provides key examples of their use as a technical textile.
Introduction
Additive Manufacturing (AM), more commonly known as 3D printing (3DP), has been labelled a disruptive technology due to the exciting impact it is having on the established areas of design and manufacturing. AM delivers this exciting potential through the fundamentally different tool-less methodology it employs for the realisation of parts and components. The current foundation of AM is Computer Aided Design (CAD) software and the three-dimensional (3D) data they generate. AM systems build directly from 3D CAD data and manufactured parts layer by layer, negating many Design For Manufacture (DFM) constraints associated with more conventional manufacturing techniques. This contrast enables designers and engineers to realise any complexity of geometry they can imagine and print their ideas as fully functional, physical 3D objects with almost no additional cost penalties. This step-change in design freedom has resulted in a serious and growing research interest, world-wide.
A novel application of AM are textile-type structures that not only incorporate drape and free-movement but also have the capability to be conformal, net-shaped and personalised to the human form - these are known as AM Textiles (Bingham et al 2007) and demonstrated in Figure 1. 
Effectively, AM Textiles approximate to the modern-day equivalent of Medieval chainmail. However, due to the application of AM techniques, the level of geometric complexity now available is virtually limitless. Additionally, the resolution capabilities of AM techniques allow the manufacture of AM textiles at the Meso and Micro-level (from 5mm to less than 1mm) in a variety of build materials including: metals, polymers, ceramics and composites.
Utilising the design freedom of AM technologies requires accurate 3D CAD data of the intended geometry and this represents the main barriers for further development and increased applications of AM textiles. Academic research has mainly concentrated in this area, developing efficient modelling strategies for the creation of 3D CAD data required for manufacture (Bingham et al 2007, Bingham 2007, and Bingham 2013). However, additional research has also been undertaken to address their mechanical properties, design and possible high-performance applications (Crookston et al 2008, and Johnson et al 2011). Most recently, research has been undertaken to successfully design and manufacture AM textiles capable of attaining the level one British standard for stab-resistance (Johnson et al 2013). While a genuine motivation exists for the further development of AM textile applications, efficient modelling of conformal and net shape AM textile artefacts remains the significant issue.
Design of AM textiles
AM textiles are considered to be hierarchical structures with the fundamental element(s) being the singular (or multiple) repeating structure (or links) that combine and tessellate to create the final AM textile itself, as demonstrated in Figure 2.
The design of individual links of AM textiles requires several considerations, initially these include:
• Intended functionality of the final textile
• Promotion of free movement and drape characteristics
• Ability to tessellate
The design of AM textile link data suitable of manufacture requires several considerations, namely the resolution capability of the AM system being utilised, which dictates the scale (or size) at which the individual links can be manufactured, and their minimum separation. The minimum separation is required to ensure that the individual links of the AM textile do not fuse during manufacture and remain separated to provide the drape and free movement characteristics in the final AM textile. However, further manufacturing considerations are also required dependant on the AM technology utilised, these include: build material properties and support structure requirements and removal. Historically, Laser Sintering has been the AM technology of choice for the manufacture of AM textiles – this is due to the absence of any support structure requirements, resolution capabilities and the desirable mechanical properties of the Polyamide build materials.
Once the design aspects of the individual links have been established, the next stage of the design process requires the modelling of the complete AM textile data for manufacture.
Modelling AM textiles with CAD
The 3D modelling of AM textiles is difficult using conventional CAD software, however, planar samples (or swatches) of AM textile structures can be generated by expert users using ‘Array’ or ‘Patterning’ functionality common to most systems. For planar AM textiles, the process requires the initial modelling of the repeating link structure (or structures) and the specification of the dimensional spacing. Generating instances of the repeating link structure at set dimensional spacing therefore generates the final AM textile planar sample. Figure 2 demonstrates both a quadrilateral-based array and triangular-based arrays of link structures.
The 3D modelling of conformal and net-shaped AM textile data is far more complicated and introduces additional considerations. Not only must link separation be maintained throughout but the surface normal direction and rotational aspects must also be considered to guarantee correct tessellation of the individual links. For simple link designs, that are rotational symmetrical, a torus for example, the rotational consideration is not an issue. However, surface normal direction remains vital for creating a conformal AM textile structure. The importance of the surface normal is demonstrated in Figure 3 which illustrates a simple curve example. The left image in Figure 3 demonstrates the surface normal of the curve at several locations, indicated by the arrow. The middle image in Figure 3 demonstrates a series of links that do not follow the surface normal of the curve, while the right image shows the surface normal correctly calculated.
A further obstacle in the 3D modelling of conformal and net-shaped AM textile structures is the ability to accurately describe the locations of the individual links in 3D space. For planar samples, the locations of the links are restricted to a 2D plane and described by the ‘Array’ function. This generates instances of the original geometry at set values in both the X and Y axis. However, there is no such functionality to accurately describe a 3D array and the only available alternative is the manual manipulation of the individual links by the CAD user. While this is achievable for very small AM textile structures, this technique becomes impractical and eventually impossible for larger structures that can incorporate hundreds or possibly thousands of individual links.
The efficiency of the ‘Array’ function described previously remains extremely desirable as it allows the 3D modelling of a single (or multiple) link structure to be copied and translated at set dimensions. This ensures that a minim
um separation is maintained throughout the AM textile data and removes any manual manipulation by the CAD user. Further investigation of this techniques revealed that the underling dimensional spacing can be considered as a surface mesh representation of the intended configuration of the AM textile. The surface mesh representations of the quadrilateral and triangular-based planar AM textiles demonstrated in Figure 2 illustrates this point and developed the idea of creating a mapping mesh system to describe the location of individual links in 3D space. Therefore, if a suitable mapping mesh can be generated that describes the intended configuration of the AM textile; this provides an efficient means of describing all the locations of the individual links. Using this principle, it was possible to generate net-shaped AM textiles, as demonstrated by the hemispherical example in Figure 4.
While this methodology allowed the generation of a hemispherical AM textile structure (Figure 4), manual manipulation of individual links was still required. A further consideration of this technique is the requirement to use triangular-based mapping meshes. While quadrilateral-based arrays can be utilised for planar and cylindrical AM textile structures, quadrilaterals do not create high-quality mesh structures with uniformly distributed mesh elements, especially for complex manifold surfaces. This is demonstrated in Figure 5, which illustrates two example mesh structures, left triangular and right quadrilateral.
The irregularity and distortion seen in the quadrilateral meshes would result in distortion in the final AM textile when utilised, where individual links would overlap and fuse during manufacture. In contrast, the triangular meshes are much higher quality, with the majority of triangular elements being equal in size (or equidistant). The use of the triangular mapping mesh enabled the generation of hemispherical AM textile structure in Figure 4, where the minimum separation of individual links was maintained throughout and the data generated entirely suitable for manufacture.
Development of a dedicated AM textile modelling strategy
In order to unlock the full potential of conformal/net-shaped AM textiles and provide practical access to an efficient modelling strategy, a dedicated technique was required. The initial modelling research highlighted two important aspects that needed to be addressed to achieve this aim.
• A geometry mapping tool capable of automatically populating individual 3D link structures to a mapping mesh (modes and element midpoints)
• A mapping meshing algorithm/system capable of generating high-quality equidistant mesh structures (mapping mesh) for a range complex surfaces
Addressing these two aspects provides an efficient means of generating complex
conformal AM textile data suitable for manufacture by AM techniques.
Geometry Mapping Methodology
The fundamental requirements of the mapping methodology were developed as follows:
• Provide the capability to map any complexity of geometry described as 3D CAD data
• Accurately calculate the surface normals and rotational requirements at all mapping locations (modes and element midpoints)
• Accurately populate all locations of the mapping mesh with specified 3D CAD data
• Export the final 3D AM textile data as STL format for manufacture
These fundamental requirements were translated into a working methodology and standalone software tool – this was undertaken in collaboration with the University of Nottingham and resulted in the creation of ChainLink (Bingham et al 2007). The core mathematics underpinning this methodology is not discussed within this paper but is available in (Bingham 2007, and Bingham 2013).
Mapping Meshing Generation
The second aspect of the modelling strategy was the generation of suitable mapping meshes. Various existing surface mesh generation algorithms were investigated including: Indirect or parametric space types (Marur 2005), Advancing front types (Borouchaki 2000), and finally direct and indirect sphere packing types (Shimada 1997). Additionally, several commercially available Finite Element (FE) pre-processors were also explored, including: MSC Patran (MSC 2011), ANSY (ANSY 2011) and Hypermesh (Hypermesh 2011). However, it was quickly established that the generation of a high quality equidistant mesh structure was impossible for all surface types. Modification of an existing surface mesh generation algorithm was also considered, however, given the exacting requirements it was decided that a dedicated surface meshing algorithm was required. Further research was undertaken to achieve this aim and a new experimental surface meshing algorithm based on Sphere Packing (Shimada 1997) was developed (Bingham 2007).
The result of combing ChainLink (mapping software) with the developed surface meshing algorithm are demonstrated in Figure 6 and illustrates the ability to generate high-quality conformal AM textile data suitable for manufacture.
The demonstrated methodology for the generation of conformal AM textile data is robust, accurate and efficient. However, the quality and range of the data generated is dependent on the input mapping mesh. Any distortion or irregularity within the mapping mesh is directly replicated in the resultant AM textile data, as demonstrated in Figure 7.
The capabilities of the surface meshing algorithm developed for this research are also limited and can only generate equidistant mesh structures from mathematically described quadrilateral surface patches (Bingham 2007). While the meshing algorithm developed allowed the discussed mapping methodology to be validated, the limitations of the meshing algorithm prevents practical access to a range of more complex surface types and manifold objects. This capability is required for present and future applications of AM textiles and addressing this limitation required a further investigation of high-quality surface mesh generation.
Mapping Mesh Generation: Re-meshing
A secondary line of investigation examined re-meshing techniques that are utilised for the quality improvement of existing mesh structures for a range of applications, including; STL quality, FEA and rendering applications in polygon modelling systems.
Re-meshing involves the manipulation of the existing mesh structure based on a set of quality variables, typically element lengths and angle or aspect ratio. Two excellent examples of re-meshing software include Geomagics (Geomagics 2013) and Meshlab (Meshlab 2013). Examples of the mesh quality delivered by both systems are demonstrated in Figure 8, based on the original STL mesh structure of a hand.
The documented re-meshed examples in Figure 8 are considered high-quality; however, they still do not attain the quality expectations required for AM textile generation. To utilise such meshes, the mapping methodology required a scaling function to be included that allowed the link geometry to be modified (scaled) based on the element lengths at each mapping location. This was eventually included within the mapping software and the results of using the Meshlab re-mesh are demonstrated in Figure 9.
The use of a scaling function within the mapping methodology and re-meshed surface meshes does allow the efficient generation of complex, net-shaped AM textile structures.
However, the irregularity within the mapping mesh structure reduces the range of possible link designs that can be utilised. The use of re-meshing also removes the ability to control the dimensions of the resultant links or their separation due to the scaling function. These limitations can result in the data generated not being suitable for manufacture by AM techniques. For some AM textile applications, the issues associated with using re-meshing can be tolerated, however, the aim of this research was the development of a method for the generation of uniform and dimensionally regular AM textile data.
Mapping Mesh Generation: Mesh Conforming
A further area of investigation was the idea of conforming a predefined mesh structure to a targeted surface. Mesh conforming has several qualities that make it an attractive alternative over re-meshing-based alternatives. The process of mesh conforming is demonstrated in Figure 10 and requires a planar mesh structure to be initially created before subsequent fitting (conforming) to a target surface using specified variables, including: mesh shear, element stretch and element compression (Autodesk 2013).
Control over these variables affects the ability to conform to the targeted surface but also affects the final distortion observed in the resultant mesh. Through careful management of these variables a compromise can be generated, resulting in a surface mesh that approximates the targeted surface with a high-quality structure and containing minimal dimensional variation.
The application of mesh conforming in combination with the new scaling function within the mapping software allows conformal AM textiles data to be efficiently generated, as demonstrated in Figure 11. This technique allows practical access to a range of complex topology far beyond the capabilities of the experimental meshing algorithm and produces higher-quality mesh structures.
Due to the uniformity of the generated mapping mesh, this technique also allows a greater range of AM textile link designs to be utilised (not shown). However, the use of mesh fitting for the generation of a mapping mesh does preclude the ability to generate net-shaped AM textile artefacts – this can only be achieved through re-meshing techniques as it is impossible to accurately conform a planar mesh to a manifold object without significant distortion and the generation of irregular seams.
AM textile applications
The fundamental research addressing the efficient generation of complex, conformal and net-shaped AM textile data has seeded further research addressing specific AM textile applications. To date, this has included AM textile research investigating fashion (Bingham 2007), contour-fashion (Continuum 2013), sports personal protective equipment (Brennan-Craddock 2008), wearable medical devices (Paterson 2013) and stab-resistance (Johnson 2013). All of the cited examples have used the design freedom of AM to generate textile structures that provide increased performance and/or combined desirable functionalities. Two specific examples include AM immobilisation wrist splints and personalised AM body armour.
AM immobilisation wrist splints
Research investigating a digitised workflow for the design and manufacture of personalised immobilisation wrist splints utilised AM as the enabling fabrication technology (Paterson 2013). The research aims to provide medical practitioners with a plausible digital alternative to the current manual/craft-based manufacture of immobilisation wrist splints. The proposed digital workflow and fabrication provides significant potential benefits, including:
• Improved patient compliance through comfort and aesthetics
• Multiple and repeatable fabrication of personalised wrist splints
• Reduced overall lead-times and costs
To aid the donning and doffing by the user, AM textiles where targeted to provide a fully integrated hinge within a Laser Sintered immobilisation wrist splint, demonstrated in Figure 12. The AM textile hinge allowed the fabrication of the splint in a single process and was designed to follow contours of the upper extremity geometry to provide seamless support.
Research investigating the design and manufacture of personalised stab-resistant body armour using AM techniques has led to the development of stab-resistant AM textiles (Johnson 2013). The research aims to address a number of issues associated with current body armour solutions through the application of body scanning techniques and a digitised workflow, including:
• The protection vs. mobility trade-off
• Restrictive and cumbersome use
• Issues and injuries related to ill-fitting armour
The research investigated Laser Sintering as the prime AM technique and required an extensive phase of stab/impact testing of Laser Sintered Polyamide samples. The research has been successful in developing AM textiles that conform to the current
Home Office and Scientific Development Branch (HOSDB) level one standard for stab-resistance (Johnson 2013). An example of a stab-resistant AM textile is demonstrated in Figure 13.
Discussion and conclusions
AM textiles provide a real opportunity for the design and manufacture of geometrically complex and potentially high-performance textile structures for a range of possible applications. However, practical access to an efficient means of generating conformal and net-shape 3D data has restricted wider-scale adoption and further investigation.
The methodologies presented here for the efficient generation of their conformal and net-shape 3D data aims to remove these restrictions and enable further fundamental research in this novel area of AM.
The limited application-driven research already underway is extremely promising and serves to demonstrate the potential of AM textiles as a technical textile solution for high performance applications. As AM technologies continually improve in resolution capabilities, build material properties and speed, the future potential applications of AM textiles is set to increase and develop.
References
- ANSYS software website, http://ansys.com/, accessed September 2011
- Autodesk 3D Studio Max software website, http://www.autodesk.co.uk/ , accessed June 2012
- Bingham, G.A., Hague, R.J.M., Tuck, C.J., Long, A.C., Crookston, J.J. and Sherburn, M.N. “Rapid Manufactured Textiles” International Journal of Computer Integrated Manufacture, Volume 20, Issue 1, 2007, pp96-105.
- Bingham, G.A., PhD thesis: Generation of 3D data for Rapid Manufactured Textiles, July 2007, Loughborough University.
- Bingham, G.A. and Hague, R.J.M., “Efficient Three Dimensional Modelling of Additive Manufactured Textiles”. Rapid Prototyping Journal, Volume 19, Issue 4, 2013.
- Borouchaki, H., Laug, P. and George, P., “Parametric surface meshing using a combined advancing-front generalized Delaunay approach”, Inter Journal of Numerical Methods in Engineering, v49, p233-259, 2000.
- Brennan-Craddock, J.P.J., Bingham, G.A., Hague, R.J.M. “Impact Absorbent Rapid Manufactured Structures (IARMS)”.The 19th Solid Freeform Fabrication Symposium, Texas, USA, August 4-7, 2008.
- Continuum fashion website, http://www.continuumfashion.com/N12.php , accessed March 2013.
- Crookston, J.J., Long, A.C., Bingham, G.A. and Hague, R.J.M.,“Finite Element Modelling of Mechanical Behaviour of Rapid Manufactured Textiles”. Proceedings of the Institution of Mechanical Engineers, Part L: Journal of Materials: Design and Applications, v 222, n 1, 2008, p 29-36.
- Geomagics, software website, http://www.geomagic.com/en/, accessed March 2013.
- Hypermesh software website, http://www.altair.com/software/hw_hm.htm , accessed August 2011.
- Johnson, A.A., Bingham, G.A. and Wimpenny, D., “Additive Manufactured Textiles for High-Performance Stab resistant Applications” – 12th RDPM conference, Lancaster, England, June 2011.
- Johnson, A.A. and Bingham, G.A., “Additive Manufactured Textiles for High-Performance Stab Resistant Applications". Rapid Prototyping Journal, Volume 19, Issue 3, 2013.
- Marur, S., “On the meshing of trimmed 3D surfaces”, Advances in Engineering Software, v36, pp 338–345, 2005.
- Meshlab, software website, http://www.meshlab.org/, accessed March 2013.
- MSC software website, http://www.mscsoftware.com/Products/CAETools/Patran.aspx, accessed July 2011.
- Paterson, A.M., PhD Thesis: Digitisation of the Splinting Process: Exploration and Evaluation of a Computer Aided Design approach to support Additive Manufacture, 2013, Loughborough University.
- Shimada, K., Yamada, A. and Itoh, T., “Anisotropic triangular meshing of parametric surfaces via close packing of ellipsoidal bubbles”, Proceedings of the 6th International Meshing Roundtable, Park City, pp 375-390, 1997