Lightweight Design worldwide

, Volume 10, Issue 4, pp 16–21 | Cite as

Scale-independent simulation of fabric-reinforced thermoplastics with class A surfaces

  • Andreas Freund
Construction Outer Skin Components

Fibre- and textile-reinforced thermoplastics provide engineers with a group of high-performance materials for commercially viable lightweight design. However, their use in outer skin applications is restricted by the way in which reinforcing fibres on the surface of the component unavoidably stand out as a result of the heterogeneous nature of the materials. Leichtbauzentrum Sachsen uses appro priate simulation methods to identify and qualify influencing technical factors in order to achieve surface quality comparable to that of sheet metal parts.

Surface Values from Simulation

The trend in car manufacturing towards developing derivatives of basic models generally results in increased sales, but these are spread over a large number of models. The high manufacturing costs that this incurs with conventional construction using sheet metal make the use of fibre composites a sensible option. Besides the established fibre composite components with a thermoset polymer matrix, automotive developers are turning their principal focus to thermoplastic matrix systems that are able to lower production time as a relevant cost factor. Statutory requirements for reduced fleet consumption also necessitate commercially viable lightweight design in which fibre-reinforced thermoplastics play a promising role.

The large range of semi-finished fibre products and matrices available on the market together with the endless possibilities for layer composition make it impossible to conduct purely experimental development for reasons of time and cost. It is therefore essential to have a thorough understanding of the mechanical and technological relationships of materials combined with a suitable strategy for simulation and assessment.

A scale-independent simulation chain can be used in conjunction with a limited number of input variables to identify the relevant material properties and manufacturing boundary conditions and provide comparable surface characteristics that reproduce the long-wave optical impression in the same way as the established wave-scan measurement methods.

Besides material-related factors that allow a pre-selection of materials based on modern mixing rules, numeric meso-models make it possible to quantify the influence of fabric-based architecture on stiffness and stability using problem-specific representative volume elements (RVEs) and to generate a displacement field for the component surface in order to assess the visual appearance.

Based on the wave-scan surface analysis method deployed in industry to experimentally assess surface quality, an analysis algorithm using Fourier transformation and subsequent band-pass filtering provides the necessary reproducible statistical indicators for a class A assessment that visualise the ripple-causing layers in the composite structure, Figure 1.
Figure 1

Fabric reinforcements stand out on the surface (left) and typical surface profile (right) (© LZS)

Perception of the Surface Quality

Subjective perception of the car has grown accustomed over many years to elaborately painted body parts whose ultra-sleek appearance thus represents an important quality requirement. The choice of phrase “sleek appearance” already points to the lack of definition of the problem in objective, measurable terms and its subjective perception. A more profound understanding of the relationships between recognisable structures based on morphological realities of the eye and the way they are processed in the brain using pattern recognition processes is necessary in order to assess all possible surface variations. Besides purely subjective perception, there are a number of measurement methods that can be used to determine objective and reproducible surface characteristics that then allow classification with regard to subjectively motivated thresholds such as class A.

The practical applicability of the calculated wave lengths was demonstrated by a comparison with wave-scan measurements.

Visual stimuli are processed at various levels in the human brain [1]. Consequently, only a fraction of the information received by the retina actually reaches our consciousness. The brain processes information only to a limited degree, for example when a visual stimulus is only visible for a split second and can therefore not be consciously perceived, like a brief glance or the famous first impression. However, certain constellations of stimuli are perceived at a more basic level by special nerve cells and regions of the brain. In this process, elementary stimulus patterns such as dots, lines and angles are analysed as attributes subconsciously and, in a second step, assembled to form more complex shapes. A linear arrangement of stimuli in particular leads to a very early recognition process in the brain and works even with the smallest of contrasts [2]. A highly concentrated arrangement of neural cells with similar response characteristics, when stimulated, leads to increased attentiveness in adjacent neural cells, comparable to a ringing mobile phone where all those present in a room listen to whether it is their device. However, this effect is also used to advantage by the human brain, as it allows the most important information to be recognised and generated, like when driving in fog with extremely poor visibility. This effect is disadvantageous when it comes to the way in which fabric structures — with their simple, linear patterns — stand out and are perceived.

Given average receptor spacing on the retina and the diffraction pattern of a point source of light on the lens of the eye at a viewing distance of e.g. 4 m, the optical resolution capacity of the human eye allows patterns with a spacing of 1.2 mm to be easily recognised. This lower limit of the wave length is a disadvantage for fabric structures, as their size is within the typical range of fibre bundles.

Non-destructive methods of optical measurement using a laser to scan surfaces are commonly used in industry to gauge surfaces. The auto-focusing, optoelectronic measuring system of a laser profilometer works with a variable focus position controlled by the diameter of the point of light recorded as a measuring signal that is virtually independent of the reflectivity of the test specimen. Nevertheless, a large number of parallel linear scans are necessary to completely scan the entire surface, making it a time-consuming process.

The wave-scan scattered light method responds predominantly to slight microscopic gradients in the surface finish which are illuminated at an angle of 60°, with the reflection being recorded by a detector on the opposite side at the same angle. The intensities of the reflected beams of light are measured and statistically analysed to provide dimensionless indicators that correlate with the rippling or roughness of the surface. The so-called long wave (LW) between 1.2 mm and 12 mm is of particular importance for assessing surface quality.

The results of all measurement methods are highly dependent on the measurement location, i.e. the choice of the evaluation path either parallel to or perpendicular to the fibre pattern. This can also be seen in the limited reproducibility of measurements with relatively large fluctuations of up to 30 %.

Scale-independent Modelling

Sophisticated modelling strategies and computational models that provide as exact a map as possible of the geometry of the reinforcement structure as well as the physical, non-linear behaviour of the material are necessary in order to simulate the shrinkage characteristics of fabric-reinforced thermoplastic composites produced using pressing methods. Virtual characterisation on a micro-scale is used to build meso-models in the form of RVEs that map areas of the matrix and fabric structures in accordance with their actual distribution. While the use of unit cells and RVEs to simulate the mechanical behaviour of fabric-reinforced composite materials independent of scale may reflect the latest research, it has so far been rarely used for the scientific investigation of shrinkage processes.

Unit cells have been rarely used for the scientific investigation of shrinkage.

Several methods are available for the discretisation of the RVEs for different fabric compounds. Commonly available software solutions (such as TexGen) can be used to generate finite element networks in the form of voxel models, tetrahedral networks or so-called dry fibre networks, Figure 2. The latter — using innovative E-RECs (enhanced representative embedded cells) — offer particular advantages in the assessment of strength behaviour, as slightly deformed elements are combined with a high level of discretisation quality [3].
Figure 2

Comparison of the network quality of the voxel method (left), tetrahedral network (centre) and E-REC method (right); matrix area (yellow) partially hidden (© LZS)

The use of periodic boundary conditions has proven to be unnecessary for the calculation of shrinkage behaviour, but is essential for modelling stability and stiffness behaviour.

An evaluation algorithm provides the indicators for a class A assessment.

When selecting temperature boundary conditions, it is assumed that shape-giving processes cease to depend on forming pressure from a certain, material-dependent temperature. The compression stresses induced by the process are smaller from this point in time than the yield point of the thermoplastic, meaning that shrinkage differences can no longer be compensated for by yielding in the matrix, resulting in local shrink marks forming on the surface.

A formulation complying with the theory of linear elasticity is sufficient to perform fundamental simulation of shrinkage behaviour. Furthermore, in order to calculate process influences such as pressure patterns and temperature distribution, it is necessary to draw on formulations from flow plasticity theory for which the flow potential function in accordance with Hill [4] offers a sufficiently precise description with a high degree of variability of the yield point for an engineering-based assessment.

Evaluation Methodology

Reflecting the wave-scan surface analysis method commonly used in industry to measure surface rippling in everyday practice, a suitable method is needed for the assessment of calculated surface rippling in fabric-reinforced thermoplastic composites that can be used to determine comparable parameters for an engineering-based assessment. To this end, an assessment methodology for linear paths using a fast Fourier transformation (FFT) and digital filtering was developed and enhanced with an area evaluation methodology for visualisation of the structures responsible for rippling.

The point of departure for the evaluation are characteristic paths in the warp and weft from which shifts in the points of intersection in the through-thickness direction are read. The characteristic paths are selected in such a way that they include the points of maximum and minimum shift, Figure 3.
Figure 3

Comparison of the profiles from the finite element method (FEM) and laser profilometer measurement as proof of the suitability of selected boundary conditions (© LZS)

Like the finite element (FE) calculation, the laser profilometer provides an elevation profile of the actual surface along the measurement path, Figure 3, meaning that the experimentally demonstrated relationship between the standard deviation from the laser profilometer data and the LW values can also be assumed for the standard deviation from the calculated values. The practical applicability of the method was demonstrated using specially prepared specimen plates with different laminate structures and a comparison of wave-scan measurements with calculated rippling. The mean difference here was only about 5 %.

Besides the dimension-free parameter for long waves, the wave spectrum also shows the relationships between wave intensity and wave length. This allows an inference to be made about the layer that causes it. As the evaluation method is based on a Fourier transformation, a uniform presentation in the wave length range cannot be made. Equidistantly distributed frequencies result in whole-number measuring length factors for the wave length. As these are negligible in the short-wave range, information is missing about the architecture of the fabric laid on a scale of several fibre bundle widths.

A variable evaluation window can be used to generate sampling lengths whose whole-number factors cover the wave length range in question and offer a considerably more granular picture of the wave spectrum, Figure 4, making an assignment of the wave length to the geometry of the origin more likely.
Figure 4

Comparison of the wave spectrum with (front) and without (back) a variable window width for a test specimen (© LZS)

The clearly defined wave ranges of the interfering signals can now be used to derive filter widths for band-pass filtering. When combined with image processing software, these provide the rippling-causing structure to be visualised, Figure 5. The simplest approach is for the surface shifts and line profiles to be decomposed and the familiar signal filtering tools to be used for them. A far more elegant processing method is provided by 2-D Fourier-transformation, presenting as the result the Fourier coeffi cients assigned to the wave lengths depending on their orientation in the evaluation area.
Figure 5

Grey-value images of band-pass filtered surface profiles based on 2-D Fourier transformation: entire elevation profile (left), broadband-pass filtered (centre), narrow-band-pass filtered to discrete wave length (right) (© LZS)

The numerical simulation and evaluation method developed was used to perform comprehensive parameter studies into the influence of material-, geometry- and process-specific design variables on surface rippling induced by internal stress in fabric-reinforced thermoplastic composites produced using pressing methods. It demonstrated in particular that, in addition to the choice of a suit able combination of materials, a gradual layering with the specific use of buffer layers can significantly reduce surface rippling. The knowledge gained about technical dependencies forms a solid basis for optimising layering in fabric- reinforced thermoplastic composites for outer skin application with class A surface quality. Conversely, the knowledge of ripple-causing relationships also enables rippling to be used to produce a drag-reducing surface like that of a golf ball using smooth tool surfaces. |


  1. [1]
    Hayes, J.-D.: Detecting deception from neuroimaging signals — a data-driven perspective. In: Trends in Cognitive Sciences, No. 12 (4), 2008, pp. 126–127Google Scholar
  2. [2]
    Grabska-Barwinska, A.; Distler, C.; Hoffmann, K.P.; Jancke, D.: Contrast independence of cardinal preference: stable oblique effect in orientation maps of ferret visual cortex. In: European Journal of Neuroscience, Nr. 29, 2009, pp. 1258–1270CrossRefGoogle Scholar
  3. [3]
    Freund, A.; Gude, M.; Vogel, C.: Simulation based process development for the novel thermo clinching joining technology. Tagung Proceedings of Euro Hybrid Materials and Structures, Kaisers lautern, 2016Google Scholar
  4. [4]
    Hill, R.: A theory of the yielding and plastic flow of anisotropic metals. In: Proceedings of the Royal Society, Nr. 193 (1033), 1948, pp. 281–297CrossRefGoogle Scholar

Copyright information

© Springer Fachmedien Wiesbaden 2017

Authors and Affiliations

  • Andreas Freund
    • 1
  1. 1.Leichtbauzentrum SachsenDresdenGermany

Personalised recommendations