OnScale - CFRP Simulation of Ultrasonic Non-Destructive Testing of Composite Materials (Whitepaper)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Whitepaper Series

Simulation of Ultrasonic Non-Destructive


Testing of Composite Materials

Executive summary
This white paper discusses how ultrasonic non-destructive testing (NDT) can be efficiently
and effectively optimized, thus reducing costs and risks, with the use of accurate engineering
simulations. With OnScale software, research and design engineers can run multiphysics
simulations of many ultrasonic inspection techniques that include piezoelectricity, electrical
circuits, structural mechanics, acoustics, and heat transfer phenomena. OnScale is fully cloud-
enabled, empowering engineers with the high-performance computing (HPC) resources
needed to explore their design space quickly and with ease. Semiconductors, MEMS, sensors,
medical devices, and 5G and IoT RF systems are among the many applications that can benefit
from design and optimization with OnScale. To learn more visit onscale.com/applications.

info@onscale.com www.onscale.com
Simulation of Ultrasonic Non-Destructive Testing
of Composite Materials
Contents
I. Introduction: Ultrasonic Non-Destructive Testing Simulation for Composite Materials 3
II. Challenges in Simulating Ultrasonic Inspections of Composite Components 4
III. Building Multiphysics Composite Inspection Models with OnScale 6
A. High-Resolution Geometrical Modeling and Boundary Conditions for CFRP Microstructure 6
B. Definition of Material Properties 7
IV. Optimization Studies with OnScale 8
A.  A-, B-, and C-Scans 8
B. Convergence Studies to Optimize Mesh Requirements 10
C. Material Property Sensitivity Studies 12
V. Conclusions 16
VI. Resources 17

© 2020 OnScale, Inc.

info@onscale.com www.onscale.com 2
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

I. INTRODUCTION: ULTRASONIC NON-DESTRUCTIVE TESTING SIMULATION FOR


COMPOSITE MATERIALS
A key consideration for materials used in the aerospace industry is to ensure the strength of aircraft
components, while limiting weight. For this reason, aerospace design challenges have driven major
advances in composite materials, which offer both increased mechanical strength and weight savings.
This is epitomized by modern aircraft such as the Boeing 787, which contains 50% composite material
by weight1. In addition to their favorable strength-to-weight ratio, composites are also quickly becoming
the material of choice in aerospace applications due to their durability, resistance to corrosion and
fatigue, and the potential they offer for new design solutions2.

The increased use of composite materials has resulted in pressure to produce new, reliable and
robust non-destructive testing (NDT) inspection techniques. NDT inspections are a crucial part of
the safety process, specifically to assess component structural integrity and to detect any material
discontinuities or defects. Components are often inspected during and after manufacture to ensure
they are acceptable to put into service, and then are routinely inspected once in service. Strict NDT
requirements are in place for high-value components found in the aerospace industry to ensure the
safety of aircraft. The introduction of complex composite materials has resulted in traditional NDT
inspections becoming insufficient due to the inspection challenges of these materials. It is therefore
crucial that NDT inspections advance in parallel with the design of composite components. The use of
simulation tools can drastically reduce both the time and money spent in development and validation
of these much needed new NDT protocols.

Composite materials are challenging to inspect due to their complex structure and anisotropic material
properties. Ultrasonic inspection techniques are commonly implemented because they have the
ability to detect defects and characterize internal structure. However, the development of ultrasonic
inspection methods for composite materials requires a numerical approach in order to characterize the
complete ultrasonic wave behavior and the spatial variation of elastic properties. These simulations
can prove challenging due to the high computational demand for capturing highly detailed component
structures. In addition, modeling ultrasonic NDT inspections typically requires multiple simulations to
recreate scanning methods used in practice.

The computational power offered by OnScale empowers engineers with a simulation platform that
can analyze ultrasonic inspections of complex composite materials in a practical time frame3. OnScale
uses a combination of highly efficient multiphysics solvers with scalable high-performance computing
(HPC) resources, removing the computational barriers common with legacy tools. In addition, the
ability to run individual simulations across multiple cloud nodes enables the recreation of complete
ultrasonic scans in a reduced time frame. The use of an integrated multiphysics approach can provide
new information on the interplay between different effects which would be unavailable when using
simpler, single physics models. This can lead to a reduction in the overall cost of the NDT process
and the development of inspections with a more reliable defect-detection capability to help prevent
catastrophic failures that can lead to fatal consequences.

info@onscale.com www.onscale.com 3
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

II. CHALLENGES IN SIMULATING ULTRASONIC INSPECTIONS OF COMPOSITE COMPONENTS


To make a composite material, two or more materials with different, usually complementary prop-
erties are combined to create a new material with desired properties. For example, carbon fiber
reinforced polymer (CFRP) materials, which are commonly used in aerospace applications, consist of
carbon filaments in an epoxy resin matrix. CFRP laminates can be constructed by layering sheets of
unidirectional carbon fibers on top of one another and adding the epoxy resin. The orientations of the
fibers can be maintained from layer to layer to create a unidirectional laminate, shown in Figure 1a,
which is extremely strong and stiff in the direction of the fiber orientation. Alternatively, the layers
can be stacked at various angles, normally 0˚, 45˚, 90˚ or -45˚, to produce cross-ply, Figure 1b, or qua-
si-isotropic laminates, Figure 1c. Through variation of layer orientation, the strength of the laminate
as a function of load direction can be tailored to provide optimum strength and stiffness 4.

0° 0° 0°

0° 90° 90°

0° 0° +45°

0° 90° -45°

0° 0° -45°

0° 90° +45°

0° 0° 90°

0° 90° 0°

90° 90° 90°


0° 0° 0°

(a) (b) (c)

Figure 1. Examples of CFRP laminate structures: (a) unidirectional, (b) cross-ply and (c) quasi-
isotropic laminate lay-up.

The complex microstructure of CFRP components creates challenges for simulations of ultrasonic
inspections. Typically, the interply resin layer is 5–10 µm thick, so a high-resolution model is required
to capture full geometrical detail. This leads to increased computational demands in terms of both
memory and solve time since such a thin layer needs to be properly discretized with a very fine mesh.

In addition to a complex microstructure, CFRP components have anisotropic and inhomogeneous


material properties. These properties, for example the material elastic modulus, are highly dependent
on fiber orientation, so accurately recreating them in a detailed three-dimensional simulation can be
challenging. For example, the anisotropic material properties of a CFRP with unidirectional plies (see
Figure 1a) can create a large variation in acoustic velocity as a function of direction. This is shown in
Figures 2 and 3 for longitudinal and shear wave velocities.

info@onscale.com www.onscale.com 4
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

10,000

8,000

m/s
[0 0 1]

6,000

4,000

[0 1 0 ] [1 0 0 ]

Figure 2. Longitudinal velocity map for a CFRP with unidirectional ply layers with fibers in the [1 0 0] direction.

2,700 1,800

2,500 1,750
[0 0 1]

[0 0 1]
m/s

m/s
2,300 1,700

2,100 1,650

1,900
[0 1 0] [0 1 0] 1,600
[1 0 0] [1 0 0]

(a) Shear vertical velocity (m/s) (b) Shear horizontal velocity (m/s)

Figure 3. Shear velocity maps for a CFRP with unidirectional ply layers with fibers in the [1 0 0] direction.

In the longitudinal velocity map, note that the acoustic velocity is strongly dependent on the fiber
orientation. In fact, in the fiber direction, the material is very stiff and so has a much larger acoustic
velocity compared to the direction perpendicular to the fibers. In each velocity map, there is a large
deviation in acoustic velocity depending on direction relative to the fiber orientation. This will cause
acoustic waves propagating in the material to behave differently depending on their incident angle. It is
essential that these types of material properties are correctly implemented in a simulation, as they can
have a profound effect on the propagation of acoustic waves through the structure.

info@onscale.com www.onscale.com 5
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

III. BUILDING MULTIPHYSICS COMPOSITE INSPECTION MODELS WITH ONSCALE


The inspection of CFRP laminates is often performed using pulse-echo techniques, where a single
element probe or array is used to send and receive an acoustic wave that propagates through the
sample. This approach is very sensitive to common defects found in laminates, such as delamination,
disbonding, and porosity, because they are in the plane of the sample. Furthermore, the generation of
ultrasonic scan datasets can offer additional characterization of the internal composite structure.

A. High-resolution geometrical modeling and boundary conditions for a CFRP microstructure


A simple simulation of a single element pulse-echo inspection of a CFRP laminate was constructed,
as shown in Figure 4. The model consists of a cross-ply CFRP laminate, containing 8 ply layers that
alternate orientation between 0˚ and 90˚, represented by red and green in Figure 4. Each ply layer
has a thickness of 250 µm and the interply resin layers are 10 µm thick. The interply resin layers are
included in the model but are not visible in Figure 4 because they are much thinner than the plies.

Ply0
Ply90
watr
Resin
tx

Figure 4. OnScale simulation: The geometry of a model for immersion inspection of a CFRP laminate.

The goal of this simulation is to model a simple pulse-echo immersion inspection, with a pressure load
representing the transducer (yellow), which is placed in a water medium (cyan). A 5 MHz wavelet signal
is applied through the pressure load across the full model width. Symmetry boundary conditions are
applied to either side of the computational domain to simulate an infinite plane wave. The average
pressure at the load source was calculated during the simulation to generate an A-scan output.
Absorbing boundary conditions were applied to the top and bottom of the domain to minimize the
computation zone without creating unwanted reflections from the boundaries.

info@onscale.com www.onscale.com 6
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

B. Definition of material properties

To implement the cross-ply CFRP model shown, the material properties of IM7/8552 were used.
IM7/8552 comprises of HexTow IM7 carbon filaments bound in a HexPly 8552 resin, and is com-
monly found in aerospace applications5. IM7 fibers possess a high tensile strength and modulus6,
while the resin forms a high-performance tough epoxy matrix7. The elastic properties used in the
simulations are presented in Table 1, and represent a single ply of unidirectional IM7/8552.

Property Value
Density (kg/m3) 1570
Elastic Modulus (GPa) E11 161
E22 11.38
E33 11.38
Shear Modulus (GPa) G12 5.17
G23 3.98
G13 5.17
Poisson's Ratio v12 0.32
v23 0.436
v13 0.32

Table 1. Elastic material properties of IM7/8552 with E11 in the fiber direction.

The elastic material properties are used to calculate the compliance matrix [8], or S, as shown below,
where ν is Poisson’s ratio and E and G are the Young’s and shear moduli, respectively. The values
from the compliance matrix are used in the simulation to define the linear elastic anisotropic material
properties of a single CFRP ply layer material.

1 v 21 v 31 0 0 0
E 11 E 22 E 33
v 12 1 v 32
E 11 E 22 E 33
0 0 0
v 13 v 23 1
0 0 0
S= E 11 E 22 E 33
1
0 0 0 G23 0 0
1
0 0 0 0 G31
0
1
0 0 0 0 0 G12

The material properties of the interply resin layers used in the simulations are shown in Table 2. The
Young’s Modulus and Poisson’s ratio are used to calculate the bulk and shear moduli K and G. The
resin material can then be defined as an isotropic linear elastic material using density, bulk modulus,
and shear modulus as inputs.

info@onscale.com www.onscale.com 7
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

Property Value
Density (kg/m3) 1301
Young's Modulus (GPa) 4.57
Poisson's Ratio 0.37

Table 2. HexPly 8552 resin material properties.

Once the geometry of the CFRP model has been defined, the material properties can be transformed
to a local coordinate system to align them with respect to the fiber orientation. The material properties
shown in Table 1 have been defined for E11 in the fiber direction, which is in our case the X axis. This
makes the out-of-plane direction the Y axis, and the in-plane direction the Z axis, as shown in Figure 5.
Unidirectional ply layers are typically defined by their in-plane fiber orientation angle, and the material
properties can be transformed to these orientations through a rotation about the Y axis. Further
transformation of material properties can be implemented about the Z axis to account for out-of-plane
orientation variances due to component geometry or ply waviness.

Figure 5. The local coordinate system relative to fiber orientation and material properties definition.

IV. OPTIMIZATION STUDIES WITH ONSCALE

A. A-, B-, and C-scans


Ultrasonic NDT data can be collected and displayed in numerous formats. Each approach provides a
different way to visualize and evaluate the test medium. Ultimately, the goal of an ultrasonic inspection
is to determine whether the test component contains any structural defects that will be detrimental
to its strength.

The simplest method to record and present ultrasonic data is in the form of an A-scan, which displays
the received signal amplitude as a function of time. This can be used to locate features or defects by
calculating the time of flight, since depth and time are correlated for this analysis. The size of features
can also be evaluated through the amplitude of its reflection, with a direct correlation between feature
size and reflection amplitude. An example of a simulated A-scan for a CFRP laminate inspection is
presented in Figure 6, where the scan is truncated to display the reflections from the front and backwall
of the component. These simulated ultrasonic measurements can be generated in a mechanical model
with OnScale by calculating the average pressure across the receiver location.

info@onscale.com www.onscale.com 8
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

Normalised Amplitude
0.5

-0.5

2 3 4 5 6 7 8
Time µS

Figure 6. OnScale simulated A-scan signal from an ultrasonic inspection of a CFRP laminate
displaying the frontwall reflection at 3µs, backwall reflection at 7.5 µs, and multiple internal
structure reflections in between.

To perform a B-scan inspection, multiple A-scan signals are collected along a line. To display the results
of a B-scan, the A-scan signals are typically combined, with time in the vertical axis and transducer
location in the horizontal axis. The B-scan method allows for a larger area of the inspected medium to
be clearly visualized. An example of a B-scan image is shown in Figure 7, for a CFRP laminate containing
some delamination defects. By simulating each transducer location in parallel, OnScale can generate all
the individual A-scan data simultaneously.

Figure 7. OnScale simulated B-scan data of an ultrasonic inspection of a CFRP laminate containing
delamination defects. The laminate frontwall and backwall are located at 3 and 7.5 µs, while delamination
defects are located at ~6.5 µs and centered at horizontal distances of 4, 9, 13, and 15 mm.

info@onscale.com www.onscale.com 9
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

An ultrasonic C-scan uses the same approach as the B-scan except individual A-scans are collected
along two axes. For display, the plane of a C-scan image is parallel to the transducer scanning plane, and
the value plotted is the signal amplitude at a single time or depth in the test medium. Figure 8 provides
an example simulated C-scan image for the same CFRP laminate containing defects. The depth in
this image is the depth at which the defects are located, to allow their footprint to be visualized. As
with B-scans, a C-scan inspection can easily be simulated in OnScale through parallel simulation of all
transducer locations.

Figure 8: OnScale simulated C-scan data of a CFRP laminate containing delamination defects. The
image is presented at the depth of the defects to illustrate their footprint.

B. Convergence studies to optimize mesh requirements


An important consideration for any numerical simulation is the mesh resolution required to properly
discretize the underlying physics equations being solved. Mesh resolution has a direct impact on both
result accuracy and the computational requirements of the simulation. Mesh convergence studies
should be utilized to investigate the performance of various mesh resolutions in terms of desired
accuracy and computational efficiency. This is achieved by varying the mesh resolution and monitoring
an output of interest. As the mesh density increases, the output will eventually become relatively
constant, and at this point it can be said the model has adequately converged to a solution. Reviewing
the deviation in the output from the converged solution, in combination with the computational solve
time, allows for a suitable mesh resolution to be selected.

An OnScale model was constructed in which mesh nodes were positioned on the interface between the
carbon fiber and the resin to ensure that the CFRP laminate layers are positioned and sized correctly.
This is important because it ensures that the model geometry is an accurate representation of the test
component and is suitably meshed for accuracy and efficiency. The mesh density can be defined using
a set number of elements per acoustic wavelength to accurately capture the wave modes of interest.
The mesh can also be adjusted by setting a minimum number of elements through a layer thickness.

info@onscale.com www.onscale.com 10
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

A mesh convergence study for the simulation model was constructed in OnScale considering two input
parameters. The first was used to represent the number of elements per wavelength. This parameter
was swept from 15 to 50 in steps of 5, using the wavelength of the shear wave in the resin material
at the center frequency of the input drive signal. The resin shear wave is used since it is the shortest
wavelength that will be generated in the model. The second input parameter implemented a minimum
number of elements through each material thickness. This parameter was swept from 1 to 4 in steps
of 1. Figure 9 provides images of mesh setups generated in the convergence study.

(a) 15 elements per wavelength and 1 minimum element (b) 15 elements per wavelength and 4 minimum element
per layer thickness per layer thickness

(c) 50 elements per wavelength and 1 minimum element (d) 50 elements per wavelength and 4 minimum element
per layer thickness per layer thickness

Figure 9. Mesh structures for the mesh convergence study, showing an interply resin layer (yellow)
between two ply layers (green, blue).

To evaluate mesh convergence, the backwall amplitude from each simulation was extracted and used
as the output for comparison. The backwall signal is selected since it is an important feature in the
derived A-scan signals and will allow for the meshing of the test component to be accurately evaluated.
The backwall amplitudes measured from each received signal were normalized against the maximum
backwall signal received, which is assumed to be the converged solution. To provide a comparison, the
amount of deviation from this converged solution was calculated for each signal. The results from this
convergence study are shown in in Figure 10.

(a) Parametric surface plot (b) Minimum elements against elements per wavelength (c) Minimum elements against percentage error (d) Elements per wavelength against percentage error

(b) Minimum elements against (c) Minimum elements against (d) Elements per wavelength
(a) Parametric surface plot
elements per wavelength percentage error against percentage error

Figure 10. OnScale mesh convergence study results for backwall reflection amplitude of CFRP laminate.

info@onscale.com www.onscale.com 11
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

As the number of elements per wavelength increases, the result converges as expected, which is clearly
shown in Figure 10d. At 20 elements per wavelength, the amplitude of the backwall is within 1.5% of
the converged result. The number of elements per wavelength determines the density of the mesh and
so will have a large impact on the simulation computation requirements. As model size increases, the
memory requirement and solve time also increase, meaning that the number of elements implemented in
the model should be selected to provide the required accuracy while also having suitable computational
demands.

In comparison to the number of elements per wavelength, the minimum number of elements per layer
thickness is shown to not have a significant impact on the derived results. This is illustrated by how
closely the curves overlap in Figure 10d. This is a key finding, as it indicates that implementing a single
element through the resin thickness for this modelling scenario is an accurate representation. This
reduces the number of elements required for the simulation and will therefore result in a faster solving
time with reduced memory requirements.

C. Material property sensitivity studies


A benefit of the OnScale simulation platform is the ability to quickly perform sensitivity studies, in
other words, to investigate how individual parameters such as material properties impact the simulated
results. When a simulation is correctly set up, incorrectly assumed material properties are often the
main source of inaccuracy. The material properties used in simulations typically come from experimental
measurements. However, the accuracy of these measurements is often unknown, and the material
properties can vary between different material batches and specimens. Performing material property
sensitivity studies in OnScale, where the massive parameter sweeps required take only a few minutes,
allows users to investigate how changes in material properties can affect inspection performance, as
well as give information on the uncertainty in simulation results based on expectations of sample-to-
sample material variability.

Using the same ultrasonic pulse-echo simulation, a material property sensitivity study was performed
to investigate the impact of the CFRP ply and resin layer densities. This was done by varying the
nominal density values by ±10% with a step size of 1% to give a total of 441 simulations. Each
simulation generates a received A-scan signal, which can be used to calculate the time of flight by
measuring the time between the frontwall and backwall reflections. Variations in material density will
result in changes to the acoustic velocity, which will in turn introduce variability in the time that signal
reflections are received. Based on the previous convergence study, the mesh was set to generate 20
elements per wavelength for the shear velocity of the resin materials, with a minimum of one element
through a layer thickness.

To visualize the results from the sensitivity study, Figure 11 shows the error in the calculated backwall
time of flight as a function of the assumed material densities. The error is calculated as the difference
from the time of flight using the nominal density values as reported in Tables 1 and Table 2, and the time
of flight calculated by assuming a given set of densities.

info@onscale.com www.onscale.com 12
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

Figure 11. OnScale calculated time of flight error resulting from variation in CFRP ply and resin densities.

From Figure 11, it can be seen that the ply density dominates the time of flight variation, which is
expected as this material makes up 96.5% of the sample thickness. For both ply and resin materials,
there is a direct correlation between changes in density and resultant time of flight, due to the inverse
relationship between material density and acoustic velocity.

Variation in ply and resin layer densities will also affect the transmission and reflection coefficients
at the laminate frontwall and backwall, as well as each ply and resin layer interface. This will cause
variation in the amplitude of the wave that is transmitted versus reflected at each interface. The
amplitude of the reflection from the frontwall, resin layers, and backwall were extracted from each
A-scan and used to calculate an amplitude error, which is shown in Figure 12. The amplitude of the
reflections was taken as the maximum instantaneous amplitude of the reflected signal. As with the
time of flight analysis, the error was calculated by comparison with the result obtained when using the
nominal density values.

(a) Frontwall reflected amplitude (b) Resin layer reflected amplitude (b) Backwall reflected amplitude

Figure 12. OnScale calculated signal reflection amplitude error resulting from variation in CFRP ply
and resin densities.

info@onscale.com www.onscale.com 13
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

In this case, changes in density lead to a variation in material impedance and, as a result, variation
in the acoustic wave transmission and reflection coefficients. On a macroscopic scale, this has the
effect of changing the attenuation of the wave propagating through the structure, due to the varying
levels of reflection at the internal material interfaces. The frontwall and backwall amplitudes have a
maximum error between 2 and 3%, while the reflection from the resin layers have a maximum error of
greater than 15%.

A second sensitivity study was performed using OnScale to investigate the out-of-plane elastic
modulus, E22, of the CFRP ply layers and the Young’s Modulus, E, of the resin layers. The out-of-
plane elastic modulus of the CFRP ply layer was selected because of its relatively large impact in the
current inspection geometry. In this geometry, the acoustic wave propagates perpendicular to the
plies, so changes to the acoustic velocity in the out-of-plane direction relative to the plies will have a
larger impact than changes in the in-plane direction. Again, the material property values were varied
by ±10% in 1% increments to generate a total of 441 simulations. The resulting errors in time of flight
and reflected signal amplitudes are shown in Figure 13.

(a) Time of Flight (b) Frontwall reflected amplitude

(c) Resin layer reflected amplitude (d) Backwall reflected amplitude

Figure 13. OnScale calculated time of flight and reflected signal amplitude error due to variation in
the out-of-plane elastic modulus of the CFRP plies and the Young’s modulus of the resin layer.

info@onscale.com www.onscale.com 14
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

In this example, the time of flight error is again dominated by the material properties of the ply, because
it makes up most of the component thickness. For scenarios where the resin layer reflection amplitude
is increased, the signal becomes more attenuated as it travels through the component, resulting in a
lower backwall reflection amplitude. Overall, in this sensitivity study, the resin layer amplitude is the
most sensitive of the calculated outputs to changes in material elastic properties, with an error of up
to 30%.

Both material sensitivity studies demonstrate the capability of OnScale to investigate the effect of
variance in material properties on the derived results. The material properties of a manufactured
composite component are often affected by many factors, including the age of the composite material,
the curing conditions, and the level of compaction [9]. Therefore, the experimentally measured
properties of a material may not be identical to the properties of that material in an as-manufactured
component and may vary from part to part. The sensitivity studies shown here demonstrate that
changes in material properties can lead to errors in the calculated time of flight. Among the simulation
outputs investigated here, the largest variance in the derived results was found in the amplitude of
reflected signals, resulting from variation in the transverse stiffness of the CFRP ply layers, and the
Young’s modulus of the resin. A change of 10% in these material properties can lead to up to 30%
error in the amplitude of the reflection from the resin layer. Because these errors can be significant,
it is important to understand and quantify the effect of variation in the material properties on both
simulation and test data. This also reinforces the importance of using accurate material properties as
inputs to simulations.

The OnScale model used for the material property sensitivity studies contained a total of 669,000
degrees of freedom. Simulations were executed in parallel with each simulation using 2 cores. Individual
simulations required 30 MB of RAM and were solved in only 82 seconds, as shown in Table 3. The
ability to run large sweeps in parallel through OnScale allows full design spaces to be solved in the
same time it takes to solve a single simulation. All 881 simulations for both material property sensitivity
studies were solved in parallel using OnScale in under 2 minutes. Performing the same studies using
legacy simulation would take over 20 hours.

RAM Requirements (MB) 30


Solution Time (Seconds) 82

Table 3. OnScale solver performance for the material property sensitivity study comprising 881 simulations.

info@onscale.com www.onscale.com 15
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

V. CONCLUSIONS
Composite materials can be used to ensure the strength of mission-critical components, while
limiting their weight. Due to the complexity of these materials, their use presents engineers with
extraordinary design and manufacturing challenges. NDT techniques are of critical importance when
evaluating components against the stringent quality requirements applied to composite materials, so
engineers need to be equipped with best-in-class modeling and simulation tools to properly simulate
and enhance these techniques. Legacy CAE tools are not suited to meet the demanding requirements
for complete 3D simulation studies, both in terms of memory and solution time. As full 3D simulations
are required to properly understand and optimize ultrasonic evaluation techniques, there is a need
for more efficient tools, such as OnScale, which can harness the computational power and scalability
offered by HPC in the cloud. This white paper provides an overview of how different NDT techniques
can be modeled and simulated, highlighting the need for modern CAE tools that enable an efficient
exploration of all variables involved. With the ability to efficiently and quickly solve time domain
problems, the multiphysics solvers provided with OnScale enable design engineers to run fast and
accurate numerical studies to optimize ultrasonic inspection techniques. This is a powerful tool for
meeting the high quality requirements imposed by the design and manufacturing of critical composite
material components for the aerospace industry.

AUTHOR

Jeff Dobson, EngD


Senior Application Engineer
OnScale

info@onscale.com www.onscale.com 16
Simulation of Ultrasonic Non-Destructive Testing of Composite Materials

VI. RESOURCES

Technical papers
1. Marsh, George. "Boeing's 787: trials, tribulations, and restoring the dream." Reinforced Plastics 53.8 (2009): 16-21.
2. Kollgaard, J. R. and S. G. LaRiviere. "NDE challenges with future commercial aircraft–a Boeing perspective." AIP
Conference Proceedings. Vol. 975. No. 1. AIP, 2008.
3. Dobson, J. An Integrated Framework for Finite Element Modelling of Ultrasonic Inspections of Carbon Fibre
Reinforced Polymer Components, EngD Thesis, University of Strathclyde, 2019.
4. Campbell, Flake C. Structural composite materials. ASM international, 2010.
5. Leckey, Cara A.C., Matthew D. Rogge, and F. Raymond Parker. "Guided waves in anisotropic and quasi-isotropic
aerospace composites: Three-dimensional simulation and experiment." Ultrasonics 54.1 (2014): 385-394.
6. “Hexcel hextow im7 eu data sheet.” Hexcel, http://www:hexcel:com/user_area/content_media/raw/IM7_HexTow_
DataSheet:pdf. Last Accessed: 12-November-2017.
7. “Hexcel hexply 8552 eu data sheet.” Hexcel, http://www:hexcel:com/user_area/content_media/raw/HexPly_8552_
eu_DataSheet:pdf. Last Accessed: 12-November-2017.
8. Boresi, Arthur Peter, Richard Joseph Schmidt, and Omar M. Sidebottom. Advanced mechanics of materials. Vol. 6.
New York et al.: Wiley, 1985.
9. Leckey, Cara A.C. et al. "Simulation of guided-wave ultrasound propagation in composite laminates: Benchmark
comparisons of numerical codes and experiment." Ultrasonics 84 (2018): 187-200.

White papers
From ultrasonic fingerprint design to HPC in the cloud, this collection of white papers provides you with the latest
information on cutting-edge technology and simulation tools:
https://onscale.com/whitepapers/
Webinars
See how fellow design engineers and simulation specialists work together to create winning designs:
https://onscale.com/webinars/
Simulation guides
Learn how to simulate complex Carbon Fiber Reinforced Polymer Composites (CFRP) in full 3D with OnScale simulation
guides:
https://onscale.com/simulation-guides/ndt-non-destructive-testing/

Videos
Learn how to model your application in OnScale with this collection of tutorial videos:
https://onscale.com/videos/
OnScale Blog
Read the OnScale Blog to stay up to date with the latest in high-performance modeling and simulation technology:
https://onscale.com/blog/
Contact information
Do you want to know if OnScale can help with your design needs? From requesting help with the software to providing
feedback or even upgrading your free account, we would be delighted to hear from you:
https://onscale.com/contact-us/

info@onscale.com www.onscale.com 17

You might also like