Final Report Summary - DOTNAC (Development and Optimization of THz NDT on Aeronautics Composite Multi-layered Structure)
Executive Summary:
The availability of light and robust structures has led to an increased use of composite materials in the aircraft industry. In order to verify and guarantee the high quality of the conventional and new composite elements, innovative approaches for non-destructive testing of these parts are required.
The European research project “DOTNAC” proposes to develop a fast, high resolution, noninvasive and non-contact inspection system for assessing aeronautic composite parts during production using terahertz waves. Conventionally two categories of systems can be discussed: pulsed and continuous wave terahertz systems. Both have been realized and their respective potential as a non-destructive inspection tool has been evaluated.
The implemented FMCW (Frequency-Modulated Continuous-Wave) system is an all-electronic THz system. It consists of three scanning heads with different frequency ranges (around 100 GHz, 150 GHz, and 300 GHz). They are used to acquire data by scanning a sample placed in front of them in reflection mode. Combining the measured in-depth information with a lateral scanning, a 3-D image can be built up. The in-depth resolution depends on the used bandwidth and the roughness of the sample surface, and varies between 2 mm and 6 mm. For the construction of the TDS (Time-Domain System), a pulsed laser system has been implemented in a fibre-optical ECOPS (Electronically Controlled Optical Sampling) pump-probe set-up. The two short-pulse lasers (one for the emitter, one for the detector) are based on Erbium-doped silica glass fibres and emit around 1560 nm centre wavelength. The in-depth resolution is significantly better than the one obtained with the FMCW THz system at the expense of a lower penetration capacity for the materials and structures tested within the scope of the project. The across-range resolution depends again on the beam focus and is comparable to the one of the FMCW system using the focused configuration. The main trade-off that needs to be considered for the TDS is the one involving measurement speed and signal-to-noise ratio.
To achieve these objectives, a series of 20 flat glass fibre reinforced plastics (GFRP) samples (solid laminates and sandwich structures), and 5 carbon fibre reinforced plastics (CFRP) were modified by artificial defects such as inserts, stucks, water inclusions, etc., to create well-controlled and well-known samples to validate the THz systems and algorithms with. In a next step the evaluation as a THz NDT method has been performed on 50 blind samples (12 GFRP and 38 CFRP samples) with embedded defects caused by an intentional miss-process. For validation and evaluation respectively, both THz systems under test have been installed on a two-dimensional scanner for sample inspection. A proof of concept has been provided using a three-dimensional motion platform to demonstrate the feasibility of THz inspection on a real aircraft object.
For the full assessment of the THz NDT performances, all the above-mentioned samples have been inspected by the following conventional techniques: thermographic NDT, radiographic testing (film and digital radiography and computed tomographic radiography testing), and ultrasound testing (2-D Inverse Wave Field Extrapolation, Phase Array, Pulse Echo, and Through Transmission). The best performances per inspection method and per sample/defect type have been identified to demonstrate complementarity between the respective NDT methods or their individual performance.
Regarding the feasibility of a THz application in aeronautics, i.e. inspection of dielectric parts, the THz
technology has proved to be a valuable technique for NDT. The technique is able to detect:
1. Delaminations and foreign inclusions in dielectric laminates such as glass fiber laminates.
2. Delaminations and disbonds in dielectric sandwich structures such as A-sandwich or C-sandwich structures with either honeycomb or syntactic foam cores.
In addition the technique is very sensitive to coating misprocess on conductive substrate such as CFRP
and probably to the same extent on dielectric substrates such as glass fiber. On the other hand THz failed to detect porosity in glass fiber laminates. The capability of THz NDT for porosity detection should be further investigated.
In comparison with the classical NDT techniques, NDT-THz looks very competitive. It does not require
high and costly radiation protection such as for X-Ray radiography and is totally non-contact in contrast
with ultrasound.
To conclude, the performed assessment shows that FMCW and TDS can be considered as complementing techniques that should be tightly associated. The same conclusion can be formulated here, i.e. a combination of the two techniques investigated in this project, FMCW and TDS, seems to be the ideal way of applying THz in NDT. FMCW can provide fast scanning at high detection sensitivity. TDS can complete the data at the critical area by providing very accurate measures of defect sizes and depth at the expense of testing time.
Project Context and Objectives:
In recent years, the development of THz systems has made a considerable progress and made possible a variety of innovative applications. Among them, Non-Destructive Testing (NDT) of composite materials can take advantage of the high transparency of certain non-conductive materials at THz frequencies. The presence of such materials is foreseen to increase in the next years, especially in aeronautics. Glass-fiber (GF) laminates or composite structures are currently being used thanks to their outstanding structural properties, durability and lightness. In response to the growing industrial interest on these materials, the European research project “DOTNAC” proposes to develop a fast, high resolution, noninvasive and non-contact inspection system for assessing aeronautic composite parts during production using terahertz waves.
Through the realization of an advanced in-process inspection and quality control during the production phase and at the same time an advanced technique for continuous health and usage monitoring of structures and systems, the DOTNAC project opens possibilities to reduce aircraft development, production and operational costs. Indeed, by detecting defects in composite material as early as possible in the production chain, the scrapping or rework of a component can be avoided, contributing immensely to the cost of part production. At the same time, the same non-invasive, non-contact THz inspection technique can be used for assessing composite part condition in service. This type of inspection can effectively and efficiently support the high standards of composite construction and repair, thereby reducing the aircraft operational cost.
The main goal of the DOTNAC project is to develop a fast, high resolution, non-invasive and non-contact inspection system for assessing aeronautic composite parts either during production or maintenance. The developed NDT tool will be easy to integrate in industrial facilities and will fill in the performance gaps that are still present amongst the existing NDT techniques. It will therefore be an extremely useful tool in NDT in terms of sensor fusion. This new method will not replace directly the existing NDT tools, but will deliver complementary results which can be sometimes more precise for some defects.
Achieving this entails the following specific objectives:
• To create an integrated (hardware-software) and optimized THz imaging system using pulsed signals and optical fibre coupling.
• To create an integrated (hardware-software) and optimized THz imaging system using continuous wave signals and electrical cable coupling.
• To demonstrate, in an industrial setting, the effectiveness of a THz NDT tool.
• To assess the performances of the 2 developed THz NDT tools for assessing aeronautic composite parts.
• To develop a user/research community for fast, high resolution, non-invasive & non-contact inspection for assessing aeronautic composite parts during production.
Project Results:
S&T results can be described at following levels:
- With respect to the hardware:
o FMCW THz system: system was acquired prior to DOTNAC but optimized for the application as described in DOTNAC
o TDS THz system: system has been developed from scratch within the frame of DOTNAC
o Motion platform: system has been developed for a tailored used within the frame of DOTNAC
- With respect to the developed software:
o Data fusion software for the focused FMCW data
o Synthetic aperture processing for the wide aperture FMCW data
o Tomographic processing for the multi-angle TDS data
- With respect to the tested NDT potential:
o Creation of a full set of samples with defects representative for aeronautics applications
o Exploitation of the results obtained with conventional NDT techniques and the THz tools
S&T results related to hardware
Originally the FMCW THz system was designed and fabricated by SynViewn during DOTNAC the system has been modified and optimized by Fh-IPM to fulfill the objectives defined for DOTNAC.
The system is 2.5 m high, 1.5 m long 1.2 m wide having a total weight of 300 kg. On its side there is the data acquisition unit, the computer and power alimentation together in a tower. The maximum sample size is limited to 700x700 mm in surface and the maximum scan range is 600x689 mm. The dimensions of the optics limit also the maximum sample thickness to 90 mm for the optic with a focus of 50 mm and to 380 mm for the optic with a focus of 200 mm.
As mentioned in the DoW the FMCW system consists of three scanning heads with different center frequencies (100 GHz, 300 GHz and 850 GHz). After the first tests, the 850 GHz scanning head was replaced by 150 GHz due to poor results. The materials used for the DOTNAC samples made this change necessary. The scanning heads were used to acquire data by scanning a sample placed in front of them in reflection mode (see Figure 1).
Figure 1- Photographs of the FMCW system. Left: side view with housing and control unit. Right: view from the front side showing the three transceivers and three receivers in transmission assembly. The z-direction between the transceivers and receivers is fix
100 GHz 150 GHz 300 GHz 850 GHz
Frequency range 0.07 THz – 0.11 THz 0.11 THz – 0.17 THz 0.23 THz – 0.32 THz 0.84 THz – 0.87 THz
Bandwidth 40 GHz 60 GHz 90 GHz
Dynamic range
(acquisition times are per pixel) > 40 dB
(at acquisition time 100 μs)
> 60 dB
(at acquisition time 10 ms)
> 70 dB
(at acquisition time 100 ms) > 40 dB
(at acquisition time 100 μs)
> 60 dB
(at acquisition time 10 ms)
> 70 dB
(at acquisition time 100 ms) > 35 dB
(at acquisition time 100 μs)
> 50 dB
(at acquisition time 10 ms)
> 60 dB
(at acquisition time 100 ms) > 25 dB
(at acquisition time 100 μs)
> 40 dB /
(at acquisition time 100 μs)
> 50 dB
(at acquisition time 100 μs)
Spatial resolution (laterally) 3 mm 2 mm 1 mm 0.4 mm
Depth resolution in air 9 mm 6 mm 3 mm 10 mm
Depth resolution in typical materials 6 mm 4 mm 2 mm 15 mm
Table 1 - System specifications of the FMCW THz system.
Fh-IPM has done measurements in order to check the manufacturer’s specifications. Three metal bars were installed as a sample and measured in transmission and reflection using different frequencies. The dynamic range was determined for two different focusing optics (50 and 200 mm). As a result one can confirm the information provided by the manu¬facturer (see Table 1).
Fh-IPM has developed a visualization software: An image-editing software for presentation of samples measured using FMCW is available. All levels in X, Y and Z direction are scaled as the signal-noise ratio. Different filters can be used for better representation. 2D and 3D representations are possible. It is possible to set and to display different cut plane. A presentation in color as well as in grey tones is possible.
Figure 2- Left: SVD view toolbox. Right: 3D illustration of a test pattern.
Setup with measuring head, lens and sample Setup with measuring head, prism, parabolic mirrors and sample
A-Sandwich with rohacell, 100 GHz, front A-Sandwich with rohacell, 100 GHz, front
Figure 3 - Testing of different optics to evaluate the trade-off azimuth resolution versus in-depth information
The focused set-up uses a set of lenses creating a focused beam with a beam waist characterized by the Rayleigh length. Depending on the type of optics, a trade-off can be made between azimuth resolution and in-depth information. For that reason different sets of optics have been test out and an assessment on the image quality has been made.
To complete the list of configurations, a wide angle set-up was tested out (omitting the lenses all together) and a specific calibration procedure for this measurement configuration has been created. The processing of the acquired data is discussed in the software section hereafter.
The complete FMCW system has been mounted on a 3D scanning system (see description hereafter), developed by Verhaert. After a fully integrated and optimized hard- and software, the developed THz NDT tool has been used to image the initially defined composite materials and structures.
As a complementary system to the FMCW THz system the TD THz system has been developed from scratch within the project. The TD THz system is equipped with a twin femtosecond laser configuration (with approximately 1560 nm centre wavelength), in which the repetition rate of one of the lasers can be either synchronized to the other or deliberately modified to induce a fixed or variable time delay between the laser pulses. Both lasers are terminated with an external fibre delivery of more than 6 m length which is internally pre-compensated with respect to dispersion and non-linear pulse broadening. The time window that can be captured in ECOPS mode at 100 Hz scan rate is 146 ps. The system itself would allow operation at higher frequencies (500 Hz was tested), but the integrated data acquisition architecture so far only runs at 10 Hz. This is, however, not a fundamental limit. The lasers, their control electronics, further electronics for generation, read-out and amplification of the THz pulses (voltage source, waveform generator, voltage amplifier) are placed in a mobile cubic rack with less than 100 kg total weight that serves as basic supply unit for the TD system to drive the remote sensor and feed the data to the DAQ card and software, which are installed on a dedicated PC. The system requires only access to a standard European 230 V AC power socket.
The sensor heads for emitter and detector contain photoconductive antennas that are permanently coupled to fibre patchcords for connecting them rigidly with the pump lasers. THz radiation is coupled out and in, respectively, by highly resistive silicon lenses. For THz beam shaping, each head includes an internal collimating and an external focusing lens (material: TPX). To match the working distance of the 3D scanner unit, 200 mm is chosen as focal length. The THz optical elements are aligned and rigidly positioned by means of an opto-mechanic cage system. The sensor heads are mounted on a common baseplate for reflection measurements under an angle of incidence to the sample normal of 10°. This baseplate also incorporates a transimpedance amplifier for pre-amplification of the signal close to the detector. The weight of the sensor assembly is less than 2 kg. For the analysis of alternative THz optics and the measurements on planar samples in an XY-stage, the focusing optics can be easily dismounted and exchanged. Additionally, a second baseplate was manufactured in which the relative angle of the measurement heads can be changed to accommodate different working distances.
The standard sensor assembly can be integrated into the 3D platform (see description of motion platform hereafter) via a mating bracket; the 6 m long fibre delivery facilitates the execution of the scan trajectories. The DAQ software was integrated with the motion platform control routines into a TD 3D scan interface.
Figure 4 - Left: Schematic of the sensor assembly (top view); the full black circle represents the test radome. Right: The assembled TD sensor mounted on an optical table.
Concerning the motion platform for the testing of complex shaped objects, the original requirements described in the DoW to build an XYθ scanning stage have been replaced since they did not take into account that the sensor has to be perpendicular and at fixed distance to the surface. Instead, the requirements evolved towards an XYZθψ scanning stage. The addition of more stages clearly increases the complexity and cost of mechanical components, hardware and software. For the testing of the motion platform, a radome has been chosen as test object.
During a first concept study phase, multiple layouts have been evaluated. In the end, a concept was chosen which offered the best balance between fulfilling the requirements stated in the previous chapter and available budget. Small concessions had to be made, e.g. by allowing that only 70% of the radome surface has to be scanned, size could be reduced and one motion stage less had to be integrated. These concessions were made to reduce complexity and cost overruns due to changes with respect to the original DoW.
Figure 2 shows the final design. It consists of 3 linear axis and 2 rotation axis. Both the sensor and sample will perform a smooth synchronized motion such the surface can be scanned with high precision and as fast as possible.
To be able to obtain high accuracy, at least part of system has to operate in real-time. However, real-time systems such as PLC’s do not offer the same flexibility as a Windows based system. Therefore, a mixed system was chosen. The host system consists of a Windows based PC which does not operate in real-time. However, the host PC incorporates a motion controller board and a DAQ board which can operate in real-time by themselves.
Communication is required between the two boards to allow synchronization between motion and data acquisition. This is done using a direct hardware link (NI RTSI bus) such that the communication happens in real-time and is not randomly delayed by the host OS. Figure 2 shows the overall system architecture.
Figure 5- Left: 3D CAD model of the 5-axis motion platform; Right: Photograph of the realized motion platform with mounted FMCW-Transceiver and radome test sample.
The motion controller board is a National Instruments PCI 7356 board and the data acquisition board a National Instruments PCI 6115 board.Following limitations have been identified: The targeted scan speed of 1 m/s cannot be reached in all cases. The linear stages have a maximum allowable speed of > 1m/s. However, due to their limited length and limits on the acceleration, this speed cannot always be reached. The effect on total scan times will be very limited.
S&T results related to software
For the FMCW focused set-up, a data fusion software has been created. In FMCW, the depth resolution depends on the bandwidth of the radiated frequency ramp.
The 300 GHz head uses a wide frequency band of 90 GHz and thus has a good depth resolution which is considerably better than that of the 100 GHz and 150 GHz heads. However much of the materials relevant to the DOTNAC project are not well transparent for 300 GHz radiation while on the other hand a depth resolution of 4 mm (150 GHz head) is not sufficient to detect some of the defects. Thus the aim of the data fusion software is to merge the data from the 100 GHz and 150 GHz heads creating SVF files with 100 GHz bandwidth resulting in a depth resolution that competes with the 300 GHz heads at the good transparency of low-frequency THz radiation.
In order to prepare data fusion, several steps are performed.
• Before the measurements, the FMCW heads are calibrated with a set of calibration mirrors.
• The frequency axes of both heads are aligned by Fourier transformation and zero padding.
• Depending on the precise hardware characteristics, there may be a small frequency gap between the 100 GHz band and the 150 GHz band. This gap is closed with a suitable algorithm.
After these preparations the SVF files from both heads are merged into a single wide-band SVF file.
Application of the data fusion on focused FMCW data has been proven to be very effective for better in-depth resolution of defects.
For the wide angle configuration of the FMCW system, appropriate data processing algorithms have been developed to compress the unfocused raw data. Conventional synthetic aperture algorithms were studied, developed and adapted to be applied into wide-beam measurements in NDT. A flexible SAT simulator was created reproducing the FMCW sensors that would be used in future real NDT situations. Theoretical and simple scenes (point targets) were simulated and reconstructed using 2-D SAT algorithms. Real measurements replicating simple scenes, using a corner reflector, were also tested out as a validation of the implemented algorithms. SAT algorithms were extended to handle 3-D SAT data volumes produced by 2-D scans of flat samples. According to the experimental results obtained (Table 2), frequency-domain algorithms were chosen since time-domain algorithms presented unfeasible running times for very similar image qualities.
SAT algorithm Time Technique
Time-Domain ~ 7 hours Convolution
Time-Domain, approx. ~ 10 min Inner product
Range-Doppler < 2 min FFT/IFFT
Omega-k < 1 min 40 sec FFT/IFFT
Table 2 - Performances of different 3-D SAT algorithms are compared.
After validation of the 3-D SAT algorithms with arbitrary scenes, real composite calibration samples on Rohacell and honeycomb cores were measured and reconstructed. Detection capabilities are limited by a combination of the reflectivity of the each defect’s material and its size. Those defects physically smaller than the system’s resolution cell will not be imaged correctly, regardless their reflectivity properties.
In parallel, the image processing methodology tool for semiautomatic target detecting in 3-D SAT imagery was developed, first on simple and more complex scenes generated by the SAT simulator.
Denoising and image contrast enhancement procedures were used to obtain such results.
As a part of the optimisation stage, the following actions were taken:
• Reducing data collection in azimuth and elevation to the minimum.
• Restricting the ranges of interest where SAT processing will be applied.
• Efficiency enhancement: single-precision data, using less volatile memory by allocating and deallocating data structures during the execution of the 3-D SAT processing, and avoiding dynamic memory allocation.
• A dedicated noise reduction and image enhancement routine.
• Image processor that promotes sensitivity vs. robustness.
Moreover, a test bed for circular SAT processing was initiated in accordance to the possibilities of performing radome measurements within the project timeframe. Using as a starting point the 2.5-D scanner delivered by Verhaert, a simulator for circular, non-constant radius scanning paths was developed.
Figure 6 - Left, schematic showing a sensor-radome configuration. Right, simulated scanning path of a sensor whose radius varies in elevation following the shape of an arbitrary radome.
Graphic user interfaces (GUI) were created for an easy, intuitive and completely guided use of the SAT processing algorithms and raw FMCW data visualisation.
Qterahertz is a software specially realised for THz imaging (by Bordeaux University). It was determined to be a robust tool for the non-destructive evaluation of multi-layered materials and volumetric opaque objects. Two principal imaging architectures (pulsed TD system and FMCW system) were developed for performing THz imaging. The TD imaging technique has the ability to gather material dependent spectroscopic information about the sample. In addition, the time-of-flight of THz electric field allows one to obtain image data that provide information about the thickness, composition and structure of the surface and hidden layers.
THz time-of-flight technique will be utilized in order to resolve defects embedded beneath and to determine the thickness of layers. The quality of the images was dependent on the contrast of the spectral responses of the media with respect to each other as well as absorption of all the covering layers.
The program makes it possible to select a specific parameter employed to build the image. In the time domain, we can select the amplitude for a specific time-delay, the maximum or minimum amplitudes, the contrast (difference between maximum and minimum values) or the time delay corresponding to the maximum amplitude (so called "phase-delay" image). In the frequency domain, we can simply select the amplitude for a specific frequency in order to perform spectroscopic THz imaging from 0.1 to 3 THz.
This new tool is especially dedicated to developers or end users. It is very efficient due to the huge quantity of data, information we have to collect, to analyse during image processing in the terahertz range.
For the 3D reconstruction of TD data, the tomosynthesis approach has been evaluated. Tomosynthesis is inspired from a medical imaging technique especially used to reconstruct mammography images. This method of acquisition allows a 3D reconstruction of small depth objects from a few numbers of projections. Unlike Tomography where projections are measured all around the object, in tomosynthesis, they are done according to a limited range angle (usually between -50° and 50°). To recover a XY plane for a specific Z position, we use an algorithm that superpose and shift all the acquired projections. All we have to do is superposing point by point all the projections. This operation gives a plane of focus. In order to have another focus, it is necessary to shift the projections before performing the superposition. The shifting process determines the depth where we focus. The OCT or time of flight principle (TOF), the signal obtained depends on the amplitude and the optical way of an electromagnetic wave reflected or backscattered. Instead of the transmission tomography where the ray crosses entirely the object, in OCT, the ray will be reflected by the internal structure of the object. As in the tomosynthesis technique each plane are acquired depth per depth. However, the only data collected are in the plane of the detectors all the other data are lost.
Moreover, we extracted the frequency dependent refractive index of a 3D target thanks to phase knowledge and a challenge would be to reconstruct a trustworthy image of the device with the filtered back projection algorithm. However, we faced the drawback which is related to the fact that this approach will be very time consuming from technological point of view but also from a numerical point of view. Probably some trade-off will have to be done between resolution and computation time. Additionally an array of detectors should be tested out with the (X,Y,_) scanning platform by CNRS in an attempt to reduce the data acquisition time by a significant factor. Efforts to improve the lateral resolution of the setup must be carried out in synchronisation with the implementation of the mechanical stage for the 3D acquisition.
Assessment of the THz NDT potential
The ultimate goal of DOTNAC was the evaluation of the THz technology as a NDT tool applied to composite materials used in aeronautics and their typical defects. To reach this objective, different groups of samples were manufactured and typical defects were intentionally produced in them. Taking into account the main characteristics of composite materials employed in the aeronautical industry and the most common defects, a variety of samples was produced to calibrate the THz system during the system integration and to compare the THz inspection results with the ones obtained with traditional NDT methods.
The tested defects were the most typical defects encountered in the aeronautical industry, i.e delaminations, foreign inclusions, debonds, porosity, surface and near subsurface scratch. Moreover coating issues on both conductive/ dielectric composites were also investigated.
These samples were manufactured in two different materials, conductive and dielectric materials, and the characteristics of the samples are briefly resumed in the following paragraphs.
After some preliminary inspections with THz technology the calibration samples were inspected by means of the different NDT technologies, as well conventional as with the FMCW and TD THz system.
The results obtained from the series of inspections were analysed and afterwards a comparative study among the conventional and THz NDT techniques was realized.
As a result of the comparative study carried out to this group of samples, different statistical graphics have been produced showing the capabilities of detection of the NDT technologies and comparing with each other. An example for the sandwich structures has been given in the figure below.
Figure 7 - Comparative analysis of the level of detection of Teflons obtained with the different NDT techniques in sandwich samples
For the effective and efficient management of the enormous amount of measured test data, a web-based centralized platform has been created. Through this platform the analysis of all the data collected from the NDT inspections could be efficiently realized. The centralized platform was programmed by CIMNE and allowed the upload of the results of the inspections as well as the necessary information regarding the test conditions. After several proposals provided by the consortium, the final solution chosen for the structure of the platform was the organisation of the data according to the type of materials, type of defects and type of NDT technique applied, enabling an advanced and automatic analysis methodology.
In the image below an overview of the graphic style and general contents of the platform is shown. The URL to access the platform is: https://www.dotnac-project.eu/dap.
Figure 8 - Statistical analysis in the DAP (right) and uploaded results shown in the DAP (left).
Last but not least, the full validation of the THz systems has been performed by applying the same series of conventional and THz techniques on a specifically created group of blind samples. Again a comparative analysis has been performed for the final assessment of the THz NDT potential. And finally an on-site test was organized to demonstrate the THz capability on a radome within an industrial environment.
Regarding the comparison of the two THz systems with the conventional NDT techniques tested (i.e. Infrared Thermography (IRT), Radiography (RX) and Ultrasounds (UT)), it has been verified that:
• The evaluated TD THz system outperforms any of the other NDT methods tested for the inspection of coating misprocesses (even with regard to the IRT).
• The FMCW THz system clearly overcomes the UT inspection difficulties for C-sandwich structures with foam or H/C core (in the case of portable UT systems) and, in some cases, even the inspection by RX as well.
• Furthermore, the results obtained by the FMCW THz system on these types of components are totally comparable to those resulting from the NDT techniques of IRT and UT non-portable systems (“in-workshop” UT systems).
• For the two developed THz systems, these have shown an equal / similar detection level that IRT, RX and some of the UT methods performed in the case of the A-sandwich foam or H/C core components. And, additionally, it has also been observed that:
o In the case of Group G (A-sandwich H/C and C-sandwich H/C panels with real debonds or delaminations), the smaller defects of less have not been well resolved, although the FMCW results are better than the corresponding TD.
o In the case of Group H (A-sandwich (Rohacell core) panels with real debonds) the better results are for TD instead of FMCW.
• For Fiberglass Solid Laminates has been verified that both THz systems are clear alternatives for the inspection by RX and that, when the obtained THz signal is subsequently processed (signal processing), both systems have yielded quite comparable results to those obtained by UT.
Regarding the application of the two THz systems as NDT tool for the different materials / structures tested, it can be stated that the THz waves have the capacity to see-through different sample configurations of FGRP composites (such as laminates, foam and sandwich structures) and, therefore, can detect most of the relevant defects which can be found in aeronautical composite parts (such as delaminations, odd materials, inclusions, among others).
Based on the different tests performed in this part of the project:
• The THz technology is capable of penetrating the FGRP sandwich panels, beyond the foam and honeycomb cores (one of the main limiting factors of the conventional NDT techniques).
• Among the two THz systems examined, FMCW (“Frequency Modulated Continuous Wave System”) seems to be the one with the widest spectral application for typical FGRP composite materials found in aeronautics.
• For composite materials made of CFRP (which are not testable in the THz region due to their transparency), the positive detections have been limited only to the case of coating misprocesses inspected by TD.
Regarding the logical end user requirements of maximum testable thickness / minimum defect size detectable the tests performed by the developed THz systems have demonstrated the good trade-off between its penetration capacity versus resolution achieved. In the case of the FMCW THz, although the penetration capacity is conditioned to the nature of the material, a depth comprised between 1cm to 3cm is easily achieved, whereas it cannot exceed 1cm for the TD THz system.
As for the resolution achieved, minima of 1mm for the spatial resolution have been obtained for both systems. But related to the depth resolution, the resulting minimums have been of 2mm for the FMCW THz system and of 20µm in the case of the TD THz system.
The radome sample selected as example for the inspection of real parts has demonstrated the capability of the THz systems (when they are integrated in a scanner) for its uses on real-life aeronautic components. Besides that, the application of several measurement sensors allows very short measuring times with comparatively high image resolution. And the tests performed with its integration in the 5-Axis Motion Platform developed have proven that data acquisition of 1.000 measurements per second (that is, 10.000 points inspected in 10 seconds) is possible.
Therefore, it can be stated that both systems of THz developed in the DOTNAC project are feasible for their use in the inspection of “aircraft parts" (such as could be the vertical stabilizer of the tail, parts of wings and parts of the fuselage) and they can be an alternative to the inspection with UT of curved parts.
In terms of the industrial capacity and the industrial feasibility of the THz systems developed in this project, they are deemed appropriate for uses such as on:
• An “On-aircraft” NDT system, that is, when the equipment can be moved to the part to be tested in order to inspect the suspicious area (a complete scan of the whole part is not usually necessary). In this configuration variety, the system is manually placed on the object under test and the measurement is performed automatically.
• An “In-workshop” NDT system, which means, the system cannot be mobilized easily or it is not possible to move it and, therefore, the part to be tested needs to be transported to the workshop where the system is located. This equipment configuration, which can be integrated in the customer’s industrial facilities, will allow the inspection and the assessment of 3-D objects with a more complex shape. And, for example, it might be equipped with custom-made FMCW THz sensor offering dual linear polarisation for enhanced defect detection and full phase information extraction for enhanced contrast in the THz image.
Lastly, the potential of combining information received from different NDT techniques will allow advantage to be taken of the detection capabilities of each one (which employ different physical properties for evaluating the state of the material) and the fusion of these techniques will provide much more information in the quality control of aeronautical specimens (remember that no single method is capable of detecting all the different defects or damage).
For example it has been proven that THz technology is capable of penetrating inside the GFRP sandwich panels, beyond the foam and honeycomb cores (one of the main limiting factors of the conventional NDT techniques). And, on the other hand, IRT is capable of inspecting any type of material (metallic or not metallic) but with a limited depth for the detection defects. These two techniques (contactless, non-destructive and non-intrusive) provide a complementary capability for defect detection, obtaining (with their fusion) a flexible NDT tool applicable to a wide range of aeronautical materials.
Compliance with the requirements of the Ethics Review
All the partners working with THz systems have guaranteed the safety of the personnel working with THz waves and laser systems. The appropriate health and safety procedures conforming to relevant local/national guidelines/legislation are followed for staff involved in this project.
Immediate exposure to the THz waves has been prohibited at all moments, and the laser beams were contained within the system so the laser system (as part of the THz wave system) could be defined as a class 1 laser.
Potential Impact:
Description of impact
DOTNAC has been designed to be responsive to the call of European authorities responsible for the aeronautics and transport policy. A highly interdisciplinary fundamental and applied research is proposed for the development of an advanced THz-NDT tool. The aims of the project meet the high-technology nature of the Transport industry, making research and innovation crucial to its further development and leading to European competitiveness. By the realization of an advanced in-process inspection and quality control during the production phase and at the same advanced technique for continuous health and usage monitoring of structures and systems, the DOTNAC project opens possibilities to reduce aircraft development, production and operational costs. Indeed, by detecting defects in composite material as early as possible in the production chain, the scrapping or rework of a component can be avoided, contributing immensely to the to the cost of
part production. At the same time, the same non-invasive, non-contact THz inspection technique can be used for assessing composite part condition in service. This type of inspection can effectively and efficiently support the high standards of composite construction and repair, thereby reducing the aircraft operational costs.
The world NDT equipment market is witnessing growth, promoted by the recovery in the economies of developed countries as well as the demand from developing markets (the in-line process NDT business segment is increasing at two times the rate of overall business). Focus on R&D in newer digital-driven NDT modalities, and translation of these into value-delivering solutions are expected to be paramount in ensuring the success of NDT equipment vendors in the market.
Following steps are foreseen to bring about these impacts:
• The proposed consortium combines the experienced skill of the participants coming from different sectors, namely from academia, SME and research centra. All of them are highly acknowledged in their area of activity.
• DOTNAC carries both scientific and industrial importance integrating trans-nationally the necessary critical mass of resources, for the development of the proposed THz-NDT tool and the global assessment of the potential of this tool is an immense challenge commanding a joint project of diverse expertise reflecting all aspects of a sensor system.
• A new concept is offered by DOTNAC in order to combine all the necessary NDT requirements, which are most of the time mutually exclusive (i.e. improving one feature will have a negative influence on others).
• The DOTNAC NDT solution will offer inspecting a range of composite materials as well as detecting virtually all types of defects of an aircraft part during the production line, before and potentially on-site after the installation on the aircraft using a safe, contact-free, high resolution, easy to integrate in industrial facilities.
• The use of two fast imaging systems, completed with a novel THz signal and imaging processor based on the tomography principles and a novel THz signal and imaging processor based on the synthetic aperture.
In terms of cost reduction and efficiency, the new THz tool will not only help to detect but also to size a defect. That means, it will be possible to determine automatically (or semi-automatically) if that defect is acceptable or not. This will reduce the human interpretation, allowing a reduction of risk and time and an increase of the quality of NDT inspection. All together this means, on one hand, cost reduction, since less time will be necessary to inspect with the same or higher quality and, on the other hand, better efficiency due to the reduction of the human interpretation factor. All the conventional NDT techniques and some advanced techniques still have a high human interpretation factor.
Main dissemination activities
First of all, a DOTNAC website has been created with two aims :
• Dissemination of the project (public part).
• Storage of information and reports, for exchange between partners (private part).
The main architecture of the website resulted from iterations between all partners based on designs provided by Innov Support. The website has been hosted through the facilities at CIMNE. Since the operational launch of the website multiple updates (mainly maintenance and improvement tasks) have been implemented.
The private interface is accessible through a tightly controlled User ID and Password scheme, with specific access rights in function of the partner role in the project. The website URL is http://www.dotnac-project.eu
Innov Support has produced the 6 monthly versions of the PuDK (Plan for using and disseminating the Knowledge), integrating the dissemination and exploitation activities and plans for all the DOTNAC partners.
Leaflets of the project have been created to hand out during the multiple conferences and workshops the DOTNAC partners participated.
The DOTNAC project has generated furthermore interesting dissemination activities through the participation to specific conferences throughout the whole project and publication in journals. The full list of publications is provided in the next section.
During the second half of the DOTNAC project, 2 Industrial and Technology Followers Group (ITFG) workshops have been organised. Attendance of ITFG members was generated through the approach by all DOTNAC partners to their industrial and technological contacts. This has ensured an attendance of approximately 25 ITFG members for each ITFG workshop and ensured the participation of a series of dedicated contact entries.
Schedule of the ITFG workshops:
• The 1st ITFG workshop was organised in Bordeaux (France) at CNRS – Université de Bordeaux on October 17th 2012.
• The 2nd workshop was organised in Barcelona (Spain) at Applus+ - LGAI Technological Center on June 27th 2013.
Both ITFG workshops included the following major constituents:
• Presentations by the DOTNAC consortium of the activities and results achieved by DOTNAC at the time of the workshop.
• Demonstration of work on real system setup
• Question and answer (Q/A) session between the ITFG members and the DOTNAC consortium.
The outcome of each ITFG workshop (primarily the feedback collected during the Q/A sessions were then further addressed by all DOTNAC partners in a ITFG workshop follow up meeting between the DOTNAC partners to analyse all the feedback and comments that were collected during the ITFG workshop. Where possible the activities of the remainder of the project have taken into account these comments. The partners have implemented the specific comments with respect to the processing of the both ITFG workshops and the potentiality of dissemination and exploitation of the DOTNAC results.
Dedicated posters were produced for the 2nd ITFG workshop. All presentations, demonstrations and posters associated to both ITFG workshops have been uploaded on the public part of the DOTNAC website.
Exploitation results
Innov Support has produced the 6 monthly versions of the PuDK (Plan for using and disseminating the Knowledge). Innov Support has, through the PuDK, collected the description of the exploitable results so far produced by the DOTNAC partners and the associated potential exploitation routes for these results. In addition the actualised exploitation plans of the various project members have been also been captured.
Based on the data collected in the various PuDKs as well as the use of a Technology Implementation Plan form, Innov Support has elaborated an exploitation strategy and scenario. This will be accompanied with the elaboration of a business plan to be executed in the context of the DOTNAC project. This information has been further worked out in deliverable D8.43 (Precompetition survey – Future technology exploitation roadmap).
List of Websites:
www.dotnac-project.eu
Marijke Vandewal, Dr. Ir
Associate Professor
Royal Military Academy
Department CISS
Hobbema straat, 8
1000 Brussels
Belgium
Tel. : +32/(0)2/742.65.63
Fax : +32/(0)2/742.64.72
Email : marijke.vandewal@rma.ac.be
The availability of light and robust structures has led to an increased use of composite materials in the aircraft industry. In order to verify and guarantee the high quality of the conventional and new composite elements, innovative approaches for non-destructive testing of these parts are required.
The European research project “DOTNAC” proposes to develop a fast, high resolution, noninvasive and non-contact inspection system for assessing aeronautic composite parts during production using terahertz waves. Conventionally two categories of systems can be discussed: pulsed and continuous wave terahertz systems. Both have been realized and their respective potential as a non-destructive inspection tool has been evaluated.
The implemented FMCW (Frequency-Modulated Continuous-Wave) system is an all-electronic THz system. It consists of three scanning heads with different frequency ranges (around 100 GHz, 150 GHz, and 300 GHz). They are used to acquire data by scanning a sample placed in front of them in reflection mode. Combining the measured in-depth information with a lateral scanning, a 3-D image can be built up. The in-depth resolution depends on the used bandwidth and the roughness of the sample surface, and varies between 2 mm and 6 mm. For the construction of the TDS (Time-Domain System), a pulsed laser system has been implemented in a fibre-optical ECOPS (Electronically Controlled Optical Sampling) pump-probe set-up. The two short-pulse lasers (one for the emitter, one for the detector) are based on Erbium-doped silica glass fibres and emit around 1560 nm centre wavelength. The in-depth resolution is significantly better than the one obtained with the FMCW THz system at the expense of a lower penetration capacity for the materials and structures tested within the scope of the project. The across-range resolution depends again on the beam focus and is comparable to the one of the FMCW system using the focused configuration. The main trade-off that needs to be considered for the TDS is the one involving measurement speed and signal-to-noise ratio.
To achieve these objectives, a series of 20 flat glass fibre reinforced plastics (GFRP) samples (solid laminates and sandwich structures), and 5 carbon fibre reinforced plastics (CFRP) were modified by artificial defects such as inserts, stucks, water inclusions, etc., to create well-controlled and well-known samples to validate the THz systems and algorithms with. In a next step the evaluation as a THz NDT method has been performed on 50 blind samples (12 GFRP and 38 CFRP samples) with embedded defects caused by an intentional miss-process. For validation and evaluation respectively, both THz systems under test have been installed on a two-dimensional scanner for sample inspection. A proof of concept has been provided using a three-dimensional motion platform to demonstrate the feasibility of THz inspection on a real aircraft object.
For the full assessment of the THz NDT performances, all the above-mentioned samples have been inspected by the following conventional techniques: thermographic NDT, radiographic testing (film and digital radiography and computed tomographic radiography testing), and ultrasound testing (2-D Inverse Wave Field Extrapolation, Phase Array, Pulse Echo, and Through Transmission). The best performances per inspection method and per sample/defect type have been identified to demonstrate complementarity between the respective NDT methods or their individual performance.
Regarding the feasibility of a THz application in aeronautics, i.e. inspection of dielectric parts, the THz
technology has proved to be a valuable technique for NDT. The technique is able to detect:
1. Delaminations and foreign inclusions in dielectric laminates such as glass fiber laminates.
2. Delaminations and disbonds in dielectric sandwich structures such as A-sandwich or C-sandwich structures with either honeycomb or syntactic foam cores.
In addition the technique is very sensitive to coating misprocess on conductive substrate such as CFRP
and probably to the same extent on dielectric substrates such as glass fiber. On the other hand THz failed to detect porosity in glass fiber laminates. The capability of THz NDT for porosity detection should be further investigated.
In comparison with the classical NDT techniques, NDT-THz looks very competitive. It does not require
high and costly radiation protection such as for X-Ray radiography and is totally non-contact in contrast
with ultrasound.
To conclude, the performed assessment shows that FMCW and TDS can be considered as complementing techniques that should be tightly associated. The same conclusion can be formulated here, i.e. a combination of the two techniques investigated in this project, FMCW and TDS, seems to be the ideal way of applying THz in NDT. FMCW can provide fast scanning at high detection sensitivity. TDS can complete the data at the critical area by providing very accurate measures of defect sizes and depth at the expense of testing time.
Project Context and Objectives:
In recent years, the development of THz systems has made a considerable progress and made possible a variety of innovative applications. Among them, Non-Destructive Testing (NDT) of composite materials can take advantage of the high transparency of certain non-conductive materials at THz frequencies. The presence of such materials is foreseen to increase in the next years, especially in aeronautics. Glass-fiber (GF) laminates or composite structures are currently being used thanks to their outstanding structural properties, durability and lightness. In response to the growing industrial interest on these materials, the European research project “DOTNAC” proposes to develop a fast, high resolution, noninvasive and non-contact inspection system for assessing aeronautic composite parts during production using terahertz waves.
Through the realization of an advanced in-process inspection and quality control during the production phase and at the same time an advanced technique for continuous health and usage monitoring of structures and systems, the DOTNAC project opens possibilities to reduce aircraft development, production and operational costs. Indeed, by detecting defects in composite material as early as possible in the production chain, the scrapping or rework of a component can be avoided, contributing immensely to the cost of part production. At the same time, the same non-invasive, non-contact THz inspection technique can be used for assessing composite part condition in service. This type of inspection can effectively and efficiently support the high standards of composite construction and repair, thereby reducing the aircraft operational cost.
The main goal of the DOTNAC project is to develop a fast, high resolution, non-invasive and non-contact inspection system for assessing aeronautic composite parts either during production or maintenance. The developed NDT tool will be easy to integrate in industrial facilities and will fill in the performance gaps that are still present amongst the existing NDT techniques. It will therefore be an extremely useful tool in NDT in terms of sensor fusion. This new method will not replace directly the existing NDT tools, but will deliver complementary results which can be sometimes more precise for some defects.
Achieving this entails the following specific objectives:
• To create an integrated (hardware-software) and optimized THz imaging system using pulsed signals and optical fibre coupling.
• To create an integrated (hardware-software) and optimized THz imaging system using continuous wave signals and electrical cable coupling.
• To demonstrate, in an industrial setting, the effectiveness of a THz NDT tool.
• To assess the performances of the 2 developed THz NDT tools for assessing aeronautic composite parts.
• To develop a user/research community for fast, high resolution, non-invasive & non-contact inspection for assessing aeronautic composite parts during production.
Project Results:
S&T results can be described at following levels:
- With respect to the hardware:
o FMCW THz system: system was acquired prior to DOTNAC but optimized for the application as described in DOTNAC
o TDS THz system: system has been developed from scratch within the frame of DOTNAC
o Motion platform: system has been developed for a tailored used within the frame of DOTNAC
- With respect to the developed software:
o Data fusion software for the focused FMCW data
o Synthetic aperture processing for the wide aperture FMCW data
o Tomographic processing for the multi-angle TDS data
- With respect to the tested NDT potential:
o Creation of a full set of samples with defects representative for aeronautics applications
o Exploitation of the results obtained with conventional NDT techniques and the THz tools
S&T results related to hardware
Originally the FMCW THz system was designed and fabricated by SynViewn during DOTNAC the system has been modified and optimized by Fh-IPM to fulfill the objectives defined for DOTNAC.
The system is 2.5 m high, 1.5 m long 1.2 m wide having a total weight of 300 kg. On its side there is the data acquisition unit, the computer and power alimentation together in a tower. The maximum sample size is limited to 700x700 mm in surface and the maximum scan range is 600x689 mm. The dimensions of the optics limit also the maximum sample thickness to 90 mm for the optic with a focus of 50 mm and to 380 mm for the optic with a focus of 200 mm.
As mentioned in the DoW the FMCW system consists of three scanning heads with different center frequencies (100 GHz, 300 GHz and 850 GHz). After the first tests, the 850 GHz scanning head was replaced by 150 GHz due to poor results. The materials used for the DOTNAC samples made this change necessary. The scanning heads were used to acquire data by scanning a sample placed in front of them in reflection mode (see Figure 1).
Figure 1- Photographs of the FMCW system. Left: side view with housing and control unit. Right: view from the front side showing the three transceivers and three receivers in transmission assembly. The z-direction between the transceivers and receivers is fix
100 GHz 150 GHz 300 GHz 850 GHz
Frequency range 0.07 THz – 0.11 THz 0.11 THz – 0.17 THz 0.23 THz – 0.32 THz 0.84 THz – 0.87 THz
Bandwidth 40 GHz 60 GHz 90 GHz
Dynamic range
(acquisition times are per pixel) > 40 dB
(at acquisition time 100 μs)
> 60 dB
(at acquisition time 10 ms)
> 70 dB
(at acquisition time 100 ms) > 40 dB
(at acquisition time 100 μs)
> 60 dB
(at acquisition time 10 ms)
> 70 dB
(at acquisition time 100 ms) > 35 dB
(at acquisition time 100 μs)
> 50 dB
(at acquisition time 10 ms)
> 60 dB
(at acquisition time 100 ms) > 25 dB
(at acquisition time 100 μs)
> 40 dB /
(at acquisition time 100 μs)
> 50 dB
(at acquisition time 100 μs)
Spatial resolution (laterally) 3 mm 2 mm 1 mm 0.4 mm
Depth resolution in air 9 mm 6 mm 3 mm 10 mm
Depth resolution in typical materials 6 mm 4 mm 2 mm 15 mm
Table 1 - System specifications of the FMCW THz system.
Fh-IPM has done measurements in order to check the manufacturer’s specifications. Three metal bars were installed as a sample and measured in transmission and reflection using different frequencies. The dynamic range was determined for two different focusing optics (50 and 200 mm). As a result one can confirm the information provided by the manu¬facturer (see Table 1).
Fh-IPM has developed a visualization software: An image-editing software for presentation of samples measured using FMCW is available. All levels in X, Y and Z direction are scaled as the signal-noise ratio. Different filters can be used for better representation. 2D and 3D representations are possible. It is possible to set and to display different cut plane. A presentation in color as well as in grey tones is possible.
Figure 2- Left: SVD view toolbox. Right: 3D illustration of a test pattern.
Setup with measuring head, lens and sample Setup with measuring head, prism, parabolic mirrors and sample
A-Sandwich with rohacell, 100 GHz, front A-Sandwich with rohacell, 100 GHz, front
Figure 3 - Testing of different optics to evaluate the trade-off azimuth resolution versus in-depth information
The focused set-up uses a set of lenses creating a focused beam with a beam waist characterized by the Rayleigh length. Depending on the type of optics, a trade-off can be made between azimuth resolution and in-depth information. For that reason different sets of optics have been test out and an assessment on the image quality has been made.
To complete the list of configurations, a wide angle set-up was tested out (omitting the lenses all together) and a specific calibration procedure for this measurement configuration has been created. The processing of the acquired data is discussed in the software section hereafter.
The complete FMCW system has been mounted on a 3D scanning system (see description hereafter), developed by Verhaert. After a fully integrated and optimized hard- and software, the developed THz NDT tool has been used to image the initially defined composite materials and structures.
As a complementary system to the FMCW THz system the TD THz system has been developed from scratch within the project. The TD THz system is equipped with a twin femtosecond laser configuration (with approximately 1560 nm centre wavelength), in which the repetition rate of one of the lasers can be either synchronized to the other or deliberately modified to induce a fixed or variable time delay between the laser pulses. Both lasers are terminated with an external fibre delivery of more than 6 m length which is internally pre-compensated with respect to dispersion and non-linear pulse broadening. The time window that can be captured in ECOPS mode at 100 Hz scan rate is 146 ps. The system itself would allow operation at higher frequencies (500 Hz was tested), but the integrated data acquisition architecture so far only runs at 10 Hz. This is, however, not a fundamental limit. The lasers, their control electronics, further electronics for generation, read-out and amplification of the THz pulses (voltage source, waveform generator, voltage amplifier) are placed in a mobile cubic rack with less than 100 kg total weight that serves as basic supply unit for the TD system to drive the remote sensor and feed the data to the DAQ card and software, which are installed on a dedicated PC. The system requires only access to a standard European 230 V AC power socket.
The sensor heads for emitter and detector contain photoconductive antennas that are permanently coupled to fibre patchcords for connecting them rigidly with the pump lasers. THz radiation is coupled out and in, respectively, by highly resistive silicon lenses. For THz beam shaping, each head includes an internal collimating and an external focusing lens (material: TPX). To match the working distance of the 3D scanner unit, 200 mm is chosen as focal length. The THz optical elements are aligned and rigidly positioned by means of an opto-mechanic cage system. The sensor heads are mounted on a common baseplate for reflection measurements under an angle of incidence to the sample normal of 10°. This baseplate also incorporates a transimpedance amplifier for pre-amplification of the signal close to the detector. The weight of the sensor assembly is less than 2 kg. For the analysis of alternative THz optics and the measurements on planar samples in an XY-stage, the focusing optics can be easily dismounted and exchanged. Additionally, a second baseplate was manufactured in which the relative angle of the measurement heads can be changed to accommodate different working distances.
The standard sensor assembly can be integrated into the 3D platform (see description of motion platform hereafter) via a mating bracket; the 6 m long fibre delivery facilitates the execution of the scan trajectories. The DAQ software was integrated with the motion platform control routines into a TD 3D scan interface.
Figure 4 - Left: Schematic of the sensor assembly (top view); the full black circle represents the test radome. Right: The assembled TD sensor mounted on an optical table.
Concerning the motion platform for the testing of complex shaped objects, the original requirements described in the DoW to build an XYθ scanning stage have been replaced since they did not take into account that the sensor has to be perpendicular and at fixed distance to the surface. Instead, the requirements evolved towards an XYZθψ scanning stage. The addition of more stages clearly increases the complexity and cost of mechanical components, hardware and software. For the testing of the motion platform, a radome has been chosen as test object.
During a first concept study phase, multiple layouts have been evaluated. In the end, a concept was chosen which offered the best balance between fulfilling the requirements stated in the previous chapter and available budget. Small concessions had to be made, e.g. by allowing that only 70% of the radome surface has to be scanned, size could be reduced and one motion stage less had to be integrated. These concessions were made to reduce complexity and cost overruns due to changes with respect to the original DoW.
Figure 2 shows the final design. It consists of 3 linear axis and 2 rotation axis. Both the sensor and sample will perform a smooth synchronized motion such the surface can be scanned with high precision and as fast as possible.
To be able to obtain high accuracy, at least part of system has to operate in real-time. However, real-time systems such as PLC’s do not offer the same flexibility as a Windows based system. Therefore, a mixed system was chosen. The host system consists of a Windows based PC which does not operate in real-time. However, the host PC incorporates a motion controller board and a DAQ board which can operate in real-time by themselves.
Communication is required between the two boards to allow synchronization between motion and data acquisition. This is done using a direct hardware link (NI RTSI bus) such that the communication happens in real-time and is not randomly delayed by the host OS. Figure 2 shows the overall system architecture.
Figure 5- Left: 3D CAD model of the 5-axis motion platform; Right: Photograph of the realized motion platform with mounted FMCW-Transceiver and radome test sample.
The motion controller board is a National Instruments PCI 7356 board and the data acquisition board a National Instruments PCI 6115 board.Following limitations have been identified: The targeted scan speed of 1 m/s cannot be reached in all cases. The linear stages have a maximum allowable speed of > 1m/s. However, due to their limited length and limits on the acceleration, this speed cannot always be reached. The effect on total scan times will be very limited.
S&T results related to software
For the FMCW focused set-up, a data fusion software has been created. In FMCW, the depth resolution depends on the bandwidth of the radiated frequency ramp.
The 300 GHz head uses a wide frequency band of 90 GHz and thus has a good depth resolution which is considerably better than that of the 100 GHz and 150 GHz heads. However much of the materials relevant to the DOTNAC project are not well transparent for 300 GHz radiation while on the other hand a depth resolution of 4 mm (150 GHz head) is not sufficient to detect some of the defects. Thus the aim of the data fusion software is to merge the data from the 100 GHz and 150 GHz heads creating SVF files with 100 GHz bandwidth resulting in a depth resolution that competes with the 300 GHz heads at the good transparency of low-frequency THz radiation.
In order to prepare data fusion, several steps are performed.
• Before the measurements, the FMCW heads are calibrated with a set of calibration mirrors.
• The frequency axes of both heads are aligned by Fourier transformation and zero padding.
• Depending on the precise hardware characteristics, there may be a small frequency gap between the 100 GHz band and the 150 GHz band. This gap is closed with a suitable algorithm.
After these preparations the SVF files from both heads are merged into a single wide-band SVF file.
Application of the data fusion on focused FMCW data has been proven to be very effective for better in-depth resolution of defects.
For the wide angle configuration of the FMCW system, appropriate data processing algorithms have been developed to compress the unfocused raw data. Conventional synthetic aperture algorithms were studied, developed and adapted to be applied into wide-beam measurements in NDT. A flexible SAT simulator was created reproducing the FMCW sensors that would be used in future real NDT situations. Theoretical and simple scenes (point targets) were simulated and reconstructed using 2-D SAT algorithms. Real measurements replicating simple scenes, using a corner reflector, were also tested out as a validation of the implemented algorithms. SAT algorithms were extended to handle 3-D SAT data volumes produced by 2-D scans of flat samples. According to the experimental results obtained (Table 2), frequency-domain algorithms were chosen since time-domain algorithms presented unfeasible running times for very similar image qualities.
SAT algorithm Time Technique
Time-Domain ~ 7 hours Convolution
Time-Domain, approx. ~ 10 min Inner product
Range-Doppler < 2 min FFT/IFFT
Omega-k < 1 min 40 sec FFT/IFFT
Table 2 - Performances of different 3-D SAT algorithms are compared.
After validation of the 3-D SAT algorithms with arbitrary scenes, real composite calibration samples on Rohacell and honeycomb cores were measured and reconstructed. Detection capabilities are limited by a combination of the reflectivity of the each defect’s material and its size. Those defects physically smaller than the system’s resolution cell will not be imaged correctly, regardless their reflectivity properties.
In parallel, the image processing methodology tool for semiautomatic target detecting in 3-D SAT imagery was developed, first on simple and more complex scenes generated by the SAT simulator.
Denoising and image contrast enhancement procedures were used to obtain such results.
As a part of the optimisation stage, the following actions were taken:
• Reducing data collection in azimuth and elevation to the minimum.
• Restricting the ranges of interest where SAT processing will be applied.
• Efficiency enhancement: single-precision data, using less volatile memory by allocating and deallocating data structures during the execution of the 3-D SAT processing, and avoiding dynamic memory allocation.
• A dedicated noise reduction and image enhancement routine.
• Image processor that promotes sensitivity vs. robustness.
Moreover, a test bed for circular SAT processing was initiated in accordance to the possibilities of performing radome measurements within the project timeframe. Using as a starting point the 2.5-D scanner delivered by Verhaert, a simulator for circular, non-constant radius scanning paths was developed.
Figure 6 - Left, schematic showing a sensor-radome configuration. Right, simulated scanning path of a sensor whose radius varies in elevation following the shape of an arbitrary radome.
Graphic user interfaces (GUI) were created for an easy, intuitive and completely guided use of the SAT processing algorithms and raw FMCW data visualisation.
Qterahertz is a software specially realised for THz imaging (by Bordeaux University). It was determined to be a robust tool for the non-destructive evaluation of multi-layered materials and volumetric opaque objects. Two principal imaging architectures (pulsed TD system and FMCW system) were developed for performing THz imaging. The TD imaging technique has the ability to gather material dependent spectroscopic information about the sample. In addition, the time-of-flight of THz electric field allows one to obtain image data that provide information about the thickness, composition and structure of the surface and hidden layers.
THz time-of-flight technique will be utilized in order to resolve defects embedded beneath and to determine the thickness of layers. The quality of the images was dependent on the contrast of the spectral responses of the media with respect to each other as well as absorption of all the covering layers.
The program makes it possible to select a specific parameter employed to build the image. In the time domain, we can select the amplitude for a specific time-delay, the maximum or minimum amplitudes, the contrast (difference between maximum and minimum values) or the time delay corresponding to the maximum amplitude (so called "phase-delay" image). In the frequency domain, we can simply select the amplitude for a specific frequency in order to perform spectroscopic THz imaging from 0.1 to 3 THz.
This new tool is especially dedicated to developers or end users. It is very efficient due to the huge quantity of data, information we have to collect, to analyse during image processing in the terahertz range.
For the 3D reconstruction of TD data, the tomosynthesis approach has been evaluated. Tomosynthesis is inspired from a medical imaging technique especially used to reconstruct mammography images. This method of acquisition allows a 3D reconstruction of small depth objects from a few numbers of projections. Unlike Tomography where projections are measured all around the object, in tomosynthesis, they are done according to a limited range angle (usually between -50° and 50°). To recover a XY plane for a specific Z position, we use an algorithm that superpose and shift all the acquired projections. All we have to do is superposing point by point all the projections. This operation gives a plane of focus. In order to have another focus, it is necessary to shift the projections before performing the superposition. The shifting process determines the depth where we focus. The OCT or time of flight principle (TOF), the signal obtained depends on the amplitude and the optical way of an electromagnetic wave reflected or backscattered. Instead of the transmission tomography where the ray crosses entirely the object, in OCT, the ray will be reflected by the internal structure of the object. As in the tomosynthesis technique each plane are acquired depth per depth. However, the only data collected are in the plane of the detectors all the other data are lost.
Moreover, we extracted the frequency dependent refractive index of a 3D target thanks to phase knowledge and a challenge would be to reconstruct a trustworthy image of the device with the filtered back projection algorithm. However, we faced the drawback which is related to the fact that this approach will be very time consuming from technological point of view but also from a numerical point of view. Probably some trade-off will have to be done between resolution and computation time. Additionally an array of detectors should be tested out with the (X,Y,_) scanning platform by CNRS in an attempt to reduce the data acquisition time by a significant factor. Efforts to improve the lateral resolution of the setup must be carried out in synchronisation with the implementation of the mechanical stage for the 3D acquisition.
Assessment of the THz NDT potential
The ultimate goal of DOTNAC was the evaluation of the THz technology as a NDT tool applied to composite materials used in aeronautics and their typical defects. To reach this objective, different groups of samples were manufactured and typical defects were intentionally produced in them. Taking into account the main characteristics of composite materials employed in the aeronautical industry and the most common defects, a variety of samples was produced to calibrate the THz system during the system integration and to compare the THz inspection results with the ones obtained with traditional NDT methods.
The tested defects were the most typical defects encountered in the aeronautical industry, i.e delaminations, foreign inclusions, debonds, porosity, surface and near subsurface scratch. Moreover coating issues on both conductive/ dielectric composites were also investigated.
These samples were manufactured in two different materials, conductive and dielectric materials, and the characteristics of the samples are briefly resumed in the following paragraphs.
After some preliminary inspections with THz technology the calibration samples were inspected by means of the different NDT technologies, as well conventional as with the FMCW and TD THz system.
The results obtained from the series of inspections were analysed and afterwards a comparative study among the conventional and THz NDT techniques was realized.
As a result of the comparative study carried out to this group of samples, different statistical graphics have been produced showing the capabilities of detection of the NDT technologies and comparing with each other. An example for the sandwich structures has been given in the figure below.
Figure 7 - Comparative analysis of the level of detection of Teflons obtained with the different NDT techniques in sandwich samples
For the effective and efficient management of the enormous amount of measured test data, a web-based centralized platform has been created. Through this platform the analysis of all the data collected from the NDT inspections could be efficiently realized. The centralized platform was programmed by CIMNE and allowed the upload of the results of the inspections as well as the necessary information regarding the test conditions. After several proposals provided by the consortium, the final solution chosen for the structure of the platform was the organisation of the data according to the type of materials, type of defects and type of NDT technique applied, enabling an advanced and automatic analysis methodology.
In the image below an overview of the graphic style and general contents of the platform is shown. The URL to access the platform is: https://www.dotnac-project.eu/dap.
Figure 8 - Statistical analysis in the DAP (right) and uploaded results shown in the DAP (left).
Last but not least, the full validation of the THz systems has been performed by applying the same series of conventional and THz techniques on a specifically created group of blind samples. Again a comparative analysis has been performed for the final assessment of the THz NDT potential. And finally an on-site test was organized to demonstrate the THz capability on a radome within an industrial environment.
Regarding the comparison of the two THz systems with the conventional NDT techniques tested (i.e. Infrared Thermography (IRT), Radiography (RX) and Ultrasounds (UT)), it has been verified that:
• The evaluated TD THz system outperforms any of the other NDT methods tested for the inspection of coating misprocesses (even with regard to the IRT).
• The FMCW THz system clearly overcomes the UT inspection difficulties for C-sandwich structures with foam or H/C core (in the case of portable UT systems) and, in some cases, even the inspection by RX as well.
• Furthermore, the results obtained by the FMCW THz system on these types of components are totally comparable to those resulting from the NDT techniques of IRT and UT non-portable systems (“in-workshop” UT systems).
• For the two developed THz systems, these have shown an equal / similar detection level that IRT, RX and some of the UT methods performed in the case of the A-sandwich foam or H/C core components. And, additionally, it has also been observed that:
o In the case of Group G (A-sandwich H/C and C-sandwich H/C panels with real debonds or delaminations), the smaller defects of less have not been well resolved, although the FMCW results are better than the corresponding TD.
o In the case of Group H (A-sandwich (Rohacell core) panels with real debonds) the better results are for TD instead of FMCW.
• For Fiberglass Solid Laminates has been verified that both THz systems are clear alternatives for the inspection by RX and that, when the obtained THz signal is subsequently processed (signal processing), both systems have yielded quite comparable results to those obtained by UT.
Regarding the application of the two THz systems as NDT tool for the different materials / structures tested, it can be stated that the THz waves have the capacity to see-through different sample configurations of FGRP composites (such as laminates, foam and sandwich structures) and, therefore, can detect most of the relevant defects which can be found in aeronautical composite parts (such as delaminations, odd materials, inclusions, among others).
Based on the different tests performed in this part of the project:
• The THz technology is capable of penetrating the FGRP sandwich panels, beyond the foam and honeycomb cores (one of the main limiting factors of the conventional NDT techniques).
• Among the two THz systems examined, FMCW (“Frequency Modulated Continuous Wave System”) seems to be the one with the widest spectral application for typical FGRP composite materials found in aeronautics.
• For composite materials made of CFRP (which are not testable in the THz region due to their transparency), the positive detections have been limited only to the case of coating misprocesses inspected by TD.
Regarding the logical end user requirements of maximum testable thickness / minimum defect size detectable the tests performed by the developed THz systems have demonstrated the good trade-off between its penetration capacity versus resolution achieved. In the case of the FMCW THz, although the penetration capacity is conditioned to the nature of the material, a depth comprised between 1cm to 3cm is easily achieved, whereas it cannot exceed 1cm for the TD THz system.
As for the resolution achieved, minima of 1mm for the spatial resolution have been obtained for both systems. But related to the depth resolution, the resulting minimums have been of 2mm for the FMCW THz system and of 20µm in the case of the TD THz system.
The radome sample selected as example for the inspection of real parts has demonstrated the capability of the THz systems (when they are integrated in a scanner) for its uses on real-life aeronautic components. Besides that, the application of several measurement sensors allows very short measuring times with comparatively high image resolution. And the tests performed with its integration in the 5-Axis Motion Platform developed have proven that data acquisition of 1.000 measurements per second (that is, 10.000 points inspected in 10 seconds) is possible.
Therefore, it can be stated that both systems of THz developed in the DOTNAC project are feasible for their use in the inspection of “aircraft parts" (such as could be the vertical stabilizer of the tail, parts of wings and parts of the fuselage) and they can be an alternative to the inspection with UT of curved parts.
In terms of the industrial capacity and the industrial feasibility of the THz systems developed in this project, they are deemed appropriate for uses such as on:
• An “On-aircraft” NDT system, that is, when the equipment can be moved to the part to be tested in order to inspect the suspicious area (a complete scan of the whole part is not usually necessary). In this configuration variety, the system is manually placed on the object under test and the measurement is performed automatically.
• An “In-workshop” NDT system, which means, the system cannot be mobilized easily or it is not possible to move it and, therefore, the part to be tested needs to be transported to the workshop where the system is located. This equipment configuration, which can be integrated in the customer’s industrial facilities, will allow the inspection and the assessment of 3-D objects with a more complex shape. And, for example, it might be equipped with custom-made FMCW THz sensor offering dual linear polarisation for enhanced defect detection and full phase information extraction for enhanced contrast in the THz image.
Lastly, the potential of combining information received from different NDT techniques will allow advantage to be taken of the detection capabilities of each one (which employ different physical properties for evaluating the state of the material) and the fusion of these techniques will provide much more information in the quality control of aeronautical specimens (remember that no single method is capable of detecting all the different defects or damage).
For example it has been proven that THz technology is capable of penetrating inside the GFRP sandwich panels, beyond the foam and honeycomb cores (one of the main limiting factors of the conventional NDT techniques). And, on the other hand, IRT is capable of inspecting any type of material (metallic or not metallic) but with a limited depth for the detection defects. These two techniques (contactless, non-destructive and non-intrusive) provide a complementary capability for defect detection, obtaining (with their fusion) a flexible NDT tool applicable to a wide range of aeronautical materials.
Compliance with the requirements of the Ethics Review
All the partners working with THz systems have guaranteed the safety of the personnel working with THz waves and laser systems. The appropriate health and safety procedures conforming to relevant local/national guidelines/legislation are followed for staff involved in this project.
Immediate exposure to the THz waves has been prohibited at all moments, and the laser beams were contained within the system so the laser system (as part of the THz wave system) could be defined as a class 1 laser.
Potential Impact:
Description of impact
DOTNAC has been designed to be responsive to the call of European authorities responsible for the aeronautics and transport policy. A highly interdisciplinary fundamental and applied research is proposed for the development of an advanced THz-NDT tool. The aims of the project meet the high-technology nature of the Transport industry, making research and innovation crucial to its further development and leading to European competitiveness. By the realization of an advanced in-process inspection and quality control during the production phase and at the same advanced technique for continuous health and usage monitoring of structures and systems, the DOTNAC project opens possibilities to reduce aircraft development, production and operational costs. Indeed, by detecting defects in composite material as early as possible in the production chain, the scrapping or rework of a component can be avoided, contributing immensely to the to the cost of
part production. At the same time, the same non-invasive, non-contact THz inspection technique can be used for assessing composite part condition in service. This type of inspection can effectively and efficiently support the high standards of composite construction and repair, thereby reducing the aircraft operational costs.
The world NDT equipment market is witnessing growth, promoted by the recovery in the economies of developed countries as well as the demand from developing markets (the in-line process NDT business segment is increasing at two times the rate of overall business). Focus on R&D in newer digital-driven NDT modalities, and translation of these into value-delivering solutions are expected to be paramount in ensuring the success of NDT equipment vendors in the market.
Following steps are foreseen to bring about these impacts:
• The proposed consortium combines the experienced skill of the participants coming from different sectors, namely from academia, SME and research centra. All of them are highly acknowledged in their area of activity.
• DOTNAC carries both scientific and industrial importance integrating trans-nationally the necessary critical mass of resources, for the development of the proposed THz-NDT tool and the global assessment of the potential of this tool is an immense challenge commanding a joint project of diverse expertise reflecting all aspects of a sensor system.
• A new concept is offered by DOTNAC in order to combine all the necessary NDT requirements, which are most of the time mutually exclusive (i.e. improving one feature will have a negative influence on others).
• The DOTNAC NDT solution will offer inspecting a range of composite materials as well as detecting virtually all types of defects of an aircraft part during the production line, before and potentially on-site after the installation on the aircraft using a safe, contact-free, high resolution, easy to integrate in industrial facilities.
• The use of two fast imaging systems, completed with a novel THz signal and imaging processor based on the tomography principles and a novel THz signal and imaging processor based on the synthetic aperture.
In terms of cost reduction and efficiency, the new THz tool will not only help to detect but also to size a defect. That means, it will be possible to determine automatically (or semi-automatically) if that defect is acceptable or not. This will reduce the human interpretation, allowing a reduction of risk and time and an increase of the quality of NDT inspection. All together this means, on one hand, cost reduction, since less time will be necessary to inspect with the same or higher quality and, on the other hand, better efficiency due to the reduction of the human interpretation factor. All the conventional NDT techniques and some advanced techniques still have a high human interpretation factor.
Main dissemination activities
First of all, a DOTNAC website has been created with two aims :
• Dissemination of the project (public part).
• Storage of information and reports, for exchange between partners (private part).
The main architecture of the website resulted from iterations between all partners based on designs provided by Innov Support. The website has been hosted through the facilities at CIMNE. Since the operational launch of the website multiple updates (mainly maintenance and improvement tasks) have been implemented.
The private interface is accessible through a tightly controlled User ID and Password scheme, with specific access rights in function of the partner role in the project. The website URL is http://www.dotnac-project.eu
Innov Support has produced the 6 monthly versions of the PuDK (Plan for using and disseminating the Knowledge), integrating the dissemination and exploitation activities and plans for all the DOTNAC partners.
Leaflets of the project have been created to hand out during the multiple conferences and workshops the DOTNAC partners participated.
The DOTNAC project has generated furthermore interesting dissemination activities through the participation to specific conferences throughout the whole project and publication in journals. The full list of publications is provided in the next section.
During the second half of the DOTNAC project, 2 Industrial and Technology Followers Group (ITFG) workshops have been organised. Attendance of ITFG members was generated through the approach by all DOTNAC partners to their industrial and technological contacts. This has ensured an attendance of approximately 25 ITFG members for each ITFG workshop and ensured the participation of a series of dedicated contact entries.
Schedule of the ITFG workshops:
• The 1st ITFG workshop was organised in Bordeaux (France) at CNRS – Université de Bordeaux on October 17th 2012.
• The 2nd workshop was organised in Barcelona (Spain) at Applus+ - LGAI Technological Center on June 27th 2013.
Both ITFG workshops included the following major constituents:
• Presentations by the DOTNAC consortium of the activities and results achieved by DOTNAC at the time of the workshop.
• Demonstration of work on real system setup
• Question and answer (Q/A) session between the ITFG members and the DOTNAC consortium.
The outcome of each ITFG workshop (primarily the feedback collected during the Q/A sessions were then further addressed by all DOTNAC partners in a ITFG workshop follow up meeting between the DOTNAC partners to analyse all the feedback and comments that were collected during the ITFG workshop. Where possible the activities of the remainder of the project have taken into account these comments. The partners have implemented the specific comments with respect to the processing of the both ITFG workshops and the potentiality of dissemination and exploitation of the DOTNAC results.
Dedicated posters were produced for the 2nd ITFG workshop. All presentations, demonstrations and posters associated to both ITFG workshops have been uploaded on the public part of the DOTNAC website.
Exploitation results
Innov Support has produced the 6 monthly versions of the PuDK (Plan for using and disseminating the Knowledge). Innov Support has, through the PuDK, collected the description of the exploitable results so far produced by the DOTNAC partners and the associated potential exploitation routes for these results. In addition the actualised exploitation plans of the various project members have been also been captured.
Based on the data collected in the various PuDKs as well as the use of a Technology Implementation Plan form, Innov Support has elaborated an exploitation strategy and scenario. This will be accompanied with the elaboration of a business plan to be executed in the context of the DOTNAC project. This information has been further worked out in deliverable D8.43 (Precompetition survey – Future technology exploitation roadmap).
List of Websites:
www.dotnac-project.eu
Marijke Vandewal, Dr. Ir
Associate Professor
Royal Military Academy
Department CISS
Hobbema straat, 8
1000 Brussels
Belgium
Tel. : +32/(0)2/742.65.63
Fax : +32/(0)2/742.64.72
Email : marijke.vandewal@rma.ac.be
