International Symposium on Computerized Tomography for
Industrial Applications and Image Processing in Radiology
March, 15 - 17, 1999 Berlin, Germany Proceedings BB 67-CD
published by DGZfP
|TABLE OF CONTENTS|
The main and tail rotors of helicopters are made more and more out of fiber composite materials because of their increased stiffness and strength to weight ratio compared to metallic structures. The earlier helicopter division of MBB, now Eurocopter Germany, has developed the first bearingless main rotor made of glass fiber reinforced materials within the 60´s. By the use of glass fiber reinforced materials the main rotor head of the BO 105 could be made without discrete flap and leadlag hinges. The newer helicopter rotor developments at Eurocopter Deutschland (ECD) have further simplifications of the rotors. The result has been that more and more functions have been integrated into the rotor blade structure, resulting in increasing structural complexity. Such complex structures of fiber composite materials, so e.g. the rotor blades of the EC135, can only be inspected restrictedly with the conventional NDT-methods, like ultrasonics or the conventional X-ray testing.
Particularly at thick-wall, monolithic composites the conventional NDT-test procedures fail. The X-ray Computed Tomography (CT) as an imaging test procedure closes this gap. CT is especially suitable for applications involving spatial analysis, differentiation, material identification, flaw analysis and structural quantification. This is made possible by the good spatial resolution and the very good resolution of density of the CT-measurements.
Subsequently the history of the CT at ECD, the measuring principle of the CT and the necessary technical equipment is described. The paper will be supported by examples, the usage of CT during the development of fiber-reinforced main rotor blades for optimisation the manufacturing techniques. In addition the methodology of the qualification of test procedure for determination the quality of fiber reinforced helicopter rotor blade structures during the manufacturing in series is presented.
|Fig 1: Comparison of the complexity - BO105 with EC135 rotor blade|
Computed Tomography is an radiographic NDT- method to locate and size planar volumetric details in three dimensions. A CT-scanner generates X-ray attenuation measurements which are used to produce computed reconstructed images of any desired part of an object. At the beginning of the 70´s Computed Tomography was developed for medical diagnostic. The inventors Hounsfield and Cormak have got for this the Nobel award for medicine in 1979. Already 1907 the German mathematician Johann Radon has put to this the mathematical foundation with the so-called Radontransformation. Since 1979 ECD has used conventional medical CT-scanners to test thick-walled composite materials of helicopter components.
The CT was used as a complementary NDT-method during the destructive testing of helicopter composite components to analyse the damage behaviour of the test specimen. Therefore it was possible to detect and knowing the manufacturing defects and anomalies inside the composite structure with their effects to the fatigue and the damage behaviour of the test specimen. Particularly during optimisation of the manufacturing of the thick-wall, monolithic main rotor blade structures the CT could successfully show its applicability. In the period of 1979 to 1992 approximately about 4000 tomograms with 100 fiber composites components were performed by CT-investigation. For the examinations medical CT-scanners of the companies EMI ( EMI 7020 ), General Electric Medical (GE 8800, GE 9800 and GE 9800 Highlight) and industrial scanners of the Bundesanstalt für Materialprüfung und -forschung (BAM) are used. Due to the good results, since 1993 ECD is using an own medical CT-scanner (GE HihLight Advantage) for development and quality control in the manufacturing process of composite helicopter rotor blades.
The CT applied within the development of new rotors made of fiber composites, in the series production of main rotor blades for quality assurance and maintenance. Since the CT-investigations started, 3000 test specimens were examined with approximately 190 000 tomograms. Eurocopter Deutschland is the first helicopter company which is using extensively the CT as a non-destructive inspection method in development, series production and maintenance of helicopter main rotor blades.
|Fig 2: ECD CT-Scanner|
Equipment and capabilities
ECD uses a medical CT-scanner (i.e. Fig. 2). An alteration of the tube energy or a customisation of the image reconstruction algorithm was not carried out. The image files will be transferred to Windows-NT computers for further image processing.
Following demands had to be realised for CT-scanner selection:
At the beginning of the selection of a CT-scanner usage of an industrial CT-scanner had been discussed. This would have the advantage that the scan techniques of X-ray transmission could be adapted optimally to the fiber composite material.
|Performance parameter of the CT equipment|
|GE Highlight Advantage|
Field size 96 - 480 mm
Image matrix 5122
Slice thickness 1.5, 3, 5, 10 mm
Number of the projections:
up to 3600 at 360° scan
Reconstruction: 11 - 16 sec per slice
Scan- and reconstruction time for 100
slices: approx. 20 min at 120 kV 170 mA
Scan- and reconstruction time for 100
slices: approx. 40 min at 140 kV and 170 mA the necessary cooling time of the X-ray tube extents here the scan- and reconstruction time
Type of X-ray tube: rotating anode
Power: max. 24 kV
Life time of the X-ray tube: approx. 40 000 shots
High voltage: 80, 120, 140 kV
Tube current: 10 to 200 mA
Focal point: 0.7x0.9
High contrast resolution: 0.38s mm @ 10% MTF (2sec scan/120kV/200mA)
CT number scale: -1024 to 3071 Units
E-glass: up to 2500; air -1000;
GFRP: up to 1400; CFRP: up to 500
Epoxy resin: ~ 150
The X-ray tube energy of 140 kV is sufficient to transmit GFRP-laminate of 50x50 mm up to 250x 60 mm by radiates to get images with a good quality for examination of the fiber structures. Primarily the short measurement- and reconstruction times (i.e. box above ) supplies economical examining times. This is reached by the high performance of 24 kVA of the rotating anode tube ensuring a high photon flux. Newer medical scanners use X-ray tubes with a performance of up to 56 kVA. To get a good quality of CT-images it is necessary that the attenuation profiles must consist of sufficient numbers of photons measured at the detectors. About three components are CT-checked at the moment per day. For a check approx. 140 min. are required. The components are mostly scanned with a tube energy of approx. 140 KV/170 mA.
For a CT-check with 120 slices approx. 40 minutes are necessary. This contains the scanning and the image reconstruction. The remaining 100 minutes are needed for the make-ready time, the image analysis and the documentation. Now a short note for the advantages of the industrial CT-scanner. The industrial CT-scanner is able to use various radiation sources . Components, made of materials of higher density, can be transmitted optimal. For high-resolution CT-inspections (spatial resolution of < 50 þm) 2-D and 3-D CT-scanners with micro focus X-ray tubes (250 kV) were developed. By comparison, industrial scanner has a 15 MV source to inspect objects up to 2.4 m in diameter. A summary of the application of industrial CT-scanners can be taken from the references [1,2,3,4,7]
After the CT inspection, all CT-images and evaluation reports will be stored by a NT-server on CD-ROM and archived into a juke box. The juke box stores 150 CD´s with approximately 150.000 tomograms. All data are accessible via the Eurocopter network. For image analysis the image processing software of the CT-equipment is used and on the NT-server the image processing tool Optimas þ is available. For the support of the tester a damage catalogue was built up. This contains all relevant component defects and damage tolerance limits. A high software integration guarantees frictionless and simple operation and economical CT- inspections.
|Fig 3: CT-network Measuring principle|
|Fig 4: Measuring principle of CT|
|dI / I = mds||(1)|
If the equation is integrated and the local dependence of µ is taken into consideration, the local dependence of the X ray intensity is:
The rest intensity of the X-ray, which is leaving the object, hits the detector. Simplified expressed, the detectors are scintillating crystal optics which the X-rays change into light. Light creates a voltage at the output of the post-connected photo multipliers (photocells with voltage gain) which is proportional to the intensity (I) to the transmitted radiation. To create cross-section images the fan-shaped X-ray beam rotates around the object (360 °). From several projection directions attenuation profiles of P(,n) are measured. P is the projection value at the detector position n and þ the angle of projection. After normalising and logarithmic operation of the equations (2) the projection value is given by:
|Fig 5: Rotation scan|
Fig 6: Scanning and image processing
Fig 7: Visualisation of calculated planes
Quantification of fiber composite structures
By the representation of the material cross-sections by the attenuation coefficients the material cross-section is available as a data record. So the material qualities becomes quantifiable. The medical scanners do not use directly the attenuation coefficients but the so-called CT-number. This number was established by Hounsfield. The Hounsfield scale standardises the reconstructed attenuation coefficients (þ) to the linear attenuation coefficient of water at a photon energy of 73 keV.
Therefore water has by definition the CT-number zero. Air has -1000 H, GFRP at 1200 up to 1400 H and CFRP approx. 300 up to 500 H.
The normalisation of the attenuation coefficients by the CT-number has the advantage, that materials of fiber composites become comparable. The CT-numbers are represented at the screen as grey value distribution. One grey value has always the same CT-number, independent of the maximum attenuation differences of the component. Therefore visually and quantitatively comparable CT-images are obtained. This is a quite essential prerequisite to analyse the CT-images automatically by image processing analysis.
In the Computed Tomography will be distinguished between the spatial (geometrical) and the density (contrast) resolution. In this paper the practical meaning of the resolution for the CT-check on fiber composites is presented. A closer description of the process engineering can be looked up at reference [4,6,13,16].
Spatial resolution is generally quantified in terms of smallest separation at witch two points can be distinguished as separate entities. The limiting value of the resolution is determined by the design and construction of the system and by the amount of data and sampling method. It must be distinguished between the spatial resolution within the scan plane (slice) and vertical to the scan plane dependent on the slice thickness. The spatial resolution affects the detectability of defects e.g. delaminations, fiber cracks or air pockets within fiber composites. The geometrical resolution within the scan plane is indicated by the MTF-curves.
|Fig 8: MTF-curve of the GE HihgLight Advantage |
Fig 9: Calibration standard, tomogram
Fig 10: Comparison of an original GFRP-prepreg cross-section with the tomogram
The MTF-curves describe the ratio of the contrast in the CT image to the contrast in the object with respect to the spatial frequency. The spatial frequency is the resolution value indicated at the threshold of 10% contrast difference. The ECD scanner reaches a value of 0.38 mm. A delamination with 0.38 mm can be resolved. By the good resolution of density delaminations of lower extensions may be detected. A calibration standard developed by ECD made of CFRP contains air-gaps, starting with 0.1 mm and increasing with 0.05 mm. The gap with 0.1 mm is still clearly visible. Why not establishing a higher resolution to detect microcracks into the fiber composites? Defect sizes less than 0.2 mm are detectable only by an industrial scanner which use a microfocus tube. This is possible on small samples and not on components which have the size of a main rotor blade (transmission cross-section up to 250x60 mm). A higher resolution is not essential because all strength relevant manufacturing defects and damages like delaminations are larger than 0.2 mm. Microcracks may not lead to a disastrous failure of the component. The main rotor blade is developed so, that damages must become indicated before failure - Fail Safe Principle-. The CT with medical scanners applies to this fact.
Contrast- or density resolution
The two Images (i.e.Fig. 10) show the good contrast resolution. Single laminates are recognised clearly. The grey values visible in the CT-images represent the attenuation differences (CT-numbers) caused by the local variable resin- and glass distributions.
This indicates a local unsteady distributed fiber volume content. The correlation between the CT-number and the fiber volume content is described briefly in the following. The characteristic material properties like the interlaminar shear strength, the modulus of elasticity and shear, the transverse contraction and the thermal expansion coefficient of a cured fiber composite are due to of the ratio of the mixture of fiber and epoxy resin.
|Fig 11: Correlation of fiber volume content and CT-number|
The fiber volume content F of fiber reinforced materials is defined as ratio of the fiber volume VF to the complete volume of the laminate VF, where VF is the volume of the fiber and VF is the volume of the epoxy matrix.
Fig 12: Influence of the beam hardening on the CT number
Fig 13: Comparative measurings on a GRFP-laminate
Measurements with GFRP and CFRP-fiber samples with various fiber volume contents have shown that there is a linear correlation between fiber volume content and the CT-number (i.e.Fig.11.),[ 2, 4]. Further examinations showed, that this will be only valid for small transmitted cross-sections. At larger transmitted cross-sections the beam hardening is noticeable [9,12,13]. However the resolution of density is sufficient. Inside the components the difference of the fiber volume content is well measured (i.e.Fig.12). So comparative measurements on series components can be performed. The absolute measurement of the fiber volume content is not possible. Pores in the laminate or fluctuations in the glass composition can distort the value. Serious process-, material alterations or manufacturing defects will be visible and quantifiable in the tomograms. Only the normalising of the attenuation coefficients allows the comparison of the series components with the master component (component without defects) by CT. The diagram (i.e.Fig.13) shows the differences of the standard deviations of the CT-numbers of a faulty CFRP-laminate (wavy fibers) to a defect-free laminate. The relative alteration, which indicates the defect within the laminate could be seen clearly. The standard deviation is a measure for the uniformity of the resin fiber distribution within a laminate cross-section.
The following listing gives a summary of defects which could be within monolithic fiber composite structures:
In a wavy laminate the load will not be carried by the fibers, therefore the resin matrix has to take over more loads. As a result high shear stresses will be introduced into the resin matrix - the resin fails. The matrix will change so strongly that the fibers will have space for bending. The fibers break.
|Fig 14: Failure of a wavy laminate||Fig 15: Wavy GFRP-laminate|
Fiber cracks are very difficult to detect in the tomograms. In the rotor blade neck cross-sections the fibers of the UD-laminate placed in the rotor blade radial direction (direction of the centrifugal force). At this point the spatial resolution is equal to the minimum thickness (1.5 mm) of the X-ray beam.
|Fig 16: Fiber crack in UD-loop after a|
|Fig 17: Delaminations|
Inspection of bondings and laminate transitions
An adhesive foil of a bonding becomes visible as the distribution of resin in the joint gap (area between the bonding parts). The compound quality (cohesion) between the bonding parts cannot be checked by CT. If it is possible the laminates should be manufactured so, that various prepreg laminates be cured together (polyaddition at the same time), so it is possible to obtain an integral meshing for the various prepreg laminates. If a good integral meshing is available the laminate transitions can be hardly recognised in the CT-images.
|Fig 18: Bondings|
Detectability of foils and films
Foils made of synthetic materials used in the production process can not be detected directly by CT. The attenuation which is caused by a foil corresponds to the attenuation by the resin matrix. Therefore they are not visible. A foil is detectable indirectly due to a stronger concentration of resin film on the foil.. Foils should contain elements which will absorb the X-rays to detect the foils better by means of CT. Here for the future the need for further actions is obvious.
Fig 19: CT on helicopter rotor blades
Fig 20: Comparative measurings on a GRFP-laminate
The life cycle of a main rotor blade can last 20 years or more. The life cycle of the helicopter rotor blade is divided into three phases, the development, the series production and the operational phase with maintenance. In front of this background the CT must be seen as non-destructive test method in the application to the rotor blades. CT is used as a test method within all three life phases. For this purpose it is necessary to qualify the CT inspection for these applications. This starts during the development phase very early.
CT during the development
The fiber composite rotor blade is a component whose strength properties depends on the design and the production quality. The sizing and the manufacturing technology must be optimally co-ordinated . Thin prepreg layers are inserted by hand into a blade mould. The mould must be filled with the prepreg layers so, that within each fiber laminate of the cross-section of the rotor blade the fiber volume content is 59%. The laminate cuts and the map to put the laminates into the mould must be arranged so, that an optimal filling of the blade mould is obtained. After closing the mould the component is cured under pressure and temperature. Nonuniform filling rates result in difference pressures in the mould. The prepreg layers can shift and wavy laminates with low-quality strength properties will be produced. These laminate deformations inside the fiber composite structure are usually not visible from outside. Only the CT makes wavy laminates visible.
Optimisation of the production process
The quality of the composite component depends on an optimal manufacturing process.
Earlier, in the prototype production the first manufactured helicopter rotor blades have been cut. Only in such way it was possible to get information about the inner distribution and set-up of the fiber laminate structures and to co-ordinate optimally the design process of the component and the productional process one to the other. Now the cutting of the components can be avoided by the use of the CT. This exhibits a great benefit if one considers that a rotor blade represents a considerable value. Each rotor blade will be CT-checked to detect production deviations after the curing process. The tomograms are visualising the manufacturing defects and deviations in the cross-sections of the fiber component. Sometimes surprising effects become visible by the CT which must be understood. Therefore an evaluation of the tomograms of new components and its information contents should always happen together with the engineers of mechanical design, production, structural and test. This guarantees an efficient interpretation of the tomograms and contributing to this, understanding the productional process. Therefore the production process can be optimised so, that the desired component qualities will be reached. During the optimisation phase must be distinguished between obvious manufacturing defects like wavy laminates within high loaded fiber layers which will result in a failure of the component and production deviations within other less loaded composites. Also production deviations occur which can only be removed by considerable cost. The judgement of visible production deviations within the tomograms with respect to structural efficiency is not often possible.
|Fig 21: Wavy laminates removed||Fig 22: Tapered laminates optimised|
|Fig 24: Fiber cracks and delaminations in a wavy UD-laminate|
|Fig 23: Fatigue test of the EC135 flexbeam in the blade bending machine|
CT during the series production
Only by the tight combination of the non-destructive CT-inspection with the destructive testing, during the fatigue tests, the judging of the CT-tomograms can be guaranteed. This methodology qualifies the CT inspection as a non-destructive test method for the quality assurance of the rotor blade in the series production. From the knowledge's of the development- and certification tests a CT-test instruction procedure for inspection was created for the series inspection of the rotor blades. The CT-test instruction contains descriptions of defects with permissible defect limitations. If new production defects are discovered, the rotor blade with the new manufacturing defects will be stored for some time in a closed store until the design engineers will have decided whether a release can be given by theoretical analysis or not. If the manufacturing defects cannot be removed by theoretical analysis or repair, the proof can be furnished only by a destructive testing of the component to certify that the manufacturing defect does not have any life time restricting effects. The destructive testing makes no sense with a component with a defect which occurred for the first time. Therefore the rotor blade must be saved in to a closed quantity store and one has to wait whether the manufacturing defect will occur further on systematically. If the defect will be present still on course of production, all sources of errors are to be removed in the production. Now a destructive testing is economical and useful to assess the effect of the defect. If the proof of the fatigue test furnishes that the defect does not cause any life time reductions, the rotor blade will be released again for flying. If an early failure will occur during the fatigue test, all rejected rotor blades with these new defects will be declared to be scraped. The most prior-ranking target will be the removal of the source of the production defect for a defect-free delivery of the blade to the customer.
CT during operational and maintenance
During the operating time of the rotor blades these are checked usually by visual inspections to external damages. The CT will be used if unusual events have been occurred and rotor strains were evoked which were above the limit loads, by examples, like loading of non-rotating rotor by over-flight and gust loads, at slop landings or manoeuvres on the ground with power or by foreign object damages or accidents. At such events the CT is used for inspection of the inner structure of the rotor blade attachment on possible damages. A further application of CT is its monitoring support during the residual strength test of helicopter rotor blades. Rotor blades with very high flight hours are brought back from the customer and the residual strength is determined in the fatigue test in comparison with the new rotor blade. The rotor blade attachment is checked by CT at the beginning, during and after the fatigue test. By that it can be found whether damages have been occurred within the structure by operational loadings and their effect on the fatigue of the structure.