Quantitative material characterisation of composites by ultrasonic scanning

·Home
·Table of Contents
·Materials Characterization and testing

Quantitative material characterisation of composites by ultrasonic scanning

Hans Erik Gundtoft
Material Research Department, Risų National Laboratory
DK-4000 Roskilde , Denmark
e-mail: hans.erik.gundtoft@risoe.dk
Contact

ABSTRACT.

Non Destructive Inspection is traditional used for finding and evaluation of defects and cracks in components and materials. However during several years ultrasonic precision scanning has also been used for measurement of material properties and quantitative characterisation of materials in different projects. All results in this paper are related to fibre reinforced compos-ite material with a plast matrix. Such composite materials are in nature more inhomogeneous than metallic materials and differences in a plate or component can be rather large. Porosity content determined by measured ultrasonic damping or attenuation in through transmission scanning is compared with destructive porosity determination. For plant fibre material a good correlation were found with porosity varying between 5 and 25%.

INTRODUCTION.

Fig 1: Ultrasonic damping measured by scanning carbon-composite /epoxy plates.

In the ultrasonic scanning examination of fibre composite plates an ultrasonic transducer can be used for measuring the damping along a raster pattern when moved relative to the plate material. In an earlier project the double through transmission technique (one transducer) was used. The plot resulting from these scans shows the ultrasonic damping values. In figure 1 the results from four different plates of carbon reinforced material are presented. One transducer measures the height of the echo from the backside of the plate. These values are shown in the plot with different colours. Yellow means low damping. This gives low numbers on the loga-rithmic colour scale. Thus the upper two plates (yellow, brown and red) have the smallest damping whereas the lover right plate (green /black) has the highest damping. The differences in porosity contents were forced using different autoclave pressure (0 to 5 atmosphere) during the curing process in the production. These measurements were performed in a BRITE EURAM project (called DESIR). Two different materials were used in the project. 12 plates were made (in two different materials) using different autoclaving pressure and afterwards ca.150 small specimens were cut from these plates based on the ultrasonic attenuation results.

The following parameters were recorded for the plates and measured on the small specimens:

Autoclave pressure during manufacturing (atmosphere)
Damping or ultrasonic attenuation (double through transmission) (dB)
Damping or ultrasonic attenuation per mm (dB/mm)
Porosity contents. Acid digestion percentage (%)
Porosity contents. Image Analyses percentage (%)
Thickness of the specimens (mm)
ILLS (Inter Laminar Shear Strength) (MPa)
Ultrasonic velocity measurements (m/s)

The large amounts of data were compared in different combinations trying to find possible correlations. The outcome was not too encouraging although some correlation was found. Especially the correlation between ultrasonic attenuation and porosity is of interest for the present work. The porosity contents were determined destructively using two different techniques, namely Acid Digestion, where the matrix material were dissolved in acid, and image analyses, where counting of pores in a cross section were used to determine the porosity percentage. Also here scattering were rather high both when using the same technique as well as from one technique to the other.

The best relationship between ultrasonic damping and porosity content (in the same material) are shown in figure 2. In this figure the average values for porosity determination found by acid digestion (or image analyses) for several small specimens are related to the (subjective judged) dB damping values for the same area from the colour in the scan Linear trend lines are drawn based on these data. The porosity determination made by image analyses show a linear correlation between porosity and ultrasonic attenuation in carbon fibre reinforced composite with epoxy matrix in the range between 0 and 10 %.

Fig 2: Porosity content and ultrasonic attenuation in carbon fibre epoxy composites measured by the double through transmission ultrasonic technique (with one transducer)

PRESENT MEASUREMENT AND RESULTS.

In a new project with plant fibre composites (Jute fibre in a polypropylene matrix) it was decided to investigate porosity content using the attenuation of ultrasonic signals measured by scanning of the plates and correlate the findings with the porosity content in the plates measured destructively. The plates were already produced and a porosity content has been determined on small samples for each plate. Other parts of the plates were used for making tensile test specimens. Thus it was assumed that the plate was so uniform that destructive porosity determination on a small specimen from the plate was representative for the whole plate. The fibres and the matrix are the same in all the plates but the type of Jute, the number of layers and the build up as well as the production parameters during fabrication varied between the plates. This contributed to the differences in porosity content.

Acid digestion could not be used for destructive porosity determination, and the conventional process for determination of porosity in glass fibre composites (burning the matrix material) was also out of the question. The plant fibre would burn. Therefore a new process was developed for the porosity determination. Here the specimen is weighted and then the matrix (poly-propylene) is dissolved in boiling Xylene. The plant fibres are filtered and weighted. Then the porosity content can be calculated. This analyse procedure is continuously refined.

Ultrasonic attenuation measurements by scanning were performed on the left over from selected plates where the rest of the plates had been used for destructive test and porosity analyses. In this project the through transmission technique (with one transducer on each side of the plate) was used instead of the double through transmission technique used in the previous mentioned project.

Scanning of the plates in water may represent a problem. Water could penetrate the composite and thus mask the porosities. Therefore each plate was cut in two halves, and one of these was protected with a thin water resistant layer on the surface. Also the weights of the plates were measured before and after the scanning to follow, if any weight gain during the scanning could be measured. Both measurements showed that protecting not was were important.

Fig 3: Relative good agreement between porosity and ultrasonic attenuation is found in the Jute-polypropylene composites.

Fig 4: Scanning results from 5 different plates in one scan. To the left the sketch show the different plates and to the right the scanning results are shown.
The ultrasonic attenuation values used in the figure 3 are subjective judged from the coulors in the scan, and the corresponding porosity values are the original values from small samples from the plate (mean value from 3 samples). In the plates used for the first measurements no destructive determined porosity values between 5 and 17 % are present (figure 3). Therefore the attenuation for other plates from the project with porosity content in this range were determined by ultrasonic scanning. Scanning plot of parts from 5 different plates (7, 10,13, 22 and 26) are shown in figure 4. Thus the range between 0 (3) and 25 % porosity's were covered. The porositys are the mean values from the plates and the attenuation's are the subjective judged values from the different scans. Dynamic damping values measured in dB from the colours in the plots must be added to the fixed gain in dB for each scanning so attenuation values from the whole dB range can be compared with the porosity measurements (Figure 5).

Fig 5: Jute-polypropylene plates. In this curve the values for the whole plates are compared. The whole range (4 to 25% porosity) is represented.

All plots showed that the composite plates have large differences regarding ultrasonic attenuation within the same plate as well as from plate to plate. Therefore this first correlation of porosity values and damping measurements only shows a cause correlation (R ² =0.75).

Fig 6: The second scanning of 3 of the 5 different plates in figure 4. Before this scanning the plates was marked with saw for positioning.

Therefore the plates were scanned again after they had been marked in the edge for positioning. (Figure 6). After scanning small areas (about 15*15mm) with almost the same damping were selected. The computer calculated the mean damping values.

Small specimens were cut at exactly the same place, and porosity analyses were performed on these specimens. Thus for the first time it was assured that the damping measurements and the destructive porosity measurements were for the same areas. The correlation (R² =0.89) between attenuation and porosity values for same positions are shown in figure 7. A linear correlation between porosity's and damping were found to fit better than a polynomial fit for most of the measurements.

Fig 7: Jute-polypropylene plates. Here the mean values for damping (calculated by the scanning program) are correlated with destructive measured porosity values from another part of the same plate. The whole range (4 to 25% porosity) is represented.

Is it possible to predict the porosity % from non-destructive ultrasonic damping measurement?

To examine this possibility some plant fibre plates were scanned. These plates were also manufactured from Jute fibres and polypropylene matrix, but the production procedure was based on manufacturing in a press instead of the autoclaving procedure.

Fig 8: A plant fibre composite plate in the water tank for ultrasonic scanning using the through transmission technique with one transducer on each side of the plate. The resulting plot is shown to the left on the figure. This plate was manufactured in a press.

Five different plates were scanned after a hole was drilled in each plate for positioning purpose. The photo and scan in figure 8 shows one of the plates with the corresponding plot from the scanning. The hole is clearly visible on both the photo and the scan. From each plate two small specimen were cut using the same selection procedure as previously described. Thus also here the damping values and the porosity values refer to exactly the same area.

Fig 9: These measurements are made on small specimens from 5 different plant-fibre composite plates made in a press. The damping values and the porosity value relate to exactly the same part of the plate. A good correlation (R ² =0.95) is found in the whole range from 5 to 27 % porosity.

Prediction of the porosity content from the non-destructive damping measurement based on the correlation between damping and porosity % found in the plates and specimens made by autoclaving gave misleading values. However the correlation between damping and porosity for the same type of plates made by the same production procedure is as good as it was for the autoclaving specimens over the same range of porosity's.

As a complement to all the previous measurement, which were performed in water in a scanning tank, tree specimen from the first plates (made by autoclaving) were scanned in air using the Through Transmission technique. These results had the same trend as found before. However transducers with much lower frequency were used and hence the damping values in general differs with 30dB from the values measured in water.

CONCLUSION

All results from our experience shows a relatively good agreement between damping measured by ultrasonic scanning, and porosity content.

In carbon composite 0 to 15 % porosity were examined whereas the range in plant-fibre composites covered 2 to 25%.
Results from one material cannot be used directly on another material, so calibration curves must be produced. The composite materials are so inhomogeneous that the destructive porosity measurements must be performed on exactly the same area as the ultrasonic damping measurement to have meaningful results.
The through transmission (one transducer on each side) gives better result than the double through transmission technique (measuring of the backecho with only one transducer on one side).
The whole range of porosity content (225%) cannot be covered by the 32 dB dynamic range in our equipment and therefor a combination of fixed gain and dynamic range must be used.
Within the same type of material a general calibration seems possible.
Also different sets of transducer (frequency, size, focussing) demands separate calibration curves.
Examination in air seems possible

ACKNOWLEDGEMENT

The author wishes to thank Kaj Kvisgaard Borum for fruitful discussions, and Erik Vogeley and Henning Frederiksen for performing the non destructive and destructive porosity measurements.