，，，，

Aerospace Research Institute of Materials and Processing Technology，Beijing 100076，P.R.China

Abstract:Carbon-fiber reinforced polymer composites have been widely used to achieve the light-weighted design and high performance due to superior performance.Internal defects in the composite materials are the main factors that determine their performance，which makes reliable and effective detection methods of internal defects essential.Nondestructive testing（NDT）methods are the most widely-used way due to their tremendous advantages.Though the theoretical background is found，experimental results could be quite complicated and confusing，especially for composite materials with complex defects characteristics.In this paper，experimental study on internal defects in composite materials based on the time of flight（ToF）are investigated.The Gaussian echo model and the parameter estimation methods are established to build a theoretical model for measurements.Then，the distance amplitude correction（DAC）method is proposed to effectively improve the signal-to-noise ratio（SNR）and to reduce distortion of the signal during measurements.Finally，the ToF is adopted to determine depth of internal defects.Experiment study is conducted to investigate the porosity defects and the anti-impact performance of composite materials，as well as defects in objects with various thicknesses.Experimental results show that the proposed method is quite helpful for obtaining the intuition and deep understanding of internal defects，thus contributing to the determination of product performance and its improvement.

Key words：ultrasonic non-destructive testing；time of flight；porosity defects；anti-impact performance；distance amplitude correction（DAC）method

0 Introduction

Composites have been widely used in aircraft to achieve the light-weighted design and high perfor?mance［1］due to their superior performance，such as high specific strength，good corrosion resistance and low weight［2］.One of their typical application is in aero engines of aircraft.The carbon-fiber reinforced polymer laminates are adopted in the rotor casing of aero engines to prevent malfunctions caused by the blade rupture，which could seriously threaten air?craft safety［3-5］，if the rotor blade fails unexpectedly during operation at a high speed.In this regard，the compressor and turbine casing are usually made of carbon fiber reinforced resin composite materials due to their outstanding anti-impact performance.To achieve satisfactory anti-impact performance，the ply order，the ply angle and hybrid modes in the interwoven laminates need to be optimized［6］，which requires the mapping relationship between internal defects and qualified anti-impact performance.Dif?ferent types of internal defects may occur in compos?ite materials.An important source of error is the forming process in manufacturing.In addition，struc?tural damages will inevitably occur in the process of use due to the influence of environmental factors，stress concentration and fatigue load［7］.These de?fects in the composites will seriously deteriorate the properties of the composites，resulting in mechani?cal failures and even safety problems［8］.Therefore，the establishment of reliable and effective detection methods is critical to the application and develop?ment of composite structures［9］.

In the testing of composite materials，non-de?structive testing（NDT）methods are the most wide?ly used due to their tremendous advantages［10］.The NDT methods involve the identification and charac?terization of damages on the surface and interior of materials without altering the original attributes or harming the object being tested［11］.Traditional NDT methods include ultrasonic testing，radio?graphic testing，magnetic particle testing，eddy cur?rent testing and penetrant testing［12］.Among them，the ultrasonic testing has the advantage of high sen?sitivity to defects，like cracks.Ultrasonic penetra?tion is strong，but harmless to materials and human body.Therefore，the ultrasonic method has been widely concerned in the field of NDT.

Great efforts have been made to carry out stud?ies on ultrasonic NDT of composite laminates by us?ing the transmission method and the pulse-echo method［13-14］.The transmission method is based on the energy change after the pulse wave penetrates the specimen［15］.The advantage of this method is that there is no blind spot in the test and it is suitable for detecting materials with a large attenuation coef?ficient.The disadvantages of this method are low de?tection sensitivity and poor ability to extract internal defect depth information.The pulse-echo method transmits the pulse into the internal structure of the specimen，and detects the internal defects of the specimen by observing the reflected wave［16］.Repre?sentative methods based on this principle include the bottom wave method and the pulse-echo method［17］.The bottom wave method uses the variance of the bottom wave amplitude to estimate the information of internal defects.This method is usually used to detect internal defects with small volume but large density［18］.As quantitative information of internal defects cannot be obtained，it can only be used as an auxiliary means of testing.The defect echo method uses the amplitude of reflected wave to determine the size and location of internal defects，and uses the transmission time of reflected wave to deduce the depth information［19］.The pulse-echo method quantifies internal defects according to the amplitude of reflected wave.But for suspicious waves，valu?able information may be missed in the experiment.In order to improve the efficiency and accuracy of the experiment，the distance amplitude correction（DAC）method［20］was introduced to improve the detection performance.

Even though the theoretical background of ul?trasonic NDT methods of composite laminates is found，experimental results could be quite compli?cated and confusing，especially for composite lami?nates with mixed layout.An advanced method［21］was proposed based on eddy current pulsed ther?mography（ECPT）to reconstruct the layers’orien?tation and to estimate the thermal and electrical con?ductivity of multiple CFRP samples.This is achieved by implementing an iterative inversion pro?cedure that processed experimental measurements together with finite element simulations of the ECPT data.Experimental results show that the time of flight（ToF）［22］of thermal waves can be used to reconstruct the fibre orientations and param?eter-based inverse reconstruction demonstrates bet?ter accuracy than the feather-based approach in terms of estimation of layer orientation.

Further，there is another reason to study the experimental results in depth.Ultrasonic detection signals are always contaminated by noises，which can be divided into two categories：Gaussian white noises caused by circuits and measurement systems，and structural noises caused by the scattering of sound in the crystal microstructure of the materi?al［23］.Noise is unavoidable and difficult to quantify，which makes in-depth experimental research diffi?cult and essential.

In this paper，the theoretical analysis and ex?perimental study on testing of internal defects of composite materials by using NDT methods are in?vestigated in-depth and in detail.First，the Gauss?ian echo model is established to represent the ultra?sonic signals.Then，the time of flight（ToF）in the Gaussian echo model is estimated accurately by ef?fective methods.Besides，approaches on how to im?prove signal to noise ratio（SNR）and how to re?duce the distortion of the signal in the ToF measure?ments are investigated.Principles and drawbacks of two typical ultrasonic methods that are the most-of?ten adopted in detecting the internal defects of com?posite materials are explored.To improve the de?tecting performance，the distance amplitude correc?tion method is proposed.Finally，experimental study is conducted to investigate the porosity de?fects，the anti-impact performance of composite ma?terials and objects with various thicknesses by using the proposed methods.Corresponding experimental results are analyzed in detail and valuable conclu?sions are derived in the end.This study provides useful guidance in the design and optimization of composite materials.

1 Theoretical Background

1.1 Gaussian echo model

An ultrasonic signal is commonly represented by the Gaussian echo model（GEM）［24］，which is expressed as

whereGidenotes the Gaussian echo signal corrupt?ed by noise；sithe Gaussian echo；x（5）the vector with five element parameters；withe white Gauss?ian noise with a variance of σ2；Lthe signal length；Tthe time of flight；fcthe center frequency；βthe bandwidth factor；Mthe magnitude；andφthe phase.

1.2 Time of flight

Once the material has been subjected to damag?es［25］，the damage can be detected by the reflected ultrasonic wave because the distance decay of re?flected wave or its frequency distribution will be changed，which will change the ToF in the GEM to a great extent.The accuracy of defect measure?ments is determined by the error of ToF estimation.To acquire precise model of the internal defects，ac?curate estimation of ToF of an ultrasonic signal is essential.In ultrasonic NDT，ToF measurements are becoming increasingly sophisticated and yield better performance.The ToF estimation is conven?tionally performed by gating and peak detection［26-29］.The gating and peak detection method can estimate ToF effectively when dealing with a clean and undis?torted signal.When the signal is noisy or distort?ed［30］（Noisy means the noise level is close to or higher than the signal amplitude，and distorted means time-overlapping ultrasound signals are often encountered whenever layer thickness is small，or distance between reflectors is short），the gating and peak detection method yields poor ToF estimation results.Other conventional methods such as overlap and phase slope methods suffer from the same prob?lem［31-32］.In this regard，experimental study on how to improve the SNR and reduce the distortion of the signal in the TOF measurements is of great impor?tance.

1.3 Principles of typical ultrasonic methods

Typical ultrasonic NDT methods include the transmission method and the pulse-echo method.In the transmission method，double transducers are placed in both sides of the composite materials，as shown in Fig.1.One transducer is used as the trans?mitting transducer，and the other is used as the re?ceiving transducer.The transmitted ultrasonic wave propagates to the receiving side via the transmitting side，while the defects are detected and evaluated by the magnitude attenuation of the transmitted ul?trasonic wave.

Fig.1 Schematic diagram of the ultrasonic transmission method

In the pulse-echo method，a transducer is used as both the transmitting and receiving transducers，as shown in Fig.2.The transmitted ultrasonic wave propagates to the bottom of the reflected surface through composite materials，and then reflects back to the incident surface.The defects are detected and evaluated by the amplitude，phase and time transit of the reflected wave.

Fig.2 Schematic diagram of the ultrasonic pulse-echo method

Although both methods are based on the char?acteristics of wave energy attenuation to assess the porosity defects，the pulse-echo method has higher sensitivity to porosity defects compared with the transmission method［33］.

1.4 DAC method

Ultrasonic NDT methods quantify the internal defects based on the amplitude of the reflected wave.Thus，valuable information might be missed in experimental study for suspicious waves，and in?ternal defects located in the deeper position are hard to find.To help acquiring the comprehensive infor?mation of internal defects，the DAC method is intro?duced，which aims to improve the SNR in the test?ing and to reduce the distortion of the signal in the TOF measurements.The DAC method is to com?pensate for the fact that the pulse-echo response of a reflector will decrease as the distance of the reflector from the ultrasonic transducer increases.The DAC method provides a means of establishing a graphic reference level sensitivity as a function of sweep dis?tance.In this method，the DAC curve is generated by plotting the amplitude of a known calibration re?flector at different distances from the transducer，thus ultrasonic signals from the same reflected sur?face will have different amplitudes at different dis?tances from the transducer.The use of DAC curve allows signals reflected from similar discontinuities to be evaluated where signal attenuation as a func?tion of depth has been correlated.

Typical DAC curve is illustrated in Fig.3.Thex-coordinate represents the ultrasonic path，and they-coordinate represents the amplitude of the echo.The highest points of different amplitudes corre?sponding to different ultrasonic paths are connected to construct a smooth curve，which is called the bus line of DAC.The DAC curve is used to distinguish the amplitude change of the reflector with the same size but different distances.Normally，after the DAC curve is plotted，the echo peak generated by the reflector of the same size will be on the same curve，regardless of the reflector’s position in the material.Similarly，a reflector smaller than the size indicated by the curve will fall under the curve，while a larger one will be located above the curve.Taking the bus line as the reference，dB values of the waste line，the quantitative line and the evalua?tion line are determined according to the correspond?ing flaw detection standards.Methods to construct the bus line of DAC are devided into three catego?ries：The linear interpolation，the least square fit?ting and the Lagrange interpolation.These three methods are shown in Fig.3，in which the linear in?terpolation is the simplest，but the bus line is not smooth.The least square fitting can yield a smooth bus line but needs to be corrected a priori.The La?grange interpolation can yield smoother curve with smaller errors，thus it is taken as a priority in con?structing the bus line of DAC.

Fig.3 Bus lines of DAC constructed by using three methods

The amplitude of the reflected echo at different depths is compensated by the decline trend of the DAC curve along the depth direction.After depth compensation of echo amplitudes，the echo ampli?tude of reflector of the same size is independent of its depth in the tested materials，thus guarantee that equivalent defects have the same sensitivity at differ?ent ultrasonic path distance.When the gain condi?tion of the detector is unchanged，it is more advanta?geous to find defects located in the deeper position of the material.

2 Experiment of Porosity Flaws in Composite Laminates

Porosity flaw is the main type of internal de?fects existing in the composite materials.Compared with the macro flaws such as lamination and inclu?sion，the porosity flaws have quite different charac?teristics in both size and dispersion distribution［34-36］.First，the porosity flaws have sizes of 10—100 μm，which are much smaller than the regular macro flaws.Second，the porosity flaws are scattered in?side the composite materials.From these two as?pects，detection methods of porosity flaws should differ from detection methods of macro defects and should be paid much more attention to.

2.1 Experiment setup

In this paper，the ring with flange composed of laminates that are made by woven fabric is studied as an example.The flange plays a critical role in the mechanism，any internal defect can cause the invalid?ity of the mechanism，even safety problems.From this respect，comprehensive information of the inter?nal defects should be derived.As presented in Fig.4，the inner diameter of the flange is 1 110 mm，the outer diameter of the flange is 1 200 mm，the thickness of the flange is 10 mm and the height of the ring is 300 mm.

Fig.4 Sketch of the ring with flange

2.2 Detection of internal defects

To detect the flaw existed in the flange，the transmission method is adopted due to its high effi?ciency.Two transducers are used in this experi?ment.One transducer is placed on one side of the flange to transmit the ultrasonic wave，while the other transducer is placed on the other side of the flange to receive the ultrasonic wave that penetrates the flange，as illustrated in Fig.1.The internal de?fects are studied by analyzing the energy change be?tween the transmitted ultrasonic wave and the pene?trated ultrasonic wave.The experimental results which adopt amplitudes to determine in-plane loca?tions of internal defects are displayed in Fig.5.Darkcolored areas represent internal defects where pene?trating signals have low energy and light-colored ar?eas represent intact regions where penetrating sig?nals have high energy.

Fig.5 Testing result in the experiment using the trans?mission method

Type A pulse testing is conducted to explore the characteristics of internal defects.The results are given in Fig.6.It can be inferred through experi?ences that the internal defects are disperse porosity.

Fig.6 Experimental results of Type A pulse reflection ultra?sonic testing

Therefore，the ToF of the flaw echo is mea?sured by using the pulse-echo method and the re?sults are shown in Fig.7.Dark-colored areas repre?sent internal defects with short transmission time of the flaw echo，which means the internal defects are close to the detection surface.And light-colored ar?eas represent internal defects with long transmission time of the flaw echo，which means the internal de?fects are far from the detection surface.The blank indicates no internal defects.The three-dimensional model of internal defects are constructed and shown in Fig.8.

Fig.7 ToF of the flaw echo in the sample of the flange

Fig.8 Three-dimensional model of internal defects

2.3 Analysis and discussion

To verify the three-dimensional model of inter?nal defects，the flange is dissected to find out its me?tallographic phase，and the result is given in Fig.9.As can be seen，the section has two porosity de?fects，and the distance from the bottom is 2.4 mm and 7.4 mm，respectively，which match the detec?tion result in Fig.8 to a good extent.

Fig.9 Metallographic phase of the flange after dissection

3 Experiment on Anti?impact Per?formance of Composite Materials

The compressor and turbine casing are usually made of carbon fiber reinforced resin composite ma?terials for their advantages in anti-impact perfor?mance［37-39］.GEnx engine first adopted two-dimen?sional three-axis braided carbon fiber reinforced res?in matrix composite fan casing［40］，as shown in Fig.10（a）.Then，this technology was applied to LEAP engine and improved into three-dimensional weaving process［41］，as shown in Fig.10（b）.Due to the existence of fiber reinforcement in the direction of thickness，3D braided/woven structure compos?ite materials can effectively resist impact layering and can directly shape parts with complex structure，which is an important application direction of the new generation of aero engines.From this respect，it is of significance to study the anti-impact perfor?mance of composite materials.

Fig.10 Aero engines adopting carbon fiber reinforced resin composite materials

3.1 Experiment setup

Ultrasonic NDT method is utilized to analyze the anti-impact performance of the rotor casing of aero engines.Typical layout of composite materials used in the casing is given in Fig.11.For fully com?parison and in detail data analysis，12 types of resinbased composite laminates are treated as specimens，which has different anti-impact performance due to their different ways of ply angles and ply orders.The experimental scheme is illustrated in Fig.12.The ad?opted specimen is of 250 mm×250 mm×8 mm，and hit by blocks made of Titanium alloy.The clamping width of each specimen is 25 mm，and the preload of bolts during each experiment is guaran?teed to be uniform by the torque spanner.

Fig.11 Typical layout of the rotor casing of aero engines

Fig.12 Schematic diagram of experiment setup

3.2 Detection of internal defects

The transmission method is firstly adopted due to its high efficiency.The results of batch #4 are giv?en as an example in Fig.13.It can be seen that inter?nal defects exhibit all dark colors.It is hard to distin?guish the dissipation of these internal defects.

Fig.13 Results detected by the transmission method

To derive the comprehensive information of this problem，the ToF of the flaw echo is measured by using the pulse-echo method.Since there are multiple layers of composite laminates in the speci?men，echoes are generated between two layers.Due to the expansion of the ultrasonic beam and the at?tenuation of materials，equivalent defects at differ?ent ultrasonic path distance may cause echo with dif?ferent magnitudes；and echos from interfaces be?tween layers may have overwhelming magnitudes than echos from internal defects，which greatly re?duces the SNR in detecting，as shown in Fig.14.

Fig.14 Comparison of echos between layers from inter?faces and from internal defects

To solve this problem，the DAC curve is intro?duced to calibrate each batch so that echos from in?ternal defects at different depths have the same mag?nitudes.In this paper，the amplitude of the flaw echo is set to be 80%，as shown in Fig.15.It can be seen that flaw echoes have much higher magnitudes even when they are located at long distances，which greatly improve the SNR and make the detection more precise.DAC curves of all batches are shown in Fig.16.

Fig.15 Calibration by setting amplitude of the flaw echo at different distances to be close to 80%

Fig.16 DAC curves of all batches

After calibration，it can be seen that echos from internal defects have overwhelming magni?tudes than echos from interfaces between layers，as illustrated in Fig.17.

Fig.17 Comparison of echos between layers from internal defects and from interfaces

The impact tests are conducted at different batches，and different batches exhibit different antiimpact performance.This can be attributed to the difference in ply orders and ply angles of each batch.

3.3 Analysis and discussion

After the impact，type C testing is used to ob?tain a more intuitive display of the internal defects distribution.ToF in all specimens are derived.From the results，it can be found that different batches ex?hibit different anti-impact performances due to their different ply orders and ply angles.The ToF in dif?ferent specimens in #4 batch is demonstrated in Fig.18 as an example.It shows that even within one batch，different specimens exhibit different anti-im?pact performances under different impact speeds，which proves the complexity of the anti-impact mechanism and the necessity for in detail experi?ments studies.

Fig.18 ToF of the six specimens under different im?pact speeds in batch #4

4 Experiment on Internal Defects in Composite Laminates with Various Thicknesses

When detecting the internal defects in compos?ite laminates，it is quite common to deal with com?posite laminates with various thicknesses.An exam?ple is an object with three different thicknesses.Since the thickness varies along the direction of ul?trasonic waves，different detection sensitivities are necessary in detecting the object with such character?istics，which brings great difficulties and time cost into test.An alternative way is to use constant sensi?tivity during the test，while post-process the mea?sured data to adapt to different thicknesses，which is also inconvenient and error prone.

In this paper，the proposed testing method is al?so adopted to deal with this problem，which elimi?nates difficulties caused by different sensitivities or post-processing.The results are demonstrated in Fig.19.The darker color than the surrounding area indicates internal defects that has been detected and has been illustrated by red and yellow circles for clear demonstration.

Fig.19 Results of object with three different thick?nesses by using the proposed method

5 Conclusions

Theoretical analysis on basis of Gaussian echo model and the parameter estimation methods is first?ly proposed to explore the information of internal de?fects in composite materials.To improve the detec?tion precision and efficiency，the DAC method in combination with the ToF measuring method are in?troduced to derive a comprehensive and deep insight of the internal defects in composite materials.Con?sidering the difference between theoretical analysis and practical detection，in-detail experimental stud?ies based on the proposed method are carried out to detect the common porosity defects in composite materials and to measure the anti-impact perfor?mance of the layered composite materials.Mean?while，the proposed method is used to detect ob?jects with various thicknesses along the direction of the ultrasonic waves.Experimental results show that the proposed method is quite helpful to obtain the intuition and deep understanding of internal de?fects，thus contributing to the determination of prod?uct performance and its improvement.Future re?search will focus on improving the detection preci?sion and efficiency while retaining its feasibility and value in practice，and further quantitative analysis and comparison will be provided in the future study.