The Pennsylvania State University. The Graduate School. Engineering Science and Mechanics Department FURTHER DEVELOPMENT OF A LOW CHANNEL PHASED ARRAY

The Pennsylvania State University The Graduate School Engineering Science and Mechanics Department FURTHER DEVELOPMENT OF A LOW CHANNEL PHASED ARRAY FOR MATERIAL DEFECT IMAGING A Thesis in Engineering Science by Mark Adam Bohenick 2008 Mark Adam Bohenick Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science May 2008

ii The thesis of Mark Adam Bohenick was reviewed and approved* by the following: Bernhard R. Tittmann Schell Professor of Engineering Science and Mechanics Thesis Advisor Clifford J. Lissenden Associate Professor of Engineering Science and Mechanics Albert E. Segall Associate Professor of Engineering Science and Mechanics Judith A. Todd P. B. Brenneman Department Head Department of Engineering Science and Mechanics *Signatures on file in the Graduate School

iii Abstract As lighter, stronger, and more affordable materials are available for industrial applications in hazardous environments, the effect of the environment on the material during operations must be studied. A material must perform adequately and degrade gracefully. Sudden failure is unacceptable. Current methods to qualify materials are expensive and do not provide real-time results. A method to monitor material characteristics and degradation while the material is in the hazardous environment is desirable. Development of a Non-Destructive Evaluation device that can function inside a high temperature, high pressure environment would provide a tool for monitoring a material specimen in real-time. A limitation is that the number of inputs to the environment should be minimized. The device should provide real-time data on the specimen s degradation. Surface and interior defects such as film growth, blistering, cracking, and corrosion should be able to be characterized. In addition, geometric changes such as shape, size, and warping should be monitored. An ultrasonic linear phased array provides the capability to examine a specimen without mechanical scanning. Linear phased arrays are essentially multiple ultrasonic transducers that are arranged along a line. Electronic time-delays provide the ability to scan a beam of acoustic energy across the specimen. Reflected pressure waves are received by the array and are used to produce real-time images of the specimen. In order to limit the number of inputs to the environment, a stepped linear phased array design was used. A physical offset is designed into the device between multiple linear phased arrays. Wire inputs to the environment connect an element on each step of the array.

iv The physical offset allows for the signals of each step to be separated in time. This allows additional sets of elements so that additional portions of a material specimen can be investigated with the same number of wires. Three prototype devices were manufactured. Each has four channels; four piezoelectric elements on each step forming a linear phased array. Prototype 1 has four steps and its elements have a central frequency of 2.25 MHz. Prototype 2 has four steps and a central frequency of 10 MHz. Prototype 3 has six steps and a central frequency of 10 MHz. The ability of the prototypes to monitor degradation of a material specimen is examined through bench-top experiments modeling defects and geometric changes to a material specimen. Beam steering and focusing were simulated using a ultrasonic phased array simulation program, Field II, to determine optimum array dimensions. The current prototypes cannot sufficiently monitor a specimen for defects. Large surface defects can be detected but smaller individual surface defects cannot be sufficiently resolved. The prototypes were unable to detect internal defects. Warping or bending of the specimen in the vertical plane has been able to be examined by the prototypes. The stepped linear phased array design parameters can be optimized to develop a device that can provide all the functionality desired. The higher frequency of Prototypes 2 and 3 did not show a marked increase in resolution or effectiveness. Sparse arrays and data analysis may be required to further optimize the stepped linear phased array design.

v Table of Contents List of Figures Acknowledgements vii ix Chapter 1 Introduction 1 Chapter 2 Literature Review and Theory 4 2.1 Piezoelectric Materials 4 2.2 Linear Phased Arrays 4 2.3 Array Element Spacing 6 2.4 Time of Flight 8 2.5 Near-Field and Far-Field 8 2.6 Beam Divergence 10 2.7 Frequency Considerations 11 2.8 Resolution 12 2.8.1 Lateral Resolution 12 2.8.2 Axial Resolution 13 2.9 Linear Phased Array Pressure Field Equations 14 2.10 Array Optimization 17 2.11 Additional Literature 19 Chapter 3 Methodology 20 3.1 Devices 20 3.1.1 Stepped Linear Phased Array 20 3.1.2 Omniscan MX System 23 3.2 Experiments 23 3.2.1 Experimental Setup 24 3.2.2 Thickness Measurement 25 3.2.3 Surface Absorbing Defect 26 3.2.4 Surface Scattering Defect 26 3.2.5 Single Surface Crater 26 3.2.6 Interior Defects 26 3.2.7 Bending of a Specimen 27 3.2.8 Effect of Temperature 28 3.3 Simulations 29 Chapter 4 Results 31 4.1 Experimental Results 31 4.1.1 Signals 31 4.1.2 Surface Absorbing Defect 34 4.1.3 Surface Scattering Defect 36

vi 4.1.4 Single Surface Crater 36 4.1.5 Interior Defects 39 4.1.6 Bending of a Specimen 39 4.1.7 Effect of Temperature 40 4.2 Simulations 41 4.3 Calculations form Linear Phased Array Theory 45 4.3.1 Inter-element Spacing 45 4.3.2 Application of Other Array Optimization Techniques 46 Chapter 5 Conclusions 47 Bibliography 50 Appendix A Non-Technical Abstract 52

vii List of Figures Figure 1: Diagram of Array Element Spacing 6 Figure 2: Near-Field and Far-Field: Distance from Transducer vs. Nominal Amplitude. 9 Figure 3: Diagram of the Beam Divergence Angle of a Transducer. 11 Figure 4: Diagram of 1.5 and 2.5 Cycle Pulses. 13 Figure 5: Table of Calculated Axial Resolution 14 Figure 6: Diagram of Element and Point P(r, θ ) in the Pressure Field. 15 Figure 7: Photographs of Prototype 1. 20 Figure 8: Schematic Drawing of Stepped Array Device 21 Figure 9: Photograph of Prototype 2. 22 Figure 10: Photograph of Prototype 3. 22 Figure 11: Photograph of the Omniscan MX System. 23 Figure 12: Photograph of Experimental Setup for Prototype 1. 24 Figure 13: Photograph of Experimental Setup for Prototypes 2 and 3. 25 Figure 14: Bending Experiment Setup. 28 Figure 15: Effect of Temperature Experimental Setup. 29 Figure 16: Comparison of Full Array Waveform and First Step Waveform. 31 Figure 17: Comparison of Single Channel Signal and First Step, Single Channel Signal. 32 Figure 18: Signal from Channel 1 of Prototype 2. 33 Figure 19: Prototype 2 S-Scan of Pristine Sample. 34 Figure 20: Surface Absorbing Defects. 35 Figure 21: Surface Scattering Defects. 36

viii Figure 22: Single Surface Crater, Prototype 1. 37 Figure 23: Single Surface Crater, Prototype 2. 38 Figure 24: Room Temperature Bending 39 Figure 25: Comparison of Bending at Room Temperature and 180 F. 41 Figure 26: Field II Simulation of 2.25 MHz Array. 42 Figure 27: Field II Simulation of 10 MHz Array. 43 Figure 28: Field II Simulations of Beam Steering. 44 Figure 29: Table of Calculated Critical Inter-element Spacing Values. 45

ix Acknowledgements I would like to thank Dr. Bernhard R. Tittmann for several years of guidance and opportunities. I would also like to thank my fellow laboratory coworkers for their contributions and camaraderie. To my fiancé and family, thank you for your love and encouragement during my collegiate years and in the years that lie ahead.

1 Chapter 1 Introduction Material properties and defects can be investigated using Ultrasonic Non- Destructive Evaluation (UNDE). This method uses high frequency pressure waves as a means to examine a specimen. Ultrasonic waves are considered sound waves whose frequency is higher than 40 khz. To generate these pressure waves, a piezoelectric transducer can be used. When a voltage is applied to the piezoelectric material within the device, the material undergoes a strain. This transduction produces a pressure wave. This wave can then be utilized to interact with a specimen. As the pressure waves travel through the specimen, the acoustic energy can be reflected, transmitted, or absorbed. The reflected pressure waves can be detected using a piezoelectric transducer. Reflected waves can be received by interacting with the same piezoelectric transducer that was used to generate the waves. This is an A-Scan and the signal is recorded as a series of data points of voltage amplitude and time. A time delay between sending and receiving of a pressure wave allow for the device and electronics to distinguish the received pressure wave from the generation of pressure waves. Phased arrays are systems of piezoelectric transducers which work in unison to perform scanning of a specimen. Working much like electromagnetic radar systems, the piezoelectric elements of the array are pulsed by electronics in time intervals which allow scanning and focusing to different points in space. The reflected ultrasonic pressure waves received by each element of the array are then interpreted by the electronics to create a composite electrical signal. The composite signals for different scan locations

2 can be combined, forming an image of the amplitude of reflected waves. This image is considered a sectorial scan (S-Scan). Understanding how materials behave in harsh environments is important so that materials can be used in new applications. Corrosion, chemical composition changes, and warping may cause unwanted property and dimension changes in the material. In order to utilize new material technology, new inspection technologies need to be developed to characterize the performance of these new materials. The current methods of study of materials in harsh environments require a great cost and lacks real-time information on material degradation. To analyze materials using conventional methods requires removing the materials from the environment so that laboratory equipment can be used to measure changes of a material that occurred in the harsh environment. This method is costly and ignores the repercussions of cycling the material between harsh environment and laboratory conditions in developing a study of the effects of the environment on the material. The performance of materials in high pressure and high temperature environments is difficult to monitor in-situ. Current devices for UNDE are made of materials that cannot withstand high pressures and, more importantly, high temperatures. The transduction characteristics of many materials such as piezoelectricity and magnetostriction are lost as the temperature rises. New smart materials are becoming available for possible use in high temperature applications. Another hindrance is the number of wires leading to a device within a harsh environment. Penetrations into a harsh environment can be costly, and a way to

3 minimize these penetrations while maintaining the effectiveness of a phased array would be beneficial. Mechanical systems for scanning surfaces of materials are also unable to withstand high temperatures and pressures. Phased arrays are able to perform scanning without having any moving parts. By utilizing both new technologies in phased arrays and piezoelectric materials, a phased array device may be designed to achieve in-situ examinations of material changes in a harsh environment in real time. A stepped linear phased array will be able to scan sections of a material specimen for the effects of degradation while also having the ability to monitor bending or warping of the overall specimen. The stepped design allows for a minimal number of wires leading to the device which will be placed in the harsh environment with the material specimen. This research work will examine the feasibility of a stepped linear phased array design and will make recommendations for the advancement of the technology.

4 Chapter 2 Literature Review and Theory 2.1 Piezoelectric Materials In some materials the influence of electrical energy causes a physical deformation of the material. These materials convert electrical energy into mechanical energy via the piezoelectric effect[1]. Piezoelectric materials have many promising applications and have been used for many years in UNDE. By stimulating piezoelectric materials with short electric pulses, a short mechanical response can be generated. This mechanical response can propagate into a material in contact with the surface of the piezoelectric material. This propagating mechanical wave can be considered an ultrasonic wave, a pressure wave with frequency higher than ~40 khz. This ultrasonic wave can be utilized to interact with materials to characterize their properties, locate defects, and monitor degradation. Piezoelectric materials can also work in the reverse, converting mechanical energy back into electrical energy. In this way, piezoelectric materials may be utilized to receive the return signals that have been generated. The piezoelectric effect can be optimized for UNDE purposes by including the piezoelectric material within a device called a piezoelectric transducer. This device improves the transmission and reception of mechanical waves and protects the piezoelectric material from environmental conditions. 2.2 Linear Phased Arrays Phased arrays are a network of piezoelectric elements which are excited electronically in a specific time sequence to allow them to simultaneously direct

5 ultrasonic energy to points in space. The angle that an array is beaming ultrasonic energy in is called the steering angle. This allows for maximum energy to interact at a specific point, and any reflected waves that return to interact with the array can be recorded. Time shifts are applied to the received signals similar to the time delays used in the initial pulsing of the array elements. The time shifts correct the received signals so that the array element signals received from specific points in the scan region are aligned in space. A graphic can then be generated showing the amplitude of the received signal from each point in space. This scanning and graphical representation can be conducted in real-time, with no mechanical parts and without translation of the device. In the 1980s, support of fracture mechanics study and the need to reduce inspection time in hazardous environments were major driving forces to the development of phased arrays[2]. Inspection techniques utilizing phased arrays were developed to take advantage of the reduced inspection time associated with replacing mechanical scanning with electronic scanning. In addition to academic fracture mechanics studies, scanning of welds to ensure their integrity was vital to industry[2]. Phased arrays require complicated electronic systems to operate. Computing power and data storage are important to fully utilize the capability of a phased array system[3]. In the past, the computing power, data storage, and graphic display capabilities were not available to fully implement high resolution phased array systems for industrial applications. Higher resolution displays and high density computing power and data storage have allowed phased arrays to become more viable for non-destructive testing and evaluation.

6 While many phased arrays directly contact the target specimen, the array of interest in this study was non-contact and couples the acoustic energy to the target specimen via a water medium. Air-coupled phased array techniques have also been studied[4]. A method using laser generated pressure waves has also been investigated[5]. This non-contact method produces the functionality of phased arrays by firing laser pulses with time delays at the surface of a specimen. The targets on the surface coincide to the location of phased array elements. A linear phased array is a series of piezoelectric elements arranged along a line on a plane. Typically they are the same size and have equal spacing. The spacing of the elements affects the ability of the array to steer and focus, while the geometry of the elements affects the frequency and beam divergence. These effects will be discussed in the following sections. 2.3 Array Element Spacing The spacing between the centers of elements is the inter-element spacing, d. The width of the elements, w, must be less than this, otherwise elements would be overlapping. Figure 1, below, illustrates inter-element spacing, d, width, w, and length, l. l d w Figure 1: Diagram of Array Element Spacing. Inter-element spacing, d, width, w, and length, l.

7 There is however a limit on the inter-element spacing relating to the wavelength of the pressure wave in the medium, λ, and the maximum steering angle, θ steering, that is required for the array s application[6]. This calculation is shown in Equation (1) below. d cr λ v = = (1) 1 + sinθ f + steering ( 1 sinθ ) steering The width of the elements, w, must be less than the inter-element spacing, d, which must be less than the critical inter-element spacing, d cr. This relation is shown in Equation (2). w < d < (2) d cr It is desirable to design the width of the elements, w, as close to the inter-element spacing, d, as possible. However, this is difficult to manufacture and the closer the elements are together, the more likely they may not maintain the ability to operate independently; the electrical connections may interfere or there may be adverse operating conditions degrade the electronics and lead to undesirable operations. An alternative method of spacing elements of a linear phased array is creating a sparse array. Sparse arrays eliminate some elements from a linear phased array, but maintain the same spacing between elements and empty spaces where elements would be placed in a linear phased array[7]. This type of array maintains some functionality of a linear phased array with more elements and provides a cost savings over fully populated linear phased arrays. Choosing which elements should be removed from a full linear phased array and which should remain needs to be optimized for each design and application[7].

8 2.4 Time of Flight Ultrasonic waves are converted back to electrical waves, and then can be analyzed by software or can be directly viewed on an oscilloscope. Typically a time versus voltage graph is used as output in real time. This allows the user to see where in time reflections of ultrasonic energy are received by the piezoelectric material. Typically reflections occur at material interfaces due to mismatches in the acoustic impedance of two materials. The reflected pressure waves associated with material interfaces can be used to measure the thickness of a material layer. The time difference between front wall and back wall reflections in the signal is directly correlated to the time the ultrasonic wave takes to travel the distance between the front and back surfaces of the specimen. 2.5 Near-Field and Far-Field The amplitude of the ultrasound emitted by a transducer wildly oscillates in a region close to the surface of the transducer. This occurs due to the interaction of many pressure waves of slightly different frequencies and amplitudes emitted from different locations on the surface of the transducer. After some axial distance, N, from the surface, the waves develops into a wave front. The region before this distance is referred to as the near-field, and the region past this distance is the far-field. The amplitude of the pressure wave decreases exponentially in the far-field. It is desirable to maintain a distance greater than N between the transducer face and the region of the target sample that is to be evaluated. If the target is evaluated in the near-field region, results will neither be useful nor accurate. Figure 2 illustrates how the amplitude of the pressure wave varies with distance from the transducer[1].

9 Near-Field Nominal Amplitude Far-Field Distance from the Transducer N Figure 2: Near-Field and Far-Field: Distance from Transducer vs. Nominal Amplitude. N is the distance to the point along the access of a transducer separating the near-field and the far-field[1]. For cylindrical pistons (transducers) emitting a continuous pressure wave, this region, called the near-field, can be estimated to be related to the diameter of the emitting surface, D, and the wavelength of the emitted pressure wave[1]. The wavelength is equal to the velocity of sound, v, in the medium divided by the frequency of the continuous pressure wave emitted, f. The distance, N, from the surface of the transducer that separates the near-field from the far-field is calculated using Equation 3. N 2 2 D D f = = (3) 4 λ 4v Although the near-field is easily calculated for continuous-wave cylindrical transducers, the actual calculation is dependent on the bandwidth, central frequency, and geometry of the transducer surface. Equation 3 can be used as a simple estimate of the near-field region of the rectangular piezoelectric elements in the linear phased arrays.

10 When the elements of the array are used in unison to perform focusing and scanning, they are unable to function as a linear phased array in the near-field of the elements due to the amplitude oscillations that take place in this region. Linear phased arrays should operate in the far-field of the piezoelectric array elements. 2.6 Beam Divergence After the pressure wave passes by the near-field distance, the wave begins to spread out. There is some angle, α, that relates the center axis of the beam and 6dB down intensity level. The 6dB down intensity is about half the intensity of the center axis intensity. For continuous-wave(cw), cylindrical transducers, α can be calculated from the first root of the far-field pressure fields. Equation 4 is developed in Rose s Ultrasonic Waves in Solid Media[1]. 1.2v α = sin 1 (4) Df Beam divergence decreases as the dimension of the element increases because the angle, α, decreases. Beam divergence is a factor in the decrease of the pressure wave amplitude along the central axis of the transducer in the far-field. The pressure field intensity spreads out over a larger region as the beam diverges. Figure 3 shows how the beam diverges after entering the far-field.

11 Near-Field 6 db down Beam Divergence Angle, α Center Axis Far-Field N 6 db down Natural Focus Figure 3: Diagram of the Beam Divergence Angle of a Transducer. The pressure field in the far-field spreads out at the beam divergence angle, α, which defines the region of the pressure field that is within 6 db of the peak pressure along the center axis of the transducer in the far-field[1]. It is desirable for linear phased arrays to maintain pressure waves in the twodimensional plane outward from the line of array elements. To minimize losses to the regions above and below this plane, the length, l, of the elements should be larger than the width, w, of the elements. See Figure 1 for illustration of these array element dimensions. Similarly it is beneficial to have a smaller array element width because this allows for more divergence in the scanning plane of the array. This allows for a larger angular region of the plane to be scanned. 2.7 Frequency Considerations Frequency plays a large role in the capability of an array. With a pulse excited element, frequency is discussed in relation to the central frequency of the element. The content of the pulse includes frequencies that are higher and lower than the central

12 frequency of the element, but to understand the overall performance of the array it is useful to consider the central frequency as the frequency of the pulse and element. Higher frequencies have a shorter wavelength and thus have a potential to interact with smaller irregularities in a sample. This allows an array to detect smaller defects that otherwise would appear not to be present. Determining the size of the anomalies that are to be monitored (the resolution required) is an important step in choosing the central frequency of the array elements. While it is beneficial to increase frequency and thus increase resolution, increasing frequency leads to many detrimental effects as well. As frequency increases, the attenuation of the pressure wave increases in most materials. This means that the amplitude of the pressure wave in the far-field decreases more rapidly with higher frequency pulses. Increasing frequency decreases beam divergence and increases the near-field distance. 2.8 Resolution The resolution of a UNDE technique is dependent on the frequency of ultrasound, the properties of materials in the system, the electronics used to operate the inspection device, and the geometry of the device and system. In the next two sections, both lateral resolution and axial resolution will be discussed. 2.8.1 Lateral Resolution Lateral resolution is most dependent on the beam directivity of the elements. The coverage of an individual element s pressure wave on the surface of the sample is the highest resolution that can be achieved. The near-field distance, N, is the point at which

13 the beam spot is the most focused. As the far-field pressure wave moves further through the sample or medium, the larger the beam spot becomes. Another factor affecting the lateral resolution is the number of elements in the row of the array. 16 elements in a row can focus to a tighter spot than 4 elements can. This is because a higher ratio of the total energy emitted can be concentrated on that spot, and likewise a higher ratio of the reflected acoustic energy will reflect from the focus. This is further expounded in Section 2.10. 2.8.2 Axial Resolution Axial resolution determines the ability of the array to detect differences in depth. Examples include the depth of a crater on the surface of the specimen, thickness of a specimen, or a change in the surface position due to bending or warping. The axial resolution is determined by the type of pressure pulse emitted by the elements of the array. A short pulse of 1.5 cycles would lead to a resolution of 1.5λ. Figure 4 shows examples of a 1.5 cycle pulse and a 2.5 cycle pulse. Voltage 1.5 cycle pulse Voltage 2.5 cycle pulse Time Time Figure 4: Diagram of 1.5 and 2.5 Cycle Pulses. For pulses of the same frequency, a pulse with fewer cycles has a shorter time length. Pulses this short can be achieved by providing sufficient electrical damping in the equipment providing the pulse or within the array itself. The table, Figure 5, summarizes

14 the axial resolution that may be achieved by piezoelectric cylindrical pistons with operating frequencies of 2.25 and 10 MHz. Axial Resolution Frequency 1.5 cycle, 2.5 cycle, 1.5 cycle, 2.5 cycle, Water* Water* Copper* Copper* 2.25 MHz 1.0 mm 1.7 mm 3.1 mm 5.2 mm 10 MHz 0.26 mm 0.38 mm 0.71 mm 1.2 mm *Speed of Sound: 1500 m/s in Water; 4700 m/s in Copper. Figure 5: Table of Calculated Axial Resolutions. These values illustrate the ideal resolutions achievable for different frequency and cycle length pulses in water and copper. 2.9 Linear Phased Array Pressure Field Equations The pressure field produced by a phased array can be determined from superimposition of the pressure fields associated with each array element. Similarly, the pressure field of an individual array element can be calculated from integration over the array element of infinitesimal elements. S.-C. Wooh and Y. Shi have extensively documented work on developing pressure field equations and application to linear phased array design[8-10].

15 field[8]. Figure 6 below shows a single array element and a specific point in the pressure r θ P(r, θ ) a x dx R Figure 6: Diagram of Element and Point P(r, θ ) in the Pressure Field. Here the width of the element is a, x is the distance from the top of the element to the differential element, dx is the width of a differential section of the element, R is the distance from the differential section to point P, r is the distance from the top of the element to point P, and theta is the angle from the element perpendicular to the line connecting point P to the top of the element[8]. The elements are estimated as line elements, with a the width of the line element. This simplifies equations minimizing the array calculations to a single twodimensional plane, and is acceptable because the length of the element is much larger than the width, a, and is also good estimate for this application because the pressure field of interest is in the plane perpendicular to the length direction of the line elements. In addition, an estimate is made to arrive at an equation for the pressure field at the point, P. For points, P, that are sufficiently far away from the array (where r/a >> 1), R can be approximated as {r x sin θ }[8]. The pressure field associated with an infinitesimal element can be approximated as the Equation (5) below[8]: 1 po 2 dp exp[ j[ ωt k( r x sinθ )]]dx (5) r

16 In Equation (5), p o is the amplitude of the initial pressure wave generated at the surface of the infinitesimal element, ω is the angular frequency of the initial pressure wave, and k is its wave number. The angular frequency, ω, and wave number, k, can also be written in terms of frequency, f, and sound speed, c, as in Equation (6) and Equation (7) below. ω = 2πf (6) k ω 2πf = = (7) c c By integrating over the entire array element, the total pressure field generated by the array element can be calculated[8]. The pressure field for an element is shown in Equation (8) and Equation (9). p a p θ dx (8) r 2 O ( r,, t) = dp exp[ j[ ωt k( r x sin θ ) ] element 0 1 1 ( kasin( θ / 2) ) k sin( θ / 2) po 2 sin jkasinθ p( r, θ, t) = exp exp ω r 2 [ j( t kr) ] (9) Again by superposition, the pressure field produced by the entire array of N elements is the summation of the pressure fields associated with each element, as shown in Equation (10) below. p N θ i (10) i= 1 ( r,, t) = p ( r, θ, t)

17 By including time delays, Δ τ, for the array element excitations and the interelement spacing, d, the pressure field can be written as Equation (11)[8]. p ( r, θ, t) = p r O 1 2 sin ( ka sin( θ / 2) ) k sin( θ / 2) sin sin [(( ωδτ kd sinθ )/ 2) N] (( ωδτ kd sinθ )/ 2) ( ωδτ kd sin θ ) ( N ) 1 exp j( jka sin θ j exp exp [ ωt kr )] (11) 2 2 Here the time delay, Δ τ, has been chosen as constant between adjacent array elements and this simplifies the array pressure field equation. Time, t, was replaced with [t - (i - 1) Δ τ ] in the summation, Equation (10). Time delays for dynamic linear phased arrays could be integrated into these equations by replacing constant Δ τ with a variable and evaluating the pressure field by superposition rather than solving for the equivalent to Equation (11). 2.10 Array Optimization S.-C. Wooh and Y. Shi have used their array pressure field calculations as a method to optimize linear phased arrays for NDE applications. They note that maximizing the effectiveness of beam steering will coincide with minimizing the main lobe, eliminating grating lobes, and minimizing energy lost to side lobes[8]. The main lobe is the angular width of the beam traveling in the direction of the steering angle. This width relates directly with the size of the beam spot that can interact at the focus of the array. If the main lobe is sharp at the steering angle, the acoustic energy is focused primarily in the steering direction. Main lobes widths at smaller steering angles are less than larger steering angles.

18 Grating lobes can occur once the steering angle of an array is increased past the maximum steering angle. Grating lobes occur at angles other than the steering angle and have acoustic energy equivalent to the main lobe. This is a large drain on the energy in the direction of the steering angle. The grating lobes acoustic energy can reflect back to the array and cause spurious return signals. This is in addition to the decreased signal from the intended steering angle reflections. From the pressure field equations of the previous section, S.-C. Wooh and Y. Shi calculated a critical element spacing, d cr, that is defined by the maximum steering angle and the wavelength of ultrasound in the medium[8]. This is shown in Equation (1) in Section 2.3. While grating lobes account for the maximum energy lost at steering angles greater than the maximum steering angle, side lobes account for loss of acoustic energy at all steering angles. Side lobes are smaller lobes of acoustic energy directed at angles other than the main lobe. A design goal is to maximize the main lobe to side lobe acoustic energy ratio. It was found that the method to increase this ratio was to increase the number of array elements, N. Once N increases past 16, the effect of minimizing the side lobe energy converged; increasing N past 16 was not beneficial to an array s effectiveness. The minimum side lobe amplitude converges to -13.5 db of the main lobe amplitude[8]. S.-C. Wooh and Y. Shi have also studied the effects of element width on array characteristics. They found that element width had little or no effect on many phased array characteristics. It is desirable to have the largest element width possible while within the limits of the optimum inter-element spacing. This is because larger element

19 widths will produce a larger acoustic pressure. It is also reiterated that 16 elements is an optimum number for linear phased array NDE applications[9]. In another article, S.-C. Wooh and Y. Shi used computer simulation methods to further investigate linear phased array design. They provide a set of guidelines summarizing their findings. One finding specifically important to linear phased arrays with a limited number of elements is that as the inter-element spacing is increased past the critical inter-element spacing the only way to avoid grating lobes is to decrease the operating frequency of the transducer[10]. E. Kühnicke used different methods for array optimization. Using wave scattering equations and considering array elements as point sources, arrays were simulated to determine optimum characteristics for angle scanning and focusing[11]. Kühnicke also examined the effect of array elements arranged along a curve instead of a flat plane[12]. This is applicable to non-contact linear phased arrays examining a fluidfilled pipe from the inside. Kühnicke notes that the method of transient and time harmonic sound fields is a faster technique to simulate arrays for optimization of array parameters[12]. 2.11 Additional Literature K. Oliver documented early design work in The Design of a Unique Two Dimensional Phased Array with Low Channel Applications for Imaging Defects on a Metal Surface[14]. Early results from experiments were documented in Investigating a stepped ultrasonic phased array transducer for the evaluation and characterization of defects[15].

20 Chapter 3 Methodology 3.1 Devices This section highlights some of the more important devices that are used n the experiments which follow. 3.1.1 Stepped Linear Phased Array The first prototype stepped transducer was built by Olympus NDT (formerly known as RD Tech) of State College, Pennsylvania. It was built with specifications determined by Kara Oliver 2. Figure 7 shows Prototype 1 lying on its side on the left and the upright on the right with an aluminum specimen. Figure 7: Photographs of Prototype 1. Prototype 1 on its side (Left) and upright in experimental setup (Right) with an aluminum specimen. The housing provides an airtight seal for the interior electronics. The piezoelectric elements are made of lead zirconate titanate (PZT). The elements lie under the gold-colored surfaces in the photograph. PZT is commonly used in the manufacture of piezoelectric transducers for UNDE.

21 Additionally, in order to show that using fewer wire inputs to the device is feasible, the elements are connected in parallel, one on each step, with four per channel. This allows for a minimal number of wires leading to the device; five wires, four channels and one ground. The stepped design minimizes wire inputs which is important for application in hazardous environments that have rules governing the number of allowable throughways into the environment. Figure 8 below shows how the elements are arranged in the array. The red lines show how the elements in a column are wired together. Front View Side View Figure 8: Schematic Drawing of Stepped Array Device. Four elements are on the face of each step. A single electronic channel is connected to four elements. A wire connects one element on each step of the array along a column. The wiring is illustrated by the red lines. The elements on the same step act as an independent linear phased array.

22 Two additional prototypes were fabricated. Prototype 2 features the same 4 steps as Prototype 1, but has different dimensions and its elements have a center frequency of 10 MHz. Prototype 3 has 6 steps and 10 MHz central frequency. In Prototypes 2 and 3, the array elements lie under the white surfaces of the steps. In addition, these prototypes are made for eventual testing in high temperature and pressure environments. Prototype 2 is shown below in Figure 9 and Prototype 3 is shown in Figure 10. Figure 9: Photograph of Prototype 2. Prototype 2 has elements with a central frequency of 10 MHz and four steps, four elements on each step. Figure 10: Photograph of Prototype 3. Prototype 3 has elements with a central frequency of 10 MHz and six steps, four elements on each step.

23 3.1.2 Omniscan MX System An Omniscan MX system was obtained from the same company which built the initial prototype, Olympus NDT. The system is commonly used in UNDE field work. The software is not specific to an individual UNDE device such as an array or single transducer but is intended to be versatile and useable by a technician. The device can be seen in Figure 11. Figure 11: Photograph of the Omniscan MX System. Manufactured by Olympus NDT (RD Tech), the portable system can interface with conventional transducers and phased arrays for Non-Destructive Testing. The goal of utilizing the Omniscan MX system is to prove that the stepped phased array works as intended. Output can be observed in an image that shows the data received in real time as absolute value of voltage received from reflections from the area in time that is scanned by the software. The electrical signal that is received by the Omniscan MX system is rectified to produce the image of the scanned area. 3.2 Experiments Experiments were conducted to evaluate the ability of the first prototype to characterize defects and material characteristics. The goal of the device is to characterize surface defects, internal defects, and geometric changes of a specimen over time.

24 3.2.1 Experimental Setup The holder for the specimen and Prototype 1 was built specifically for these experiments. Prototype 1 and the specimen sit on the holder within cutouts that do not allow them to move along the surface of the holder. This holder is placed at the bottom of a glass beaker filled to a level above the top of the array with distilled water. Care is taken to remove air bubbles from the surfaces of the specimen and array before experiments are conducted. Prototype 1 s experimental setup can be seen below in Figure 12. Figure 12: Photograph of Experimental Setup for Prototype 1. This setup maintains the position of the prototype and specimen, allows for heating of the water and system with a hot plate, and translation of the top of the specimen. Prototypes 2 and 3 were built differently than Prototype 1. The wire leads come out of a different side of the array and the holder for Prototype 1 experiments would not be useful with Prototypes 2 and 3. Figure 13 below shows how Prototypes 2 and 3 were fixed in place and their orientation with the specimen held in place in front of them.

25 Figure 13: Photograph of Experimental Setup for Prototypes 2 and 3. This setup maintains the position of the prototype and specimen and also allows for precise rotation of the specimen. The array prototypes have been manufactured to interface with the Omniscan MX system. This system is used to operate the array and to receive and analyze signals for the following experiments. 3.2.2 Thickness Measurement The thickness of a specimen can be monitored by determining the speed of sound in the material and monitoring the distance between front and back wall reflections from the specimen. As discussed in Section 2.3, pressure waves can reflect at material interfaces. The reflections of the front wall and the back wall of a specimen are separated in time as the signals are received by the array. This time difference between reflections is the additional time that the sound waves travel in order to interact with the back wall of the specimen. This physical distance traveled is two times the thickness of the specimen. In order to monitor this distance, one can measure the initial thickness of the specimen conventionally and calculate the sound speed of the specimen, or the sound speed of the

26 known material can be read from available material literature or previously compiled tables of material properties. 3.2.3 Surface Absorbing Defect Some material defects that may arise on the surface will be chemical transformations of the surface material. Such materials may have the tendency to absorb sound energy and discharge this energy to some other form such as heat. These corrosive films or deposits on the outer layer of a material may be important to monitor to understand the life cycle of materials as they undergo chemical reactions. 3.2.4 Surface Scattering Defect Scattering surface defects may occur when a specimen s smooth surface becomes dimpled and/or blistered with small growths or bubbling. This was simulated by shallow drill holes on the surface of specimens. 3.2.5 Single Surface Crater To determine the ability of the array to resolve a crater on its surface, a small drill hole is made partway into the surface of specimens at various locations across from the steps of the array. 3.2.6 Interior Defects Interior voids caused by blistering within a material may lead to catastrophic failure of a specimen. Mechanical characteristics could be severely hampered by such an unseen defect growing within a material. Such defects may be caused by crack growth and gaseous expansion.

27 The ability of the array to detect defects interior to specimens was simulated by drilling holes through the side of samples. These holes could then be either plugged with air or be allowed to fill with water. 3.2.7 Bending of a Specimen Warping or bending of a specimen could occur with inhomogeneous stresses or inhomogeneous expansion due to heating. These geometric changes could affect the ability of a material to perform in a harsh environment. To test the ability of the Prototype 1 to monitor the warping of a specimen, the specimen s top was moved horizontally using a precision translator, simulating the effect of bending or warping without the need to bend the specimen with force. This allows for precise movement of the specimen top while the bottom of the specimen rested in a grove in the specimen holder, free to rotate but not to translate up and down. Figure 14 illustrates the movement of the specimen.

28-7 mm to +7 mm Fixed position but free to rotate Figure 14: Bending Experiment Setup. The bottom of the specimen is fixed while the top of the specimen is precision translated to lean the specimen in relation to the stepped array. At every 0.2 mm movement of the top of the specimen, the time of the front wall reflections for each step was recorded. This monitored how the position of the specimen surface facing the array moved as the specimen bent away or towards the array. Prototypes 2 and 3 could utilize their experimental setup to rotate the specimen by angle instead of translation. 3.2.8 Effect of Temperature The effect of raising the temperature of the water medium was also investigated. To raise the temperature of the system, the beaker is placed upon a hot plate, and the temperature is raised slowly to approximately 180 degrees F. The altered experimental setup of the system can be seen in Figure 15.

29 Figure 15: Effect of Temperature Experimental Setup. The experiment is conducted on top of a hot plate and the temperature of the system is monitored with a thermocouple. This temperature was chosen such that little boiling would occur in the water medium. Gas bubbles would negatively affect the performance of the array because sound waves have difficulty passing through a composite medium such as water and gaseous water bubbles. The bubbles disrupt the pressure waves because of the impedance mismatch between the bubble and the surrounding liquid. At approximately 180 F, the temperature of the system was held constant and the bending test was repeated, Section 3.2.7. The results were compared to determine the effect that temperature had on the experiment. 3.3 Simulations Computer simulations were conducted to determine how the next series of prototypes would perform compared to the first prototype. The program used was Field II, a program run through MatLab, by J. Jensen. Codes that were altered for 10 MHz array simulations were initially developed by K. Oliver from codes packaged with Field

30 II[6]. Modifications were made to the codes to allow higher frequency ultrasonic waves to be simulated.

31 Chapter 4 Results 4.1 Experimental Results 4.1.1 Signals A better understanding of how the Omniscan MX system displayed the signals from the entire prototype was desired. One step of the array was isolated by covering the active surfaces with sound absorbing stick tack, effectively turning off the other steps. The covered steps do not affect the signals received by the uncovered step, so the received signals are only comprised of reflections due to the ultrasonic waves sent by the uncovered step. The red waveform in Figure 16 below, overlayed on the black waveform, is the waveform displayed when only the first step is left uncovered. The black waveform is displayed when all four steps are in use. Figure 16: Comparison of Full Array Waveform and First Step Waveform. The isolated waveform of the first step of the array is shown in red. The full waveform, with all four steps active, is displayed in black. The figure shows how the reflections associated with the first step appear in the full array waveform.

32 Figure 10 shows that echoes from the first step affect the waveforms associated with successive steps. This means that with the current setup, array, and software, it is difficult to generate reliable detection of defects and other sources of reflections. This is especially true for reflections internal to the specimen. For future work, an adapter was procured to activate individual channels of the array prototype. When a single channel is operated using a pulser/receiver, the array receives a similar signal as those received using the Omniscan MX system. This adapter also allows the use of the array without the Omniscan MX system. The adapter was used to replicate the simple results of the Omniscan MX system with a conventional pulser/receiver. The graph below, Figure 17, compares the signals of all four steps of the second channel active and only the first step uncovered. Figure 17: Comparison of Single Channel Signal and First Step, Single Channel Signal. The isolated signal from a single element on the first step is shown in red. The full signal, with all four elements of the channel active, one on each step, is displayed in blue. The figure shows how the reflections associated with the single element on the first step appear in the full signal of the channel.

33 This confirmed that the signals seen on the Omniscan MX system were similar to those that could be obtained using a pulser/receiver. This adapter will be useful to future work when monitoring prototype element responses to different electronic pulses. A pulse-echo signal of Prototype 2 s first channel was also captured for comparison to Prototype 1. The waveform can be seen in Figure 18. 5 4 3 Prototype 2, 4 Step 10 MHz Channel 1 2 Voltage 1 0 0.00000-1 0.00005 0.00010 0.00015 0.00020 0.00025 0.00030-2 -3-4 Time (s) Figure 18: Signal from Channel 1 of Prototype 2. Prototype 2 has four steps and its elements have a central frequency of 10 MHz. This reflection signal of channel 1 shows significantly more noise and higher frequency content than the signals seen in Figure 17. This waveform shows that Prototype 2 behaves similarly to Prototype 1. the same reflected front and back wall reflections can be seen, as well as additional echoes of the reflections. The main difference that can be seen is that Prototype 2 has sharper minima and maxima which are due to the higher central frequency of 10 MHz. There are many more cycles in the pulse reflections. The pulse cycle length needs to be shortened to

34 several cycles in order to have any benefit of the increased central frequency of the array elements. Figure 19 below also illustrates the poor quality of Prototype 2. Figure 19: Prototype 2 S-Scan of Pristine Sample. This image shows a rectified waveform (top) and a S-Scan image of a defect-free specimen (bottom). Four front wall reflections from the four steps of Prototype 2 can be seen but back wall reflections are significantly lower in magnitude. Front wall reflections of Prototype 2 scans are large but back wall reflections are almost nonexistent. The large pulse length of the front wall reflections also contributes to an inability of Prototypes 2 and 3 to resolve surface and interior defects. 4.1.2 Surface Absorbing Defect Absorbing surface defects were simulated by applying commercially available stick tack, a rubbery adhesive, to the surface of specimens. This defect was applied in front of individual steps of the array on a sample metal specimen and is circled in red in Figure 20. The decreased amplitude of the signal received by the array is seen for the step that has the sound absorbing stick tack in front of it.

Figure 20: Surface Absorbing Defects. Prototype 1 was used to show the applicability of the array to detect surface defects that absorb acoustic energy. A piece of stick tack was placed on the surface of the specimen in front of each step of the array. Each step was able to experience a decrease in the amplitude of the reflected wave associated with that step. 35

36 The stick tack decreases the amplitude of the reflected waves significantly. This can be seen visually in the S-Scan produced by the Omniscan MX system and the amplitude can be seen to decrease in the A-Scan signal as well. 4.1.3 Surface Scattering Defect Surface scattering mock defects were made by drilling small shallow holes into the surface of a specimen in front of a step of the array. An example of these mock defects can be seen in the photograph on the right of Figure 21. On the left is the signal associated with this defect. Figure 21: Surface Scattering Defects. A surface defect that scatters pressure waves was produced on a specimen at the location across from the second step of Prototype 1. The defect lowered the amplitude of the reflected waves of the second step of the stepped array. The amplitude of the second step s front wall reflection decreased significantly due to the shallow divots on the specimen in front of the second step of the array. 4.1.4 Single Surface Crater A single surface crater defect was made in a specimen. A larger (1/8 diameter) shallow hole was drilled into the surface of a specimen in front of the first step of

37 Prototype 1. Figure 22 shows the output of a single surface crater in front of the center of the first step of Prototype 1, and below this, a screenshot of a pristine specimen. Figure 22: Single Surface Crater, Prototype 1. A single shallow drill hole was produced in the specimen surface, centered in front of the first step of Prototype 1. The effect in the front wall reflection (top) shows a shift of the front wall reflection, possibly reflecting off the bottom of the hole in addition to the surround surface. A defect-free specimen reflection image is also shown for comparison (bottom). While Figure 22 shows that a crater type defect can be detected by the system, it is difficult to determine the dimensions of the defect. The depth that the crater penetrates into the specimen and the width across the face of the specimen cannot be directly correlated to the features of Figure 22.

38 Prototype 2 was also used to investigate single surface defects. Figure 23 below shows an A-Scan of a single surface crater in front of the first step of Prototype 2. The same method of covering steps 2 through 4 as was used in Section 4.1.1 was used to isolate the signals of step 1. Front wall Bottom surface of crater defect Figure 23: Single Surface Crater, Prototype 2. Similar to Figure 22, a shift of the front wall reflection can be seen with a similar defect with Prototype 2. The A-Scan shows a reflection from the surface of the specimen as well as a reflection from the bottom surface of the crater. While the signal of the entire array had more background noise and interferences that marred the ability to distinguish a single surface crater, the crater could be seen in Figure 23. Prototype 2 s ability to resolve a single surface defect was greater than Prototype 1. Additional data analysis techniques would be beneficial to detecting surface defects. Removing noise and spurious reflections would benefit defect detection and characterization.

39 4.1.5 Interior Defects Prototype 1 could not detect interior defects filled with air or water. No reflections could be seen at the location of the defect within the depth of the specimen. In addition, noise levels and interference with front and back wall reflections may have obscured any reflections from an interior defect. Prototype 2 similarly could not detect interior defects. Additional signal processing techniques may provide the means to detect such internal defects. 4.1.6 Bending of a Specimen As discussed in Section 3.2.7, it is desirable to monitor bending or warping of a specimen in a hazardous environment. Figure 24 below shows how the position of the front wall reflections moved in time as the specimen leaned away from Prototype 1. Room Temperature Bending - 0 to 7 mm Time difference from zero deflection (microseconds) 4.00 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0.00 0 2 4 6 Translation of Top of Specimen (mm) 1st step 2nd step 3rd step 4th step Figure 24: Room Temperature Bending. Prototype 1 was used to monitor a specimen leaning away from the stepped array. The graph shows how the location in time of the front wall reflection of each step changed with 0.2 mm incremental movements of the top of the specimen.

40 As the specimen leaned away from Prototype 1, the distance the pressure waves traveled to reflect off of the surface of the specimen increased. The fourth step s distance increased the most because the specimen was fixed closest to the first step. This experiment shows that the stepped array is applicable for monitoring bending and warping of a specimen along its vertical axis. 4.1.7 Effect of Temperature The same experiment as Section 4.1.6 was conducted at an elevated temperature, 180 F. The system temperature was raised with a hot plate and held constant around 180 F. The specimen was again leaned in 0.2 mm increments and the results were compared to the room temperature experiment conducted in Section 4.1.6. The results can be seen in Figure 25. The open circles are the elevated temperature results and the closed circles track the data obtained at room temperature.

41 Comparison of bending, 0 - +7 mm 4.00 Time difference from zero deflection (microseconds) 3.50 3.00 2.50 2.00 1.50 1.00 0.50 1st step 2nd step 3rd step 4th step 1st step 180F 2nd step 180F 3rd step 180F 4th step 180F 0.00 0 1 2 3 4 5 6 7 Translation of Top of Specimen (mm) Figure 25: Comparison of Bending at Room Temperature and 180 F. Prototype 1 maintained functionality at both room temperature (solid circles) and 180 F (open circles). The results for leaning of the specimen show that Prototype 1 was effective at both temperatures. By overlaying the results of the bending experiment for both ambient and 180 F, it can be seen that the bending can be monitored at either temperature. There is little change in the times of the front wall reflections for the two experiments. The array maintains its effectiveness at either temperature. 4.2 Simulations Field II simulation results are provided in this section to provide insight in the differences of the array prototypes. Prototype 1 has a center frequency of about 2.25 MHz and the first step is about 6 mm from the specimen. When the array focuses at 6 mm, the specimen plane, the simulation shows that the energy profile at the specimen surface would look as follows in Figure 26.

42 Pressure XY Cross Section Response 0-30 -10-20 -20-10 y position [mm] 0-30 10-40 20-50 30-60 -20-10 0 10 20 x position [mm] Figure 26: Field II Simulation of 2.25 MHz Array. The image on the left shows the amplitude of the pressure wave in a 2-D plane corresponding to the surface of a specimen. On the right are the inputs to the simulation: the width of the elements, wth, height, ht, and space between elements, k_x. In addition, the number of elements is set at 4 in the next input, and the focus and axis capture (2-D plane to display) are set at 6 mm from the array. The central frequency of the elements is set at 2.25 MHz in the final input. The intensity of the pressure wave at the focus point is white in the figure. As the intensity decreases in the figure, the representative color goes to yellow, red, and then black. The graphic on the right is the user input boxes that are used to alter the parameters of the program. For comparison, a 10 MHz linear phased array was simulated for central focus at a distance of 8 mm away from the specimen. Figure 27 shows the outputs of the simulation for this case.