Rapid Non-Destructive Residual Stress Analysis of Steel Structures

Transcription

1 Copy No. Defence Research and Development Canada Recherche et développement pour la défense Canada DEFENCE & DÉFENSE Rapid Non-Destructive Residual Stress Analysis of Steel Structures Phase 2, Assessment and Testing of the Prototype Magnetic Barkhausen Noise Analysis System Thomas W. Krause, P. McNairnay, V. Babbar, A. Samimi, P. Weetman Royal Military College of Canada Lynan Clapham Queen's University Prepared By: Royal Military College of Canada PO Box 17000, Station Forces Kingston, Ontario, Canada Contract Project Manager: Dr. Thomas Krause, , ext 6415 PWGSC Contract Number: FE S1431CIA01 CSA: Dr. Shannon P. Farrell, Defence Scientist, The scientific or technical validity of this Contract Report is entirely the responsibility of the contractor and the contents do not necessarily have the approval or endorsement of Defence R&D Canada. Defence R&D Canada Atlantic Contract Report DRDC Atlantic CR March 2014

2 This page intentionally left blank.

3 Rapid Non-Destructive Residual Stress Analysis of Steel Structures Phase 2, Assessment and Testing of the Prototype Magnetic Barkhausen Noise Analysis System Thomas W. Krause, P. McNairnay, V. Babbar, A. Samimi, P. Weetman Royal Military College of Canada Lynan Clapham Queen's University Prepared By: Royal Military College of Canada PO Box 17000, Station Forces Kingston, Ontario, Canada Contract Project Manager: Dr. Thomas Krause, , ext 6415 PWGSC Contract Number: FE S1431CIA01 CSA: Dr. Shannon P. Farrell, Defence Scientist, The scientific or technical validity of this Contract Report is entirely the responsibility of the Contractor and the contents do not necessarily have the approval or endorsement of Defence R&D Canada. Defence R&D Canada Atlantic Contract Report DRDC Atlantic CR March 2014

4 Approved by Original signed by Dr. Leon M. Cheng Dr. Leon M. Cheng Head / Dockyard Laboratory Atlantic Approved for release by Original signed by Dr. Leon M. Cheng Dr. Leon M. Cheng Chair / Document Review Panel Her Majesty the Queen in Right of Canada, as represented by the Minister of National Defence, 2014 Sa Majesté la Reine (en droit du Canada), telle que représentée par le ministre de la Défense nationale, 2014

5 Abstract.. The Royal Military College of Canada (RMCC) was contracted by DRDC to develop a prototype portable Magnetic Barkhausen Noise Analysis (MBNA) system for qualitative rapid residual stress analysis of structural ferromagnetic steels. The unique design offers rapid interrogation of ferromagnetic materials and a tunable depth of analysis. During phase I, a prototype laboratory MBNA measurement system was designed and constructed. This report describes progress achieved during phase II of the three phase project. During phase II (current work) a deeper understanding of the MBNA functionality and inner workings was achieved through experimental testing and modelling. The laboratory system was improved. Experimental testing and modeling were used to achieve greater accuracy of stress values at varying depth within ferromagnetic steel. This analysis provides the basis for design of the portable MBNA system that is to be delivered at the end of phase III. The laboratory MBNA has been demonstrated to be effective for rapid identification of regions of high tensile stress and stress gradients a precursor to crack initiation. A field-ready portable system will be supplied to DRDC Atlantic by March Résumé... RDDC a donné comme contrat au Collège militaire royal du Canada (CMRC) d'élaborer un prototype de système portatif d analyse du bruit magnétique Barkhausen (ABMB) qui permet une étude qualitative rapide de contraintes résiduelles au sein d aciers ferromagnétiques structuraux. Grâce à sa conception unique, ce système permet d analyser des matériaux ferromagnétiques rapidement et à diverses profondeurs. Au cours de la première phase d élaboration, un prototype de laboratoire a été conçu et fabriqué. Le présent rapport porte sur les progrès réalisés pendant la seconde phase d un projet de conception tripartite. La seconde phase (travaux en cours) du projet a permis de mieux connaître la capacité et le fonctionnement du système de laboratoire au moyen de modèles et d essais expérimentaux, ainsi que de perfectionner le système. Les modèles et les essais expérimentaux ont servi à accroître l exactitude des mesures de contraintes à diverses profondeurs au sein d aciers ferromagnétiques, de même qu à établir le concept de base du système portatif d ABMB devant être livré au terme de la troisième phase. On a démontré que le système de laboratoire permet d identifier efficacement et rapidement des zones présentant de fortes contraintes de traction et de fortes variations de contraintes, lesquelles constituent des signes précurseurs de fissuration. Un système portatif utilisable sur le terrain sera fourni à RDDC Atlantique d ici mars DRDC Atlantic CR i

6 This page intentionally left blank. ii DRDC Atlantic CR

7 Executive summary Rapid Non-Destructive Residual Stress Analysis of Steel Structures: Phase 2, Assessment and Testing of the Prototype Magnetic Barkhausen Noise Analysis System T.W. Krause; P. McNairnay; V. Babbar; A. Samimi; P. Weetman; L. Clapham; DRDC Atlantic CR ; Defence R&D Canada Atlantic; March Introduction: Residual stress analysis surveys represent a means to gain insight into the extent of pre-existing fabrication-induced stress and the stress-redistribution in response to repair processes (such as weld cladding). While similar approaches (ex., X-ray diffraction, hole drilling) offer more accurate results, these approaches are labor intensive. The extensive surface preparation also restricts the scope of analysis, density of data points and modifies the stress within the material. There is a need for development of a portable, non-destructive (no surface preparation) and rapid interrogation technique that offers tunable depth of analysis in ferromagnetic submarine steels. The Royal Military College of Canada (RMCC) was contracted to develop a prototype portable Magnetic Barkhausen Noise Analysis (MBNA) system. During phase I, a prototype laboratory MBNA measurement system was designed, constructed and tested. The goal for Phase II was to improve the functionality of the laboratory system, conduct stress dependent measurements under elastic strain conditions, and design and assemble components for a portable MBNA system. This document describes activities undertaken by RMCC during the second year of a three year project. Results: During phase II (current work) a deeper understanding of the MBNAs functionality and inner workings was achieved through experimental testing and modelling. The laboratory system was improved. Experimental testing and modeling were used to achieve greater accuracy of stress values at varying depth within ferromagnetic steel. This analysis provides the basis for design of the portable MBNA system that is to be delivered at the end of phase III. The laboratory MBNA has been demonstrated to be effective for rapid identification of regions of high tensile stress and stress gradients a precursor to crack initiation. Significance: This MBNA system will enable an increase to the periodicity and scope of inspections of critical components and structures. The non-destructive nature of the MBNA system will allow the timely return to service of inspected structures/components in an unmodified (by surface preparation) condition. Explicit understanding of the residual stress fields on naval structures and components will provide the requisite data to enhance the validity of numerical models and experimentation supporting operations and design life. Future plans: Further experimental and theoretical work is required to define physical system parameters for quantitative stress calculations. A field-ready portable system will be supplied to DRDC Atlantic by March DRDC Atlantic CR iii

8 Sommaire... Rapid Non-Destructive Residual Stress Analysis of Steel Structures: Phase 2, Assessment and Testing of the Prototype Magnetic Barkhausen Noise Analysis System T.W. Krause; P. McNairnay; V. Babbar; A. Samimi; P. Weetman; L. Clapham; DRDC Atlantic CR ; R & D pour la défense Canada Atlantique; mars Introduction : L analyse des contraintes résiduelles constitue un moyen de connaître l importance de contraintes préexistantes provoquées en cours de fabrication, ainsi que la nouvelle répartition des contraintes après des réparations (p. ex. placage par soudure). Bien que des résultats plus exacts puissent être obtenus grâce à des approches similaires (p. ex. diffraction des rayons et perforation), ces dernières exigent davantage de main-d œuvre, de même qu une préparation poussée des surfaces analysées, ce qui restreint la portée des analyses, limite la densité des points de données et transforme les contraintes dans les matériaux étudiés. Il serait utile de disposer d un système portatif d analyse rapide et non destructive (aucune préparation des surfaces) qui permet d étudier à diverses profondeurs des aciers ferromagnétiques de sous-marins. RDDC a donné comme contrat au Collège militaire royal du Canada (CMRC) d'élaborer un prototype de système portatif d analyse du bruit magnétique Barkhausen (ABMB). Au cours de la première phase d élaboration, un prototype de laboratoire a été conçu, fabriqué et éprouvé. La seconde phase visait à améliorer la capacité du système, à prendre des mesures en fonction des contraintes dans diverses conditions de déformation élastique, ainsi qu à concevoir et à assembler les composants d un système d ABMB portatif. Le présent document porte sur les activités entreprises par le CMRC au cours de la seconde année d un projet triennal. Résultats : La seconde phase (travaux en cours) du projet a permis de mieux connaître la capacité et le fonctionnement du système d ABMB au moyen de modèles et d essais expérimentaux, ainsi que de perfectionner le système. Les modèles et les essais expérimentaux ont servi à accroître l exactitude des mesures de contraintes à diverses profondeurs au sein d aciers ferromagnétiques, de même qu à établir le concept de base du système portatif d ABMB devant être livré au terme de la troisième phase. On a démontré que le système de laboratoire permet d identifier efficacement et rapidement des zones présentant de fortes contraintes de traction et de fortes variations de contraintes, lesquelles constituent des signes précurseurs de fissuration. Portée : Le système d ABMB permettra d accroître la fréquence et la portée des inspections visant des composants et des structures essentiels. Les analyses non destructives que l on peut effectuer avec celui-ci favorisent une remise en service opportune des composants et des structures inspectés dans leur état d origine (aucune préparation des surfaces). En connaissant de manière approfondie la nature des zones de contraintes résiduelles formées au sein de structures et de composants navals, il est possible de recueillir des données qui permettent d améliorer la validité d expériences et de modèles numériques sous-tendant des opérations et l établissement de la durée de vie utile. iv DRDC Atlantic CR

9 Recherches futures : D autres expériences et travaux théoriques doivent être exécutés pour établir des paramètres de systèmes physiques relatifs aux calculs quantitatifs des contraintes. Un système portatif utilisable sur le terrain sera fourni à RDDC Atlantique d ici mars DRDC Atlantic CR v

10 This page intentionally left blank. vi DRDC Atlantic CR

11 Table of contents Abstract..... i Résumé i Executive summary... ii Sommaire iv Table of contents... vii 1 Introduction References Magnetic Barkhausen Noise Analysis System Design and Components Components and Assembly Probe Assembly Probe Coils Connections Cables Preamplifier BNC Flux Control System (FCS) and Power Supply FCS Power Supply PC and Software National Instruments Cards Software Operation Powering Up the System Setting up the Probe on a Sample Defining Universal Software Settings Defining the Measurement Settings Defining the Analysis Settings Running the Measurement Results Analysis and Troubleshooting No Convergence References Evaluation of Stress Dependence of Magnetic Barkhausen Noise in an HY-80 Steel Sample Introduction Experimental Set-up and procedure DRDC Atlantic CR vii

12 3.2.1 Uni-Axial Testing Gripping Mechanism Strain Gages Flux-Controlled Magnetic Measurement System Measurement Analysis Results Stress-strain plot MBN Envelopes MBN Energy Angular MBN Measurements and Magnetic Anisotropy Summary and Future Work References Biaxial Stress Models Background Obtaining Magnetization Curves as a Function of Applied Magnetic Field Model I: The Phenomenological Model Theory Results: Application to uniaxial stress Model II: The Magnetic Object Model Theory Boltzmann Distribution Expectation Values Extension to Multiply Oriented Grains Results: Application of Model II to uniaxial stress Conclusions References Conclusions and Future Work Annex A Additional Information for Operation A.1 FCS potentiometer/gain calibration A.2 Data Set Descriptions Distribution list viii DRDC Atlantic CR

13 List of figures Figure 1: Block diagram of system components Figure 2: CAD models (a and b) and photo (c) of assembled probe Figure 3: Pin Layout for excitation/feedback coils Figure 4: Pin Layout for excitation/feedback coils on probe Figure 5: Picture of pickup coil pins Figure 6: Cable pictures and pin layouts Figure 7: Preamplifier with probe coaxial input Figure 8: BNC-2110 with preamp coax input Figure 9: Circuit diagram for a single channel of the FCS[7] Figure 10: FCS board and box Figure 11: Picture of preamp at described settings Figure 12: Probe mounted on sample Figure 13: Initial screen of MBN Acquire program Figure 14: Picture of Universal Probe Settings (R and L of coils, etc.) Figure 15: Measurement tab with step numbers overlaid Figure 16: Analysis tab with step numbers overlaid Figure 17: This is the caption for the figure shown above Figure 18: Typical MBN measurement results shown on the analysis tab Figure 19: Typical MBN measurement results shown on the data display tab Figure 20: Monsanto Tensometer T20 tensile test machine Figure 21: Wedge clamp Figure 22: HY-80 Sample with the bonded aluminum tabs and strain gage Figure 23: Strain gage with grid length and width of 1.57 mm Figure 24: Schematic diagram of the flux-controlled MBN system Figure 25: MBN signal Figure 26: Stress-Strain curve Figure 27: MBN envelope variations with stress parallel (upper) and transverse (lower) to the direction of uni-axial stress Figure 28: MBN energy variations with stress parallel and transverse to the direction of uniaxial stress Figure 29: Angular MBN variations with stress DRDC Atlantic CR ix

14 Figure 30: Dog-bone shape for high-stress deformations (upper) and octagonal-shape sample for bi-axial stress study (lower) Figure 31: Schematic of MBN apparatus with feedback flux control [1] Figure 32: The instantaneous MBN voltage signal and the resulting RMS voltage from a sinusoidal applied magnetic field [2] Figure 33: The RMS Barkhausen voltage over one period for a quasi-static, sinusoidal applied magnetic field of 12Hz, 1018 carbon steel [2] (upper). The corresponding hysteresis (dashed line) and anhysteretic (solid line) curves (lower). M-, and M+ are the top and bottom of the M vs. H hysteresis curves Figure 34: Plot of the normalized γ versus σx for the experimental data Figure 35: A representation of the magnetic object Figure 36: Change in the magnetic object when an external magnetic field is applied. The red and green lines represent possible domain wall motions due to this field Figure 37: The average number of 180 degree domain walls as a function of tensile stress with zero applied magnetic field Figure 38: Average number of 180 degree domain walls versus applied magnetic field at three tensile stresses Figure 39: The anhysteretic curve of magnetization versus magnetic field for the same stresses as Figure x DRDC Atlantic CR

15 1 Introduction Identification of significant residual stress variations and characterization of the resident stress state within naval structures can contribute valuable information to risk assessments. This provides some assurance that in-service induced stresses will not compromise structural integrity under normal operating conditions. At present, X-ray diffraction (XRD) methods are used for in-situ sampling of the residual stress state of structural steel surfaces. Although XRD techniques are believed to offer the most accurate in-situ stress surveys, the need for surface preparation restricts the scope of their application. A complementary technology for stress measurement is based on Magnetic Barkhausen Noise Analysis (MBNA). MBNA can provide a more rapid interrogation of a steel surface with the intention of identifying local maxima in residual stress, which can then be confirmed with XRD. Barkhausen noise from ferromagnetic steel materials is the result of abrupt domain wall motions that arise during the magnetization process. The level of Barkhausen noise is affected by domain wall configurations within the steel material and their interaction with pinning sites. These domain wall configurations are strongly affected by the material s stress state [1]. The measurement of residual stresses in conventional steels has demonstrated good agreement with X-ray determined residual stress [2]. The scope of the overall project will be to optimize the magnetization capability and configuration of the MBNA measurement system for rapid measurement of stress in ferromagnetic steels. This shall require three phases that are anticipated to take place over three years [2]. During the first phase of this project, a prototype laboratory MBNA measurement system was designed and constructed [3]. The phase II goal was to improve the functionality of the laboratory system, conduct stress dependent measurements under elastic strain conditions and, design and assemble components for a portable MBNA system. This report describes progress achieved in Phase II of a project to develop a MBNA capability for qualitative rapid residual stress analysis of structural ferromagnetic steels, particularly naval steels such as HY-80. This includes modification of probes to provide greater proportional sampling at varying depth within ferromagnetic steel (>0.25 mm), experimental testing and a theoretical analysis of preliminary measurements. 1.1 References [1] T.W. Krause, A. Pattantyus and D.L. Atherton, Investigation of Strain Dependent Magnetic Barkhausen Noise in Steel, IEEE Trans. Magn. 31, (1995) [2] Development of Magnetic Barkhausen Noise Analysis Residual Stress Measurement System: for qualitative rapid residual stress analysis of structural ferromagnetic steels. SLA #: RMCC Serial # SLA ANNEX #: PA [3] T.W. Krause, P. McNairnay, V. Babbar, A. Samimi, P. Weetman and L. Clapham, Rapid Non-Destructive Residual Stress Analysis of Steel Structures: Phase 1, The Prototype Magnetic Barkhausen Noise Analysis System, DRDC Atlantic CR , September DRDC Atlantic CR

16 2 Magnetic Barkhausen Noise Analysis System Magnetic Barkhausen Noise (MBN) is the discrete jumps in magnetization that can be observed during the magnetization process of ferromagnetic materials [1,2]. MBN results from the abrupt motion of domain walls between pinning sites and therefore, depends on the intrinsic domain structure within the sample [3-5]. The domain structure itself is affected by grain size [5], texture[3,4], microstructure [6] and the presence of residual and applied stresses [5,6]. Therefore, characteristics of the measured signal can be correlated with various physical properties of a target ferromagnetic sample [3-6]. For the system described here, the stress induced changes to the domain structure that affect the MBN signals are the target application. 2.1 Design and Components The tetrapole MBN system described in this manual measures MBN signals under flux controlled conditions, where a specified flux density at each of the poles of the U-core excitation magnet is maintained [7]. The four pole configuration of the probe allows for rapid angular measurements by means of flux superposition, while the flux control circuitry and software allows for reproducible results under varying test conditions. The system consists of three main components; the probe, the flux control system (FCS) and the LabVIEW software. The probe generates a magnetic field using excitation coils wound around magnetic U-cores. The combined probe and core is placed on the material under inspection creating a magnetic circuit. The alternating magnetic field in the sample generates the desired MBN signal, which is sensed by a small pickup coil at the centre of the probe. The generated MBN signal is sent to a low noise preamplifier and subsequently to a National Instruments DAQ card in a PC where it is recorded. The probe consists of two Supermendur [7,10] U-core magnets, oriented with respect to each other so that they complete two orthogonal magnetic flux paths when applied to a target steel sample. The magnetic field generating system consists of an excitation coil, on each of the poles that generates the time dependent variation of flux in the magnetic circuit, along with feedback coils at the pole ends, which respond to changes in flux closest to the sample surface and which are input to the FCS. The FCS controls the voltages driving the four excitation coils and reads the voltage induced in the feedback coils. The feedback coil voltage is fed into an inverting summing amplifier, which acts as an analog error correction that stabilizes the excitation waveform [7-9]. The FCS in combination with the four poles of the two U-cores facilitate the generation of angular varying flux generation via flux superposition within the sample. Once the FCS has achieved the specified waveform accuracy, the software applies a digital error correction (DEC) algorithm to increase accuracy even further. The DEC samples the feedback coil voltage and iteratively changes the excitation signal until the feedback signals match the target feedback signal. After the DEC has achieved its accuracy threshold, the pickup voltage (MBN signal) is sampled over a number of cycles. The software also allows the user to process the raw waveform with several analysis tools. 2 DRDC Atlantic CR

17 2.2 Components and Assembly This section lists the major components that make up the system and describes how it is assembled. Each component is briefly described and its relation to other components outlined. This section should allow the user to assemble all components of the MBN lab system in the proper configuration. A block diagram of the major system components is shown in Figure 1 below. Each of these components will be described in detail below. Figure 1: Block diagram of system components Probe Assembly Probe The probe consists of a 3D printed ABS housing, which holds two orthogonal U-cores each with an excitation and feedback coil. In the centre of the probe is a pickup coil wound around a ferrite core. Both the cores and pickup coil are spring loaded to ensure minimal liftoff from the sample surface. Figure 2a is a CAD model of all probe components. Figure 2b is a CAD model of the assembled probe, while 2c shows a photo of the assembled probe. The tall core is held in the main housing over the short core, which screws into the main housing with four springs. The pickup coil sits in a copper shield holder, which slides into the pickup housing using a spring to hold it in place Coils The excitation coils are wound from 36 AWG bond wire, have 500 turns and are 5.5 mm long. The feedback coils are also wound from 36 AWG bond wire, have 50 turns and are 1 mm long. The pickup coil is wound with 44 AWG bond wire, has 100 turns and is 1 mm long. DRDC Atlantic CR

18 A B C Figure 2: CAD models (a and b) and photo (c) of assembled probe. 4 DRDC Atlantic CR

19 Connections The probe has two sets of connections. The excitation and feedback coils are connected to a two by eight header grid as shown in Figure 3. The pickup coil is connected separately (to avoid interference) and has a two header receptacle with sockets as shown in Figures 4 and 5. 1 Vex 1 9 VS 1 2 VF 1 10 GND 1 3 Vex 2 11 VS 2 4 VF 2 12 GND 1 5 VF 3 13 GND 1 6 Vex 3 14 VS 3 7 VF 4 15 GND 1 8 Vex 4 16 VS 4 Figure 3: Pin Layout for excitation/feedback coils. Figure 4: Pin Layout for excitation/feedback coils on probe. DRDC Atlantic CR

20 Pins Figure 5: Picture of pickup coil pins Cables The probe uses two cables. The main cable carries the excitation feedback signals between the probe and FCS. It is a shielded cable with sixteen 28 AWG colour coded cables, with banana plug connectors on one end for the FCS and a two by eight header receptacle with sockets. The pickup coil cable is a shielded coaxial cable with one end converted to a two pin header that connects to the probe. Figure 6: Cable pictures and pin layouts. 6 DRDC Atlantic CR

21 Preamplifier The signal from the pickup coil is very weak and thus requires a low noise preamplifier. The SR560 from Stanford Instruments, shown in Figure 7, is a configurable low noise preamplifier, with multiple gain and filter settings. The coaxial connection from the probe plugs into the single input of the preamplifier and the output connects to the position A1 on the BNC Figure 7: Preamplifier with probe coaxial input BNC-2110 The BNC-2110 is a National Instruments adapter that converts coaxial input from the preamplifier to the 68 pin standard National Instruments connection. The signals are then transmitted to the National Instruments PCI 6133 DAQ [11] in the computer Flux Control System (FCS) and Power Supply FCS The FCS contains the flux feedback circuitry which, as mentioned, is an inverting summing amplifier feedback loop. There are a total of four channels, one for each excitation/feedback coil pair. The circuit diagram for a single channel is shown in Figure 9. DRDC Atlantic CR

22 Figure 8: BNC-2110 with preamp coax input. 8 DRDC Atlantic CR

23 Figure 9: Circuit diagram for a single channel of the FCS[7]. DRDC Atlantic CR

24 The FCS is a custom printed circuit board powered by a ±24 V source. Regulators on the board provide ±15 V and +5 V rails for various circuit components and cooling fans. Each channel has two potentiometers that control the gain of the circuit and are calibrated prior to running the system. The calibration procedure is described in Appendix A1. The FCS has two main sets of connections; two NI 68 pin connectors that run to the PCI 6259 DAQ in the PC and 16 banana plug connectors that are connected to the probe. Figure 10: FCS board and box Power Supply The power supply is a Power One HCC AG with low noise, low ripple and multiple voltage outputs. The power supply provides ±24 V at a maximum current of 2.4 A each. To run the system at ±29 V, the HCC5-6-AG, which provides ±5 V, can be connected in series PC and Software National Instruments Cards The PC contains two NI DAQs, which control the FCS and record the MBN signal. The PCI 6259 DAQ has two cables that connect to the FCS. The 6259 outputs the analog reference excitation voltages and samples the actual excitation voltages, the feedback voltages and the shunt resistor voltages. It also uses digital output to control mute functions as well as select voltage following or flux feedback mode. The PCI 6133 DAQ is used solely to sample the MBN signal. 10 DRDC Atlantic CR

25 Software The system is controlled and operated using a LabView code on the PC. Depending on the application, settings can be changed in the program, which is highly customizable. The program is split into three main sections; measurement, analysis and data display. The user sets the measurement parameters in the measurement tab and chooses the automated analysis settings in the analysis tab. Processed data is displayed in the data display tab. There are also universal parameters that must be set during calibration of the system. 2.3 Operation This section provides a step-by-step guide for making a basic MBN measurement. This section should allow the user to power up all the equipment, setup the probe on a sample and run the software to take a MBN measurement Powering Up the System 1. Turn on the PC and open the LabVIEW program named MBN Acquire v Turn on the power supply to the FCS. 3. Turn on the FCS. 4. Turn on the Preamplifier. Allow the components to heat up for a few minutes as temperature changes can affect the system operation. Also ensure the FCS case fan is running to prevent overheating. The preamplifier should be set to a gain of 60 db (1000x) and the low pass filter should have a cut off frequency set at 100 khz (Figure 11). DRDC Atlantic CR

26 Figure 11: Picture of preamp at described settings Setting up the Probe on a Sample 1. Orientate the probe axes relative to a reference direction on the sample. 2. Gently apply pressure to the probe using a clamp or other pressure application device until all four poles are in contact with the sample surface. 3. Ensure that any sources of interference (AC power lines, magnets etc.) are far from the probe and the cables connected to the probe. The probe can also be mounted on the sample after the software settings have been chosen by the user. The probe can also be applied to the sample manually, but will require a constant application of pressure. 12 DRDC Atlantic CR

27 Figure 12: Probe mounted on sample Defining Universal Software Settings The LabVIEW program front panel opens up on the measurement tab, Figure 13. Above this tab are several universal settings that do not change from measurement to measurement. The standard settings and values are shown in Figure 14 below. The table on the right contains values specific to the probe being used. If the probe is replaced these values should be changed to reflect the new probe s coil inductances and impedances. All other universal settings do not need to be changed for basic measurements. DRDC Atlantic CR

28 Figure 13: Initial screen of MBN Acquire program. Figure 14: Picture of Universal Probe Settings (R and L of coils, etc.) Defining the Measurement Settings 1. Ensure the measurement tab is selected. 2. On the left side of the measurement tab select the Waveform tab. 3. Select signal type as sine wave. 14 DRDC Atlantic CR

29 4. Choose a signal frequency. Note: the convergence of the control algorithm is very sensitive to the signal frequency. In general the frequency should be between 10 Hz and 50 Hz. 5. Choose the tetrapole angle (the angle at which the flux is generated in the sample). See Figure On the Waveform tab select either the Voltage or Flux tab to choose voltage control or flux control. 7. Voltage Control: Select the excitation voltage (Maximum voltage of excitation waveform). 8. Flux Control: Select the flux density (Maximum flux through feedback coils) and ensure that the Digital Error Correction (DEC) is enabled. 9. Select the number of periods over which the software will take the measurement. 10. On the bottom right side of the measurements tab set the RMS threshold error. Note: it is initially set to 1% and setting it above 3% often leads to inconsistent data. Inconsistent data is defined as an MBN signal that is either not reproducible or indistinguishable from MBN signals taken obtained from different samples or/and at different flux/voltage settings. As a result, data taken above 3% RMS threshold error is of little value in characterizing a sample. Figure 15: Measurement tab with step numbers overlaid Defining the Analysis Settings 1. On the measurements tab select auto analyze (located at the top left). DRDC Atlantic CR

30 2. Select the Analysis Setup tab. 3. At the bottom of the screen enter directory in which the data will be saved and enter a sample name. 4. Click the yellow button next to the data sets you wish to save (located at the bottom of the tab). Note: a list with brief descriptions of each data set is in Appendix A2. 5. Select the Data Display tab. 6. Click the yellow button next to the data set you wish to save (located at the bottom of the tab). Figure 16: Analysis tab with step numbers overlaid. 16 DRDC Atlantic CR

31 Figure 17: This is the caption for the figure shown above Running the Measurement 1. Select the Measurements tab. 2. Click Run Measurement (located in the top left corner on the Waveform tab). On the measurements tab, the operation of the system can be observed as it attempts to converge toward the target voltage or flux waveform. The RMS error of the measured signal to the target is displayed in the graph on the bottom right of the measurements tab. Once this drops below the threshold the pickup coil s signal will be sampled over the number of selected cycles. Once the signal has been sampled, the software will analyze the data and display the results in the Data Display tab. The data will also be saved to the previously specified directory. 2.4 Results After the system has sampled the pickup coil signal, the software will save the data as well as perform some initial analysis, which can be viewed in the Analysis and Data Display tabs. Figures 18 and 19 show a sample of the raw and processed data in the two tabs, respectively. The central graph in Figure 18 shows the power spectra of the pickup signal with and without background noise and both time and frequency averaged. The window below this shows the time waveforms of the pickup signal. The three graphs on the right show the dynamic power spectra for the background pickup signal, the raw signal and the signal with background removed. DRDC Atlantic CR

32 Figure 18: Typical MBN measurement results shown on the analysis tab. Figure 19: Typical MBN measurement results shown on the data display tab. 18 DRDC Atlantic CR

33 The central graph in Figure 19 can be changed to show several different datasets but its main function is to display the Barkhausen signal energy as a function of angle for sweep measurement. The left side of the figure displays the reference voltages, excitation voltages, excitation currents, time rate change of flux, flux density and the calculated Barkhausen noise envelope. The right side of the figure shows some calculated values for a single measurement. 2.5 Analysis and Troubleshooting This section covers the most common problems encountered while operating the system and how to resolve them No Convergence When the system is taking a measurement, the user should monitor the RMS Error graph on the Measurement tab. This graph tracks the error in the voltage or flux waveform as a percentage of the target. Once all four waveforms are beneath the specified threshold the pickup coil is sampled. Occasionally the error will remain above the threshold for a long time indicating that the waveforms are not converging on the specified target waveform. In such an event the measurement should be stopped in order to prevent over heating of the probe coils. The FCS is limited in its ability to control waveforms at all flux densities and frequencies and thus non-convergence does not necessarily indicate a hardware problem, the following is a list of possible hardware sources for no convergence. 1. Loose connections: A poor electrical connection can prevent the FCS from properly controlling the waveform. The following connections should be checked: - NI 68-PIN Cables to FCS should not be loose. - Excitation/Feedback banana plugs at FCS. - Excitation/Feedback connection at probe. Dust, dirt or other residue can prevent a good connection. 2. Broken or damaged coil: - Check the resistances of each excitation and feedback coil. All values should be similar and should not change when the probe is moved (indicating a partial connection). 3. Short circuit: - Check for short circuits between excitation/feedback coils and the magnetic cores. DRDC Atlantic CR

34 2.6 References [1] Chikazumi, S. and Charap, S. H. Physics of Magnetism, Krieger, Florida, 1964, p [2] B. D. Cullity and C. D. Graham, Introduction to Magnetic Materials, A John Wiley & Sons, Inc., Publication, Second Edition (2009), p [3] T.W. Krause, L. Clapham, and D.L. Atherton, J. Appl. Phys. 75, pp (1994). [4] T. W. Krause, K. Mandal and D. L. Atherton, Modelling of Magnetic Barkhausen Noise in Single and Dual Easy Axis Systems in Steel, J. Magn. Magn. Mater. 195, (1999) [5] T. W. Krause, L. Clapham, A. Pattantyus and D. L. Atherton, Investigation of the Stress- Dependent Magnetic Easy Axis in Steel Using Magnetic Barkhausen Noise, J. Appl. Phys. 79, (7), 15 April (1996) [6] D.C. Jiles, Dynamics of Domain Magnetization, Czecholsovak J. of Physics, 50, pp (2000). [7] White, S. A. A Barkhausen Noise Testing System for CANDU Feeder Pipes. Queen s University, Kingston, Ontario, Canada. 2009, p 60. [8] S. White, T. W. Krause and L. Clapham, Measurement Science and Technology, 18, pp (2007). [9] S. White, L. Clapham and T.W. Krause, A multi-channel magnetic flux controller for periodic magnetizing conditions. IEEE Trans. Instr. and Meas. 61, no. 7, (2012) pp [10] H.L.B Gould and D.H. Wenny, Supermendur a new rectangular loop magnetic material, Elect. Eng., vol. 76, no. 3, pp ). [11] National Instruments Corporation, DAQ M Series, M Series User Manual, DRDC Atlantic CR

35 3 Evaluation of Stress Dependence of Magnetic Barkhausen Noise in an HY-80 Steel Sample The dependence of magnetic Barkhausen noise (MBN) on uniaxial tensile stress in a HY-80 steel sample was investigated. The HY-80 plate sample had dimensions of mm. HY-80 is a high strength steel with a typical grain size of about 7 micron and yield strength of 550 MPa. The goal of this study was to evaluate the stress dependence of Barkhausen noise for measurement of in situ residual stress of submarine hull. The experimental apparatus consists of a tensometer and strain gages for uni-axial tensile test, and a flux-controlled MBN probe for magnetic measurements. The results indicate an increase of the MBN signal with increasing applied uniaxial tensile stress, which is consistent with previously measured behavior of steels under applied tensile stress conditions. 3.1 Introduction Magnetic Barkhausen Noise (MBN) results from abrupt local changes in magnetization that may be sensed by a pickup coil during the magnetization of a ferromagnetic material [1, 2]. MBN is most often associated with motion of domain walls between pinning sites, and is therefore a function of the domain structure within ferromagnetic materials [3-5]. The domain structure itself is affected by grain size [5], texture [3, 4], microstructure [6] and the presence of residual and applied stresses [5, 6]. Therefore, characteristics of the measured signal can potentially be an indication of various physical properties of a particular ferromagnetic material [3-6], with the potential to be sensitive to the residual stress state of the material. This report documents the effects of a uniaxial tensile stress, applied to an HY-80 steel sample, on MBN measurements, for development of MBN as a stress measurement tool [7]. The MBN measurements are performed under flux controlled conditions, where a specified flux density at each of the poles of the U-core excitation magnet is maintained [8-10]. 3.2 Experimental Set-up and procedure Uni-Axial Testing Uni-axial tensile stress is applied using a Monsanto Tensometer T20 tensile test machine, which features digital display of force and crosshead speed. As shown in Figure 20, the tensometer is a horizontal screw-driven device with a constant crosshead speed of 1 mm/min and accuracy of 1% for both crosshead speed and applied load. This speed provides a quasi-static tensile test with a low strain rate, which is required to achieve uniform strain in steel samples and to minimize the errors in the load cell output [11, 12]. DRDC Atlantic CR

36 Figure 20: Monsanto Tensometer T20 tensile test machine Gripping Mechanism The tensile machine is equipped with wedge grips as shown in Figure 21. According to ASTM standards [13], for proper gripping and alignment of samples, the entire length of each wedge face must be in contact with sample ends. Therefore, roughed-surface square aluminum tabs were bonded to sample ends using epoxy as shown in Figure 22. These tabs provide proper gripping, which prevents slippage, minimizes stress concentration induced by the grips, and accommodates a nonmagnetic component in the grips, important for decoupling potential magnetization in the primarily steel tensometer. Figure 21: Wedge clamp. 22 DRDC Atlantic CR

37 Figure 22: HY-80 Sample with the bonded aluminum tabs and strain gage Strain Gages For accurate strain measurements, one uni-axial strain gage (EA AQ-350, Micro-Measurements Group) with a gage factor of 2.115±0.15% was mounted on each of the samples. These gages had a grid length and width of 1.57 mm and are mounted in the middle of the samples. Figure 23: Strain gage with grid length and width of 1.57 mm Flux-Controlled Magnetic Measurement System Control of flux in the magnetic circuit is found to be an effective method to obtain consistency in MBN measurements by reducing the distortions in the periodic magnetic flux waveforms [9, 14, 15,]. This study uses a novel hybrid flux controller, which was designed in our research group [10]. The controller combines the real time control of analog feedback with the accuracy of an iterative digital feedback algorithm to allow rapid measurement and effectively minimize errors in the control loop. This is in contrast to conventional flux controllers, which use either digital or analog feedback, leading to slow or inaccurate measurements. DRDC Atlantic CR

38 The deviations from a sinusoidal flux waveform, i.e., the form factor error, is <0.1% for excitation frequencies higher than 2 Hz, and the root-mean-square (rms) flux rate error is <1% for excitation frequencies higher than 10 Hz [10]. Figure 23 shows a simplified schematic diagram of the flux-controlled MBN system. A digital-to-analog converter (DAC) generates a reference voltage, V ref, which is then fed into an analog feedback amplifier. Excitation coil voltage, V ex, and measured feedback coil voltage, V F, are sampled by an analog-to-digital converter (ADC) and processed through a personal computer (PC). V ref is modified in each iteration and the control loop repeats until V F meets the target feedback coil voltage. The probe is a Supermendur tetrapole [10], which has two U-shaped cores with 500 turn excitation coil and a 50 turn feedback coil on each of its poles. The feedback coils monitor the flux at the excitation magnet poles, which are at the closest distance to the sample surface. The cylindrically symmetric pick-up coil is spring loaded with its solenoid winding axis normal to the sample surface. The pick-up assembly has 100 turns of 44 AWG copper coils with a ferrite core and a conductive brass shield to minimize the sensing radius for the probe configuration. The 90% sensing radius of the pick-up coil is about 2.4 mm at 30 Hz. The pick-up coil couples to magnetization changes projected out of the sample surface. The voltage induced into the pick-up coil is fed to an Ithaco 1201 preamplifier and the low frequency excitation signal is filtered out. Therefore, it is the high frequency Barkhausen emissions (>1 khz), which contribute to the voltage induced in the pick-up coil. The probe is mounted about 1 cm away from the strain gage Figure 24: Schematic diagram of the flux-controlled MBN system. Measurement Analysis A typical MBN signal has two bursts of energy in each full cycle of the sinusoidal applied magnetic flux rate with both positive and negative components as shown in Figure 25. The envelope signal labeled MBN is calculated to simplify data analyses. MBN energy (BN E ) is then defined as the time integral of the voltage squared signal. The system uses a statistical approach, which is implemented in LabVIEW 8.2, to produce a root-mean-square MBN envelope (BN env = V BN (rms)) of the induced voltage. The total energy is then calculated as [16]: BN E = (BN env ) 2 dφ, (1) in which φ is the feedback signal phase corresponding to time within the cycle. 24 DRDC Atlantic CR

39 Figure 25: MBN signal. 3.3 Results Stress-strain plot The sample was progressively stressed up to 50% of its yield strength (about 227 MPa) and surface MBN measurements were taken. Figure 26 plots the engineering stress (force divided by the initial cross sectional area of the sample) versus strain. Linearity of stress-strain plot was monitored throughout the measurements in order to ensure a uniform stress along the sample, and to avoid any misalignment due to improper gripping. Figure 26: Stress-Strain curve. DRDC Atlantic CR

40 3.3.2 MBN Envelopes The variations in MBN emissions with stress were investigated at a 30 Hz excitation frequency and 100 mt flux density measured at the excitation magnet poles. This frequency and flux level is low enough to maintain a constant static value for the domain wall width, and to minimize the disturbance of domain wall geometry, while providing rapid MBN measurements. The sample was demagnetized prior to each measurement to remove any residual magnetization. The results were reproducible with less than 1% variation. Figure 27 plots the envelope variations with stress for directions parallel and perpendicular to applied stress. The main feature of the envelopes is the increase in the peak height and sharpness under conditions of increasing tensile stress. Due to Poisson s effect, a transverse compressive strain is induced, which decreases the peak height and widens the envelope shape. 26 DRDC Atlantic CR

41 Figure 27: MBN envelope variations with stress parallel (upper) and transverse (lower) to the direction of uni-axial stress. DRDC Atlantic CR

42 3.3.3 MBN Energy The variations in peak height and shape of the envelopes were quantified by introducing the total energy (BN E ), as described by Eq. 1 and plotted in Figure 28. The observed MBN energy variations are similar to energy variations of high-strength SAE 9310 steel, as observed earlier by Mierczak et al [17]. As shown in Figure 28, the energy level transverse to the applied stress direction is higher than the energy parallel to the applied tensile stress with sharper variations. This suggests a magnetic anisotropy in the sample, attributed to roll magnetic anisotropy [18]. Figure 28: MBN energy variations with stress parallel and transverse to the direction of uni-axial stress Angular MBN Measurements and Magnetic Anisotropy Angular surface MBN measurements were taken with the excitation field progressively oriented at eighteen equally spaced angles over 180 (i.e., 10 apart) using the tetrapole. Figure 29 shows the polar plot of surface MBN energies. Under zero stress condition, there is a strong magnetic easy axis along the width of the sample. This easy axis is likely due to residual compressive stresses within the as-received sample. By further increasing the uni-axial stress, another easy axis emerged along the stress direction, and the initial easy axis due to residual stress becomes smaller. 28 DRDC Atlantic CR

43 Figure 29: Angular MBN variations with stress. 3.4 Summary and Future Work Preliminary flux-controlled Barkhausen measurements were performed on a HY-80 steel sample under applied uni-axial tensile stress conditions. Variations of MBN signal were examined as a function of stress up to 50% of the sample s yield strength. For the next step of our studies, we have prepared a standard dog-bone shape sample to study the influence of higher stress levels and plastic deformation on Barkhausen signals as shown in Figure 30(a) below. A more complex case of bi-axial stress will also be investigated, in which an octagonal shaped sample of HY-80 as shown in Figure 30(b) will be used. DRDC Atlantic CR

44 (a) 3.5 References (b) Figure 30: Dog-bone shape for high-stress deformations (upper) and octagonal-shape sample for bi-axial stress study (lower). [1] Chikazumi, S. and Charap, S. H. Physics of Magnetism, Krieger, Florida, 1964, p [2] B. D. Cullity and C. D. Graham, Introduction to Magnetic Materials, A John Wiley & Sons, Inc., Publication, Second Edition (2009), p [3] T.W. Krause, L. Clapham, and D.L. Atherton, Characterization of the magnetic easy axis in pipeline steel using magnetic Barkhausen noise, J. Appl. Phys. 75, pp (1994). [4] T. W. Krause, K. Mandal and D. L. Atherton, Modelling of Magnetic Barkhausen Noise in Single and Dual Easy Axis Systems in Steel, J. Magn. Magn. Mater. 195, (1999) [5] T. W. Krause, L. Clapham, A. Pattantyus and D. L. Atherton, Investigation of the Stress-Dependent Magnetic Easy Axis in Steel Using Magnetic Barkhausen Noise, J. Appl. Phys. 79, (7), 15 April (1996) [6] D.C. Jiles, Dynamics of Domain Magnetization, Czecholsovak J. of Physics, 50, pp (2000). 30 DRDC Atlantic CR

45 [7] Development of Magnetic Barkhausen Noise Analysis Residual Stress Measurement System: Phase II, Development of Portable MBNA System, SLA#: RMCC Serial# SLA, ANNEX #: PA11029, Oct [8] White, S. A. A Barkhausen Noise Testing System for CANDU Feeder Pipes. Queen s University, Kingston, Ontario, Canada. 2009, p 60. [9] S. White, T. W. Krause and L. Clapham, Measurement Science and Technology, 18, pp (2007). [10] S. White, L. Clapham and T.W. Krause, A multi-channel magnetic flux controller for periodic magnetizing conditions. IEEE Trans. Instr. and Meas. 61, no. 7, (2012) pp [11] ASTM E112-12, Standard Test Methods for Determining Average Grain Size, (2013). [12] Joseph R. Davis, Tensile Testing, ASM International, 2nd edition (2004). [13] ASTM E8/E8M-11, Standard Test Methods for Tension Testing of Metallic Materials, (2012). [14] O. Stupakov, J. Pala, T. Takagi, and T. Uchimoto, Governing conditions of repeatable Barkhausen noise response, J. Mag. Mag. Mater., 321 (2009) [15] H. Patel, S. Zurek T. Meydan, D. Jiles and L. Li, A new adaptive automated feedback system for Barkhausen signal measurement, Sensors and Actuators A, Vol. 129 (2006) [16] H. Kwun, Investigation of the dependence of Barkhausen noise on stress and the angle between the stress and magnetization directions, J. Magn. Magn. Mater. 49 (1985) [17] L. Mierczak, D. C. Jiles, G. Fantoni, A New Method for Evaluation of Mechanical Stress Using the Reciprocal Amplitude of Magnetic Barkhausen Noise, IEEE. Trans. Mag. 47 (2) (2011) [18] L. Clapham, C. Heald, T. Krause and D.L. Atherton, The Origin of a Magnetic Easy Axis in Pipeline Steel, J. Appl. Phys. 86, (1999) DRDC Atlantic CR

46 4 Biaxial Stress Models Magnetic measurement of residual stress, or at a minimum, comparison of residual stresses over an area such as a submarine hull, will require a means of determining relevant magnetic parameters and how they are modified by stress. This report summarizes the current progress of two models that will be used to estimate the stress in high strength steels using surface Magnetic Barkhausen Noise (MBN) measurements. In both models, the MBN is first used to construct a local magnetization (M) versus applied magnetic field (H) curve. The first model (I) is phenomenological and based on the magnetization model of Jiles and Atherton [D. Jiles and D. Atherton, Theory of Ferromagnetic Hysteresis, J. of Mag. and Mag. Mat., 61, 48-60, (1986).]. All physical details are described by various fitting coefficients. Data from the companion report [A. Samimi, T. W. Krause and L. Clapham, Evaluation of Stress Dependence of Magnetic Barkhausen Noise in a HY-80 Steel Sample, DRDC Report, March 2013] is used in the results section of this model. The second model (II) is the more physical magnetic object model [T. Krause, L.Clapham, A. Pattantyus and D. Atherton, Investigation of the stress-dependent magnetic easy axis in steel using magnetic Barkhausen noise, J. Appl. Phys., 79, 4242, (1996)] which creates idealized magnetic domain configurations for each grain. Using statistical mechanical arguments, we can then model the anhysteretic M versus H curve. 4.1 Background Magnetic Barkhausen Noise (MBN) arises from discontinuous changes in the magnetization of a material when subjected to an applied magnetic field. An excitation coil produces a magnetic field in the sample. The resulting discontinuous magnetization changes generate a voltage in the pickup coil (V BN ). A schematic of a surface Barkhausen apparatus which uses a feedback coil for flux control is shown in Figure 31 [1,2]. It is well known that stress is a significant factor in the magnetization response, primarily through the magnetostrictive effect [3]. Conceptually therefore, the magnetization response of a system can give information on its stress state. Extracting this information from surface MBN would provide a valuable non-destructive evaluation tool. In order to use this, one must first model the relation between stress and magnetization response. Models I and II to be discussed in this report are two such possibilities. Model I is a phenomenological model that is simpler to implement than Model II. However, it requires a number of fitting parameters to be determined. Although it can provide useful information of the system, it lacks the predictive power to answer more fundamental physical questions such as the effect of grain size on magnetization response. Model II requires more theoretical and computational resources than Model I, it is based on the magnetic object model and directly accounts for magnetic domain structures in the material. As such, one can examine more physical effects than Model I. 32 DRDC Atlantic CR

47 Figure 31: Schematic of MBN apparatus with feedback flux control [1]. After some signal processing, V BN and the resulting RMS (V RMS ) values will look like that of Figure 32. [ 2]. Figure 32: The instantaneous MBN voltage signal and the resulting RMS voltage from a sinusoidal applied magnetic field [2]. DRDC Atlantic CR