Nondestructive Testing with 3MA—An Overview of Principles and Applications

applied sciences

Article Nondestructive Testing with 3MA—An Overview of Principles and Applications

Bernd Wolter *, Yasmine Gabi and Christian Conrad Fraunhofer Institute for Nondestructive Testing IZFP, Campus E3 1, 66123 Saarbrücken, Germany; [email protected] (Y.G.); [email protected] (C.C.) * Correspondence: [email protected]; Tel.: +49-681-9302-3883

Received: 1 February 2019; Accepted: 8 March 2019; Published: 14 March 2019

Abstract: More than three decades ago, at Fraunhofer IZFP, research activities that were related to the application of micromagnetic methods for nondestructive testing (NDT) of the microstructure and the properties of ferrous materials commenced. Soon, it was observed that it is beneficial to combine the measuring information from several micromagnetic methods and measuring parameters. This was the birth of 3MA—the micromagnetic multi-parametric microstructure and stress analysis. Since then, 3MA has undergone a remarkable development. It has proven to be one of the most valuable testing techniques for the nondestructive characterization of metallic materials. Nowadays, 3MA is well accepted in industrial production and material research. Over the years, several equipment variants and a wide range of probe heads have been developed, ranging from magnetic microscopes with µm resolution up to large inspection systems for in-line strip steel inspection. 3MA is extremely versatile, as proved by a huge amount of reported applications, such as the quantitative determination of hardness, hardening depth, residual stress, and other material parameters. Today, specialized 3MA systems are available for manual or automated testing of various materials, semi-finished goods, and final products that are made of steel, cast iron, or other ferromagnetic materials. This paper will provide an overview of the historical development, the basic principles, and the main applications of 3MA.

Keywords: 3MA; case-depth; hardness; micromagnetic; NDT; residual stress; steel; tensile strength; yield strength

1. Introduction The laboratory usually determines the quality-related material properties of ferrous materials. Here, several measuring techniques are available, allowing for the microstructure to be directly analyzed, e.g., by x-ray diffraction and optical or electron microscopy. In addition, destructive testing methods are available for measuring the mechanical properties, such as hardness, tensile strength, etc., of metallic materials. However, today’s industry production is characterized by a high degree of automation, which is why it is also necessary to automate quality inspection. The quality characteristics of raw materials, semi-finished, and final products should be determined not only in material laboratories but also parallel to or embedded into production. Critical material characteristics should be continuously and completely registered in order to monitor or even control the manufacturing processes with the aim of optimizing them in terms of quality, efficiency, and costs. These objectives can only be achieved by the application of appropriate nondestructive testing (NDT) methods. In the beginning, NDT techniques have been almost exclusively used for defect detection in order to prevent the catastrophic failure of safety-relevant structures after they have been in service [1]. However, nowadays the role of NDT is extended and it includes all phases of the product’s life-cycle, from raw material over production to operation, disposal, and recycling [2].

Appl. Sci. 2019, 9, 1068; doi:10.3390/app9061068 www.mdpi.com/journal/applsci Appl. Sci. 2019, 9, 1068 2 of 29

Meanwhile, the application of NDT in production has become an indispensable part of modern industry and characterizing materials in terms of microstructure, condition, and properties are important tasks for today’s techniques of production-integrated NDT [3]. In this context, the so-called “micromagnetic” methods are probably the most prominent representatives of such techniques. Micromagnetic methods allow for the nondestructive characterization of all ferromagnetic materials, showing magnetic interactions with the microstructure and stresses on a microscopic level.

2. Historical Development Strictly speaking, the development of micromagnetic methods started more than 80 years ago [4]. In this early work, the electromagnetic properties of the material were correlated to other material properties, like elastic parameters. However, the name Friedrich Förster dominated the early days of the development of micromagnetic methods [5–7]. The idea to use micromagnetic NDT for production-integrated testing—in that case, material sorting—was initially reported in 1955 [8]. At least since 1964, micromagnetic methods have been used for the NDT of case depth in carburized steels [9]. Only shortly later, it was shown that magnetic Barkhausen Noise measurements could be used for investigating the stress condition in the magnetic materials [10]. Later, there have been some basic physical investigations, e.g., from Kronmüller, providing a better understanding of the relations between the electromagnetic properties of a material and its microstructure [11]. In the early 1970s, Förster and his co-researchers discussed the relationship between Rockwell hardness type C (HRC) and magnetic permeability in technical steels [12]. In addition, they first promoted electromagnetic tools for the quality assurance of metals by characterizing internal stresses via magnetic methods [13]. In the late 1970s, the refinement of the Barkhausen Noise technique marked major advances in the enhancement of microstructure and stress sensitivity. Since then, micromagnetic methods for nondestructive evaluation have attracted the attention of several research groups [14,15]. At about the same time, the researcher of Fraunhofer IZFP started to investigate and develop micromagnetic NDT methods. A first project was dedicated to the “determination of microstructure in pressure vessel steels with magnetically induced measuring parameters” on behalf of the German Federal Ministry for Research and Technology (BMFT) [15]. The main results led to the development of a specific surface sensor with an open U-shaped yoke design and the use of a Hall sensor in the control of a tangential magnetic field in the specimen. The correlation between the micromagnetic measuring parameter and the macro-stress level as the target parameter was reported [16]. It was recognized that different measuring parameters are sensitive to different target quantities, i.e., microstructure and stress parameters in the material. It was concluded that the micromagnetic measuring parameters complement each other in regard to their measuring information [17]. Based on this approach, the separation of microstructure effects from the influence of residual stresses could be discussed for the first time [18]. This research project resulted in the construction of the so-called EM8101—the first prototype of a micromagnetic testing equipment that was developed at Fraunhofer IZFP (see Figure1). This equipment combined two micromagnetic methods, namely Barkhausen Noise (BN) and incremental permeability (IP). Two micromagnetic measuring parameters, HCM and MMAX, have been derived from BN and two parameters, HCµ and µMAX from IP. By combining both techniques, it was possible to independently quantitatively characterize residual stress from changes in microstructure [19]. In this early micromagnetic research at IZFP, it became apparent that the use of one single micromagnetic parameter and also the use of parameters from only one micromagnetic method could lead to ambiguous or unreliable results. In fact, each micromagnetic measuring parameter is influenced by a multitude of microstructure- and/or stress-related material features. In some “simple” cases, demonstrated by Tiitto, a quantitative evaluation using only one measuring quantity could be successful [20]. This simple approach is restricted to applications where other influences remain constant or their changes can be neglected. Appl. Sci. 2019, 9, 1068 3 of 29

Appl. Sci.Since 2019 the, 9, x midFOR PEER 1980s, REVIEW the philosophy of IZFP was to extend the combined methods by3 of the 29 implementation of other techniques, such as Eddy Current. This was the birth of the Micromagnetic Multiparametric Microstructure and and stress Analyzer, abbreviated as 3MA [21,22]. [21,22]. Later, Pitsch and DobmannAppl. Sci. introduced 2019, 9, x FOR Harmonic PEER REVIEW Analysis as a fourth micromagnetic method in 3MA [[23].23]. From 3 of 29 that point, thethe 3MA 3MA success success story story started. started. Meanwhile, Meanwhile, a multitude a multitude of applications, of applications, probe heads,probe andheads, devices and devicesof theMultiparametric 3MA of the combination 3MA combination Microstructure technique technique hasand beenstress has developed. Analyzer, been developed. abbreviated In the following, In as the 3MA following, the[21,22]. basic Later,the principles basic Pitsch principles and the Dobmann introduced Harmonic Analysis as a fourth micromagnetic method in 3MA [23]. From that andapplications the applications of 3MA of will 3MA be described. will be described. point, the 3MA success story started. Meanwhile, a multitude of applications, probe heads, and devices of the 3MA combination technique has been developed. In the following, the basic principles and the applications of 3MA will be described.

FigureFigure 1. FirstFirst micromagnetic micromagnetic testing device EM8101 of Fraunhofer IZFP. Figure 1. First micromagnetic testing device EM8101 of Fraunhofer IZFP. 3. Basic Principles of 3MA 3. Basic Principles of 3MA 3MA is a methodical and technical combination of four micromagnetic methods, namely 3MA 3MAis a ismethodical a methodical and and technical technical combination combination of fourfour micromagneticmicromagnetic methods, methods, namely namely Barkhausen Noise (BN), Harmonic Analysis of the tangential magnetic ﬁeld strength (HA), BarkhausenBarkhausen Noise Noise (BN), (BN), Harmonic Harmonic Analysis Analysis of of the the tangentialtangential magneticmagnetic field field strength strength (HA), (HA), multi multi‐ ‐ frequencymulti-frequencyfrequency Eddy Eddy Current Eddy Current Current analysis analysis analysis (EC), (EC), and and (EC), Incremental Incremental and Incremental PermeabilityPermeability Permeability (IP) (IP) [25,26]. [25,26]. (IP)Figure Figure [24 2, 25shows 2]. shows Figure the the2 designshowsdesign theof the design of 3MA the of3MA probe the probe 3MA head. head. probe In theIn head. the following, following, In the following,the the four four methods the four are are methods described. described. are described.

(a) (b)

FigureFigure 2. Micromagnetic 2. Micromagnetic multiparametric multiparametric microstructure microstructure and stress analyzer analyzer (3MA) (3MA) probe probe head head with with (a) (b) its components:its components: (a) image (a) image and ( band) technical (b) technical sketch; sketch; 1-Housing, 1‐Housing, 2-Electronic 2‐Electronic board board (preamp), (preamp), 3-Magnetic 3‐ Magnetic yoke, 4‐Magnetization coils, 5‐Connection cable, 6‐Hall sensor, 7‐Transmitter coils, and 8‐ yoke,Figure 4-Magnetization 2. Micromagnetic coils, multiparametric 5-Connection cable,microstructure 6-Hall sensor, and 7-Transmitterstress analyzer coils, (3MA) and probe 8-Receiver head coils.with Receiver coils. its components: (a) image and (b) technical sketch; 1‐Housing, 2‐Electronic board (preamp), 3‐ Magnetic yoke, 4‐Magnetization coils, 5‐Connection cable, 6‐Hall sensor, 7‐Transmitter coils, and 8‐ Receiver coils. Appl. Sci. 2019, 9, 1068 4 of 29

3.1. Harmonic Analysis (HA) Method

A low frequency (fLF = 10–1000 Hz) sinusoidal voltage excitation is supplied into a magnetization coil (see Figure2, no. 4), wounded around a magnetic yoke (no. 3), which is placed on or in vicinity to the specimen. This generates a variable magnetization in the specimen. This magnetization can be chosen in the amplitude range between 10 and 70 A/cm. The amplitude is controlled via a Hall sensor (no. 6), which is placed in the middle of the detection zone, between the pole shoes of the magnetic yoke (no. 3). This Hall sensor is also used to record the entire tangential magnetic field development during the hysteresis cycle. A Fourier analysis processes this recorded signal. Due to the hysteresis symmetry, only odd higher harmonics can be determined. The amplitudes and phase shifts of these upper harmonics are observed. Further measuring parameters can be determined such as UHS (upper harmonics sum) and the signal distortion factor K, which are described in Table1. In total, the HA method provides 11 measuring parameters. Table1 shows all of the detected parameters. A1,A9, and P9 are not used as own parameters. The HA method permits the analysis of deeper material ranges and it can even be applied to shallower structural and stress (tension) gradients, as occurring in surface hardened parts, for example. For a specific excitation (yoke + magnetization coil) system, the electromagnetic interaction depth is only dependent on the magnetic field strength (Ht) and the excitation frequency (fLF) and it can reach depths of up to 10 mm. On the other side of the coin, the HA method only has a limited sensitivity to the near-surface properties of the material, when compared to the BN (see Section 3.3) or IP (see Section 3.4) method.

Table 1. Measuring parameters from Harmonic Analysis (HA).

Measuring Parameter Description

A1 Amplitude of the fundamental wave. A3,A5,A7,A9 Amplitudes of the 3rd, 5th, 7th and 9th harmonics. P3,P5,P7,P9 Phases of the 3rd, 5th, 7th and 9th harmonics. UHS Sum of all upper harmonics, UHS = A3 + A5 + A7 + A9 √ 2 2 2 2 K Distortion factor, K = 100%∗ A3 + A5 + A7 /A1 HCO Coercive magnetic ﬁeld from harmonic analysis Hro Harmonic content of the magnetic ﬁeld strength at zero crossing Vmag Final stage voltage of the electromagnet

3.2. Eddy Current (EC) Method A sinusoidal current with low amplitude, typically of the order of mA, and high frequency (fHF = 10 kHz − 1 MHz) is fed into the transmitter coil with a diameter of only a few mm (see Figure2, no. 7). In the specimen, this excitation results in a low level of induction of a few µT (Rayleigh domain). The magnetic ﬁeld induces electrical currents (eddy currents) in the specimen according to Faraday’s Law of electromagnetic induction. A second coil, the receiver coil (no. 8) is used to detect the magnetic ﬂux through its windings as an induced voltage. Based on a signal analysis, the real and imaginary part, as well as the modulus and phase of the coil impedance, are recorded. The 3MA system offers to apply up to four different EC frequencies in a single measuring cycle, resulting in 16 measuring parameters, as shown in Table2. These measuring parameters are dependent on the frequency f HF and on both the conductivity (σ) and the excitation-dependent permeability (µ(H)) of the material.

Table 2. Measuring parameters from multi-frequency Eddy Current (EC).

Measuring Parameter Description Re1–Re4 Real parts of EC signals at frequencies 1, 2, 3 and 4. Im1–Im4 Imaginary parts of EC signals at frequencies 1, 2, 3 and 4. Mag1–Mag4 Signal magnitudes at frequencies 1, 2, 3 and 4. Ph1–Ph4 Signal phases at frequencies 1, 2, 3 and 4. Appl. Sci. 2019, 9, 1068 5 of 29

3.3. Barkhausen Noise (BN) Method In order to induce the reorganization of the magnetic microstructure and thus to provide the Barkhausen Noise (BN) response of the specimen, the application of a variable magnetic field is necessary. This magnetic field is applied by the U-shaped yoke (see Figure2, no. 3) with the magnetization coil (no. 4), as in case of the HA method, as described in Section 3.1. The applied magnetization amplitudes are sufficient to excite the inspected ferromagnetic materials up to their saturation levels. The magnetization frequency is adapted in order to avoid local eddy currents, thus not changing or distorting the magnetic hysteresis behavior in the specimen. During the reorganization of the ferromagnetic microstructure, Bloch wall displacements are generated, which occur by discreet jumps. These Bloch wall jumps result in sudden changes of magnetization, which causes magnetic flux variations. These variations can be detected with a receiver coil (no. 8) that is placed above the sample. After signal processing, which consists of several steps, like amplification and filtering, the 7 measuring parameters that are shown in Table3 can be derived from the BN signal. The BN method uses analyzer frequencies mostly in the high kHz range. Therefore, only the near-surface area of the material can be inspected. A further disadvantage of the method is its sensitivity to electromagnetic interference, which is a result of the broadband signal analysis in conjunction with high amplification.

Table 3. Measuring parameters from Barkhausen Noise (BN).

Measuring Parameter Description

MMAX Maximum amplitude

MMEAN Amplitude averaged over one magnetization cycle

MR Amplitude at remanence point Coercive magnetic ﬁeld, derived from Barkhausen Noise HCM (magnetic ﬁeld strength at M = MMax)

DH25M Curve width at 25% of MMAX

DH50M Curve width at 50% of MMAX

DH75M Curve width at 75% of MMAX

3.4. Incremental Permeability (IP) Method The Incremental Permeability (IP) method provides measuring parameters, which are equivalently deﬁned to the BN measuring parameters (see Table4).

Table 4. Measuring parameters from Incremental Permeability (IP).

Measuring Parameter Description

µMAX Maximum amplitude

µMEAN Amplitude averaged over one magnetization cycle

µR Amplitude at remanence point Coercive magnetic ﬁeld, derived from Incremental Permeability HCµ (magnetic ﬁeld strength at µ = µMax)

DH25µ Curve width at 25% of µMAX

DH50µ Curve width at 50% of µMAX

DH75µ Curve width at 75% of µMAX

The IP method requests two simultaneously acting excitation sources. These are a ﬁrst high-amplitude low-frequency (fLF) excitation, as in case of the HA method (see Section 3.1) and a second low-amplitude high-frequency (fHF) excitation, as in case of the EC method (see Section 3.2). Appl. Sci. 2019, 9, 1068 6 of 29

Appl. Sci.The 2019 fLF, 9excitation, x FOR PEER by REVIEW the magnetization yoke (see Figure2, no. 3 and 4) generates magnetic6 of 29 hysteresis cycles in the material. Simultaneously, the EC transmitter coil (see Figure2, no. 7) generates detectedminor asymmetric with the receiver hysteresis coil loops (no. 8) that is linked are superimposed to incremental to thepermeability. major hysteresis In contrast curve. to TheBN, signalthe IP methoddetected shows with the only receiver little fluctuations coil (no. 8) isduring linked the to hysteresis incremental cycle. permeability. In addition, In contrastdue to the to BN,applied the narrowIP method‐band shows signal only filtering, little fluctuations the IP method during is relatively the hysteresis insensitive cycle. to In electromagnetic addition, due to distortions. the applied narrow-band signal filtering, the IP method is relatively insensitive to electromagnetic distortions. 3.5. Correlations to Microstructure and Material Properties 3.5. Correlations to Microstructure and Material Properties Coercive field, Hc, remanence, Br, as well as initial and maximum permeability, μrinit and μrmax are “hysteresisCoercive field, parameters”, Hc, remanence, i.e., measuring Br, as well parameters as initial that and are maximum determined permeability, from the majorµrinit andhysteresisµrmax curve.are “hysteresis Variations parameters”, in the microstructure i.e., measuring of a parameters ferromagnetic that arematerial determined are reflected from the by major changes hysteresis in the hysteresiscurve. Variations morphology in the and microstructure in changes ofof athese ferromagnetic parameters. material Therefore, are reflected the hysteresis by changes parameters in the showhysteresis correlations morphology to microstructure and in changes data. of In these an early parameters. work, Adler Therefore, and Pfeiffer, the hysteresis who reported parameters a linear correlationshow correlations between to Hc microstructure and the grain data. size In or an the early concentration work, Adler of and nonmagnetic Pfeiffer, who inclusion reported in a steels, linear showedcorrelation this between [24]. Researchers Hc and the at grain Fraunhofer size or theIZFP concentration have shown of correlations nonmagnetic between inclusion Hc inand steels, the approximateshowed this [value24]. Researchersof skin depth at in Fraunhofer grinded parts IZFP [25]. have Later, shown such correlations correlations between were Hcused and for the to determineapproximate the value quantitative of skin hardening depth in grindeddepth in partssteel and [25]. cast Later, iron suchcomponents, correlations while were using used 3MA for [26]. to determineBesides the “conventional” quantitative hardening measuring depth parameter,s in steel and such cast ironas coercive components, fields while (HCO using, HCM, 3MA HCμ) [ 26or]. maximumBesides amplitudes “conventional” (MMAX, μ measuringMAX), additional parameters measuring such parameters as coercive have fields been (H COdefined,,HCM as,H alreadyCµ) or describedmaximum in amplitudes Tables 1–4. (M SuchMAX, “unconventional”µMAX), additional measuringparameters parameters could be useful, have been as it defined,becomes as apparent already indescribed observing in Tables the diagrams1–4. Such in “unconventional” Figure 3. The diagram parameters (a) shows could hysteresis be useful, asmeasurements it becomes apparent on a steel in sampleobserving before the diagrams (I) and after in Figure (II) plastic3. The deformation. diagram (a) Measurements shows hysteresis with measurements the Incremental on a Permeability steel sample (IP)before method (I) and on after the same (II) plastic sample deformation. are shown in diagram Measurements (b). The with shape the of Incremental hysteresis drastically Permeability changes (IP) frommethod I to on II the due same to samplethe increased are shown dislocation in diagram density (b). The and shape altered of hysteresis residual drasticallystress level. changes Especially from noticeableI to II due tois thesome increased kind of dislocation“bulging” around density the and coercive altered residual field in hysteresis stress level. II. Especially This corresponds noticeable to isa distinctsome kind change of “bulging” in the width around of the the IP coercive signal. field Therefore, in hysteresis the widths II. This of corresponds the IP signal to at a 25, distinct 50, and change 75% ofin the widthmaximum of the amplitude IP signal. contain Therefore, valuable the widths additional of the measuring IP signal at information. 25, 50, and 75% of the maximum amplitude contain valuable additional measuring information.

(a) (b)

Figure 3. Measurements on a steel sample before (I) and after (II) plasticplastic deformation:deformation: (a) hysteresis measurements and (b) IP measurements.

3.6. Set Set-up‐up Optimization and Calibration

The 3MA software offers the possibility of running sweeps of several set set-up‐up parameters (e.g., H t,, fLF,, and and f fHFHF) )in in order order to to select select their their values, values, which which are are optimized optimized according according to to the the inspection inspection situation. situation. In general, thethe optimumoptimum set-up set‐up parameters parameters should should provide provide the the clearest clearest signal signal differences differences between between the thegrades grades of the of the interesting interesting target target quantity, quantity, e.g., e.g., hardness, hardness, residual residual stress, stress, or or case case depth. depth. For example, for case hardening depth (CHD), it is beneﬁcialbeneficial if the interactioninteraction depth of the used electromagnetic field is in the range of the highest value of CHD. The set‐up frequencies (fLF, fHF) and magnetization amplitude (Ht) can adjust this interaction depth. If the interaction depth is much higher than the maximum CHD, then the 3MA measurement explores a large amount of unhardened bulk material, which impairs the sensitivity to CHD variations. Appl. Sci. 2019, 9, 1068 7 of 29

field is in the range of the highest value of CHD. The set-up frequencies (fLF, fHF) and magnetization amplitude (Ht) can adjust this interaction depth. If the interaction depth is much higher than the maximum CHD, then the 3MA measurement explores a large amount of unhardened bulk material, which impairs the sensitivity to CHD variations. A preceding calibration is mandatory for using 3MA to non-destructively determine the quantitative values of a target quantity. Here, calibration means the determination of a mathematical model—the so-called calibration function, which describes the functional relationship between the target quantity and the 3MA measuring parameters. At the early stages of micromagnetic characterization, several scientists tried to develop physical models that describe the functional dependency between micromagnetic parameters and microstructure features or material properties. Becker and Karsten both described a linear dependency between the coercive field and Bloch wall diameter. They observed that the coercive field is the field that is necessary to obtain an irreversible displacement of the Bloch walls [27,28]. Later, Néel extended this work by considering the demagnetizing field in the calculation of total energy. He noticed that the volume and size of inclusions affects the coercive field [29]. Other scientists have developed mathematical functions describing the coercive field, while taking into account the shape and size of particles [30], the grain size [31], and anisotropy effects [32,33]. Hoselitz investigated the influence of alloying elements, like Al, Ni, and Co, as well as measuring and calculating the coercive field for Alcomax® isotropic steel [34]. Such theoretical calculations of microstructure-property relations have been shown to be reliable for simple materials and processes. However, most of the ferromagnetic materials that were used in industry, e.g., steels and cast iron, show complex microstructures and behavior. In order to theoretically develop a model describing the interrelation between targets and measuring parameters, the inverse problem has to be solved, i.e., the microstructure characteristics as a function of the micro-magnetic measuring parameters must be theoretically described. This is not usually possible. Moreover, in the case of industrial 3MA application, calibration functions have to be determined in a fast uncomplicated manner. However, the development of simulation models for calibration is time consuming and expensive. Therefore, the application of physical models based on theoretical considerations for 3MA calibration remains limited so far. Currently, 3MA is usually calibrated based on empirical data. In order to determine a calibration function, the values of the target quantity, as well as the values of the 3MA measuring parameters, must be measured on a set of parts with graded values of the target quantity. To measure these target values (mostly destructive), the reference methods have to be applied, like hardness indenter tests, tensile tests, or x-ray diffraction for residual stress. The measuring values of both, 3MA parameters and target quantity are stored in a common database, and then the correlations between both are analyzed. The 3MA software offers different methods for the calculation of the calibration functions, including regression analysis and pattern recognition. In the case of regression analysis, the calibration function can be written as: Y = a0 + a1.X1 + a2.X2 + ··· + an.Xn, (1) where Y is the target quantity. The parameters an (n = 1, 2, 3, ... ) are the coefficients that have to be determined by the least squares method error minimization and Xn (n = 1, 2, 3, ... ) are the 3MA measuring parameters and their modifiers, such as square, square root, etc. Several calibration functions for different target quantities can be determined at once. Besides the 3MA measuring parameters, also parameters that are measured by external sensors, such as temperature, sensor lift-off, sheet thickness, strip velocity, and tension could be included in the calibration functions. Currently, different strategies are developed in order to reduce the experimental effort for calibration. For some special materials and components (e.g., for PHS, see Section 4.3.1) large calibration databases already exist, which allow for developing universally applicable calibration functions. In addition to experiments, simulation results can also be used to fill the database. Such artificially Appl. Sci. 2019, 9, 1068 8 of 29 generated data also allow for minimizing the risk for experimental calibration errors due to “outliers”. In addition, machine-learning algorithms are applied for determining the calibration functions.

4. Applications

4.1. Overview Meanwhile, a broad range of different applications on different components of ferromagnetic materials has been developed based on the 3MA approach. Table5 summarizes these applications. Subsequently, some of the main important 3MA applications will be described in more detail.

Table 5. Overview of possible 3MA applications.

No. Component Main Application/Description Ref. Components in nuclear Materials damage and degradation; 1 [35,36] power plants Neutron-induced embrittlement Welds (laser, etc.), turbine 2 Hardness and residual stress [37–39] blades, bearing rings, etc. Machined parts (gear Thermally induced material damage due to 3 [40–42] wheels, etc.) machining Surface hardened or nitrated Hardening depth, depth proﬁles of hardness and 4 [43–45] parts residual stress, retained austenite content Classiﬁcation of steel grades and microstructure 5 Steel grades [46,47] variants Residual stress, cementite content, primary chill, 6 Cast iron [48–50] microstructure gradient Mechanical properties, microstructure features 7 Strip steel [51,52] (texture, grain size) and residual stress Heavy plate and forged Mechanical properties, microstructure features, 8 [52,53] parts residual stress, hard spots 9 Cold formed steel sheets Residual stress and spring back angle [54,55] Mechanical properties of steel, coating thickness, 10 Hot formed steel sheets [56,57] resistance spot weld size 11 Electrical steel Inspection of cut edges; cutting quality [57,58] Hydrogen-induced embrittlement, fatigue, 12 Miscellaneous [59–62] toughness, notch impact strength, creep damage

One of the earliest applications of 3MA was the inspection of components in nuclear power plants. Such components are subjected to different ageing phenomena. For the steel components in the reactor pressure vessel, heat exchangers, pipe lines, etc., the main important aging phenomena are thermal ageing, fatigue, and neutron embrittlement. In the past, it was shown that 3MA is an important tool in evaluating material degradation due to ageing in such components [35]. Besides neutron degradation, the superimposed impact from thermal ageing and low-cycle fatigue was investigated. In the case of austenitic stainless steels when exposed to mechanical static or cyclic loads, the material reacts with local phase transformations to generate bcc α’ martensite. 3MA can also detect such local martensite [36]. The nondestructive determination of hardness and residual stress distribution in the weld seam and in the heat-affected zone of welded components was another early 3MA application [37]. Later, 3MA was applied to determine the hardness and residual stress in heat-treated (turbine blades) or grinded components (e.g., bearing rings) [38,39]. Another well-developed 3MA application is the determination and characterization of grinding damage (“grinding burns”) [40–42]. 3MA is able to Appl. Sci. 2019, 9, 1068 9 of 29 determine the depth-proﬁles of hardness, residual stress, and retained austenite in surface-hardened components [43–45]. The classiﬁcation tasks can also be solved with 3MA [46,47]. Besides steel, cast iron is another material that can be characterized with 3MA [48–50]. Section 4.2 describes the 3MA applications on strip steel in detail [51,52]. For thin strip, heavy plate, and large forged steel components, 3MA is mainly used to determine the parameters of the tensile test, e.g., tensile strength Rm and yield strength Rp0.2 [52,53]. In further processing, when the steel sheets are cold or hot formed, 3MA is used for determining the mechanical properties and also the residual stress or spring back angle [54–57]. An interesting application of 3MA is the determination of electromagnetic properties (e.g., iron losses, maximum induction) in electrical steel, because 3MA is able to locally determine these properties, which is not possible with Epstein frames, etc. [57,58]. Furthermore, 3MA can be used to characterize hydrogen-induced embrittlement, fatigue, toughness, notch impact strength, and creep damage in steel [59–61].

4.2. Applications in the Steel Industry

4.2.1. Inline Strip Steel Testing Cold rolled and recrystallizing annealed sheet steel is a high-value product that is used, among other uses, for automotive body shell parts. Material parameters, such as yield strength Rp0.2, tensile strength Rm, and anisotropy parameters rm and ∆r must be met with sufficient accuracy. In production, the strips are welded one after another to a coil of several kilometers in length. Typically, the material properties from above are tested by destructive tensile tests. It is clear that these tests cannot be done over the entire strip length. Therefore, only samples from the beginning and the end of the coil are taken for these tests. These samples undergo extensive and time-consuming tensile tests before the coil can be released for shipping. The time delay between sample collection and testing can reach up to several hours. A quick detection of, and reaction to, deviations in the production process is not possible with this time delay between sampling and testing. In order to overcome these limitations, nondestructive testing (NDT) systems for the continuous inline determination of mechanical parameters in the entire steel strip are required. More than 20 years ago, Fraunhofer IZFP started research activities that were related to the development of such NDT equipment for strip steel testing. A first prototype system is described in a PhD thesis from 1997 [63]. This system, as shown in Figure4, was installed in a production line of ThyssenKrupp Steel in Duisburg, Germany. The prototype system was integrated after annealing and skin-path-rolling processes and just before coiling. At this point, the strip feed is in the range of 300 m/min. In this equipment, measuring quantities, as derived from the Incremental Permeability (IP) method, were used as the only micromagnetic parameters. Here, the narrow-band-filtering suppressing most of the wide-band environmental noise, which is typically present in strip steel production lines, has turned out to be particularly advantageous. Measuring quantities, as derived from the IP signal, were µMAX, Hcµ, as well as DH25µ, DH50µ, and DH75µ (see Section 3.4). These parameters have shown to be sensitive to the yield strength Rp0.2 of the strip steel. A further disturbance to be considered is the lift-off variation due to vibrations or ripple of the running strip. The electromagnetic signal detection depends on the distance between the sensor coil and the material’s surface, i.e., the lift-off. In order to reduce its disturbing influence to an acceptable level, controlling the lift-off in the range to 2 mm ± 0.5 mm was necessary, which was realized by the pretension of the strip with two rollers that were integrated into a roller-table. Further attempts were made to determine correlations between the nondestructive measuring parameters and anisotropy parameters of the steel, like the mean vertical anisotropy rm and the planar anisotropy ∆r. For this purpose, it turned out to be advantageous to also implement the so-called EMAT technique (electromagnetic acoustic transducer), in addition to the IP technique. The EMAT principle is described elsewhere [64]. Appl. Sci. 2019, 9, x FOR PEER REVIEW 9 of 29 electromagnetic properties (e.g., iron losses, maximum induction) in electrical steel, because 3MA is able to locally determine these properties, which is not possible with Epstein frames, etc. [57,58]. Furthermore, 3MA can be used to characterize hydrogen‐induced embrittlement, fatigue, toughness, notch impact strength, and creep damage in steel [59–61].

4.2. Applications in the Steel Industry

4.2.1. Inline Strip Steel Testing Cold rolled and recrystallizing annealed sheet steel is a high‐value product that is used, among other uses, for automotive body shell parts. Material parameters, such as yield strength Rp0.2, tensile strength Rm, and anisotropy parameters rm and r must be met with sufficient accuracy. In production, the strips are welded one after another to a coil of several kilometers in length. Typically, the material properties from above are tested by destructive tensile tests. It is clear that these tests cannot be done over the entire strip length. Therefore, only samples from the beginning and the end of the coil are taken for these tests. These samples undergo extensive and time‐consuming tensile tests before the coil can be released for shipping. The time delay between sample collection and testing can reach up to several hours. A quick detection of, and reaction to, deviations in the production process is not possible with this time delay between sampling and testing. In order to overcome these limitations, nondestructive testing (NDT) systems for the continuous inline determination of mechanical parameters in the entire steel strip are required. More than 20 years ago, Fraunhofer IZFP started research activities that were related to the development of such NDT equipment for strip steel testing. A first prototype system is described in a PhD thesis from 1997 [63]. Appl.This Sci.system,2019, 9 ,as 1068 shown in Figure 4, was installed in a production line of ThyssenKrupp Steel10 of in 29 Duisburg, Germany.

Appl. Sci. 2019, 9, x FOR PEER REVIEW 10 of 29

A further disturbance to be considered is the lift‐off variation due to vibrations or ripple of the running strip. The electromagnetic signal detection depends on the distance between the sensor coil and the material’s surface, i.e., the lift‐off. In order to reduce its disturbing influence to an acceptable level, controlling the lift‐off in the range to 2 mm ± 0.5 mm was necessary, which was realized by the pretension of the strip with two(a) rollers that were integrated into( ba) roller‐table. Further attempts were made to determine correlations between the nondestructive measuring parameters and anisotropy Figure 4. First prototype system for continuous testing of strength (Rp0.2) andand deep-drawingdeep‐drawing (r(rmm and m  parameters∆r)r) parameters of the insteel,in strip strip like steel steel the based based mean on on a combinationvertical a combination anisotropy of the of Incrementalthe r Incremental and the Permeability planar Permeability anisotropy (IP) and (IP) the and EMATr. theFor this purpose,EMATmeasuring it measuring turned technique: outtechnique: (ato) Topbe ( a view)advantageous Top of view the sensor,of the to sensor, showing also showingimplement the IP sensorthe IP the sensor (IP) so as ‐ well(IP)called as as well theEMAT twoas the EMATtechnique two (electromagneticEMATsensors sensors (EMAT_RD (EMAT_RDacoustic and EMAT_45transducer), and EMAT_45°);◦); and, in ( and,baddition) Complete (b) Complete to table the table carrierIP carriertechnique. with with two two stabilizationThe stabilization EMAT rollers. principle rollers. is described elsewhere [64]. TheThe prototypeEMAT sensors sensors system were were was used integrated used to tomeasure measureafter annealingthe thephase phase and velocity skin velocity‐ pathof the‐ ofrolling first the symmetricalprocesses ﬁrst symmetrical and shear just‐ beforeshear-horizontalhorizontal coiling. plate At platethismode point, mode of anthe of ultrasound anstrip ultrasound feed is (US) in the (US) wave range wave that of that 300 was wasm/min. travelling travelling In this in inequipment, the the rolling measuring directiondirection ◦ ◦ (EMAT_RD)quantities,(EMAT_RD) as and andderived 4545° totofrom thisthis the (EMAT_45(EMAT_45°). Incremental). Based Permeability onon thesethese (IP) measurements,measurements, method, were rrmm usedandand ∆asrr couldthecould only bebe micromagneticdetermineddetermined [ 51[51].]. parameters. Investigations Investigations Here, were were the conducted conducted narrow on‐band on a soft a‐filtering soft deep-drawing deep suppressing‐drawing material material most of interstitial-free of of theinterstitial wide‐band steel‐free (IF-steel).environmentalsteel (IF‐steel). Three noise,Three different whichdifferent grades is typicallygrades of IF steel of present IF with steel differences inwith strip differences steel in nitrogen, production in nitrogen, niobium, lines, niobium, and has titaniumturned and out contentstitanium to be wereparticularlycontents used. were Theadvantageous. used. inline The calibration inlineMeasuring calibration was quantities, carried was out ascarried on derived 353 out strips from on of the353 these IPstrips signal, steel of grades. werethese μ steelMAX Destructive, Hcgrades.μ, as welltensileDestructive as testsDH25 (Rp0.2)tensileμ, DH50 andtestsμ, strain and(Rp0.2) DH75 tests andμ (rm(see strainand Section∆ testsr) were 3.4).(rm used andThese inr) orderparameters were to used determine have in order shown the to reference todetermine be sensitive values the toforreference the calibration. yield values strength Figures for calibration.Rp0.25 and of the6 show Figures strip the steel. 5 typicaland 6 show calibration the typical results. calibration These results. calibration These plots calibration show theplots nondestructively show the nondestructively predicted valuespredicted of Rp0.2values (from of Rp0.2 3MA (from and 3MA from and EMAT) from asEMAT) a function as a function of the destructivelyof the destructively measured measured values values (from ref.). (from ref.).

(a) (b)

FigureFigure 5.5. ResultsResults ofof thethe calibrationcalibration toto yieldyield strengthstrength Rp0.2Rp0.2 basedbased onon measuringmeasuring parametersparameters fromfrom IPIP measurements,measurements, combinedcombined withwith sheetsheet thickness;thickness; nondestructivelynondestructively predictedpredicted valuesvalues ofof yieldyield strengthstrength (from(from IP) IP) are are shown shown as as a functiona function of of destructively destructively measured measured values values (from (from ref.): ref.): (a) for (a) a for steel a steel grade grade IF_1 andIF_1 ( band) for (b a) steelfor a gradesteel grade IF_2. IF_2.

(a) (b)

Figure 6. Results of the calibration to deep‐drawing parameters based on measuring parameters from electromagnetic acoustic transducer (EMAT) measurements; nondestructively predicted values (from Appl. Sci. 2019, 9, x FOR PEER REVIEW 10 of 29

Appl. Sci. 2019, 9, 1068 11 of 29

(a) (b) microstructureFigure 5. Results gradient of the resulting calibration from to skin-pathyield strength rolling. Rp0.2 The based electromagnetic on measuring parameters ﬁelds that from were IP used for IPmeasurements, measurements combined cannot with completely sheet thickness; penetrate nondestructively the material predicted thickness values (skin of depth). yield strength Therefore, the microstructure(from IP) are shown at the as surface a function of the of destructively material affected measured by skin-path values (from rolling ref.): will (a) for have a steel a more grade or less pronouncedIF_1 and effect (b) for on a steel the IP grade signal, IF_2. depending on the sheet thickness.

Appl. Sci. 2019, 9, x FOR PEER REVIEW 11 of 29

EMAT) are shown as a function of destructively measured values (from ref.): (a) for vertical

anisotropy rm and (b) for the planar anisotropy r.

It was apparent that the best correlations could be found if an individual calibration model for each steel grade is used, as shown(a) in the examples of Figures 5 and 6.(b It) was also observed that sheet thicknessFigureFigure has 6.6. Results a distinct of the effect calibration on the tomeasured deep-drawingdeep‐drawing IP signals. parameters This basedbasedis due onon to measuring the material’s parametersparameters microstructure fromfrom gradientelectromagneticelectromagnetic resulting acousticacousticfrom skin transducertransducer‐path (EMAT)rolling.(EMAT) measurements;measurements;The electromagnetic nondestructivelynondestructively fields predictedpredictedthat were valuesvalues used (from(from for IP measurementsEMAT) are showncannot as acompletely function of destructivelypenetrate the measured material values thickness (from ref.): (skin (a) for depth). vertical Therefore, anisotropy the microstructurerm and (b) for at thethe planar surface anisotropy of the material∆r. affected by skin‐path rolling will have a more or less pronounced effect on the IP signal, depending on the sheet thickness. For thatthat reason,reason, itit makesmakes sensesense toto eithereither useuse anan individualindividual calibrationcalibration modelmodel forfor eacheach sheetsheet thicknessthickness or toto integrateintegrate thethe sheetsheet thicknessthickness asas aa parameterparameter inin thethe calibrationcalibration functions.functions. TheThe latterlatter variantvariant waswas applied applied in thein the work work that isthat described is described here. As here. shown As inshown Figure in5, theFigure correlation 5, the coefﬁcientscorrelation 2 Rcoefficientswere quite R2 were high quite (0.7–0.8) high and (0.7–0.8) the root and mean the root square mean errors square RMSE errors were RMSE quite were low quite (4–6 low MPa). (4–6 2 TheMPa). calibrations The calibrations for rm andfor r∆mr and provided r provided values forvalues R that for R were2 that in were the range in the between range between 0.6 and 0.80.6 andand RMSEs0.8 and betweenRMSEs between 0.04 and 0.04 0.06. and Nevertheless, 0.06. Nevertheless, it has to be it has considered to be considered that the value that the ranges value are ranges quite low are forquite these low calibrations, for these calibrations, which promotes which smallpromotes RMSE small values. RMSE values. Based onon the the above above calibrations, calibrations, continuous continuous inline inline measurements measurements of Rp0.2, of Rp0.2, rm, and rm,∆ andr were r made.were Figuresmade. Figures7 and8 show 7 and some 8 show of thesome results. of the In results. the measuring In the measuring plot of Figure plot 7of, itFigure can be 7, observed it can be thatobserved the stripthat the leaves strip the leaves acceptance the acceptance range (130–180 range (130–180 MPa) at MPa) position at position x = 2100 x m. = 2100 This m. is laterThis is conﬁrmed later confirmed by the destructiveby the destructive measurements. measurements. The results The ofresults the inline of the measurements inline measurements of rm and of ∆rmr inand the r same in the coil same are showncoil are in shown Figure in8 .Figure 8.

Figure 7.7. Inline measurement of yield strength Rp0.2 inin aa coilcoil ofof 23002300 mm inin length. length. DestructiveDestructive measurements of Rp0.2 at the beginning and at the end of thethe stripstrip areare alsoalso shownshown asas whitewhite squaressquares withwith errorerror bars.bars.

Figure 8. Inline measurement of vertical anisotropy rm and planar anisotropy r in a coil of 2300 m in length. Appl. Sci. 2019, 9, x FOR PEER REVIEW 11 of 29

EMAT) are shown as a function of destructively measured values (from ref.): (a) for vertical

anisotropy rm and (b) for the planar anisotropy r.

It was apparent that the best correlations could be found if an individual calibration model for each steel grade is used, as shown in the examples of Figures 5 and 6. It was also observed that sheet thickness has a distinct effect on the measured IP signals. This is due to the material’s microstructure gradient resulting from skin‐path rolling. The electromagnetic fields that were used for IP measurements cannot completely penetrate the material thickness (skin depth). Therefore, the microstructure at the surface of the material affected by skin‐path rolling will have a more or less pronounced effect on the IP signal, depending on the sheet thickness. For that reason, it makes sense to either use an individual calibration model for each sheet thickness or to integrate the sheet thickness as a parameter in the calibration functions. The latter variant was applied in the work that is described here. As shown in Figure 5, the correlation coefficients R2 were quite high (0.7–0.8) and the root mean square errors RMSE were quite low (4–6 MPa). The calibrations for rm and r provided values for R2 that were in the range between 0.6 and 0.8 and RMSEs between 0.04 and 0.06. Nevertheless, it has to be considered that the value ranges are quite low for these calibrations, which promotes small RMSE values. Based on the above calibrations, continuous inline measurements of Rp0.2, rm, and r were made. Figures 7 and 8 show some of the results. In the measuring plot of Figure 7, it can be observed that the strip leaves the acceptance range (130–180 MPa) at position x = 2100 m. This is later confirmed by the destructive measurements. The results of the inline measurements of rm and r in the same coil are shown in Figure 8.

Figure 7. Inline measurement of yield strength Rp0.2 in a coil of 2300 m in length. Destructive Appl. Sci.measurements2019, 9, 1068 of Rp0.2 at the beginning and at the end of the strip are also shown as white squares12 of 29 with error bars.

FigureFigure 8.8. Inline measurement of vertical vertical anisotropy anisotropy r rmm andand planar planar anisotropy anisotropy r∆ inr in a coil a coil of of 2300 2300 m min Appl. inSci.length. length. 2019 , 9, x FOR PEER REVIEW 12 of 29

AtAt aa laterlater point,point, thethe prototypeprototype thatthat isis describeddescribed aboveabove waswas substitutedsubstituted byby anan inlineinline measuringmeasuring systemsystem basedbased onon thethe completecomplete 3MA3MA technique,technique, i.e.,i.e., besidesbesides IPIP alsoalso BN,BN, HA, and EC were implemented inin thethe system.system. The use of EMATEMAT sensorssensors forfor measuringmeasuring deep-drawingdeep‐drawing parametersparameters waswas nono longerlonger necessary.necessary. Based Based on on the the 3MA 3MA technique, technique, an an online online testing testing system system for for the the continuous continuous determination determination of tensileof tensile strength strength (Rm), (Rm), yield yield strength strength (Rp0.2), (Rp0.2), and and other other mechanical–technological mechanical–technological characteristics characteristics in in a broada broad range range of steel of steel grades grades was developed.was developed. The 3MA The probe3MA headprobe is head integrated is integrated into a movable, into a movable, rotatable holderrotatable (see holder Figure (see9). TheFigure orientation 9). The orientation between the between main magnetic the main ﬁeld magnetic and the field rolling and direction the rolling of ◦ ◦ ◦ thedirection strip can of the be adjustedstrip can tobe 0 adjusted, 45 and to 90 0°,, 45°and which offers 90°, which the possibility offers the of possibility determining of determining the anisotropy the parametersanisotropy parameters rm and ∆r fromrm and the r 3MA from measurements.the 3MA measurements. Again, the Again, probe the holder probe is holder mounted is mounted onto a hydraulicallyonto a hydraulically adjustable adjustable table carrier, table which carrier, was equippedwhich was with equipped distance with rollers. distance A fast loweringrollers. A of fast the entirelowering construction of the entire in theconstruction case of an in escape the case warning of an escape is necessary warning to avoidis necessary damaging to avoid the probedamaging due tothe weld probe seams due orto zincweld deposits seams or on zinc the deposits strip. The on external the strip. electronic The external equipment electronic that equipment is integrated that into is theintegrated measuring into cabin the measuring controls the cabin measuring controls procedure. the measuring The electronic procedure. equipment The electronic mainly equipment consists of themainly 3MA consists measuring of the module, 3MA measuring the electro-pneumatic module, the device electro control,‐pneumatic interfaces device to control, supporting interfaces signals, to likesupporting weld seam signals, recognition, like weld meter seam pulse, recognition, or strip meter velocity pulse, signals, or strip and velocity interfaces signals, to the and plant interfaces process controlto the plant (PCS) process and data control management (PCS) and (DMS)data management systems of (DMS) the strip systems steel production of the strip steel line. production line.

Figure 9.9. Rotatable 3MA probe headhead holderholder andand tabletable carriercarrier ofof thethe 3MA3MA inlineinline stripstrip testingtesting system.system.

ForFor calibration,calibration, about about 11,000 11,000 strips strips from from 12 12 different different steel steel grades grades were were used. used. In contrast In contrast to the to ﬁrst the IP+EMATfirst IP+EMAT equipment equipment described described above, hereabove, it washere not it necessarywas not tonecessary develop individualto develop calibrations individual forcalibrations each steel for grade. each steel Rather, grade. it was Rather, found it thatwas thefound 12 steelthat the grades 12 steel could grades be combined could be combined into only four into calibrationonly four calibration classes, which classes, are which shown are in Table shown6. in Table 6.

Table 6. Steel grade classes and calibration results.

Steel Grade Class RMSE of Rm [MPa] RMSE of Rp0.2 [MPa] No. of Strips IF, conventional 5.4 8.2 2667 IF, high strength 11.3 12.3 7764 Bakehardening 5.8 8.8 1294 Structural 7.9 10.1 164

Besides standard steel qualities (normal steel and construction steel), interstitial‐free (IF) steel, and high strength material were investigated. The 3MA data that were determined at the beginning and at the end of the strip were associated to the destructively determined data of Rm and Rp0.2 data in order to determine the calibration functions. Table 6 also shows the results of this calibration in terms of the RMSE of the correlation between the 3MA data and destructive data. In order to verify the calibration, i.e., to check the quality of the developed calibration functions and to evaluate the measurement accuracy of this 3MA application, a series of comparing measurements was carried out with the calibrated 3MA system. The nondestructively (nd) Appl. Sci. 2019, 9, 1068 13 of 29

Table 6. Steel grade classes and calibration results.

Steel Grade Class RMSE of Rm [MPa] RMSE of Rp0.2 [MPa] No. of Strips IF, conventional 5.4 8.2 2667 IF, high strength 11.3 12.3 7764 Bakehardening 5.8 8.8 1294 Structural 7.9 10.1 164

Besides standard steel qualities (normal steel and construction steel), interstitial-free (IF) steel, and high strength material were investigated. The 3MA data that were determined at the beginning and at the end of the strip were associated to the destructively determined data of Rm and Rp0.2 data in order to determine the calibration functions. Table6 also shows the results of this calibration in terms of the RMSE of the correlation between the 3MA data and destructive data. In order to verify the calibration, i.e., to check the quality of the developed calibration functionsAppl. Sci. 2019 and, 9, x to FOR evaluate PEER REVIEW the measurement accuracy of this 3MA application, a series of comparing13 of 29 measurements was carried out with the calibrated 3MA system. The nondestructively (nd) determined valuesdetermined of Rm values and Rp0.2 of wereRm and compared Rp0.2 withwere the compared values that with were the measured values that from were destructive measured (d) tensile from testsdestructive at over (d) 2700 tensile strips. tests The diagramsat over 2700 of Figure strips. 10 The show diagrams the results of Figure of this 10 comparison. show the Theresults data of from this allcomparison. four steel gradeThe data classes from are all combined four steel ingrade these classes diagrams. are combined in these diagrams.

(a) (b)

Figure 10.10. ResultsResults of of the the verification veriﬁcation of the calibration; comparison of the values from 3MA measurements with destructively measured values: ( a)) for tensile strength, Rm and (b) forfor yieldyield strength,strength, Rp0.2.Rp0.2.

InIn orderorder to to evaluate evaluate the the deviation deviation between between nd nd data data and and d data, d data, the RMSEthe RMSE values values for the for four the steelfour gradesteel grade classes classes are shown are shown in Table in 7Table. 7.

Table 7.7. Steel grade classes and resultsresults ofof veriﬁcationverification measurementsmeasurements forfor RmRm andand Rp0.2.Rp0.2.

Steel Grade Grade Class Class RMSE of of Rm Rm [MPa] [MPa] RMSERMSE of of Rp0.2 Rp0.2 [MPa] [MPa] No. No. of Strips IF,IF, conventional conventional 5.35.3 7.47.4 1985 IF,IF, high high strength strength 6.86.8 9.39.3 68 BakehardeningBakehardening 11.211.2 15.615.6 55 Structural 7.5 12.4 34 Structural 7.5 12.4 34

According toto the the rules rules of of error error propagation, propagation, the the measuring measuring uncertainties uncertainties of both of both methods methods (nd and(nd d)and will d) contributewill contribute to the to RMSE, the RMSE, i.e., i.e.,

q 𝑅𝑀𝑆𝐸 𝑢2 𝑢2 (2) RMSE = und + ud (2)

where und and ud are the uncertainties of the nondestructive measurement and of the destructive measurement, respectively. The values of the RMSE, as shown in Table 7, are close to the known measuring uncertainties of the destructive tensile test, indicating that the measuring uncertainty of 3MA is in the same range. Inline measurements of Rm and Rp0.2 with 3MA can be used in order to precisely determine the acceptance range of a coil, as is shown in Figure 11. The threshold level (acceptance line) of 260 MPa is not exceeded in the strip position range x > 200 m and x < 2400 m. Appl. Sci. 2019, 9, 1068 14 of 29

where und and ud are the uncertainties of the nondestructive measurement and of the destructive measurement, respectively. The values of the RMSE, as shown in Table7, are close to the known measuring uncertainties of the destructive tensile test, indicating that the measuring uncertainty of 3MA is in the same range. Inline measurements of Rm and Rp0.2 with 3MA can be used in order to precisely determine the acceptance range of a coil, as is shown in Figure 11. The threshold level (acceptance line) of 260 MPa is notAppl.Appl. exceeded Sci.Sci. 20192019,, 99 in,, xx FOR theFOR strip PEERPEER positionREVIEWREVIEW range x > 200 m and x < 2400 m. 1414 ofof 2929

FigureFigureFigure 11. 11.11. InlineInlineInline measurementmeasurementmeasurement ofof yieldyieldyield strengthstrengthstrength Rp0.2Rp0.2Rp0.2 ininin aaa coil coilcoil of ofof 2500 25002500 m mm in inin length; length;length; destructive destructivedestructive measurementsmeasurementsmeasurements of ofof Rp0.2 Rp0.2Rp0.2 at atat the thethe beginning, beginning,beginning, at at theat thethe end endend and andand in the inin thethe middle middlemiddle of the ofof the stripthe stripstrip are also areare alsoalso shown shownshown as red asas dots;redred dots;dots; the acceptance thethe acceptanceacceptance line indicating lineline indicatingindicating the threshold thethe thresholdthreshold level levellevel of Rp0.2 ofof Rp0.2Rp0.2 = 260 == MPa 260260 MPaMPa is shown. isis shown.shown.

InInIn later laterlater developments, developments,developments, the thethe 3MA 3MA3MA systems systemssystems for forfor inline inlineinline strip stripstrip steel steelsteel testing testingtesting were werewere also alsoalso calibrated calibratedcalibrated to toto surface surfacesurface hardnesshardnesshardness and andand residual residualresidual stress stressstress atat differentdifferent depths.depths. These These calibrationcalibration resultsresults results areare are shownshown shown inin in FigureFigure Figure 12.12. 12 .

((aa)) ((bb))

FigureFigureFigure 12. 12.12. Results ResultsResults of ofof the thethe calibration calibrationcalibration to: to:to: ( (a(aa))) hardness hardnesshardness and andand ( (b(bb))) residual residualresidual stress stressstress at atat the thethe surface surfacesurface and andand at atat depths depthsdepths ofofof 50, 50,50, 100 100100 and andand 200 200200µ μ μmmm (all (all(all RMSE RMSERMSE values valuesvalues in inin MPa). MPa).MPa).

CalibrationCalibrationCalibration to toto hardness hardnesshardness and andand residual residualresidual stress stressstress was waswas made mademade with withwith different differentdifferent steel steelsteel grades gradesgrades from fromfrom tool tooltool and andand springspringspring steels steelssteels (75Cr1, (75Cr1,(75Cr1, CK75, CK75,CK75, 80CrV2), 80CrV2),80CrV2), low lowlow and andand unalloyed unalloyedunalloyed steels steelssteels (C55, (C55,(C55, C75), C75),C75), and andand heat-treatable heatheat‐‐treatabletreatable steel steelsteel (50CrMo4),(50CrMo4),(50CrMo4), providing providingproviding hardness hardnesshardness levels levelslevels that thatthat were werewere between betweenbetween 40 4040 and andand 54 5454 HRC HRCHRC (Rockwell (Rockwell(Rockwell hardness hardnesshardness type typetype C). C). UsingUsingUsing X-ray XX‐‐rayray diffraction diffractiondiffraction determined determineddetermined the thethe reference referencereference values valuesvalues for forfor residual residualresidual stress. stress.stress. TheTheThe different differentdifferent depths depthsdepths for forfor the thethe x-ray xx‐‐rayray measurements measurementsmeasurements were werewere uncovered uncovereduncovered by byby electropolishing. electropolishing.electropolishing. Again, Again,Again, the thethe measuringmeasuringmeasuring results resultsresults from fromfrom the thethe calibrated calibratedcalibrated 3MA 3MA3MA system systemsystem were werewere compared comparedcompared with withwith the thethe results resultsresults from fromfrom destructive destructivedestructive measurements.measurements.measurements. For ForFor hardness, hardness,hardness, the thethe RMSE RMSERMSE was waswas in inin the thethe range rangerange of ofof 1 11 HRC. HRC.HRC. For ForFor residual residualresidual stress, stress,stress, RMSE RMSERMSE values valuesvalues thatthatthat were werewere between betweenbetween 1 11 nd ndnd 47 4747 MPa MPaMPa were werewere determined, determined,determined, dependent dependentdependent on onon the thethe steel steelsteel grade gradegrade and andand the thethe measuring measuringmeasuring depthdepthdepth (see (see(see Table TableTable8 ). 8).8).

TableTable 8.8. SteelSteel gradesgrades andand resultsresults ofof verificationverification measurementsmeasurements forfor residualresidual stress.stress.

ResidualResidual Stress,Stress, RSRS SteelSteel GradeGrade AtAt thethe SurfaceSurface AtAt 5050μμmm DepthDepth AtAt 100100μμmm DepthDepth AtAt 200200μμmm DepthDepth RMSERMSE [MPa][MPa] RMSERMSE [MPa][MPa] RMSERMSE [MPa][MPa] RMSERMSE [MPa][MPa] 75Cr175Cr1 4747 11 33 44 80CrV280CrV2 1515 11 99 3131 50CrMo450CrMo4 44 44 44 1111 Appl. Sci. 2019, 9, 1068 15 of 29

Table 8. Steel grades and results of veriﬁcation measurements for residual stress. Appl. Sci. 2019, 9, x FOR PEER REVIEW 15 of 29 Residual Stress, RS BasedSteel on Grade these calibrations,At the Surface the hardness At 50 andµm Depth residual Atstress 100 µcanm Depth be determined At 200 µ asm Deptha function of strip position. Figure 13RMSE shows [MPa] a typical example. RMSE [MPa] At the beginning RMSE [MPa] of the strip, RMSE strong [MPa] variations of residual stress can be observed. The hardness is increased. After x = 60 m, the strip has calmed down, 75Cr1 47 1 3 4 i.e., hardness80CrV2 and residual stress 15 have reached their 1 “normal” values. 9 31 Meanwhile,50CrMo4 a wide range 4 of 3MA installations 4 for inline strip 4 steel inspection exists. 11 Figure 14 shows two of them. In the image of Figure 14a, the 3MA probe is applied above a large deflection Appl. Sci. 2019, 9, x FOR PEER REVIEW 15 of 29 roller.Based On the on other these side, calibrations, the probe the head hardness may andalso residualbe applied stress with can a robot, be determined as shown as in a Figure function 14b. of Meanwhilestrip position.Based 3MA on Figure these applications calibrations,13 shows on a typical hotthe strip,hardness example. up toand 300 Atresidual °C the can beginning stress be realized, can ofbe the determined if the strip, probe strong as head a function variations is actively of of cooledresidualstrip withposition. stress pressurized can Figure be observed. 13 air. shows Other The a typicaloptimizations hardness example. is increased. of At the the probe beginning After head x = designof 60 the m, strip, the allow strip strong for has a liftvariations calmed‐off of down, up of to 5i.e., mm.residual hardness stress and can residual be observed. stress haveThe hardness reached is their increased. “normal” After values. x = 60 m, the strip has calmed down, i.e., hardness and residual stress have reached their “normal” values. Meanwhile, a wide range of 3MA installations for inline strip steel inspection exists. Figure 14 shows two of them. In the image of Figure 14a, the 3MA probe is applied above a large deflection roller. On the other side, the probe head may also be applied with a robot, as shown in Figure 14b. Meanwhile 3MA applications on hot strip, up to 300 °C can be realized, if the probe head is actively cooled with pressurized air. Other optimizations of the probe head design allow for a lift‐off of up to 5 mm.

Figure 13. Inline measurement of hardness and residual residual stress at the the surface and at depths of of 50, 50, 100, 100, and 200 μµm.

Meanwhile, a wide range of 3MA installations for inline strip steel inspection exists. Figure 14 shows two of them. In the image of Figure 14a, the 3MA probe is applied above a large deﬂection roller. On the other side, the probe head may also be applied with a robot, as shown in Figure 14b. Meanwhile 3MA applications on hot strip, up to 300 ◦C can be realized, if the probe head is actively Figure 13. Inline measurement of hardness and residual stress at the surface and at depths of 50, 100, cooled withand 200 pressurized μm. air. Other optimizations of the probe head design allow for a lift-off of up to 5 mm.

(a) (b)

Figure 14. Further variants of 3MA probe head application on the running strip: (a) on a large deflection roller and (b) guided by a robot.

4.2.2. Heavy Plate Testing (a) (b) Heavy plate with a slab thickness of up to 400 mm or more is produced in nearly all unalloyed and alloyedFigureFigure steel 14. grades 14.Further Further today. variants variants The of diverse 3MA of 3MA probe range probe head ofhead application properties, application on e.g., the on running yield the running strengths strip: (strip:a) onfrom a(a large) 200on deﬂection ato large more than 1100 rollerN/mmdeflection and2 ensures (b )roller guided theand by precise(b a) guided robot. suitability by a robot. for use in bridges, high rise buildings, offshore platforms, ship building, line pipes, pressure vessels, heavy‐duty machines, and much more. In order to reduce 4.2.2. Heavy Plate Testing the constructive thicknesses and, with it, the self‐weight of a construction, higher‐strength steel grades are increasinglyHeavy plate requested, with a slab while thickness the weldability of up to 400 and mm the or coldmore toughness is produced must in nearly still be all acceptable. unalloyed and alloyed steel grades today. The diverse range of properties, e.g., yield strengths from 200 to more than 1100 N/mm2 ensures the precise suitability for use in bridges, high rise buildings, offshore platforms, ship building, line pipes, pressure vessels, heavy‐duty machines, and much more. In order to reduce the constructive thicknesses and, with it, the self‐weight of a construction, higher‐strength steel grades are increasingly requested, while the weldability and the cold toughness must still be acceptable. Appl. Sci. 2019, 9, 1068 16 of 29

4.2.2. Heavy Plate Testing Heavy plate with a slab thickness of up to 400 mm or more is produced in nearly all unalloyed and alloyed steel grades today. The diverse range of properties, e.g., yield strengths from 200 to more than 1100 N/mm2 ensures the precise suitability for use in bridges, high rise buildings, offshore platforms, ship building, line pipes, pressure vessels, heavy-duty machines, and much more. In order to reduce the constructive thicknesses and, with it, the self-weight of a construction, higher-strength steel grades Appl. Sci. 2019, 9, x FOR PEER REVIEW 16 of 29 are increasingly requested, while the weldability and the cold toughness must still be acceptable. For the production of suchsuch steelsteel grades,grades, modernmodern manufacturingmanufacturing processes,processes, suchsuch asas water/airwater/air quenching andand tempering, tempering, accelerated accelerated cooling, cooling, normalizing, normalizing, and thermo-mechanicaland thermo‐mechanical rolling arerolling applied. are applied.The customer The customer asks for geometrical asks for geometrical and mechanical and properties,mechanical which properties, are uniform which across are uniform product across length productand width, length especially and width, for these especially high-value for these grades. high‐value Consequently, grades. Consequently, high demands high are demands made on theare madedetermination on the determination and documentation and documentation of product quality. of product quality. Comprehensive dimensional, shape, shape, and and surface surface inspections inspections already already take place within the production flowflow directly on the plate. On the other hand, it is still necessary to extract tests coupons from the edges of the un un-trimmed‐trimmed plate in order to determine the mechanical properties. Tensile Tensile and toughness tests are performed according to the codes and delivery from from highly qualified qualified and certified certified personnel andand theythey cannotcannot be be integrated integrated into into online online closed closed loop loop control control with with direct direct feedback. feedback. Costs Costs in inthe the range range of of several several thousands thousands of of Euros Euros per per year year arise arise in in a a mid-sized mid‐sized heavy heavy plateplate plant from coupon scrap alone. For these reasons, heavy plate producers are eager to replace the mechanical testing of coupons by nondestructive technology, allowing forfor these tests to be transferred transferred from the laboratories laboratories of quality assurance to the shop floor.floor. In a firstfirst project, a 3MA system was integrated into into a trolley, allowing for 3MA tests to be performed on plates moving on the roller conveyor line in the steel steel plant. plant. Figure Figure 1515 showsshows this this device. device. A brushingbrushing device device was was integrated integrated into into the trolleythe trolley in order in toorder the removeto the scaleremove that scale otherwise that representsotherwise representsa strong disturbing a strong disturbing influence. influence.

(a) (b)

Figure 15. Remote application of the 3MA system in the roller conveyor line of a steel plant: ( a)) 3MA inspection trolley and (b) remote control desk.

Calibration was done on coupons that were taken from the running production. It was apparent that the best results could be achieved if individual calibration functions were used for each steel grade and each plate thickness. In this case, the RMSE was in the range of 4 4 HB HB (Brinell (Brinell hardness) hardness) for hardness determination, 10 MPa for Rm, and 20 MPa for Rp0.2. The main objective of a second project was to avoid pseudo pseudo-scrap‐scrap in heavy plate by the precise detection of the so so-called‐called cold ends in the mother plate plate after after rolling rolling [65]. [65]. This This is is shown shown in in Figure Figure 16.16. Cold ends are developing due to altered cooling conditions and at the plate ends. The cold ends are characterized byby impaired impaired mechanical mechanical properties properties (Rm (Rm and Rp0.2).and Rp0.2). Therefore, Therefore, they have they to behave removed to be removed in order to assure a plate that fulfills the quality requirements over the entire length. For this purpose, large sections—usually much larger than the cold ends—are cut off. Due to these safety margins, pseudo‐scrap is produced, i.e., material, which still fulfills the requirements on material conformity, is unnecessarily removed and scrapped. In order to avoid this pseudo‐scrap, the extension of the cold ends has to be precisely detected. This can be accomplished by 3MA. Appl. Sci. 2019, 9, 1068 17 of 29 in order to assure a plate that fulﬁlls the quality requirements over the entire length. For this purpose, large sections—usually much larger than the cold ends—are cut off. Due to these safety margins, pseudo-scrap is produced, i.e., material, which still fulﬁlls the requirements on material conformity, is unnecessarily removed and scrapped. In order to avoid this pseudo-scrap, the extension of the cold

Appl.ends Sci. has 2019 to, be9, x precisely FOR PEER detected.REVIEW This can be accomplished by 3MA. 17 of 29

Figure 16. Avoiding pseudo‐scrap by precise detection of the extension of cold ends. Figure 16. Avoiding pseudo-scrap by precise detection of the extension of cold ends.

The plates that were investigated in this project were steel grades showing a pronounced anisotropy of of material material properties properties due due to to a a rolling rolling texture. texture. Therefore, Therefore, it turned it turned out out to tobe beuseful useful to combineto combine the the measuring measuring information information from from 3MA 3MA with with measuring measuring parameters parameters that that are derived are derived from from the timethe time-of-ﬂight‐of‐flight (TOF) (TOF) measurements measurements of ultrasound of ultrasound (US) (US) waves. waves. It Itcould could be be shown shown that, that, in in highly texturizedtexturized steel grades, adding relative relative differences of the US TOFs to the calibration functions could reduce the measuring error. This is shown in Table9 9..

TableTable 9. ResultsResults of different types of calibration—without and with ultrasound (US) parameters.

ParametersParameters in in RMSE ofRMSE Rm RMSERMSE of Rp0.2 of RMSERMSE of Hardnessof CalibrationCalibration Functions Functions [MPa]of Rm [MPa] Rp0.2[MPa] [MPa] Hardness[HB] [HB] OnlyOnly 3MA 3MA parameters parameters 1212 2323 5 5 3MA3MA and USand parameters US parameters 1010 1111 4 4

The calibrated 3MA + US combination system makes it possible to determine the cold ends The calibrated 3MA + US combination system makes it possible to determine the cold ends before cutting by precisely detecting the positions of the acceptance lines, which are separating OK before cutting by precisely detecting the positions of the acceptance lines, which are separating OK and NOK quality material at the head and the root of the plate. Figure 17 shows the corresponding and NOK quality material at the head and the root of the plate. Figure 17 shows the corresponding measuring procedure. In a first step, the value of the assessing measuring quantity (here: tensile measuring procedure. In a ﬁrst step, the value of the assessing measuring quantity (here: tensile strength, Rm) is measured in the reference point in the middle of the plate. Subsequently, the 3MA + strength, Rm) is measured in the reference point in the middle of the plate. Subsequently, the 3MA + US system is moved stepwise from the edge to the middle until the value of the measuring quantity US system is moved stepwise from the edge to the middle until the value of the measuring quantity falls below a preset threshold level, which is the value at the reference point plus a specified margin falls below a preset threshold level, which is the value at the reference point plus a speciﬁed margin (here: Rm = 15 MPa). Afterwards, the acceptance line has been reached and it can be drawn on the (here: ∆Rm = 15 MPa). Afterwards, the acceptance line has been reached and it can be drawn on plate. the plate.

Figure 17. Measuring procedure for determining the cold ends at the plate head and at the plate root. Appl. Sci. 2019, 9, x FOR PEER REVIEW 17 of 29

Figure 16. Avoiding pseudo‐scrap by precise detection of the extension of cold ends.

The plates that were investigated in this project were steel grades showing a pronounced anisotropy of material properties due to a rolling texture. Therefore, it turned out to be useful to combine the measuring information from 3MA with measuring parameters that are derived from the time‐of‐flight (TOF) measurements of ultrasound (US) waves. It could be shown that, in highly texturized steel grades, adding relative differences of the US TOFs to the calibration functions could reduce the measuring error. This is shown in Table 9.

Table 9. Results of different types of calibration—without and with ultrasound (US) parameters.

Parameters in RMSE RMSE of RMSE of Calibration Functions of Rm [MPa] Rp0.2 [MPa] Hardness [HB] Only 3MA parameters 12 23 5 3MA and US parameters 10 11 4

The calibrated 3MA + US combination system makes it possible to determine the cold ends before cutting by precisely detecting the positions of the acceptance lines, which are separating OK and NOK quality material at the head and the root of the plate. Figure 17 shows the corresponding measuring procedure. In a first step, the value of the assessing measuring quantity (here: tensile strength, Rm) is measured in the reference point in the middle of the plate. Subsequently, the 3MA + US system is moved stepwise from the edge to the middle until the value of the measuring quantity falls below a preset threshold level, which is the value at the reference point plus a specified margin (here: Rm = 15 MPa). Afterwards, the acceptance line has been reached and it can be drawn on the Appl. Sci. 2019, 9, 1068 18 of 29 plate.

FigureFigure 17. 17.Measuring Measuring procedure procedure for for determining determining the the cold cold ends ends at theat the plate plate head head and and at theat the plate plate root. root. 4.3. Applications in the Automotive Industry

4.3.1. Car Body Parts It is well accepted that the use of hot-stamped steel or Press-Hardened Steel (PHS) in reinforcement elements of the car body structure offers the possibility for significant reduction of vehicles’ weight at moderate costs [66]. However, the production of PHS parts is also very challenging. Today, OEMs and suppliers offer a broad range of PHS products that are based on tailored steel properties and adapted coating systems, including monolithic and patched blanks, tailor tempered blanks (TTB), tailor welded blanks (TWB), and tailor rolled blanks (TRB) [67]. For that reason, there is a high demand for methods that allow for the precise monitoring of the PHS production process in order to maintain it to fixed targets and to ensure that PHS products in all variants meet the high quality requirements. This is crucial for the automotive industry in order to maintain and improve its competitiveness in a global market [68]. The mechanical properties of the PHS are adjusted not in the steel plant during the process, but only afterwards during hot forming. In order to meet the material specifications, all process parameters during heating, forming, and quenching must be kept within strict tolerance bands. Nevertheless, process disturbances, such as furnace malfunctions, wear on the pressing tool, or breakdown of the cooling circuit in the press could affect the material quality. Deteriorated material properties could also result if the heated blank cools down too fast during transfer between the furnace and press. In the case of AlSi-coated blanks, a complicated coating layer structure is developed in the furnace. In order to achieve the optimum press-hardening results and paint adhesion, this layer structure has to meet strict specifications. The application possibilities of 3MA for quality assurance and process monitoring in PHS production have been researched, since the first use of PHS in large-scale automotive production in 2004 [56]. Based on the calibrated 3MA technique, quantitative values of hardness, tensile strength (Rm), elastic limit (Rp0.2), fracture elongation (A50) and uniform elongation (Ag) of the PHS material can be determined. The 3MA calibrations are generally steel-grade specific. Thus, different calibration functions for the hardenable and the nonhardenable parts in a TWB have to be used. Additionally, each coating type requires an individual calibration. The diagrams in Figures 18 and 19 show the result of a calibration for an aluminum–silicon (AlSi)-coated steel of grade 22MnB5. Usually, for hardness and Rm, the 3MA parameters provide the best correlations, whereas the correlations for A50 and Ag (not shown here) are the worst. Appl. Sci. 2019, 9, x FOR PEER REVIEW 18 of 29

4.3. Applications in the Automotive Industry

4.3.1. Car Body Parts It is well accepted that the use of hot‐stamped steel or Press‐Hardened Steel (PHS) in reinforcement elements of the car body structure offers the possibility for significant reduction of vehicles’ weight at moderate costs [66]. However, the production of PHS parts is also very challenging. Today, OEMs and suppliers offer a broad range of PHS products that are based on tailored steel properties and adapted coating systems, including monolithic and patched blanks, tailor tempered blanks (TTB), tailor welded blanks (TWB), and tailor rolled blanks (TRB) [67]. For that reason, there is a high demand for methods that allow for the precise monitoring of the PHS production process in order to maintain it to fixed targets and to ensure that PHS products in all variants meet the high quality requirements. This is crucial for the automotive industry in order to maintain and improve its competitiveness in a global market [68]. The mechanical properties of the PHS are adjusted not in the steel plant during the process, but only afterwards during hot forming. In order to meet the material specifications, all process parameters during heating, forming, and quenching must be kept within strict tolerance bands. Nevertheless, process disturbances, such as furnace malfunctions, wear on the pressing tool, or breakdown of the cooling circuit in the press could affect the material quality. Deteriorated material properties could also result if the heated blank cools down too fast during transfer between the furnace and press. In the case of AlSi‐coated blanks, a complicated coating layer structure is developed in the furnace. In order to achieve the optimum press‐hardening results and paint adhesion, this layer structure has to meet strict specifications. The application possibilities of 3MA for quality assurance and process monitoring in PHS production have been researched, since the first use of PHS in large‐scale automotive production in 2004 [56]. Based on the calibrated 3MA technique, quantitative values of hardness, tensile strength (Rm), elastic limit (Rp0.2), fracture elongation (A50) and uniform elongation (Ag) of the PHS material can be determined. The 3MA calibrations are generally steel‐grade specific. Thus, different calibration functions for the hardenable and the nonhardenable parts in a TWB have to be used. Additionally, each coating type requires an individual calibration. The diagrams in Figures 18 and 19 show the result of a calibration for an aluminum–silicon (AlSi)‐coated steel of grade 22MnB5. Usually, for Appl.hardness Sci. 2019 and, 9, 1068Rm, the 3MA parameters provide the best correlations, whereas the correlations19 of for 29 A50 and Ag (not shown here) are the worst.

(a) (b)

FigureFigure 18. Results of the calibration for the mechanical properties in AlSi AlSi-coated‐coated Press Press-Hardened‐Hardened Steel Appl.Appl. Sci. (PHS):(PHS):Sci. 2019 2019 (,(a ,a9 9)), , tensilex tensilex FOR FOR PEER PEER strengthstrength REVIEW REVIEW andand ((bb)) yieldyield strength.strength. 1919 ofof 2929

((aa)) ((bb))

FigureFigure 19.19. Results Results of of thethe calibration calibrationcalibration forfor mechanical mechanicalmechanical propertiesproperties inin AlSi-coated AlSiAlSi‐‐coatedcoated PHS:PHS: ((aa)) hardness hardness and and (((bb))) fracture. fracture.fracture.

ToTo aaa lesser lesserlesser extent, extent,extent, the thethe steel steelsteel sheet sheetsheet thickness thicknessthickness also alsoalso inﬂuences influencesinfluences the thethe 3MA 3MA3MA signals. signals.signals. Therefore,Therefore, twotwo separateseparate calibrations calibrations for for sheetsheet thicknessesthicknesses thatthat areare smaller smaller andand largerlarger than than 1.0 1.0 mm mm havehave provedproved to to bebe beneﬁcial.beneficial.beneficial. On On the the other other hand, hand, it it was was shown shown that that batch batch or or supplier supplier variations variations have have no no detectable detectable effect effect onon the the calibration calibration functions.functions. TheThe widespread widespread AlSi AlSi coating coating often often develops develops a a layer layer structurestructure with with a a diffusion diffusion layer layer that that consistsconsists ofof ferromagnetic ferromagnetic Al–Fe Al–Fe phases. phases. Besides Besides the the streel streel substrate,substrate, this this diffusiondiffusion layer layer will will produceproduce a a secondsecond contributioncontribution to to the the 3MA 3MA signals, signals, which which isis distinguishable distinguishable fromfrom the thethe steel’s steel’ssteel’s signal signalsignal contribution. contribution.contribution. This This is is shownshown in in Figure Figure 20 20.20..

FigureFigure 20.20.20. BarkhausenBarkhausen NoiseNoise (BN)(BN) signal signal (left) (left) and and IP IP signal signal of of anan AlSi-coatedAlSiAlSi‐‐coatedcoated PHSPHS part;part;part; thethe characteristiccharacteristic double-peak doubledouble‐‐peakpeak structure structurestructure is isis clearly clearlyclearly visible. visible.visible.

TheThe BarkhausenBarkhausen NoiseNoise (BN)(BN) signalsignal and,and, toto somesome extent,extent, thethe IncrementalIncremental PermeabilityPermeability (IP)(IP) signalsignal showshow aa characteristiccharacteristic doubledouble‐‐peakpeak structure.structure. WhenWhen comparingcomparing thethe measurementsmeasurements onon coatedcoated andand uncoateduncoated sheetssheets havehave shownshown thatthat thethe outerouter peaks,peaks, appearingappearing atat higherhigher magneticmagnetic fieldsfields H,H, areare generatedgenerated byby thethe steelsteel substrate,substrate, whereaswhereas thethe innerinner peakspeaks atat lowerlower fieldsfields HH resultresult fromfrom thethe ferromagneticferromagnetic diffusiondiffusion layer.layer. TheThe heightheight ofof thethe innerinner peakpeak isis dependentdependent onon thethe amountamount ofof thisthis diffusiondiffusion layer.layer. Therefore,Therefore, somesome ofof thethe 3MA3MA measuringmeasuring parametersparameters areare correlatedcorrelated toto thethe diffusiondiffusion layerlayer thicknessthickness (DLT).(DLT). Moreover,Moreover, therethere areare alsoalso 3MA3MA measuringmeasuring parametersparameters thatthat areare dependentdependent onon thethe liftlift‐‐offoff betweenbetween thethe 3MA3MA probeprobe andand steelsteel substrate.substrate. TheseThese parametersparameters cancan bebe usedused toto determinedetermine thethe totaltotal thicknessthickness (TLT)(TLT) ofof thethe AlSiAlSi coating.coating. TheThe valuesvalues ofof DLTDLT andand TLTTLT areare relevantrelevant forfor thethe furtherfurther processingprocessing ofof thethe PHSPHS parts,parts, becausebecause theythey affectaffect weldabilityweldability andand paintability.paintability. TheThe AlSiAlSi coatingcoating isis burnedburned intointo thethe steelsteel duringduring aa timetime‐‐consumingconsuming furnacefurnace process.process. DuringDuring thisthis time,time, thethe diffusiondiffusion layerlayer isis developing.developing. Therefore,Therefore, thethe valuesvalues ofof DLTDLT andand TLTTLT areare importantimportant measuresmeasures forfor controllingcontrolling thethe properproper processingprocessing inin thethe furnace.furnace. ForFor thesethese reasons,reasons, PHSPHS producersproducers alsoalso areare veryvery interestedinterested inin measuringmeasuring thesethese coatingcoating properties.properties. ThisThis isis possiblepossible withwith 3MA,3MA, asas isis shownshown inin thethe calibrationcalibration diagramsdiagrams ofof FigureFigure 21.21. TheThe accuracyaccuracy (RMSE(RMSE == 1.01.0 μ μmm andand 1.71.7 μ μm)m) isis sufficientsufficient inin orderorder toto evaluateevaluate thethe coatingcoating quality.quality. Appl. Sci. 2019, 9, 1068 20 of 29

The Barkhausen Noise (BN) signal and, to some extent, the Incremental Permeability (IP) signal show a characteristic double-peak structure. When comparing the measurements on coated and uncoated sheets have shown that the outer peaks, appearing at higher magnetic fields H, are generated by the steel substrate, whereas the inner peaks at lower fields H result from the ferromagnetic diffusion layer. The height of the inner peak is dependent on the amount of this diffusion layer. Therefore, some of the 3MA measuring parameters are correlated to the diffusion layer thickness (DLT). Moreover, there are also 3MA measuring parameters that are dependent on the lift-off between the 3MA probe and steel substrate. These parameters can be used to determine the total thickness (TLT) of the AlSi coating. The values of DLT and TLT are relevant for the further processing of the PHS parts, because they affect weldability and paintability. The AlSi coating is burned into the steel during a time-consuming furnace process. During this time, the diffusion layer is developing. Therefore, the values of DLT and TLT are important measures for controlling the proper processing in the furnace. For these reasons, PHS producers also are very interested in measuring these coating properties. This is possible with 3MA, as is shown in the calibration diagrams of Figure 21. The accuracy (RMSE = 1.0 µm and 1.7 µm) is sufficientAppl. Sci. 2019 in, 9 order, x FOR to PEER evaluate REVIEW the coating quality. 20 of 29

(a) (b)

Figure 21. Results of the calibration to coating layer thickness in AlSi-coatedAlSi‐coated PHS: ( a) total layer thickness (TLT) andand ((bb)) diffusiondiffusion layerlayer thicknessthickness (DLT).(DLT).

This is shown in in Figure Figure 22a,22a, which which is is a a screenshot screenshot of of the the 3MA 3MA results results that that re re displayed displayed after after a ameasurement measurement on on a PHS a PHS part. part. The The measuring measuring time time was was only only three three seconds. seconds. Due Due to the to thefact factthat that the thefield field-of-view‐of‐view of the of 3MA the 3MA probe probe is only is onlya 1–2 a mm 1–2 in mm diameter, in diameter, 3MA 3MAcan be can used be to used inspect to inspect sharp sharptransitions transitions of material of material properties, properties, as they as are they preferred are preferred in tailor in tailortempered tempered blanks blanks (TTB). (TTB). The Thedevelopment development of hardness of hardness and other and mechanical other mechanical properties properties in such in a sharp such atransition sharp transition can be detected can be detectedwith 3MA, with as shown 3MA, in as Figure shown 22b. in Figure Such measuring 22b. Such profiles measuring of hardness, profiles Rm, of hardness, Rp0.2, etc., Rm, cannot Rp0.2, be etc.,determined cannot beby determinedconventional by destructive conventional testing destructive methods. testing Therefore, methods. 3MA Therefore, offers 3MAadditional offers additionalinformation information that is not that available is not availablefrom destructive from destructive tests. 3MA tests. can 3MA be canmanually be manually applied applied in PHS in PHSproduction. production. Meanwhile, Meanwhile, automated automated 3MA 3MA systems systems that that are are based based on on conventional conventional or or collaborative collaborative robots have been developed. Meanwhile, there is a comprehensive calibration database for for the 3MA application on on PHS. PHS. Calibrations are available for all steel blank and coating coating types types (see (see above) above) and and usual usual sheet sheet thicknesses. thicknesses. When calibrationcalibration is completed,is completed, all ofall the of calibrated the calibrated target quantities target quantities can be simultaneously can be simultaneously determined indetermined a single measuring in a single step. measuring Joining of step. PHS Joining always bearsof PHS the always risk of irregularities bears the risk that of affect irregularities the static that and dynamicaffect the strengthstatic and of dynamic the joint. strength Additionally, of the for joint. PHS Additionally, welding, 3MA for canPHS provide welding, valuable 3MA can measuring provide informationvaluable measuring for quality information assurance for and quality process assurance monitoring. and In process the case monitoring. of Resistance In Spot the Welding case of (RSW),Resistance the strengthSpot Welding of the (RSW), welds is the mainly strength defined of the by welds the size is mainly and quality defined of the by weld the size nuggets. and quality It was shownof the weld that 3MAnuggets. allows It was for theshown size that of the 3MA weld allows nugget for to the be size accurately of the weld determined nugget [ to69 ].be accurately determined [69].

(a) (b)

Figure 22. Results of the application of a calibrated 3MA system: (a) screenshot of the results displayed after a single 3MA measurement and (b) profiles of mechanical properties in a sharp transition zone. Appl. Sci. 2019, 9, x FOR PEER REVIEW 20 of 29

(a) (b)

Figure 21. Results of the calibration to coating layer thickness in AlSi‐coated PHS: (a) total layer thickness (TLT) and (b) diffusion layer thickness (DLT).

This is shown in Figure 22a, which is a screenshot of the 3MA results that re displayed after a measurement on a PHS part. The measuring time was only three seconds. Due to the fact that the field‐of‐view of the 3MA probe is only a 1–2 mm in diameter, 3MA can be used to inspect sharp transitions of material properties, as they are preferred in tailor tempered blanks (TTB). The development of hardness and other mechanical properties in such a sharp transition can be detected with 3MA, as shown in Figure 22b. Such measuring profiles of hardness, Rm, Rp0.2, etc., cannot be determined by conventional destructive testing methods. Therefore, 3MA offers additional information that is not available from destructive tests. 3MA can be manually applied in PHS production. Meanwhile, automated 3MA systems that are based on conventional or collaborative robots have been developed. Meanwhile, there is a comprehensive calibration database for the 3MA application on PHS. Calibrations are available for all steel blank and coating types (see above) and usual sheet thicknesses. When calibration is completed, all of the calibrated target quantities can be simultaneously determined in a single measuring step. Joining of PHS always bears the risk of irregularities that affect the static and dynamic strength of the joint. Additionally, for PHS welding, 3MA can provide valuable measuring information for quality assurance and process monitoring. In the case of Resistance Spot Welding (RSW), the strength of the welds is mainly defined by the size and quality Appl.of the Sci. weld2019, nuggets.9, 1068 It was shown that 3MA allows for the size of the weld nugget to be accurately21 of 29 determined [69].

(a) (b)

Figure 22. ResultsResults of of the the application application of a a calibrated calibrated 3MA 3MA system: system: ( (aa)) screenshot screenshot of the the results results displayed displayed after a single 3MA measurement and (b)) proﬁlesprofiles ofof mechanicalmechanical propertiesproperties inin aa sharpsharp transitiontransition zone.zone.

If the PHS should be laser welded, the aluminum–silicon coating could be a problem. Therefore, the coating is partially ablated by laser. Again, the 3MA technique can be used in order to verify the success of the partial ablation by determining the thickness of the residual coating layer [70]. Meanwhile, PHS testing is the most successful 3MA application by far. Almost all of the PHS-producing OEMs and suppliers from the automotive industry use 3MA.

4.3.2. Surface-Hardened Parts and Machined Parts Surface heat treatments and machining processes are necessary to provide materials with the desired strength, shape, and surface quality. Hardening the material’s surface by case-hardening, induction hardening, or nitriding will increase the resistance to surface wear without impairing the material’s toughness in the bulk. Usually, the workpiece is already formed or milled to its final shape when surface hardening is applied. Therefore, unavoidable quenching distortions that are due to hardening usually have to be removed afterwards. In order to ensure a high-dimensional workpiece of accurate form, a subsequent fine machining of the workpiece’s surface by means of turning, grinding, honing, or polishing is often required after hardening. However, this surface machining process can, again, result in unwanted changes in the material’s microstructure. Machining can lead to a local temperature increase on the surface of the workpiece and thus thermally induced damage to the near-surface edge zones—the so-called “grinding burns”. Depending on the degree of damage, this leads to an unfavorable change in the depth profiles of the hardness and residual stress, even resulting in hardening cracks. The most important quality characteristics of surface-hardened workpieces are the hardening depths, i.e., the surface-hardening depth (SHD) after induction hardening, the case-hardening depth (CHD) after carburizing, the nitriding hardness depth (NHD) after nitriding, and the fusion hardness depth (FHD), e.g., after laser hardening. In addition, the hardness and residual stress values at the surface and in the bulk or even the depth-profiles of hardness and residual stress are relevant in order to assure the quality of such workpieces. All of these material parameters can be determined with 3MA after appropriate calibration. Figure 23 shows the CHD and SHD calibration results. For all surface-hardening mechanisms, 3MA not only allows the hardening depth to be determined, but also the hardness at the surface, the core hardness, and the hardness values between. With verification measurements on parts not used in these calibrations, the RMSE values, i.e., the average deviations between the nondestructive and destructive measuring data of 0.07 mm for CHD and 0.13 mm for SHD could be determined. Applications for the determination of FHD with 3MA have also been previously described [71,72]. Especially impressive are the results of 3MA for Appl. Sci. 2019, 9, x FOR PEER REVIEW 21 of 29

If the PHS should be laser welded, the aluminum–silicon coating could be a problem. Therefore, the coating is partially ablated by laser. Again, the 3MA technique can be used in order to verify the success of the partial ablation by determining the thickness of the residual coating layer [70]. Meanwhile, PHS testing is the most successful 3MA application by far. Almost all of the PHS‐ producing OEMs and suppliers from the automotive industry use 3MA.

4.3.2. Surface‐Hardened Parts and Machined Parts Surface heat treatments and machining processes are necessary to provide materials with the desired strength, shape, and surface quality. Hardening the material’s surface by case‐hardening, induction hardening, or nitriding will increase the resistance to surface wear without impairing the material’s toughness in the bulk. Usually, the workpiece is already formed or milled to its final shape when surface hardening is applied. Therefore, unavoidable quenching distortions that are due to hardening usually have to be removed afterwards. In order to ensure a high‐dimensional workpiece of accurate form, a subsequent fine machining of the workpiece’s surface by means of turning, grinding, honing, or polishing is often required after hardening. However, this surface machining process can, again, result in unwanted changes in the material’s microstructure. Machining can lead to a local temperature increase on the surface of the workpiece and thus thermally induced damage to the near‐surface edge zones—the so‐called “grinding burns”. Depending on the degree of damage, this leads to an unfavorable change in the depth profiles of the hardness and residual stress, even resulting in hardening cracks. The most important quality characteristics of surface‐hardened workpieces are the hardening depths, i.e., the surface‐hardening depth (SHD) after induction hardening, the case‐hardening depth (CHD) after carburizing, the nitriding hardness depth (NHD) after nitriding, and the fusion hardness

Appl.depth Sci. (FHD),2019, 9, 1068 e.g., after laser hardening. In addition, the hardness and residual stress values22 at of the 29 surface and in the bulk or even the depth‐profiles of hardness and residual stress are relevant in order to assure the quality of such workpieces. All of these material parameters can be determined with determining3MA after appropriate the NHD, e.g.,calibration. in piston Figure rings 23 [44 shows]. The the typical CHD values and SHD of NHD calibration were between results. 60 For and all 100surfaceµm.‐hardening It was found mechanisms, that, for the 3MA NHD not values only allows determined the hardening with 3MA, depth the to reproducibility be determined, is but better also thanthe hardness for the values at the determined surface, the by core metallographic hardness, and investigations the hardness (Nht700values between. test).

(a) (b)

FigureFigure 23. 23. ResultsResults ofof thethe calibrationcalibration toto hardeninghardening depth:depth: ((aa)) case-hardeningcase‐hardening depthdepth (CHD)(CHD) andand ((bb)) surface-hardeningsurface‐hardening depth depth (SHD). (SHD).

InWith the caseverification of machined measurements workpieces, on not parts only not the used hardness, in these but calibrations, also the residual the RMSE stress invalues, its depth i.e., distributionthe average will deviations be of interest. between In orderthe nondestructive to provide quality and informationdestructive regardingmeasuring the data microstructure of 0.07 mm for of aCHD grinded and component, 0.13 mm for the SHD nital could etching be determined. method is often Applications used. This for method the determination does not work of completely FHD with nondestructively.3MA have also been Grinding previously defects, described e.g., tempering [71,72]. Especially zones and re-hardeningimpressive are zones, the results will be of indicated 3MA for bydetermining the discoloration the NHD, of e.g., the surface in piston after rings the [44]. etching The typical process. values This of method NHD were is useful between so long 60 and as the100 informationμm. It was found from thethat, material’s for the NHD surface values is sufficient determined to describewith 3MA, the the overall reproducibility quality of is the better grinded than workpiece.for the values Nevertheless, determined thermally by metallographic induced changesinvestigations in the (Nht700 microstructure test). and residual stress due to grinding could reach up to 100 µm material depth or more. Moreover, such deep grinding defects could be “hidden”, e.g., covered by material with appropriate microstructure and residual stress. In such cases, it will not be possible to detect all grinding burns with nital etching. For years, 3MA has been used for the nondestructive detection and characterization of grinding burns. Such applications of 3MA have been developed in cooperation with industrial partners [40]. As part of a research project (FVA No. 453) with research partner FZG from Technical University of Munich, the relationships between the load-bearing capacity of gearwheels and the hardness (Hd) and residual stress (RS) in the gear wheel tooth flanks that are influenced by grinding were examined [73]. On the basis of the 3MA technique, the gearwheels were nondestructively tested, so that a tested gearwheel could be systematically examined in running tests with regard to its tooth flank load capacity and with regard to the change of the material properties in the tooth flanks. A specially adapted and miniaturized 3MA probe was developed and then integrated into a multi-axis manipulation system (see Figure 24). This so-called “gearwheel scanner” allows for the tooth flanks to be scanned in meander-shaped surface scans. For calibration, the 3MA measuring parameters were recorded on undamaged gear wheels with different degrees of grinding damage. Subsequently, the gradients of Hd and RS in these gearwheels were measured with the reference methods (Vickers hardness and X-ray diffractometry). On this database, the 3MA system was calibrated to hardness (Hd) at the material depths of 0.1, 0.2, 0.3, 1.0, and 3.0 mm and to the residual stress (RS) at the depths of 0.00, 0.01, 0.02, 0.04, and 0.06 mm (see Figure 25). The results show remarkably good correlations between 3MA parameters and Hd and RS; the RMSE value for the Hd calibration was 3HV and for the RS calibration 10 MPa, respectively. Appl. Sci. 2019, 9, x FOR PEER REVIEW 22 of 29

In the case of machined workpieces, not only the hardness, but also the residual stress in its depth distribution will be of interest. In order to provide quality information regarding the microstructure of a grinded component, the nital etching method is often used. This method does not work completely nondestructively. Grinding defects, e.g., tempering zones and re‐hardening zones, will be indicated by the discoloration of the surface after the etching process. This method is useful so long as the information from the material’s surface is sufficient to describe the overall quality of the grinded workpiece. Nevertheless, thermally induced changes in the microstructure and residual stress due to grinding could reach up to 100 μm material depth or more. Moreover, such deep grinding defects could be “hidden”, e.g., covered by material with appropriate microstructure and residual stress. In such cases, it will not be possible to detect all grinding burns with nital etching. For years, 3MA has been used for the nondestructive detection and characterization of grinding burns. Such applications of 3MA have been developed in cooperation with industrial partners [40]. As part of a research project (FVA No. 453) with research partner FZG from Technical University of Munich, the relationships between the load‐bearing capacity of gearwheels and the hardness (Hd) and residual stress (RS) in the gear wheel tooth flanks that are influenced by grinding were examined [73]. On the basis of the 3MA technique, the gearwheels were nondestructively tested, so that a tested gearwheel could be systematically examined in running tests with regard to its tooth flank load capacity and with regard to the change of the material properties in the tooth flanks. A specially adapted and miniaturized 3MA probe was developed and then integrated into a multi‐axis Appl.manipulation Sci. 2019, 9, 1068 system (see Figure 24). This so‐called “gearwheel scanner” allows for the tooth flanks23 of 29 to be scanned in meander‐shaped surface scans.

Appl. Sci. 2019, 9, x FOR PEER REVIEW 23 of 29

Appl. Sci. 2019, 9, x FOR PEER REVIEW(aFigure) Figure 24. 24. 3MA3MA gearwheel scanner scanner system. system.(b ) 23 of 29

FigureFor 25.calibration, Results of the the 3MA calibration measuring to depth parameters‐dependent were material recorded properties: on undamaged (a) hardness gear at wheelsdepths ofwith different0.1, 0.2, degrees0.3, 1.0, and of grinding 3.0 mm and damage. (b) residual Subsequently, stress at depths the gradients of 0, 0.01, of 0.02,Hd and 0.04, RS and in 0.06 these mm. gearwheels were measured with the reference methods (Vickers hardness and X‐ray diffractometry). On this database,The application the 3MA ofsystem the calibrated was calibrated 3MA tosystem hardness on gearwheels (Hd) at the materialwith different depths degrees of 0.1, 0.2, of 0.3,grinding 1.0, damageand 3.0 is shownmm and in to Figure the residual 26. Table stress 10 shows (RS) at the the meaning depths of of 0.00, the marking0.01, 0.02, codes 0.04, ofand the 0.06 curves mm in(see the diagramsFigure 25).of Figure The results 26. show remarkably good correlations between 3MA parameters and Hd and RS; theBecause RMSE valuethe application for the Hd ofcalibration 3MA is completelywas 3HV and nondestructive, for the RS calibration such measurements 10 MPa, respectively. of hardness and residual‐stress profiles could be repeated during the stress tests of the grinded gearwheels, as is often as desired. These investigations allowed for detailed analysis of the relation between the extent of grinding damage and the behavior of the gearwheels during the stress tests. Characteristic features in the Hd and RS profiles that are relevant to the durability of the gearwheel could be identified, allowing for the load‐bearing capacity (pitting fatigue strength) of grinded gearwheels to be (a) (b) predicted based on measurements with 3MA. TheFigure determination 25. Results of of the depth calibration ‐profiles to depth depth-dependent of hardness‐dependent (Hd) material and residual properties: stress ( a) hardness (RS) is useful, at depths not of only for gearwheels,0.1, 0.2, 0.3, 1.0,but and also 3.0 for mm other and ( belaborately)) residual stressstress processed atat depthsdepths components, ofof 0,0, 0.01,0.01, 0.02,0.02, such 0.04,0.04, andasand valve 0.060.06 mm.mm. springs. The production process of these high‐performance parts includes several steps that result in a final product with complicatedThe application Hd andof the RS calibrated depth‐profiles. 3MA Thesesystem can on begearwheels nondestructively with different determined degrees with of the grinding 3MA “valvedamage spring is shown scanner”, in Figure which 2626. is. Table shown 10 in shows Figure the 27a. meaning As shown of thein Figure marking 27b, codes H can of be the determined curves in the at materialdiagrams depths of Figure of 0.25, 2626.. 0.50, 1.00, and 2.00 mm and RS at depths of 0.00, 0.05, 0.10, 0.20, and 0.30 mm. Because the application of 3MA is completely nondestructive, such measurements of hardness and residual‐stress profiles could be repeated during the stress tests of the grinded gearwheels, as is often as desired. These investigations allowed for detailed analysis of the relation between the extent of grinding damage and the behavior of the gearwheels during the stress tests. Characteristic features in the Hd and RS profiles that are relevant to the durability of the gearwheel could be identified, allowing for the load‐bearing capacity (pitting fatigue strength) of grinded gearwheels to be predicted based on measurements with 3MA. The determination of depth‐profiles of hardness (Hd) and residual stress (RS) is useful, not only for gearwheels, but also for other elaborately processed components, such as valve springs. The production process of these high‐performance parts includes several steps that result in a final product with complicated Hd and RS depth‐profiles. These can be nondestructively determined with the 3MA “valve spring scanner”, which is shown in Figure 27a. As shown in Figure 27b, H can be determined at material depths of 0.25, (0.50,a) 1.00, and 2.00 mm and RS at depths of 0.00, 0.05,(b) 0.10, 0.20, and 0.30 mm. FigureFigure 26. 26. ResultsResults of of measurements measurements with with the the calibrated calibrated 3MA 3MA in in gearwheels gearwheels showing showing different different degrees degrees ofof grinding grinding damage damage with with designations designations FA0 FA0-II‐II…FE3 . . . FE3-II:‐II: (a) ( ahardness) hardness profiles proﬁles and and (b ()b residual) residual stress. stress.

Table 10. Results of different types of calibration—without and with US parameters.

Light tempered zones FA0‐II No grinding damage FB1‐II at < 10% of the flank Light tempered zones Light tempered zones FB2‐II FB3‐II at 10 to 25% of tooth flank at > 25% of tooth flank

(a) (b)

Figure 26. Results of measurements with the calibrated 3MA in gearwheels showing different degrees of grinding damage with designations FA0‐II…FE3‐II: (a) hardness profiles and (b) residual stress.

Table 10. Results of different types of calibration—without and with US parameters.

Light tempered zones FA0‐II No grinding damage FB1‐II at < 10% of the flank Light tempered zones Light tempered zones FB2‐II FB3‐II at 10 to 25% of tooth flank at > 25% of tooth flank Appl. Sci. 2019, 9, 1068 24 of 29

Table 10. Results of different types of calibration—without and with US parameters.

Light tempered zones FA0-II No grinding damage FB1-II at < 10% of the flank Light tempered zones Light tempered zones FB2-II FB3-II at 10 to 25% of tooth flank at > 25% of tooth flank FD3-II Strong tempered zones FE3 Re-hardening

Because the application of 3MA is completely nondestructive, such measurements of hardness and residual-stress profiles could be repeated during the stress tests of the grinded gearwheels, as is often as desired. These investigations allowed for detailed analysis of the relation between the extent of grinding damage and the behavior of the gearwheels during the stress tests. Characteristic features in the Hd and RS profiles that are relevant to the durability of the gearwheel could be identified, allowing for the load-bearing capacity (pitting fatigue strength) of grinded gearwheels to be predicted based on measurements with 3MA. The determination of depth-profiles of hardness (Hd) and residual stress (RS) is useful, not only for gearwheels, but also for other elaborately processed components, such as valve springs. The production process of these high-performance parts includes several steps that result in a final product with complicatedAppl. Sci. 2019, 9 Hd, x FOR and PEER RS REVIEW depth-profiles. These can be nondestructively determined with the24 3MA of 29 “valve spring scanner”, which is shown in Figure 27a. As shown in Figure 27b, H can be determined at material depths ofFD3 0.25,‐II 0.50,Strong 1.00, tempered and 2.00 mm zones and RS atFE3 depths ofRe 0.00,‐hardening 0.05, 0.10, 0.20, and 0.30 mm.

(a) (b)

FigureFigure 27.27. 3MA3MA valve valve spring spring scanner scanner system: system: (a) measuring (a) measuring system system and (b and) application (b) application for depth for‐ depth-dependentdependent determination determination of hardness of hardness (Hd) and (Hd) residual and residual stress stress(RS). (RS).

5.5. LimitationsLimitations TheThe applicationapplication of 3MA is is limited limited to to only only ferromagnetic ferromagnetic materials. materials. Due Due to the to theskin skin effect, effect, the theanalyzation analyzation depth depth of ofthe the 3MA 3MA application application is is always always restricted. restricted. As As already already described, described, this maximummaximum electromagneticelectromagnetic interactioninteraction depthdepth isis dependentdependent onon probeprobe headhead designdesign andand onon appliedapplied magneticmagnetic ﬁeldfield strengthstrength andand excitationexcitation frequency.frequency. Moreover,Moreover, thethe materialmaterial propertiesproperties (conductivity(conductivity andand magneticmagnetic hysteresis),hysteresis), as as well well as as the the size size and and the shapethe shape of the of inspected the inspected specimen, specimen, will have will an have inﬂuence. an influence. Usually, theUsually, maximum the maximum interaction interaction depth is not depth previously is not previously known, leading known, to anleading uncertainty to an uncertainty in depth-dependent in depth‐ measurements.dependent measurements. Fortunately, powerfulFortunately, simulation powerful tools simulation have been tools developed have been at Fraunhofer developed IZFP, at allowingFraunhofer for IZFP, a realistic allowing prediction for a realistic of the maximum prediction interaction of the maximum depth [ 74interaction]. depth [74]. InIn general,general, 3MA3MA calibrationcalibration cancan bebe challenging.challenging. Usually, aa seriesseries ofof calibrationcalibration samplessamples withwith gradedgraded targettarget valuesvalues isis required.required. EvenEven ifif thethe 3MA3MA calibrationcalibration measurementsmeasurements onlyonly representsrepresents littlelittle effort,effort, the the production production of of these these calibrationcalibration partsparts andand theirtheir (destructive)(destructive) testingtesting withwith referencereference methodsmethods could be expensive in time and costs. However, these costs must be faced with the enormous savings potential of the application of the calibrated 3MA system. Current research activities aim to reduce the experimental effort for 3MA calibration by developing generalized material databases and by using simulation models that are combined with machine learning algorithms (see Section 3.6). Further optimization of the 3MA probe head should be dedicated to the increase of its robustness and wear resistance in harsh environments, because, today, probe head damage due to its intensive usage cannot be avoided. Furthermore, there is a need for the realization of large probe heads allowing for lift‐offs higher than 5 mm. On the other side, smaller probe heads with a sensitive area not bigger than 0.5 × 0.5 mm2 are required for testing parts with restricting geometry (for example, in the radius of crankshafts). Additional measuring information could result from a deeper analysis of BN and IP double peak signals and IP droplet curves [75,76].

6. Conclusions The fundamental idea behind 3MA was born more than three decades ago. The main 3MA principle, i.e., the multiparametric approach, is based on the combination of measuring information from several micromagnetic methods and several measuring parameters. This allows for avoiding measuring ambiguities and offering the possibility to not only detect qualitative changes, but also to determine the quantitative values of the target quantities. The combined methods differ in their interaction mechanisms and interaction depths, and therefore their combination allows for the influence of disturbances (e.g., temperature, batch variations) to be minimized and different material depths to be sampled at once. The authors are convinced that the method combination is crucial in Appl. Sci. 2019, 9, 1068 25 of 29 could be expensive in time and costs. However, these costs must be faced with the enormous savings potential of the application of the calibrated 3MA system. Current research activities aim to reduce the experimental effort for 3MA calibration by developing generalized material databases and by using simulation models that are combined with machine learning algorithms (see Section 3.6). Further optimization of the 3MA probe head should be dedicated to the increase of its robustness and wear resistance in harsh environments, because, today, probe head damage due to its intensive usage cannot be avoided. Furthermore, there is a need for the realization of large probe heads allowing for lift-offs higher than 5 mm. On the other side, smaller probe heads with a sensitive area not bigger than 0.5 × 0.5 mm2 are required for testing parts with restricting geometry (for example, in the radius of crankshafts). Additional measuring information could result from a deeper analysis of BN and IP double peak signals and IP droplet curves [75,76].

6. Conclusions The fundamental idea behind 3MA was born more than three decades ago. The main 3MA principle, i.e., the multiparametric approach, is based on the combination of measuring information from several micromagnetic methods and several measuring parameters. This allows for avoiding measuring ambiguities and offering the possibility to not only detect qualitative changes, but also to determine the quantitative values of the target quantities. The combined methods differ in their interaction mechanisms and interaction depths, and therefore their combination allows for the inﬂuence of disturbances (e.g., temperature, batch variations) to be minimized and different material depths to be sampled at once. The authors are convinced that the method combination is crucial in achieving correct and reliable results in the application of micromagnetic methods. Without this approach, the 3MA technique would not be so accepted in this dimension in research and industry. So far, a broad range of applications on a variety of materials and components has been reported. 3MA was successfully used to quantitatively determine hardness, hardening depth, residual stress, parameters of static and dynamic tests (tensile, bending, fatigue, creep, impact testing), and microstructure features (texture, cementite, retained austenite). Meanwhile, some applications have been developed to an impressively high degree of technical maturity. Today, specialized 3MA systems are available for the testing of semi-ﬁnished products, such as strip steel and heavy plates, as well as all types of ferromagnetic steel or cast iron components, such as cold- or hot-formed parts, machined (milled, turned, grinded, etc.) parts, and heat-treated (through, induction, case hardened, etc.) parts. Nevertheless, there are still application possibilities to be explored and to be turned into industrial-suited 3MA systems. Current research activities are concerned with the implementation of 3MA into closed-loop control schemes. In the future, advanced 3MA systems will be able to identify individual products. The experimental effort towards 3MA calibration will be reduced, and the 3MA hardware and software will be further developed in regard to costs and large-scale manufacturability. Hence, ample work remains, at least for the next three decades.

Author Contributions: B.W. and Y.G. wrote the manuscript and contributed to the methodical development. C.C. and B.W. developed the optimization and calibration set-ups for 3MA application. All authors were involved in experimental investigations. Funding: Part of this research was funded by Bundesministerium für Bildung und Forschung, (BMBF, former BMFT), Project Number: BMFT RS 150.334, European Research Fund for Coal and Steel (RFCS, former ECSC), Contract Numbers: 7215-PP/055, 7210-PR/248, 7210-PA/339, RFSR-CT-2005-00044 and Forschungsvereinigung Antriebstechnik e.V., FVA-Forschungsvorhaben Nr. 453. The APC was funded by Fraunhofer Gesellschaft. Acknowledgments: The authors would like to respectfully thank the “inventors” of 3MA, who are former colleagues of Fraunhofer IZFP, like Paul Höller, Gerd Dobmann, Holger Pitsch, Werner Theiner, Rolf Kern and Iris Altpeter. But special thanks go also to all current colleagues who, for many years, have supported the research and the further development of the 3MA technique. These include, among others, Ute Maisl, Thomas Lambert, Andreas Haas, Florian Weber, Joachim Grimm, Thorsten Müller, David Böttger, Klaus Szielasko, Ralf Tschuncky and Melanie Kopp. The authors also want to thank Benjamin Straß, Andreas Haas and Florian Weber for advice and corrections in writing the manuscript. Appl. Sci. 2019, 9, 1068 26 of 29

Conﬂicts of Interest: The authors declare no conﬂict of interest.

References

1. Ono, K.; Dobmann, G. Nondestructive testing, 1. general. In Ullmann’s Encyclopedia of Industrial Chemistry; Ono, K., Erhard, A., Eds.; Wiley: Hoboken, NJ, USA, 2000. [CrossRef] 2. Hanke, R. Fraunhofer institute for non-destructive testing IZFP—Expanding the potential of NDT across the entire product life cycle. In Proceedings of the 55th Annual Conference of the British Institute of Non-Destructive Testing, NDT 2016, Nottingham, UK, 12–14 September 2016; pp. 305–308. 3. Wolter, B.; Dobmann, G.; Boller, C. NDT Based Process Monitoring and Control. J. Mech. Eng. 2011, 57, 218–226. [CrossRef] 4. Kersten, M. Special behavior of modulus of elasticity of ferro-magnetic materials. Z. für Metallkunde 1935, 5, 97–101. 5. In Memoriam Friedrich Förster. Available online: https://www.ndt.net/article/ndtnet/2009/foerster.pdf (accessed on 6 March 2019). 6. Föster, F. Ein messgerät zu schnellen bestimmung magnetischer größen. Z. für Metallkunde 1940, 32, 184–187. 7. Förster, F.; Stambke, K. Magnetische untersuchungen innerer spannungen. Z. für Metallkunde 1941, 33, 97–104. 8. Hainz, R. Beispiel einer einführung der elektromagnetischen sortierverfahren in die fertigung von dieseleinspritzpumpen und -düsen. Z. für Metallkunde 1955, 46, 358–370. 9. Hatch, H.P.; Fowler, K.A. Electromagnetic Method for Nondestructively Examining Components for Excessive Decarburization. Springfield Armory; Technical Report SA-TR19-1508; Springfield Armory: Springfield, MA, USA, 14 April 1964. 10. Leep, R.W.; Pasley, R. Method and System for Investigating the Stress Condition of Magnetic Materials. U.S. Patent No. US3427872, 18 February 1969. 11. Kronmüller, H. Magnetic Techniques for the Study of Dislocations in Ferromagnetic Materials. Int. J. NDT 1972, 3, 315–350. 12. Förster, F. Messung physikalischer und mechanisch-technologischer Werkstoffeigenschaften mit zerstörungsfreien Methoden. In Proceedings of the Jahrestagung der DGZfP-Jahrestagung 1973, Salzburg, Austria, 18–19 October 1973. 13. Förster, F.; Köster, W. Messung physikalischer und technologischer Materialeigenschaften mit Hilfe magnetischer und elektromagnetischer Meßmethoden. Industrie Anzeiger 1974, 96, 687–690. 14. Tiitto, S. On the mechanism of magnetization transitions in steel. IEEE Trans. Magn. 1978, 14, 527–529. [CrossRef] 15. Matzkanin, G.A.; Beissner, R.E.; Teller, C.M. The Barkhausen Effect and Its Applications to Nondestructive Evaluation; NTIAC Publications 79-2; Southwest Research Institute, SwRI: San Antonio, TX, USA, 1979. 16. Theiner, W.A.; Grossmann, J.; Repplinger, W. Zerstoerungsfreier Nachweis von Spannungen insbesondere Eigenspannungen mit magnetostriktiv angeregten Ultraschallwellen, Messung der Magnetostriktion mit DMS bzw. mittels des Barkhausenrauschens. In Proceedings of the Internationales Symposium Neue Verfahren der zerstoerungsfreien Werkstoffpruefung und deren Anwendung insbesondere in der Kerntechnik, Berlin, Germany, 17–19 September 1979. 17. Theiner, W.A.; Schorr, P.; Reimringer, B.; Deimel, P.; Herz, K.; Kuppler, D. Bestimmung der Mikrogefüge von Druckbehälterstählen mit Magnetisch Induzierten Messgrößen; Berichts Nr. 800413-TW; Fraunhofer IZFP: Saarbrücken, Germany, 1980. 18. Theiner, W.A.; Ayere, Q.; Salzburger, H.-J.; Herz, K.; Kuppler, D.; Deimler, P. Neue Verfahrensansaetze fuer die magnetische und magnetoelastische Gefuegepruefung. In Proceedings of the Internationale Konferenz über Zerstörungsfreie Prüfung in der Kerntechnik, Lindau, Germany, 1 April 1981. 19. Theiner, W.A.; Altpeter, I.; Becker, R.; Herz, R.; Repplinger, W. Eigenspannungsmessung an Stahl der Guete 22 NiMoCr 3 7 mit magnetischen und magnetoelastischen Pruefverfahren. In Proceedings of the Internationale Konferenz über Zerstörungsfreie Prüfung in der Kerntechnik, Lindau, Germany, 1 April 1981. 20. Tiitto, S. Über die Zerstörungsfreie Ermittlung der Eigenspannungen in ferromagnetischen Stählen. In Eigenspannungen: Entstehung, Berechnung, Messung, Bewertung; Deutsche Gesellschaft für Metallkunde: Oberursel, Germany, 1980; pp. 261–270. Appl. Sci. 2019, 9, 1068 27 of 29

21. Dobmann, G. Physical basics and industrial applications of 3MA—Micromagnetic multiparameter microstructure and stress analysis. In Proceedings of the 10th European Conference on Nondestructive Testing, ECNDT 2010, Moscow, Russia, 7–11 June 2010. 22. Becker, R.; Dobmann, G.; Rodner, C. Quantitative eddy current variants for micromagnetic microstructure multiparameter analysis-3MA. Rev. Prog. Quant. Nondestruct. Eval. 1987, 7B, 1703–1707. 23. Dobmann, G.; Pitsch, H. Magnetic tangential field-strength-inspection, a further ndt-tool for 3MA. In Proceedings of the 3rd International Symposium on Nondestructive Characterization of Materials, Saarbrücken, Germany, 3–6 October 1988; pp. 636–643. 24. Adler, E.; Pfeiffer, H. The influence of grain size and impurities on the magnetic properties of the soft magnetic alloy 47.5%NiFe. IEEE Trans. Magn. 1974, 10, 172–174. [CrossRef] 25. Höller, P. Nondestructive analysis of structure and stresses by ultrasonic and micromagnetic methods. In Nondestructive Characterization of Materials II; Green, R.E., Kozaczek, K.J., Ruud, C., Eds.; Springer: Boston, MA, USA, 1987; pp. 211–212. 26. Altpeter, I.; Dobmann, G.; Theiner, W.A. Quantitative hardening-depth-measurements up to 4 mm by means of micromagnetic microstructure multiparameter analysis-3MA. Rev. Prog. Quant. Nondestruct. Eval. 1987, 7B, 1471–1475. 27. Becker, R.; Döring, W. Ferromagnetismus; Springer: Berlin, Germany, 1939. 28. Kersten, M. Zur Theorie der Koerzitivkraft. Z. Phys. 1948, 124, 714–741. [CrossRef] 29.N éel, L. Effet des cavités et des inclusions sur le champ coercitif. Cah. Phys. 1944, 25, 21–44. 30. Kittel, C. Theory of the structure of ferromagnetic domains in films and small particles. Phys. Rev. 1946, 70, 965–971. [CrossRef] 31. Néel, L. Théorie des lois d’aimantation de Lord Rayleigh. 2ère partie: Multiples domaines et champ coercitif. Cah. Phys. 1943, 13, 18–30. 32. Guillaud, C. Propriétés ferromagnétiques des alliages manganèse-antimoine et manganèse-arsenic. Ann. Phys. 1949, 12, 671–703. [CrossRef] 33. Weil, L.; Marfoure, S. Variation thermique du champ coercitif du nickel aggloméré. J. Phys. Radium 1947, 8, 358–361. [CrossRef] 34. Hoslitz, K.; Sucksmith, A. Magnetic study of the two-phase iron-nickel alloys. Proc. R. Soc. Lond. 1943, 181, 303–313. 35. Dobmann, G. Non-Destructive Testing for Ageing Management of Nuclear Power Components. In Nuclear Power-Control, Reliability and Human Factors; Tsvetkov, P., Ed.; InTech: Zagreb, Croatia, 2011; pp. 311–338. 36. Alpeter, I.; Dobmann, G.; Hübschen, G.; Kopp, M.; Tschuncky, R. Nondestructive characterization of neutron induced embrittlement in nuclear pressure vessel steel microstructure by using electromagnetic testing. In Electromagnetic Nondestructive Evaluation XIV, Proceedings of the International Workshop on Electromagnetic Nondestructive Evaluation, Szczecin, Poland, 13–16 June 2010; Chady, T., Gratkowski, S., Takagi, T., Udpa, S.S., Eds.; IOS Press: New York, NY, USA, 2011; pp. 322–329. 37. Theiner, W.A.; Deimel, P. Non-destructive testing of welds with the 3MA-analyzer. Nuclear Eng. Des. 1987, 102, 257–264. [CrossRef] 38. Becker, R.; Dobmann, G.; Theiner, W.A. Progress in the micromagnetic multiparameter microstructure and stress analysis—3MA. In Nondestructive Characterization of Materials III, Proceedings of the International Symposium on Nondestructive Characterization of Materials, Saarbruecken, Germany, 3–6 October 1988; Höller, P., Ed.; Springer: Berlin, Germany, 1989; pp. 515–523. 39. Wolter, B. Zerstörungsfrei prüfen. Automobil Entwicklung 2004, Mai 2004, 58. 40. Wolter, B.; Theiner, W.A.; Kern, R.; Becker, R.; Rodner, C.; Kreier, P.; Ackeret, P. Detection and quantification of grinding damage by using EC and 3MA techniques. In Proceedings of the ICBM4—4th International Conference on Barkhausen Noise and Micromagnetic Testing, Brescia, Italy, 3–4 July 2003. 41. Wolter, B. Zerstörungsfreie charakterisierung von schleifbrand. In Proceedings of the VDI-Fachtagung Windkraftanlagen—Sicherheit und Zuverlässigkeit, Darmstadt, Germany, 4 November 2004. 42. Alpeter, I.; Boller, C.; Kopp, M.; Wolter, B.; Fernath, R.; Hirninger, B.; Werner, S. Zerstörungsfreie Detektion von Schleifbrand mittels elektromagnetischer Prüftechniken. In Proceedings of the DGZfP-Jahrestagung 2011, Bremen, Germany, 30 May–1 June 2011. Appl. Sci. 2019, 9, 1068 28 of 29

43. Alpeter, I.; Kröning, M. Nondestructive determination of the hardening depth in inductive hardened steels. In Nondestructive Characterization of Materials VI; Green, R.E., Kozaczek, K.J., Ruud, C., Eds.; Plenum Press: New York, NY, USA, 1994; pp. 659–668. 44. Dobmann, G.; Altpeter, I.; Wolter, B.; Kern, R. Industrial applications of 3MA—Micromagnetic multiparameter microstructure and stress analysis. In Electromagnetic Nondestructive Evaluation (XI)—ENDE 2007; Tamburino, A., Ed.; IOS Press: Amsterdam, The Netherlands, 2008; pp. 18–25. 45. Cosarinsky, G.; Kopp, M.; Rabung, M.; Seiler, G.; Petragalli, A.; Vega, D.; Sheikh-Amiri, M.; Ruch, M.; Boller, C. Non-destructive characterisation of laser-hardened steels. Insight 2014, 56, 553–559. [CrossRef] 46. Dobmann, G.; Kröning, M.; Koblé, T.D. Steel-grading by micromagnetic techniques. In Nondestructive Characterization of Materials V; Green, R.E., Ed.; Plenum Press: New York, NY, USA, 1992; pp. 615–621. 47. Ewen, M.; Blaes, N.; Bokelmann, D.; Braun, P.; Conrad, C.; Gabi, Y.; Kern, R.; Kopp, H.; Wolter, B. Nondestructive determination of mechanical properties of open-die forgings and potentials for full implementation in production process chain. In Proceedings of the 11th European Conference on Nondestructive Testing, ECNDT 2014, Prague, Czech Republic, 6–10 October 2014. 48. Alpeter, I. Nondestructive evaluation of cementite content in steel and white cast iron using inductive Barkhausen noise. J. Nondestruct. Eval. 1996, 15, 45–60. [CrossRef] 49. Maisl, U.; Kopp, M.; Altpeter, I. Electromagnetic testing methods for the detection of tendency to chilling. Cast. Plant Technol. Int. 1999, 15, 44–47. 50. Maisl, U.; Frauendorfer, R.; Kopp, M.; Altpeter, I. Zerstörungsfreies prüfverfahren zur bestimmung von werkstoffeigenschaften von gußeisen. Gießerei Praxis 2000, Heft 3, 113–121. 51. Borsutzki, M.; Dobmann, G.; Theiner, W.A. On-line ND-characterization and mechanical property determination of cold rolled stell strips. In Advanced Sensors for Metals Processing; Brusey, B.W., Bussière, J.F., Dubois, M., Moreau, A., Eds.; Canadian Institute of Mining: Montreal, QC, Canada, 1999; pp. 77–84. 52. Wolter, B.; Dobmann, G. Micromagnetic testing for rolled steel. In Proceedings of the 9th European Conference on Nondestructive Testing, ECNDT 2006, Berlin, Germany, 25–29 September 2006. 53. Gabi, Y.; Wolter, B.; Gerbershagen, A.; Ewen, M.; Braun, P.; Martins, O. FEM simulations of incremental permeability signals of a multi-layer steel with consideration of the hysteretic behaviour of each layer. IEEE Trans. Magn. 2014, 50, 1–4. [CrossRef] 54. Wolter, B.; Müller, T.; Behrens, B.-A.; Hübner, S.; Gaebel, C.M. Prüfsysteme zur prozessüberwachung beim Kragenziehen—Blechteilen geht es an den “Kragen”. QZ 2015, 60, 44–47. 55. Wolter, B.; Straß, B.; Müller, T.; Behrens, B.-A.; Hübner, S.; Wölki, K. Einsatzmöglichkeiten von zerstörungsfreien sensortechniken innerhalb der wertschöpfungskette blechverarbeitung. In Proceedings of the 38. EFB-Kolloquium Blechverarbeitung 2018, Bad Boll, Germany, 17–28 April 2018. 56. Wolter, B. Pressgehärtete karosserieteile zerstörungsfrei geprüft. Blechnet 2013, 5, 114–117. 57. Conrad, C.; Goebel, M.; Kern, R.; Lambert, T.; Wolter, B. Nondestructive testing for quality assurance and monitoring in the press hardening process. In Proceedings of the Strategien des Karosseriebaus 2015, Bad Nauheim, Germany, 17–19 March 2015. 58. Gabi, Y.; Wolter, B.; Kern, R.; Conrad, C.; Gerbershagen, A. Simulation of electromagnetic inspection techniques using FEM analysis. In Proceedings of the 19th World Conference on Nondestructive Testing—WCNDT 2016, Munich, Gemany, 13–17 June 2016. 59. Gabi, Y.; Böttger, D.; Straß, B.; Wolter, B.; Conrad, C.; Leinenbach, F. local electromagnetic investigations on electrical steel FeSi 3% via 3MA micromagnetic NDT system. In Proceedings of the 12th European Conference on Nondestructive Testing, ECNDT 2018, Gothenburg, Sweden, 11–15 June 2018. 60. Dobmann, G.; Seibold, A. First attempts towards the early detection of fatigued substructures using cyclic-loaded 20 MnMoNi 5 5 steel. Nucl. Eng. Des. 1992, 137, 363–369. [CrossRef] 61. Dobmann, G.; Kröning, M.; Theiner, W.A.; Willems, H.; Fiedler, U. Nondestructive characterization of materials (ultrasonic and micromagnetic techniques) for strength and toughness prediction and the detection of early creep damage. Nucl. Eng. Des. 1995, 157, 137–158. [CrossRef] 62. Lang, M.A. A way to detect hydorgen-induced stress corrosion cracking (HISCC) at an early stage. Matériaux Techniques 1998, 7–8, 27–32. [CrossRef] 63. Borsutzki, M. Prozeßorientierte Ermittlung der Streckgrenze und der Tiefziehkenngrößen rm und Delta r an Kaltgewalztem, Feuerverzinktem Feinblech. Ph.D. Thesis, Saarland University, Saarbruecken, Germany, 1997. Appl. Sci. 2019, 9, 1068 29 of 29

64. Salzburger, H.-J.; Niese, F.; Dobmann, G. Emat Pipe inspection with guided Waves. Weld World 2012, 56, 35–43. [CrossRef] 65. Meilland, P.; Marmulev, A.; Wolter, B.; Conrad, C.; Hofmann, U.; Dussaulx, J.; Weber, W. European Commission. Combined Ultrasound and Micromagnetic Measurements for Non-Destructive Assessment of Textured Heavy Plate Properties (PLATEND); Final Report, Research Fund for Coal and Steel, Project ID: RFSR-CT-2005-00044, 2005-07-01 to 2008-06-30; Publications Office of the European Communities: Luxemburg, 1 June 2010. 66. Karbasian, H.; Tekkaya, A.E. A review on hot stamping. J. Mater. Process. Technol. 2010, 210, 2103–2118. [CrossRef] 67. Neugebauer, R.; Schieck, F.; Polster, S.; Mosel, A.; Rautenstrauch, A.; Schönherr, J.; Pierschel, N. Press hardening—An innovative and challenging technology. Arch. Civ. Mech. Eng. 2012, 12, 113–118. [CrossRef] 68. Wolter, B.; Conrad, C.; Kern, R.; Wicke, T.; Tonne, J. Industrial demands and nondestructive testing (NDT) solutions for process monitoring and quality control in hot and cold formed steel production. In Proceedings of the 5th International Conference Hot Sheet Metal Forming of High-Performance Steel, Toronto, ON, Canada, 31 May–3 June 2015. 69. Neugebauer, R.; Wienera, T.; Zösch, A. Quality control of resistance spot welding of high strength steels. Procedia CIRP 2013, 12, 139–144. [CrossRef] 70. Bastuck, M. In-Situ-Überwachung von Laserschweißprozessen Mittels Höherfrequenter Schallemissionen, Dissertation. Ph.D. Thesis, Saarland University, Saarbruecken, Germany, 2016. 71. Kern, R.; Theiner, W.-A.; Höller, P. Residual stress- and microstructure anisotropies in laser-hardened or shot peened surfaces. In Proceedings of the 9th International Conference on Experimental Mechanics, Lyngby, Denmark, 20–24 August 1990. 72. Kern, R.; Theiner, W.-A. Non-destructive magnetic testing of laser hardening parameters. In Laser Treatment of Materials; Mordlike, B., Ed.; Deutsche Gesellschaft für Materialkunde (DGM): Oberursel, Germany, 1987; pp. 427–434. 73. Schwienbacher, S.; Wolter, B. Randzonentragfähigkeit-Zahnflanke-Ermittlung und Charakterisierung von Randzonen-Kennwerten und -eigenschaften und deren Einfluss auf die Flankentragfähigkeit einsatzgehärteter, geschliffener Zahnräder-Abschlussbericht Forschungsvorhaben Nr. 453 (AVIF-Nr. A200); Forschungsvereinigung Antriebstechnik e.V. (FVA): Frankfurt am Main, Germany, 2007. 74. Gabi, Y.; Martins, O.; Wolter, B.; Strass, B. Combination of electromagnetic measurements and FEM simulations for nondestructive determination of mechanical hardness. AIP Adv. 2017, 8.[CrossRef] 75. Gabi, Y. Modélisation FEM du Système de Contrôle non Destructif 3MA en Ligne de Production des Aciers Dual Phase. Ph.D. Thesis, Grenoble University, Grenoble, France, 2012. 76. Meilland, P.; Lombard, A. Improved modelling of the 3MA system’s incremental permeability for on-line steel strip property assessment. In Proceedings of the 19th World Conference on Nondestructive Testing—WCNDT 2016, Munich, Gemany, 13–17 June 2016.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).