Dynamics of chip formation during the cutting process using imaging techniques: A review

Imaging techniques have been widely implemented to study the dynamics of chip formation. They can offer a direct method and a full field measurement of the cutting process, providing kinematic information of the chip formation process. In this article, the state of the art of the imaging techniques reported in the literature has been summarized and analyzed. The imaging techniques have been applied to study the chip formation mechanism, friction behavior, strain/strain rate, and stress fields. Furthermore, the study of surface integrity has been advanced by deriving the thermo‐mechanical loading, subsurface deformation, and material constitutive model from the imaging technique. Finally, achievements in the area of imaging techniques have been summarized, followed by future directions for their application in the study of surface integrity.

a powerful and comprehensive tool to study the metal cutting process. Its vast applications have been reported in previous publications. 24,25 However, accurate determination of complex factors and validation of these numerical-based models on a global scale remain limited. 26 To reduce the gap between modeling and the real cutting process, several experimental techniques have been used. Early efforts can be traced back to the microgrids method. Elaborated microgrids are inscribed on the side of the workpiece using different techniques, and then the deformation can be determined according to the distorted grid after machining. Combined with quick-stop devices, Stevenson and Oxley measured the strain rate field within the primary shear zone. 27 Roth and Oxley constructed the slip-line field based on the experimentally measured flow field. 28 Despite the success achieved by the quick-stop device, its main drawback is the nonnegligible time required to separate the tool from the workpiece, especially in high-speed cutting, causing the deformation state to be disrupted.
Therefore, the microgrids method is modified by using imaging techniques instead of a quick-stop device to freeze the cutting process. Recording the deformation of the grid as a series of images captured by high-speed cameras, the in situ information can be obtained by further postprocessing. Childs used the double-exposure technique to deduce the velocity, strain rate, and even slip-line field in the chip formation zone; the cutting speed was limited to 254 µm/min. 29 Due to the development of high-intensity lighting systems and high-speed filming devices, Pujana et al. 30 were able to measure the strain and strain rates in the cutting process when the speed reached 300 m/min. Sela et al. 31 proposed a novel methodology for measuring the strain and strain rate fields in the primary shear zone using only a single image. Recently, two image correlation techniques called digital image correlation (DIC) and particle image velocimetry (PIV) have been developed. Similar to the microgrids method, a high-speed camera or a double-shutter camera is used to observe the material's deformation. Moreover, the DIC technique does not require use of complex grids on the lateral surface of the workpiece.
This article focuses on the imaging technique (PIV or DIC) coupled with the microgrids method, and it is organized as follows.  detectors. CCD architecture captures higher-quality images, while CMOS enables superior integration and responsivity. 33 A survey 34 of high-speed cameras based on frame rate and CCD and CMOS technologies is presented in Figure 1, which shows that a CMOS-based higher frame rate camera sacrifices the resolution, while for CCD-based cameras, the resolution and frame rate are independent.
Light sources are another important factor of the imaging system.  Table 1. The two most popular lighting sources are laser and light-emitting diodes (LEDs). They both belong to "cold light" and do not generate high temperatures, which could affect the image quality. 35 Moreover, the pulsed laser and LEDs offer the possibility to considerably reduce the interframe time. 18,36 There exist two main methods to record images. One is using continuous light sources, and the exposure time is controlled by the camera shutter. 37,38 The other is with the shutter open, and a synchronized pulsed light allows determination of the exposure time. 18,36 It is worth noting that the latter system needs to be used in a dark room during the imaging period to reduce the noise.

| Advanced in situ imaging systems
To understand local phenomena during chip formation, several experimental setups using imaging systems had been developed. An interesting study on kinematic field measurements was performed by Baizeau et al. 39 They used a double-frame imaging device with pulsed lasers to obtain highly resolved images during orthogonal cutting and used a ×10 lens with a resolution of 0.66 µm/pixel. As shown in Figure 2A, a LaVision Imager sCMOS double-frame camera and a dualpulsed laser were used to capture the images. This equipment captured image pairs before, during, and after the cutting at a frequency of 15 Hz using the synchronization system shown in Figure 2B.
Another well-organized imaging device for orthogonal cutting analysis was developed by Meurer et al. 40 They conducted the experiments on a vertical external broaching machine. As shown in Figure 3A,B, their imaging system includes an sCMOS double-shutter camera offering a resolution of 2560 × 2160 pixels at a dynamic range A/D of 16 bit and a double-pulse laser of Quantel EverGreen 70 mJ working at a wavelength of 532 nm at a 15 Hz repetition rate.
Harzallah et al. 41 proposed an imaging system for coupled kinematic and thermal field measurements, as shown in Figure 4.
The imaging system was equipped with two continuous highpower LEDs, a high-speed camera, Fastcam SA3, and an infrared camera, FLIR SC7000. The kinematic and thermal images can be obtained at the same time, but the only drawback is that the image quality is not good enough to distinguish between the cutting tool and the chip.
Zhang et al. 3,18,38,42,43 conducted several investigations to obtain chip formation and shear zone information based on in situ imaging.
They developed several imaging systems, including (1) a high-speed camera (PCO.dimax HD) with a continuous white laser for low-speed cutting as shown in Figure 5 and (2) a double-shutter camera (PCO.pixelfly) with pulsed LEDs for high-speed cutting as shown in Figure 6. Regarding the image quality, they found that the LED lighting source is better than laser. A comparison of the apparatus and the setting parameters is shown in Table 2.

| Digital image processing technique
Digital image processing techniques are used to obtain the dynamic information from a series of images recorded by imaging systems. PIV 44 and DIC 45 are two of the most effective techniques.
Traditionally, PIV has been widely used to track the instantaneous velocity of particles in fluids. Meanwhile, high-speed PIV can be applied in aerodynamics, 46 hydrodynamics, 47 and even biomedical research. 48 DIC is very similar to PIV in terms of the principle and the F I G U R E 1 Survey of high-speed cameras based on frame rate and window size. Reproduced with permission. 34  Local subset-based DIC 52-55 is a classical approach to calculate points of the region of interest (ROI) one by one to determine displacements. Namely, each speckle pattern is analyzed independently.
A schematic illustration of the local subset-based DIC approach principle is shown in Figure 7. In this approach, the analysis starts by selecting a reference subset centered at the point in ROI and matching its target counterpart in the deformed image with maximum similarity. Then, a displacement mapping function that quantifies the degree of similarity between two subsets is used to optimize the position and shape of the target subset. Finally, the desired

| DYNAMICS OF CHIP FORMATION
A better understanding of chip formation mechanisms is necessary.
Chip formation leads to improvements in the surface integrity and functional performance and life of the components. Imaging systems combined with digital imaging processing techniques are suitable for analyzing the dynamic cutting processes due to their accessibility and processing time. In this section, chip formation mechanisms, tribological behavior in metal cutting, strain, strain rate, and stress fields in the deformation are discussed.

| Finite element modeling to study chip formation
Metal cutting is a material removal process, in which external energy is applied to the cutting system, causing the separation of the layer being removed. The finite element method (FEM) is an effective approach adopted in the manufacturing field. The FE approach not only reduces the expensive and time-consuming experimental testing but also allows analysis of the mechanisms of chip formation. Obikawa et al. 64 reported that FEM can be applied to simulate and analyze the mechanism of discontinuous/serrated chip formation process in orthogonal cutting. Childs 65  suggested that the causes and effects of chip segmentation need to be considered in the simulation to improve tool life prediction and select the optimal cutting parameters for improved product quality. Guo et al. 68 investigated the chip formation mechanism using FEM considering a damage model. They presented to simulate discontinuous chips in high-speed machining ( Figure 9A). Using numerical simulation, Duan et al. 69 found that the degree of chip segmentation and the space between each chip segment increase with decreasing tool rake angle ( Figure 9B). The effect of cutting speed on chip characteristics was investigated by Li et al., 70 and the simulation of microcrack during serrated chip formation was achieved through FEM ( Figure 9C). Xu et al. 71 predicted the microstructure of serrated chips by FEM and cellular automata methods considering grain refinement ( Figure 9D).

| Experimental methods to observe the chip formation process
There are a large number of experimental studies on the chip formation process using several techniques. In the earlier days of metal found discontinuous chips for cutting speeds below 30 m/min ( Figure 10B). The chip segmentation process was found to be the result of instabilities in the cutting process and this is further augmented by the dynamic response of part of the machine tool structure. Jaspers et al. 74 indicated that the quick-stop device was well suited to study the chip formation process ( Figure 10C). Vyas and Shaw 75 used a quick-stop device and drew the conclusion that the tendency for saw-tooth chip formation increases with the cutting speed, owing to the increase of the strain rate ( Figure 10D).
Using a quick-stop device also enabled investigation of the deformation in the primary and secondary deformation zones, and observation of the built-up edge and the stagnation point. However, the design of the quick-stop device required a significant amount of work, since a large force is required to retract the cutting tool or workpiece during experiments. In addition, the time-varying strain rate, strain, and temperature cannot be captured using this method.
With the development of digital imaging techniques, the dynamics of the metal cutting process can be captured easily. Sutter et al. 76 Figure 12B). They noticed that the built-up edge structure obtained at the highest cutting speed was thinner than the lowest one. The differences in chip morphology during machining were investigated by Davis et al. 84

| Tribological behavior in metal cutting
Contact between bodies is encountered in many engineering applications. In metal cutting, the tool-chip and tool-workpiece contact significantly influences chip formation. Numerous analogy experiments have been developed to obtain the friction coefficient and to On the other hand, Pujana et al. 30 used high-speed filming to obtain a sequence of frozen images of chip formation using a workpiece marked with square grids. The results are shown in Figure 16A.
Recently, Sela et al. 31 created physical microgrids in a workpiece and measured strain and strain rate fields in the primary deformation zone using a single image with the distorted grids, as shown in Figure 16B. The microgrids were observed to have severe distortion, and the results of strain and strain rate were not smooth enough.
Moreover, the fabrication of microgrids on the workpiece surface is complex, and due to the low spatial resolution, calculation of the strain near the tool cutting edge can be inaccurate.

| In situ imaging with the DIC technique
In situ imaging and the DIC technique have been successfully implemented for characterizing the strain and strain-rate distributions in metal cutting. 100 Using these techniques, Guo et al. 37,101,102 obtained high strain ( Figure 17A) with a low cutting speed during chip formation.
They found that chip flow is steady and laminar at large negative rake angles. As the rake angle increases, the chip morphology changes from discontinuous to segmented, and then to a continuous chip. Crack initiation occurs at the prow-free surface for a strain of approximately 0.8 in brass, providing a strain-controlled failure criterion. List et al. 103 focused on high-speed cutting of a mild steel to analyze the strain and strain rate distributions by observing the chip using a high-speed camera system. Thimm et al. 104 studied the chip formation and the shear strain rate of AISI 1045 steel at several cutting speeds, uncut chip thickness, and rake angle using double-frame camera devices. Harzallah et al. 41 described the use of a bispectral imaging apparatus to obtain the kinematic and thermal fields simultaneously in orthogonal cutting. They found that the region of high strain rate is fixed in space and their values only vary slightly over time (see Figure 17B) due to the cyclic nature of chip formation. Davis et al. 105 studied the strain progression within the primary deformation zone in the chip flow direction during machining ( Figure 17C). A similar work was presented by Yadav and Sagapuram 106 ; they captured the strain history along the chip flow, which is shown in Figure 17D. Zhang et al. 18 described the workpiece behaviors accurately during machining by means of in situ imaging.
Different levels of strain/strain rate were obtained by varying the rake F I G U R E 15 (A) Material velocity calculated using particle image velocimetry and (B) sliding velocities and local friction coefficients. Reproduced with permission. 95 Copyright 2021, Elsevier angle, cutting speed, and initial workpiece temperature. Figure 17E shows an example of the strain rate field obtained by these researchers.
Bergs et al. 107 also presented the in-process analysis of the strain/strain rate in the primary deformation zone using the imaging technique, as shown in Figure 17F. The in situ imaging and DIC techniques have contributed toward better understanding of the chip formation mechanism by making it convenient and easy to obtain intermediate physical information, including velocity, strain, and strain rate fields.

| Stress field
The stress fields generated during cutting are also of great value for studying cutting mechanisms, improving surface integrity, and reducing tool wear. However, although the research on strain and strain rate in the cutting process is extensive, as previously mentioned, mapping the stress field in the deformation zone using the in situ imaging technique remains a challenge because of its considerable sensitivity to experimental result noise.
To address this issue, some researchers used Oxley's model, 97,108 as shown in Figure 18A, in orthogonal cutting. Using this model, the primary deformation zone is simplified and represented by a parallelsided zone centered about the straight-line AB shown in Figure 18A.
Based on this assumption, the von-Mises stress can be calculated based on the DIC-measured equivalent strain and strain rate with a given constitutive model. Combined with the shear angle determined by the strain rate field, the plastic stress distribution in the primary shear zone is obtained, as shown in Figure 18B. 109 In addition, using the parallel-sided shear zone to characterize the deformation in the primary shear zone, Huang et al. 63 proposed a model-based DIC algorithm to reconstruct the stress field. On the other hand, efforts have been made to modify the image technique for stress field determination. Zhang 38 adopted mechanical equilibrium equations to modify the hydrostatic pressure field to compensate for the inaccuracy of the measured elastic deformation. In a subsequent publication by the same author, 3 a FEM analysis was performed to optimize the velocity field determined by DIC and to generate the stress field, as shown in Figure 19A,B, respectively.

| Thermo-mechanical loadings
Traditionally, most studies on machining surface integrity start from the machining parameters and aim to establish their relationship with In the last decade, to describe the underlying mechanism of the surface integrity generation more conveniently and adequately, a new concept named "Process Signature" was proposed by Brinksmeier et al. 121 in 2011. Instead of the machining parameters or external loads, process signature only takes the internal loads such as stress, strain, temperature, and their gradients into account, and then physics-based correlations between these internal loads and the resulting material modifications are established (Figure 21 correlation 3). It is claimed that process signature can not only predict the surface integrity in the machining process but can also provide a knowledge-based approach to determine the optimum process parameters for the desired surface integrity. As an effective method to obtain in situ kinematic information, the imaging technique can be used to determine the internal loads during the cutting process and correlate them with the corresponding material modifications ( Figure 22). It is reported that the large initial strain in the workpiece and the high wedge incidence angle promote the formation of localized deformation in the form of a shear band as shown in Figure 23B.

| Subsurface plastic deformation
Furthermore, Guo et al. 37 studied the occurrence of this instability and crack initiation in both sliding and cutting conditions, and developed a universal mechanism to determine the chip morphology.
Using a high-resolution imaging technique, many investigations have explored the dependence of subsurface deformation on the cutting parameters. For example, it was found that the more negative the tool rake angle, the more severe the subsurface plastic deformation. 36,[125][126][127][128][129] It was reported that different undeformed chip thicknesses generate a similar in-depth profile of the subsurface plastic strain with respect to the normalized depth, which was obtained by dividing the depth by the undeformed chip thickness 36,125,126 as shown in Figures 24 and 25. Moreover, similarities between the deformation history of the chip and the near-surface region have also been reported. [125][126][127]129 In addition to enabling the study of the surface integrity directly, the imaging technique can also offer full-field subsurface information for metal cutting modeling and simulation. On the one hand, it can be an effective way to validate or evaluate both the analytical 130 and numerical-based 128 models by comparing the predicted results with those measured experimentally. On the other hand, it can be widely used to improve hybrid models of surface integrity. 131 As discussed above, use of imaging techniques has mainly been focused on studying the kinematic state on the subsurface, such as material flow and deformation. However, some recent publications have demonstrated direct links between the plastic strain components or eigenstrain in the subsurface and the residual stress 112,113 and between the equivalent plastic strain and white layer formation. 132 Therefore, it is expected that the imaging technique will have an extended range of applications in surface integrity research in the future. reproduced by simple SHB compression tests using cylindrical specimens. 4,18 To tackle these issues, some attempts have been made to use the cutting process itself as the material testing method to determine the coefficients of the constitutive model using an inverse approach. [133][134][135] The main idea is to conduct the cutting experiment for different cutting parameters to generate the required range of strain, strain rate, cutting temperature, and state of stresses in the primary shear zone. Then, coefficients of the constitutive model can be identified by matching experimental measured stress distribution in primary shear zone to analytical results.

| Identification of the constitutive parameters
An analytical model assuming uniform distribution of the strain along the primary deformation zone is not consistent with reality.
Therefore, the imaging techniques are used to accurately determine the strain and strain rate distributions in the first deformation zone.
A summary of the imaging techniques used in the inverse approach to determine the coefficients of the constitutive model is presented in Table 3.

| SUMMARY AND FUTURE DIRECTIONS
This review summarizes the current state of the art of the dynamics of the chip formation using imaging techniques along with their applications for surface integrity analysis. Significant progress has been made in the development of imaging techniques to study the strain and strain rate during the chip formation process. It has been implemented in the orthogonal cutting process of continuous and segmented chip formation and milling processes. In addition, the research community is actively engaged in research on stress calculation from the kinematic fields. The achievements can be summarized based on progress made in the following areas: • Clear images of the cutting process to study serrated chip formation processes and workpiece stagnation point.
• Velocity fields to identify the built-up edge and the sliding speed between the tool and the chip.
• Strain rate and strain, particularly in the primary shear zone and the machined surface. The success of the investigations conducted in the following research areas is attributed to the measured information: • Inverse identification of the work material constitutive parameters.
With the measured strain, strain rate, stress, and derived temperature in the primary shear zone, the material behavior in the high strain rate, strain, and the rate of temperature increase during the cutting process have been studied.
• Identification of tool-chip friction. Through measurement of the sliding velocity of the work material near the cutting tool edge and the derivation of the stress through stepwise increment of the cutting depth, the sliding velocity-dependent friction coefficient has been determined.
• Slip-line construction. Because of the measured strain rate field, the slip-line field in the main deformation zone has been successfully identified.
• Numerical confrontation. The finite-element simulated strain rate field has been validated by the in situ imaged result.
Nevertheless, the imaging techniques need further development as below: • Measurement of the plastic deformation along the adiabatic shear zone, tool-chip friction, and machined surface. In these regions, extensive plastic deformation occurs in a narrow zone compared with the primary shear zone, which requires higher resolution and frame rate as well as lighting.
• Determination of the separation zone. The separation of the chip from the work material near the cutting tool edge poses a major challenge to the image correlation algorithm since it is normally continuous subset-based. Therefore, subset-splitting-based image correlation should be adopted to address this challenge.
• Lubricants and cooling. In the practical machining process, lubricants or cooling are generally used. The possible obstruction and contamination to the imaging process should be considered to study the effects of the lubricants or cooling on the cutting process.
• Three-dimensional cutting process. The in situ imaging techniques should be updated and adjusted to study the oblique cutting process that occurs in the practical turning and milling processes.
• Experimental validation of measurement results. There are many factors influencing the measurement accuracy of the in situ imaging techniques. Experimental validation using more accurate methods such as the microgrids method should be conducted.
In future research, the eigenstrains or plastic strain components during the machining process, which are critical to the prediction of machined surface integrity, 112 can be determined using the derivation in situ imaging technique. Future studies can be focused on the following topics: • Derivation of the thermo-mechanical loading. With the calculated stress field and temperature fields, the thermo-mechanical loading applied on the primary and tertiary shear zones can be calculated.