1 Introduction

The development of manufacturing directly affects the speed of economic growth and the progress of human civilization. In addition, the damage to the environment (such as resource scarcity, greenhouse effect, sewage discharge, etc.) has become more and more apparent. The concept of green manufacturing has gradually emerged. The green manufacturing has changed the original end-of-pipe approach to environmental protection. It protects the environment from the source and considers the basic attributes of the product. With this technology, the product meets the requirements of the environmental objectives while ensuring the basic performance, service life, quality, etc., that the product should have [1]. With the help of various advanced technologies, green manufacturing innovates the manufacturing mode, manufacturing process, and manufacturing organization. Its goal is to construct the product from the design, manufacturing, packaging, transportation, and use, and to recycle the disposal of the entire life cycle without any environmental pollution, high resource utilization, and low energy consumption. Finally, green manufacturing achieves the coordination and optimization of economic and social benefits [2, 3].

Methods to achieve green manufacturing include an effective material utilization, process monitoring, waste elimination or minimization, and material recycling [2], where process monitoring is critical because it links the entire green manufacturing system and determines the level of green manufacturing development. An acoustic emission (AE) technology of a low-cost, and a high-efficiency engineering application technology, enables a quality assurance throughout the product life cycle [4]. The AE technology detects stress waves using sensitive sensors or sensor arrays, which is vital for the green manufacturing.

AE is a direct transient elastic energy that is generated by stress waves of mechanical deformation, which is caused by various materials. These stress waves on the surface of material structures can be analyzed and detected using AE monitoring systems [5]. The monitoring system presented is shown in Fig. 1.Holford [6] proposed that damage and fracture mechanisms can be located, identified, quantified, and detected by AE techniques. The AE technology is a real-time non-destructive testing technology. It has many advantages. The AE sensors are small and portable, and only need to be bonded to the object under the test, using a coupling agent. The placement of the sensor is free, so it can be used for different shapes of the body under test. Also, the sensor’s environmental adaptability is extreme, and it can be used in a variety of harsh environments. It does not require an external supply of detection signals, the signal that comes from the internal material can reflect the deformation and crack expansion of the material. It is very suitable for real-time monitoring of material damage during processing and of the health status of a mechanical equipment. With the rapid development of the sensor and computer technology, the AE signal identification and acquisition has become more accurate and has been used to detect, monitor, and analyze various parameters that are related to the cutting process (such as tool wear, chip breakage and fracturing, chatter vibration, and buildup edge (BUE) formation. AE technology is cost-effective while attaining monitoring objectives. This attribute further enhances its viability and desirability as a sustainable solution. AE technology plays a crucial role in ensuring sustainable economic development and promoting green manufacturing practices [7]. Its proven sustainability can be observed in various sectors, including aerospace, and civil [8, 9].

Fig. 1
figure 1

AE monitoring system

Full size image
Fig. 2
figure 2

The application of AE in green manufacturing

Full size image
Fig. 3
figure 3

The AE phenomena during machining

Full size image

This paper reviews the framework of Fig. 2, which shows the existing application areas of the AE technology in green manufacturing. The first section of this paper examines the research on surface integrity, tool wear, chip shape, and chip size for AE monitoring, with the purpose of improving material utilization. Section 2 describes the applications of AE technology in monitoring different processes, with the purpose to improve the energy efficiency. Section 3 introduces the application of AE technology in monitoring the green cutting field. The amount of cutting fluid is then controlled by the AE signal to achieve the purpose of pollution reduction. Finally, this review proposes a future development direction of the AE technology applications in the field of green manufacturing.

2 Improvement of Material Utilization

With the rapid progress in science and technology, the contradiction between human and social development, and the loss of natural resources, has gradually emerged. The accompanying concept of sustainable development has become popular worldwide [10]. Sustainable development requires the refinement of manufacturing processes to reduce scrap rates, increase productivity, and use resources efficiently. In addition, the achievement of a sustainable development needs the support of high-precision monitoring and measurement systems [11, 12]. When a material or structure is defective or damaged, stress waves are generated from within. According to Rodriguez, P., AE sensors can identify these stress waves [13], and their sensitivity and reliability are sufficient to complete the investigation and analyses of processing characteristics and failures [14, 15].

The AE phenomena during machining (as shown in Fig. 3) are vibrant. Plastic deformation of materials, friction between chips and workpiece surfaces, and tool wear are all expressed as AE signals with different characteristics [16,17,18]. The AE signal is defined as the elastic energy that is released by the machining process. Furthermore, the AE energy is represented in the graph in two forms: continuous and abrupt. The continuous type of signal is collected from plastic cutting, and the abrupt type of signal comes from damaged fracture, crack extension, edge cutting, chip adhesion, etc. The original AE signal cannot be directly identified by pattern recognition but needs to be analyzed in the time-frequency domain to extract its features, such as peak value, mean value, root mean square, energy density, intensity, etc. [19, 20]. The standard analysis methods include Fourier transform and wavelet transform [21,22,23,24]. Artificial intelligence has also shown excellent results in the level of AE signal analysis [25, 26].

This section will explain how AE monitoring technology can scientifically improve the resource utilization at three levels: surface integrity, chip formation mechanism, and tool wear mechanism.

2.1 Surface Integrity

If the product processing is not qualified, it will not function properly, which results in scrap and wasted material resources. This subsection focuses on applying AE to assist machining in achieving an excellent surface integrity. The surface integrity of a part is the critical factor that affects the material properties and fatigue life. Surface integrity [27] is a comprehensive indicator that characterizes, evaluates, and controls the various changes that may occur within the surface layer during processing and manufacturing, and its impact on the performance of the final product in use. Plastic deformation, microscopic cracks, material phase changes, microhardness, tears and folds, and residual stress distribution affect the surface integrity. These surface characteristics result from changes in variables within the material [28], and generate AE abrupt signals. Dragos et al. verified that the change in surface integrity, using an AE sensor, or sensor arrays, can be used for accurate source localization of mutation-type events [29]. Advances in sensor technology and intelligent algorithms have led to a more accurate localization of AE sources [30, 31]. Table 1.

Table 1 Cases of Improving material utilization using AE technology
Full size table

The machining surface quality is the most common surface integrity characteristic. It includes surface morphological characteristics, and physical and mechanical properties. It is usually characterized by its surface roughness and finish. Ref. [32] demonstrated that the root mean square (RMS) of the AE is correlated with the surface quality. Sutowski et al. [33] evaluated the surface quality of an abrasive waterjet machining using AE signals. They found a significant correlation between the signal values (AE amplitude and RMS) and the machined surface quality. As the line speed increased, the machining quality changed, and the AE amplitude and AE RMS increased. They developed a statistical dependence model between surface quality and AE signal parameters.

AE monitoring of surface qualities is not limited to machining methods and has been demonstrated in various machining fields. De Agustina et al. [34] investigated the correlation between the machined surface integrity and AE signals for robot-assisted polishing. They established an AE-based monitoring system, demonstrating the applicability of AE monitoring technology for ultra-precision machining applications. To obtain an accurate roughness prediction model for grinding, Guo et al. [35] completed the processing of sensor signals, such as selecting AE feature combinations and developing prediction models. After a correlation analysis of many features with surface roughness, they were handed over to a long- and short-term memory neural network for learning. The accuracy of the roughness prediction model was, finally, experimentally verified. Kwak et al. [36] used a multi-sensor approach, combined with AE sensors and power meters, to study the fault diagnosis techniques of chatter and burn during grinding. This technology has a 95% accuracy rate, and can effectively avoid workpiece burn and chatter phenomena. The abrasive concentration has a significant influence on the surface integrity of correct cutting. Buj-Corral et al. [37] studied the effects of different abrasive concentrations on the surface roughness and tool wear of an accurate cutting. They collected AE signals during the process for analysis, and demonstrated that the selection of abrasive concentrations by AE values could achieve a good machining. Arul et al. [38] conducted drilling experiments on glass-fiber-reinforced plastics (GFRP) and investigated the effect of drilling parameters on thrust and flank wear. They used an AE sensing technology to monitor the condition of the workpiece online in reducing the associated defects and improving the process stability and workpiece quality. Neugebauer et al. [39] studied the position of the drilling tool during the drilling of GFRP using AE. They identified and improved the drilling quality and tool life by changing the drilling speed and feed rate.

Oliveira et al. [40] collected AE signals to monitor the response of the workpiece in the cutting process. AE counts and energy characteristics were strongly related to the cutting path, and the amplitude was related to the cutting mechanism and damage phenomena during the cutting process. The developed monitoring system could predict the surface quality to avoid certain damages. Marinescu et al. [41] used AE to monitor anomalies in the milling process, thereby ensuring the productivity and milling surface integrity of the aerospace safety workpieces. It was found that the cutting force, AFxy, could not accurately identify the damage by machining. A combined AFxy and AEMARSE map (as shown in Fig. 4) was instead used to identify the undamaged surface and surface anomalous moments. The ability of AE was demonstrated to identify milling machining anomalies and detect surface defects on the workpieces. In addition, advanced processing techniques were used, such as the Choi-Williams distribution (CWD), Zhao-Atlas-Marks distribution (ZAMD), and a formant analysis. A novel monitoring system based on AE was developed to identify surface anomalies, such as edge damage caused by the damaged cutting-edge contact with the workpiece. This novel monitoring technique process is based on a combination of time-frequency domain analysis of AE and was validated by the inconsistency [8]. An accurate cutting depth control guarantees the dimensional accuracy of the machined workpiece, with the purpose to avoid catastrophic failures during application. Haythem Gaja et al. [42] proposed a real-time cutting depth monitoring system that was based on AE sensors and prediction models. Regression models and neural networks were used to characterize the relationship between AE signals and the cutting depth, which could effectively predict the cutting depth.

Hard machining refers to the use of CBN tools or ceramic tools to process hardened workpieces. As compared with grinding, it has advantages such as low cost, high productivity, and environmental protection. Hard machining has a large competitive advantage in the field of precision machining. However, the white layer that occurs during machining affects the surface integrity and the tool life, which limits the applications of hard machining. Guo et al. [43] studied the correlation between the root-mean-square, frequency, amplitude, and count rate of the AE signal and the white layer. A monitoring system was then developed based on the research results, which advanced the development of hard processing.

Fig. 4
figure 4

Combined AFxy vs. AEMARSE plot (a) against damage-free (b, d, f) and anomalous (c, e, g) surfaces using different inspection techniques. [41]

Full size image

Additive manufacturing is an advanced green manufacturing technology, but its process reliability and product quality limit its development. Shevchik et al. [44] conducted a feasibility study for in situ and real-time quality monitoring using AE and machine learning. It is found that the classification accuracies using SCNN are as high as 83, 85 and 89% for high, medium and poor workpiece qualities. Wu et al. [45] proposed a non-intrusive condition monitoring method for Fused deposition modeling (FDM) machines based on AE. Support vector machines are used for the state identification of FDM machines using the time domain characteristics of AE signals as indicators. It was suggested that AE technology could diagnose machine abnormalities to processing quality. To process the large amount of AE data collected by high-fidelity sensors, Liu et al. [46] used the method of dimension reduction and clustering to significantly improve the efficiency and accuracy of FDM state identification. Figure 5 shows the basic flow of proposed healthy condition monitoring method.

It has been found that many structures have damages in the microscopic range of a few tens of microns. The changes in the mechanical properties have been caused by this damage, which impact the serviceability and fatigue life of the parts. To ensure the service life of a part, it is necessary to not only control the conventional surface geometric characteristics, such as surface roughness, but also to comprehensively consider changes in physical and mechanical properties. Lloyd et al. [47] investigated the AE in Al2O3 and Al2O3–Mo fiber composites in terms of rate, total volume, amplitude, and spectrum of events by performing a three-point bending test. It was found that AE due to crack growth phenomena existed in both materials before fracturing. After the formation of microscopic cracks in the composites, the characteristics of the AE signals differed significantly from the beginning of the test. This was caused by deformed Mo fibers that were passing through the cracks. It has also been demonstrated that AE monitoring can analyze crack initiation and propagation in materials. Furthermore, surface and subsurface damage of optical glass is an essential factor limiting its performance and lifetime. Zhao et al. [48] investigated the surface integrity of a diamond-ground optical glass by using AE. They found that the smaller the amplitude and root mean square value of the AE signal, the better the surface and subsurface integrity. Still, the subsurface cracking was inevitable without an ultra-precision grinder with a very low vibration amplitude. Composites have excellent mechanical properties that are not found in single-phase materials, but their damage mechanisms are also more complex and challenging to study. Read [49] demonstrated the applicability of AE for damage detection by performing AE-based damage detection experiments. Marec et al. [50] investigated the AE-induced damage mechanisms of composites, and developed unsupervised pattern recognition analysis (i.e., fuzzy C-mean clustering) tools for monitoring AE event classifications. It uses a continuous and discrete wavelet transform technique that can effectively identify and classify the composite damage (matrix cracking, fiber-matrix debonding, etc.) and ensure the well-machined surface integrity. Microdefects sometimes form in workpieces fabricated by selective laser melting (SLM). Ito et al. [51] proposed a method to real-time monitor the generation of microdefects during SLM by AE technology. The pores and microcracks formed under high-laser-intensity conditions appear simultaneously with AE counting, with errors of a few millimeters in source localization. According to his study, the AE technique is very suitable for detecting and localizing the generation of microcracks in the additive manufacturing process.

Fig. 5
figure 5

Basic flow of proposed healthy condition monitoring [46]

Full size image

2.2 Chip Formation

Some complex materials cannot be made to correlate the AE directly with the surface integrity, so their chip formation mechanism (i.e., surface formation mechanism) had to be investigated by AE monitoring. Many parameters in material machining (such as cutting force, temperature, tool wear, frictional contact between tool surface and chip, and cutting power) are related to the chip formation process. In machining, the chip formation is influenced by many factors, such as tool, workpiece, cutting dosage, and cutting media. An in-depth study of the chip formation law under different cutting conditions has an important theoretical significance and practical value for the realization of green manufacturing.

Fig. 6
figure 6

Relationship between specific AE energy and maximum undeformed chip thickness a for a 1 mm and b 3 mm width of cut [53]

Full size image

Chiou et al. [17] investigated the relationship between AERMS and the chip thickness. They found that the AERMS responds to dynamic changes in the chip thickness during cutting, whereby the RMS AE dynamics of the tool/workpiece geometry were derived. In micromachining, the determination of the minimum chip thickness, and avoidance of machining features below this threshold, is the hinge to ensure a dimensional machining accuracy. Mian et al. [52] proposed a method to determine the minimum chip thickness value by analyzing the AE signals that were generated in orthogonal micro-milling experiments. It was suggested that the increase in the AE signal might be related to the material extrusion mechanism. Mian et al. [53] analyzed the AE signals of cutting materials that were difficult to machine by a wavelet transform. It was found that the maximum undeformed chip thickness was positively correlated with the specific AE energy(as shown in Fig. 6), which was important for solving the size effect of micromachining.

The monitoring of minimum chip thicknesses is not enough for materials that are difficult to machine. Instead, a thorough understanding of the chip formation mechanism is required. Prakash et al. [54] used AE signals to investigate the effect of tool wear on the surface roughness (Ra), chip formation mechanism, and chip morphology during micromachining of an aluminum alloy (AA 1100) material. The AE RMS was found to be sensitive to the formation of stacked edges. The AE signal energy was decomposed into higher and lower orders using a discrete wavelet transform. The higher order of the AE-specific energy was related to the chip formation mechanism (e.g., shear and microfracture). Also, the relationship between AE and chip morphology was investigated using Fourier transformation. A low frequency and low AE amplitude led to tightly curled chips, while a high frequency and high AE amplitude led to elemental chips. The present study provides new and vital insights into tool wear, chip formation mechanisms, and chip morphology in the micromachining of aluminum alloys.

Barry et al. [55] performed orthogonal cutting tests in studying the chip formation mechanism of the Ti-6 Al-4 V alloy and to evaluate its effect on the AE. The study resulted in a variable pattern of the chip morphology, which was based on machining parameters. In addition, the chip formation mechanism of the Ti-6 Al-4 V alloy was obtained from an analysis at a low cutting speed of 0.25 m/s. The damage in the upper region of the main shear zone occurred by cracking. At a high cutting speed, the damage occurred by ductile fracturing (void formation, growth, and coalescence), which proved the existence of thermal softening in the main shear zone. Another characteristic of machining the Ti-6 Al-4 V alloy, using P-type carbide tools, was the occurrence of a bonding between the chip and the tool. The degree of bonding increased with the cutting speed. At a cutting speed higher than 0.5 m/s, the fracture of this joint appeared to be the main source of AE. The mechanism of chip formation in hard machining was also studied by Barry et al. [56]. They obtained a two order of magnitude higher AE signal for hard machining than for normal machining, which is a result that was caused by the change in hardness and cutting parameters of the workpiece material. The periodic rapid release of elastic strain energy occurred during the transition from a continuous chip formation to an irregular chip formation, especially the catastrophic damage in the main shear zone.

Pawade et al. [57] evaluated the dependence of machining deformation on the chip generation mechanism in terms of energy, number of counts, and average frequency amplitude of the AE signal. It was found that the morphology of chips generated during high-speed turning of Inconel 718 could be correlated with the AE waveform. Highly burned chips with significant cracking were produced that exhibited a continuous AE signal. However, silvery uniformly sized chips with some evidence of wear showed abrupt AE signals, and the corresponding machined surfaces showed lower roughness values.

Hase et al. [58] investigated the correlation between the cutting phenomena and AE in turning and found that the chip formation process, chip type, and shear angle significantly affected the AE signal. A strong negative correlation was found between the shear angle and the level of the AE signal. It was shown that the amplitude of the AE signal can be used to identify the chip formation process during turning, and, thus, to identify and predict the degree of tool wear and the quality of the machined surface.

2.3 Tool wear

Tool wear affects the quality of machining and productivity, so tool wear monitoring is essential for improving the resource utilization. AE tool wear monitoring has the following implications;

  • Accurate tool life prediction: Tool wear directly affects machining quality and productivity. With the tool wear monitoring technology, the tool wear can be monitored in real-time. In addition, the tool life can be accurately predicted, and tool replacement can be carried out to avoid any production downtime or waste caused by tool life being too short, or too long.

  • Improve machining accuracy: The tool wear can decrease the machining accuracy and affect the product quality. With the tool wear monitoring technology, the tool wear can be detected in time for tool replacement or repair, thus improving the machining accuracy and ensuring the product quality.

  • Reduce production costs: The tool is an important consumable during machining, and the cost of a tool replacement and repair is high. With help of the tool wear monitoring technology, the tool wear can be detected in time. The time and number of tool replacements and repairs can be reasonably arranged to avoid an excessive replacement and waste, thus reducing the production costs.

  • Improve productivity: The tool wear monitoring technology enables a real-time monitoring and evaluation of the machining process, thus avoiding machining stops and adjustments caused by tool wear and improved productivity.

Kuntoğlu et al. [59] found that when machining AISI 1050, AE sensors captured the signal of tool breakage so that appropriate machining parameters could be selected based on the information provided by the AE. Also, the occurrence of tool breakage failures could be identified in advance for the machining process. Maia et al. [60] found a correlation between the AE signal spectrum and the tool wear. The tool wear led to an increase in the average amplitude of the AE power spectrum density. A mechanism for detecting wear based on AE signals was proposed, and so was a method for determining the end of tool life. The accuracy of the method was verified by turning hardened AISI 4340 steel.

Gómez et al. [61] studied the torque and AE parameters in drilling. It was shown that the Mean Power (MP) moving average (MA) correlated better than the traditional AE RMS. It was found that MP and MA for different wear states clustered around different centers of mass. This method could effectively circumvent the randomness of AE and facilitate monitoring tool conditions to obtain a well machined surface quality during repetitive manufacturing processes. Heinemann et al. [62] used the average and variance of the spindle power and AE RMS to successfully predict the final stage of the fine hole drilling tool life. Tool life utilization reached 84%.

Fig. 7
figure 7

The setup for separating the chip formation occurrences a schematic view, b real view [65]

Full size image

Cao et al. [63] proposed a new method based on the lifting scheme and Mahalanobis distance (MD) for monitoring AE at different periods of tool wear of the end milling cutter. A lifting scheme meant the construction of a double orthogonal wavelet with shock characteristics, and to perform a wavelet transformation in separating the components from the original signal of AE. The signal envelope of the wavelet coefficients was then demodulated using Hilbert transform, and the salient features representing the real-time state of the tool (i.e., normal state, light wear, and heavy wear) were extracted. Finally, the pattern recognition of the tool wear state was performed by MD. The effectiveness of this method for monitoring the tool wear state during end milling has been experimentally demonstrated. Olufayo et al. [64] studied the tool wear during the milling of H13 tool steel. They found an AE RMS and AE average that were highly correlated with the tool life when using the wavelet transforms, statistical analysis, and neural networks.

As described in the previous subsection, the AE could monitor the chip formation. The adhesion of chips affected the tool wear and life in some machining. Bhuiyan et al. [65] developed an original tool holder device (Fig. 7) that facilitated the separation of chip separation events from the frequency of other events, so that the extracted signal could be used directly for tool wear status monitoring. The response of the tool condition to the AE raw signal, RMS, and other characteristics were obtained. The chip formation was related to the cutting parameters, and the tool wear rate and plastic deformation rate increased with an increasing cutting speed, feed rate, and depth of cut until chip breakage. Furthermore, the tool wear decreased with an increasing chip breakage rate. Bhuiyan et al. [66] performed experiments on carbide tools grinding ASSAB 705 steel. The tool wear frequency (67 − 471 kHz) was decomposed from the total frequency and it was distinguished from the plastic deformation of the workpiece. This study has been useful for the analysis and prediction of the tool wear condition. Neslusan et al. [67] proposed a method to detect a catastrophic tool failure during hard turning of 100cr6 bearing steel, which was based on multiple AE sensor signals. A low-frequency AE sensor was then responsible for receiving signals from shear zone cracks, and a high-frequency AE sensor was responsible for identifying the transformation process. It was found that the direct AE raw signal was less effective in identifying the tool wear condition, while using R1, and R2 could effectively identify the tool wear condition. Also, the use of R3 could predict the arrival of the tool CTF [67].

$$R1 = {{AE_{{rms(D9241)}} } \mathord{\left/ {\vphantom {{AE_{{rms(D9241)}} } {AE_{{rms}} \left( {WD} \right)}}} \right. \kern-\nulldelimiterspace} {AE_{{rms}} \left( {WD} \right)}}$$
(1)
$$R2 = \frac{{AE_{{absolutee\,energy(D9241)}} }}{{AE_{{absolutee\,energy(WD)}} }}$$
(2)
$$R3 = {{AE_{{strength(D9241)}} } \mathord{\left/ {\vphantom {{AE_{{strength(D9241)}} } {AE_{{strength(WD)}} }}} \right. \kern-\nulldelimiterspace} {AE_{{strength(WD)}} }}$$
(3)

Chethan et al. [68] studied the mechanism of tool wear during Nimonic-75 turning. The wear area, perimeter, AE RMS, and AE COUNT that were obtained from orthogonal experiments correlated with the tool wear. The machining parameters were optimized using the Taguchi technique and it was found that the tool wear area and circumference were reduced at 450 rpm, 0.07 mm/min, and 0.2 mm. The machining results were better than the initial machining conditions. By combining data from the AE sensors with photos taken by the machine vision, the machining parameters were optimized to reduce the effect of tool wear and to achieve a better surface quality.

The metal matrix composites had a high specific strength to stiffness and were very suitable for aerospace applications. The problem with a tool wear in metal matrix composite machining is a constraint to its applications. Mukhopadhyay et al. [69] used an AE technique to monitor tool wear during the turning of silicon carbide (20 wt%) aluminum matrix composites. The AE values increased abruptly after a certain distance of tool cutting. Correlation analysis of the multiple eigenvalues of the AE signal showed that the skewness and kurtosis varied most significantly with cutting time. This result was attributed to different wear mechanisms in the metal matrix composites. The combined use of an AE amplitude, skewness, and kurtosis effectively predicted a tool wear in aluminum-based silicon carbide.

2.4 Feasibility Analysis of AE Technology

Compared to optical, electromagnetic and other measuring instruments, acoustic instruments offer the advantage of continuous monitoring without the need for downtime for inspection, thus minimizing energy and resource losses. AE technology is an environmentally friendly, non-destructive testing technique that has no adverse effects on operators or the environment [7] AE sensors offer a cost-effective solution with affordable pricing and minimal installation, removal, and maintenance costs [70, 71]. In the quest to monitor machining quality and tool wear, the implementation of AE sensors proves to be significantly more economical, costing only one-third to one-tenth of other sensor options, such as force sensors. This significant cost advantage strengthens the feasibility and attractiveness of using AE sensors for effective monitoring.

3 Saving Energy

In the manufacturing systems, the energy consumption of machine tools is relatively small compared to the proportion of the thermal industrial fields. With the gradual appearance of the decreasing energy and related environmental problems, researchers began to re-evaluate the energy efficiency of machining. It was found that the energy efficiency of the machining process was generally low [72], and the total energy consumption was huge. The machine tool was the main component of the machining system, the carrier to accomplish the machining tasks, and also the primary cause of the main energy consumption of the machining system. How to improve the energy efficiency of machine tools has become an urgent problem in the manufacturing industry. There are two major challenges in solving the machine tool energy consumption: one is the difficulty in modeling the machine tool energy consumption models, and the other is the difficulties with online monitoring of the machine tool energy consumption. Several studies have presented a correlation between the AE phenomena and energy dissipation [73, 74]. This chapter provides an overview of the use of AE techniques to improve the energy efficiency in different processing areas.

Fig. 8
figure 8

Spectrum analysis of acoustic emission signals [78]

Full size image

3.1 Cutting Process

The energy efficiency is affected by a variety of factors (such as cutting conditions, tool wear, and machine conditions), which makes it difficult to model and monitor the milling energy consumption based on traditional signal processing strategies [75]. Cai et al. [9] established the experience system of energy efficiency state identification for milling processes based on multi class sensors, such as AE sensors. Time-domain and frequency-domain analysis were then, conducted on the collected signals, and the relationship between the signals and energy efficiency state characteristics were obtained. Based on the energy efficiency state mechanism and experimental results, a milling energy efficiency evaluation method was proposed, which was important for energy saving.

Chatter is a self-excited vibration caused by the interaction between the dynamic characteristics of the cutting process and the modal characteristics of the machine tool-tool-workpiece system. Chatter reduces the number of cuts and the surface quality of the workpiece, and generates a lot of noise and can even lead to premature tool scrap [76]. The process to reduce the cutting chatter and improve the energy efficiency was, therefore, optimized. Kuljanic et al. [77] used a multi-sensor fusion approach of accelerometers, rotational dynamometers, and AE sensors for machine tool chatter identification. The time and frequency domain features of the signal were compressed, and a statistical approach was, thereafter, used to obtain a chattering identification system. The accuracy of this system was verified to be higher than that of conventional monitoring methods. Li et al. [78] analyzed the chattering phenomenon during high-speed robotic milling of aluminum alloys (cutting speed of 678 m/min) using AE RMS and fast Fourier transform methods. The occurrence of robot milling chatter were found to lead to variations in energy waves, surface quality, and AE signal amplitudes. The maximum value of the RMS value in the time domain was reduced when the milling state changed from chatter to stable. As shown in Fig. 8, the maximum amplitude (> 0.0160 V) of the groups A5, A8, A9, A11, and A12 were higher than those of the groups A6, A7, and A10. The results showed that the fast Fourier transform and the RMS value of the time domain of the AE signal, could be effectively used to detect and verify the chatter in the robot milling process. This helped to select reasonable cutting parameters to avoid chatter in the high-speed robot milling process and improve the efficiency.

During the turning process, the less number of faults, the higher the energy efficiency. Chiou et al. [79] investigated the AE signal caused by the chatter in the tool wear state, which was based on the energy dissipation principle. A dynamic model between the tool side wear and the effective value of AE was established by comparing the cutting energy of the sharp cutting edges (generated by chatter during cutting) with the energy dissipated due to tool edge wear. This dynamic model could successfully explain the chatter phenomenon near the tool wear frequency. Based on this model, a tool wear and chatter process monitoring system was developed, which improved the process parameter constraint problem while increasing the cutting efficiency and saving energy. Filippov et al. [80] used AE techniques for the steady and chatter mode turning to reveal the AE response of the workpiece chatter. The experimentally obtained amplitude, average frequency, and power spectrum of the AE signals were compared with the results of molecular dynamic (MD) simulations. It was concluded that the high amplitude vibrations increased the amplitude of the AE signal by changing the structure of the AE signal (i.e., increasing the wave-like signal). The spectral characteristics were altered such that the median frequency decreased in parallel with the vibration, and the power spectrum was shifted toward the low-frequency range. Based on his study, a power spectrum-based cutting stability monitoring system was finally developed [81].

Ribeiro et al. [82] investigated the relationship between process conditions such as feed rate, cutting depth, cooling system, and the type of tool and AE response in microturning that is based on the AE technique. By using these research results, it could improve the machining process of microturning, improve the surface roughness and microhardness of the workpiece, and increase the machining energy efficiency.

3.2 Abrasive Machining

Abrasive machining is a highly energy-intensive process, and the energy efficiency of abrasive machining must be addressed. Studies have shown [83] that the correct choice of a proper process strategy for abrasive machining can significantly reduce the energy consumption.

Opoz et al. [84]collected AE signals from scratch tests. The experimental results demonstrated that material removal, plowing, and cutting produced AE signals with different power intensities, amplitudes, and frequencies. By monitoring these characteristics and identifying the material removal phase of the grinding process, it could help to improve the process performance and energy efficiency. Pilov [85] used the AE technique to establish a correlation between the energy of the AE signal and the size of the ground particle distribution. By detecting the AE amplitude and frequency, the search for a maximum grinding efficiency could be made. It was demonstrated that the AE signal characteristics could be used to predict the grinding efficiency. This study contributed to the improvement of the grinding process methods and energy efficiency.

As an important machining parameter, the grinding time is closely related to the machining quality and efficiency. How to optimize and select the machining time at each stage of the grinding process is also of a large importance for the achievement of energy saving and efficiency in grinding. Jiang et al. [86] proposed a mathematical model of the AE signal during the grinding cycle. In addition, Jiang et al. [87] established an online minimum dwell time estimation method, which was based on a material removal model and a surface roundness model using AE technique measurements and analyses. This method helped to control the grinding process precisely, optimized the process parameters, and improved the energy efficiency.

A specific grinding energy refers to the energy consumed per unit time when removing a unit volume of the workpiece material, which directly reflects the energy dissipation of the grinding process. Adibi et al. [88] investigated the variation of the specific grinding energy and AE for different grinding parameters. The relationships between grinding parameters and specific grinding energy, AE signal characteristics, and surface roughness were obtained from the grinding of a nickel-based super alloy with CBN grinding wheels. It was demonstrated that the AE RMS can predict the grinding specific grinding energy and surface roughness online, which helped to improve the grinding parameters and reduce the energy loss. Denkena et al. [89] found a correlation between AE and the grinding power. The work performed during grinding was mainly converted into thermal energy. AE was used as an indicator of the thermal changes in the workpiece, which integrated the energy consumption with the machining quality and helped to adjust the process parameters in optimizing the grinding process, reducing the energy consumption, and improving the energy efficiency.

Gradišek et al. [90, 91] proposed an automatic chatter detection method using the entropy and the coarse-grained information rate (CIR) as indicators. The entropy was calculated from the power spectrum, and the CIR was obtained from the fluctuation of the signal. The grinding force and RMS of the AE could be used to monitor the chattering phenomenon that occurred during grinding. The threshold values of the entropy and CIR had been experimentally obtained to achieve an accurate monitoring of the vibration state of the machine tool (i.e., for grinding).

The dissipation of processing energy was difficult to accurately quantify, which was due to the influence of various variables such as cutting parameters, workpiece materials, tool parameters, process conditions, and the external environment. A study of the mechanism of energy dissipation helped to quantitatively analyze the energy consumption. Mohan et al. [92] monitored the hydro abrasive machining (HAM) process by AE. Any change in parameters that tended to increase the energy dissipation during machining was found that result in an upward shift of the PSD curve. The AE signal energy, given by the area enclosed by the PSD curve, had a square root relationship with the dissipated energy. A physical model(Eq. 4)was proposed to quantify the energy dissipation of the HAM such as plastic deformation, material fracture, wall friction, cavitation, turbulent vortex, and damping.The parameters in Eq. (4) are either known or could be measured,and that the variables,α, φ, and pT depend on other process parameters.

$$E_{{DISS}} = \frac{{\alpha ^{2} \varphi ^{2} d_{F} (\dot{m}_{p} + \dot{m}_{W} )}}{{v\rho w(1 + \frac{{\dot{m}_{p} }}{{m_{W} }})^{2} }}\left[ {p - p_{T} } \right]$$
(4)

In abrasive water jet machining of titanium, ceramic, and fiber-reinforced materials, the nozzle condition directly affects the machining quality and efficiency. Kim et al. [93] developed a nozzle monitoring system that was based on the AE. The AE RMS values, which were due to the nozzle condition changes, were calculated by experimentally measuring the nozzle condition at different water injection pressures and abrasive feed rates. It was found that the AE RMS values were proportional to the material removal rate (MRR) and the material removal energy was proportional to the AE energy. Therefore, the processing parameters could be optimally selected by the AE technique to avoid any nozzle failure and improve the energy efficiency.

4 Solving Ecological Pollutions and Hazards

To increase the productivity and reduce the costs, higher cutting speeds and feeds are required. This can lead to large amounts of heat being generated during machining and affect the dimensional accuracy and tool condition of the workpiece. As a result, the cutting fluids that have been used as an adjunct to the machining process, have greatly increased the productivity [94]. Although the emergence of cutting fluids could effectively reduce the workpiece surface roughness and finish and improve the tool life, the large amount of discharge of mineral-based cutting fluids caused serious environmental pollution and damaged the ecology while endangering the human health. Some studies have shown that the contact between the operators and the cutting fluids can increase the pathogenicity of lung cancer, respiratory diseases, skin diseases, and genetic diseases [95]. In addition, the cutting fluids are expensive and not easily degradable, causing high economic costs for their purchase and disposal [96]. The emergence of degradable cutting fluids has alleviated this situation [97], but to fundamentally address the pollution they cause, a green cutting technology must be vigorously developed. Green cutting means using as little cutting fluid as possible to minimize the harm of cutting fluid to people and environment, while ensuring the product quality. The AE monitoring technology has the unique advantage of revealing the processing mechanism of green cutting. After understanding its mechanism, the production process can adaptively change the supply of cutting fluid according to the change in cutting state to meet the production requirements and reduce the amount of cutting fluid used, thus realizing the concept of green manufacturing.

4.1 Dry Cutting

Dry cutting is a machining method that does not use any cutting fluid in the machining process. Due to the absence of a cutting fluid, dry cutting can eliminate the environmental and human hazards that are generally caused by a cutting fluid, but the conditions to be met by this machining method are also quite demanding. It should maintain a high productivity, ensure the qualified product quality and long tool life, and ensure the reliability of the cutting process [98]. The dry cutting causes a rapid tool wear when machining highly adhesive materials such as metal matrix composites. Much research has, therefore, been made to improve the tool wear in dry cutting.

The absence of a cutting fluid in dry cutting makes the machining process more complex and variable. The risk of tool wear, chipping, and excessive temperature increases significantly. Therefore, researchers have paid more and more attention to the condition monitoring and analysis of the machining process in the field of dry cutting. During the tool wear process, the tool’s back surface’s friction and chip impact, fracture, and plastic deformation in the shear zone will generate stress waves while releasing part of the elastic energy. These signals are transmitted to the cutting area as elastic waves, finally detected by the AE sensor [99]. The interference of low-frequency machining noise can be effectively avoided due to the wide sampling band of the AE sensor. Wang et al. [100] used a signal acquisition system based on the Labview software to collect vibration (i.e., acceleration) and AE signals during dry milling experiments, and the results showed that the AE signals were more sensitive to changes in dry milling machining parameters (e.g., machining speed). Liao et al. [101] investigated the effect of milling parameters on aluminum dry milling using an AE monitoring technique. They found that the AE signal amplitude increased with an increasing rotational speed, and increased with an increasing cutting depth. When the spindle speed was constant, the increasing feed rate did not affect the AE signal’s amplitude variation. Zhong et al. [102] found that dry cutting still had some defects and could only be applied in the roughing field by multi-signal fusion. A small quantity of lubrication, or even a minimal quantity of lubrication, should be used in the finishing field.

Fig. 9
figure 9

Experimental set-up and AE sensors location [106]

Full size image

4.2 Minimal Quantity Lubrication Cutting

Dry cutting completely abandons a cutting fluid, chip removal, cooling, lubrication, and other functions, which leads to tool wear and speed up the reduction in machining efficiency. These problems limit the application range of dry cutting, so the use of cutting fluids is not abandoned. A Minimal Quantity Lubrication (MQL), which limits the use of cutting fluids, has been studied more in detail, and its application is more extensive than that of dry cutting (especially for grinding [103]). This technology mixes a compressed air with a very small amount of lubricant and vaporizes it to form millimeter- and micrometer-sized vapor mists, which are sprayed into the cutting area to cool and lubricate the contact surfaces between the tool and chip, or between the tool and workpiece. As a typical green cooling lubrication method, the micro-lubrication technology has the advantages of a small amount of cutting fluid, low cutting force, anti-sticking, long tool life, and improvement of the surface quality of the workpiece [104]. As discussed above, the AE signal can make a good pattern recognition for tool wear and chip formation. Therefore, the processing conditions caused by different lubrication and cooling conditions have a strong correlation with the value of AE [105]. A large number of researchers have carried out MQL machining research using AE. Wang et al. [106] used the device in Fig. 9 to acquire the AE signals in the experiment. The noise was at first eliminated by setting appropriate AE parameters such as threshold, rise time, and duration. The relationship between tool flange wear and three types of AE burst signals, as caused by MQL, material fracture, and plastic deformation, was, thereafter, established by clustering the AE energy.

Fig. 10
figure 10

Processing of AE data for calculation of power consumed [116]

Full size image

The use of a wheel cleaning jet (WCJ) has triggered the use of the MQL technique instead of a traditional coolant cutting [107]. Javaroni et al. [108] measured the AE signal (RMS.) under different cooling-lubrication conditions. The AE results improved as the flow rate of the MQL increased. The AE signal that was obtained with the MQL method was the worst. In contrast, the MQL + WCJ method effectively improved the AE condition. The reason was that the grinding with MQL accelerated the wheel clogging adherence to the wheel bore, with chips scratching the workpiece surface and an increasing friction and plowing. At the same time, the addition of WCJ effectively reduced the occurrence of wheel clogging. Barros et al. [109] found that MQL grinding with an aluminum oxide rod, or a Teflon block, produced significantly lower AE values than for MQL grinding without cleaning. This demonstrates that an aluminum oxide rod, or a Teflon block, can effectively reduce the chip adhesion, friction, and noise. Rodriguez et al. [110] concluded that an excessive lubrication with PMQL + WCJ (i.e., pure oil) techniques does not reduce adhesion but affects the friction on the grinding wheel surface, thereby producing higher AEs. The HMQL + WCJ at the highest flow rate of 120 ml/h produces approximately the same AEs as the conventional cutting fluid machining method.

The composition, concentration, and additives of the MQL have a significant effect on the machining quality and tool wear. By comparing the AE of MQL grinding wheels with different water-oil ratios, Moretti et al. [111] learned that the MQL solution was added with the right amount of water (1:5). With the help of compressed air, this water reduced the ductility of the material during the machining process. Due to its lower viscosity, it penetrated more easily into the cutting zone. The energy consumption was then reduced by the adhesion and friction between the plug and the workpiece surface during the cutting process. With a significant increase in water in the MQL solution (1:10), its lubricating capacity was compromised, leading to an increased friction and plowing and higher AEs [112]. Thus, a diluted MQL could effectively compensate for the shortcomings of pure MQL. Machining efficiencies were obtained that were closer to those of conventional cutting, and some sustainability and savings at the machining level were also achieved. Garcia et al. [113] have investigated the grinding performance of MQL-bearing steels at different dilution levels.

Talon et al. [114] proposed a biodegradable inhibitor, a V-active VCI fluid, which was used for MQL machining. At lower feed rates, the V-active VCI fluid performed significantly better than the base fluid, for which the process severity increased. Since AE decreased with improved process lubrication, the V-active VCI fluid proved to have excellent properties for improved lubrication. Talon et al. [115] compared the magnitude of AE of different media VCI cutting fluids, and selected the VCI cutting fluid with the best-performing media. Machai et al. [116] evaluated the effect of CO2 snow as an innovative cooling method for machining titanium alloys, using quantifying AE, cutting forces, etc., at three different temperatures. The powers consumed by the processing AE data was then estimated. Figure 10 illustrates the above method, integrating the amplitude of the AE (in the left panel) over all recorded frequencies for a given time interval. The result of the integration is the process power, as shown in the two-dimensional curve to the right. By using this method, the pre-cooling of the work material by the CO2 snow was concluded to lead to an increase in strength, which was gradually reduced to normal levels by the heat of the cutting process. It helped to eliminate the health hazards and recycling costs associated with the use of coolants.

In the work by Iqbal et al. [117], the AE signals that were emitted during cutting of β-titanium alloys under different cooling and lubrication conditions, were processed and calculated to evaluate the total cutting energy consumed in a given time period or for a given cutting length. They found a strong upward relationship between the tool damage and tool displacement area, indicating that the application of CO2 snow at different positions of the tool is a better choice than conventional water cooling in terms of tool damage, cutting energy, and dynamic process stability. Lopes et al. [118] found that the use of non-friable grinding wheels does not frequently produce cutting edges during machining and tends to wear by microcracking in the abrasive grains. This increases the contact area between the abrasive grains and the workpiece surface. Thus, the compressive stresses will also increase, which increases the friction and produces higher AEs. Regarding optimum performances, the most brittle wheels showed 13.8%, 11.5%, and 13.6% lower AE values compared to brittle wheels when using water immersion, pure MQL, and diluted MQL methods, respectively. Javaroni et al. [119] found that heat dissipation in the cutting zone during machining is an important factor that affects the AE values. Since it is difficult for the MQL to dissipate heat from the cutting zone, the heat removal is mainly achieved by heat transfer from the grinding wheel. By considering that the feed rate determines the equivalent cutting thickness, the highest equivalent cutting thickness has the highest cutting feed rate. Therefore, a larger wheel-workpiece contact area will result in a larger heat transfer. Sato et al. [120] compared CBN and alumina wheels for MQL applications, where the lower AE values of CBN were reflected at the macro level by a relative reduction in the roughness of 29% and roundness error of 9%, while the cost of alumina was reduced by 72%. The choice of a grinding wheel depends on the trade-off between the workpiece accuracy and machining cost.

Fig. 11
figure 11

Diagram of the proposed tool wear monitoring strategy [121]

Full size image

In addition to grinding, the MQL has also been applied in various processes, such as milling and drilling, to improve the cutting fluid contamination. Hu et al. [121] proposed a strategy (Fig. 11) that used AE to monitor the tool wear during titanium alloy Ti-6 Al-4 V milling under MQL conditions. The time and frequency domain features extracted from signals such as AE were used for model training and prediction. By using ν-SVM, a model to predict the tool wear during MQL milling was established. After dimensionality reduction by using linear discriminant analysis(LDA), the model had a higher prediction accuracy than the model using original features. By using this model, an effective prediction of the tool wear status was finally achieved. Heinemann et al. [122] used AE signals to evaluate the life of a drill under MQL conditions. They found that the amplitude of the AE signal increased significantly in the last third of the drill life. This was because tool wear, increased friction, and poor lubrication lead to an increasingly difficult chip evacuation.

5 Conclusion and Outlook

As a common non-destructive monitoring technology, the AE technology is conducive to the progress of green manufacturing by making full use of its low cost and high sensitivity. This paper has reviewed and evaluated the application of the AE technology in various fields of green manufacturing.

  1. (1)

    The AE technology has become quite mature in improving the material utilization. Based on the sensitivity of the AE signals to product failures and the high correlation between AE features and product quality, the AE monitoring system could effectively predict and identify common machining failures (such as poor surface quality and tool wear), reduce the number of product scraps, and improve material utilization.

  2. (2)

    There have been some researchers that used the AE technology to study the energy dissipation mechanism and online monitoring energy consumption faults. It was found that the AE and energy dissipation had an intrinsic link. By using the AE technology, the process methods were optimized and the processing anomalies were reduced to improve the energy efficiency. The AE technology has broad prospects and an important significance in solving the machine tool energy consumption.

  3. (3)

    AE contains information about many material cutting processes, and its study can reveal subtle differences between different cutting processes. Compared with dry friction, the cutting fluid lubrication reduced the static and dynamic components of friction in the second and third deformation zones, which resulted in corresponding changes in the amplitude and frequency components of the AE signals. Therefore, the study of the green cutting with the AE technology, to make the machining effect close to conventional machining with a minimum environmental pollution, is very suitable. On one hand, the experimental study of dry cutting and MQL using the AE technology fills a gap in the theory of green cutting. On the other hand, it provides a guidance and practical significance for applying dry cutting and MQL cutting in production.

AE technology brings cost and sustainability benefits to the machining process. It uses low-cost sensors to monitor machine tool conditions, detect tool wear and prevent costly machine downtime in real time. In addition, its non-destructive nature reduces environmental impact, optimizes the machining process, and supports sustainable manufacturing practices and predictive maintenance strategies.

There are limitations to the application of AE in the field of green manufacturing. The future research and development directions toward further improvements, are as follows:

  1. (1)

    When extending the AE technology to other green manufacturing fields, a re-correlation analysis was needed which pose both time and economic challenges. An in-depth study on the physical interpretation of the AE technology was used to solve the problem of locating the source of AE signals in processing complex materials.

  2. (2)

    In the study of AE feature correlations, optimized intelligent algorithms have been used, such as a deep learning, to improve the accuracy of AE fault identifications.

  3. (3)

    The AE signal is affected by the medium attenuation, and it must be used in the process of its reasonable design of the location of the arrangement. Modeling research on the location and number of sensors in the AE monitoring process has been performed to achieve the best monitoring state.

  4. (4)

    Green manufacturing that is based on AE needs to analyze and integrate various materials, equipment, components, etc., in the application process. Based on the careful consideration of each processing link, one should select the most suitable solution.

  5. (5)

    The AE signal database has been established to realize data interaction and sharing in various green manufacturing fields so that the use of AE technology in green manufacturing can be more integrated and systematic.