A Review on Recent Advances in the Energy Efficiency of Machining Processes for Sustainability

Abstract

The pursuit of energy efficiency in machining processes is a critical aspect of sustainable manufacturing. A significant portion of global energy consumption is by the industrial sector; thus, improving the energy efficiency of machining processes can lead to substantial environmental and economic benefits. The present study reviews the recent advancement made for improving the energy efficiency of machining processes. First the energy consumption of the machining processes was explored and then the key areas and developments in their energy consumption modeling were identified. Following this, the review explores various strategies for achieving energy savings in machining. These strategies include energy-efficient machine tools, the accurate modeling of the energy consumption of machining processes, the implementation of optimization techniques and the application of artificial intelligence (AI). Additionally, the review highlights the potential of AI in further reducing energy consumption within machining operations and achieving energy efficiency. A review of these energy-saving strategies in machining processes reveals impressive potential for significant reductions in energy consumption: energy-efficient design can achieve up to a 45% reduction, optimizing cutting parameters may minimize consumption by around 40%, optimizing tool paths can reduce consumption by approximately 50%, optimizing non-cutting energy consumption and sequencing can lead to savings of about 30% and employing AI shows promising energy efficiency improvements of around 20%. Overall, the present review offers valuable insights into recent advancements in making machining processes more energy-efficient. It identifies key areas where significant energy savings can be achieved.

1. Introduction

The rapid increase in the global population has indeed put immense pressure on the planet’s resources, leading to critical challenges in ensuring sustainable food, water, and energy supplies. The reliance on fossil fuels has exacerbated environmental issues, primarily due to the significant carbon emissions contributing to climate change. Even with the increased use of renewable energy, fossil fuels are still projected to be a significant source of primary energy in 2050 [1]. Fossil fuels (coal, oil and natural gas), the engine of rapid globalization and industrialization, are nearing depletion. Environmental concerns over carbon emissions are now pushing the world to focus on improving energy efficiency and developing alternative energy sources [2]. The use of fossil fuels is expected to rise to 35.4 Gigatons in 2035 [3]. To prevent global warming and its negative consequences, achieving net-zero greenhouse gas (GHG) emissions by 2050 is a critical goal for many countries and organizations. The Intergovernmental Panel on Climate Change (IPCC) has repeatedly cautioned about the severe effects of rising temperatures on ecosystems [4,5,6,7]. According to the US Energy Information Administration, global energy demand is expected to double by 2050 compared to that in 2020 [1], with the industrial sector being a major energy consumer worldwide [8]. According to the U.S. Environmental Protection Agency, the industrial sector accounts for 29% electricity-related carbon dioxide emissions (one of the primary greenhouse gases) [9]. A significant reduction in energy consumption is a must for industrial sector across the globe for reducing environmental burdens and ecological benefits.

The rational utilization of energy and its sources with improving energy efficiency is a crucial step in reducing our reliance on fossil fuels and achieving net-zero emissions. Manufacturing is the back-bone of the industrial sector and thereby the growth of the economy, but it also accounts for more than 30% of global total energy consumption [10] and 36% of GHG emissions [11]. Generally, a significant amount of energy is consumed in manufacturing while converting raw material into products, and it also releases a large amount of waste. Machining, a crucial operation of many manufacturing processes, presents a significant opportunity for reducing energy consumption and achieving sustainable manufacturing practices. Machining processes consume significant energy, often with low efficiency. Sometimes the non-cutting operations in modern machine tools consume the majority of the energy and the actual machining contributes only around 15% to the total energy used [12]. Energy-efficient machining requires research on three fronts: designing for energy efficiency, analyzing the power usage of individual components and developing software tools that optimize energy usage during operation [13]. Renewable energy is an important solution, but economic challenges and limited global capacity limit its implications, so improving energy efficiency remains an important strategy. It will always be valuable to reduce supportive consumption and user-end efficiency. This focus has given rise to important research attempts made by various researchers across the globe.

2. Scope and Objectives of the Review

Recognizing the high energy consumption and environmental impact of machining, researchers and manufacturers have long focused on improving energy efficiency in these processes. This focus is a key to achieving sustainable manufacturing practices. Driven by the need for sustainability, this review explores the topic of improving energy efficiency in machining processes. It begins by examining the current state of global industrial energy consumption, specifically focusing on the manufacturing sector. To understand the specific energy usage within machining processes, the review investigates the factors contributing to energy consumption during the machining process. Following this, the review explores various strategies for achieving energy savings in machining. Figure 1 highlights the various strategies focused on in the present literature on the energy efficiency of machining processes for sustainability.

These strategies include the design of energy-efficient machine tools, the accurate modeling of the energy consumption of machining processes and the implementation of optimization techniques. Additionally, the review highlights the potential of artificial intelligence in further reducing energy consumption within machining operations. Overall, the present review offers valuable insights into recent advancements in making machining processes more energy-efficient. It identifies key areas where significant energy savings can be achieved.

3. Understanding Energy Usage in Machining Processes and Modeling

Manufacturing metal parts such as shafts and turbine blades poses a substantial energy challenge. Machining methods discard as much as 97% of the raw material during shaping processes that can span days or even weeks, all while consuming considerable energy. However, despite these efforts, the overall energy efficiency of such processes typically hovers below 30%. This highlights the urgent need for innovation in energy-efficient technologies within the metal machining manufacturing processes [14].

Some investigations into energy consumption in machining processes aimed to develop predictive models for the energy used at the tool tip–work interface. This refers to the energy necessary for material removal from the workpiece in the form of chips [15]. As shown in Figure 2, Wang et al. [16] categorized the cutting energy at the tooltip into three components and proposed predictive models, as given in Equation (1), for the tool-tip (machining) energy consumption ($E_{tooltip}$).

$$E_{tooltip}=E_p+E_f+E_k$$

(1)

where $E_p$, $E_f$ and $E_k$ represent the energy associated with the primary shear zone, frictional forces and kinetic energy of chips flow, respectively.

These types of models were significant; they were limited to estimating energy consumption at the tool–workpiece interface (chip removal). The chip removal energy, though significant, only contributes a portion of the overall machine tool’s energy demand [15]. Machine tools, particularly Computer Numerical Controlled (CNC) machines, stand out as a significant component in modern manufacturing, with high precision and productivity. However, it is also associated with substantial energy consumption, which has become a critical concern in the pursuit of sustainable manufacturing practices. The energy consumed by CNC machines not only impacts the operational costs but also contributes to the environmental footprint of the manufacturing process. CNC machines, despite their advantages, are complex systems that consume a significant amount of energy. This energy usage stems from the interplay between the various electrical and mechanical components within the machine, each with its own power demand. Furthermore, the total energy consumption also depends on the operational state of the machine tool. Whether it is starting up, idling or actively cutting material, the energy needs will vary. A typical power profile of a machine tool during a turning process is shown in Figure 3.

To better understand machine tool energy consumption, researchers have categorized machine tool energy consumption into different classes, such as ‘modes’ (e.g., idle, run-time, production) and ‘states’ (e.g., basic, cutting, ready) [18].

These modes, introduced by Dahmus and Gutowski [19], define different operational stages based on the activity level, and they proposed a model, as shown in Equation (2). In the “Idle” mode, the machine is powered on but not working, using a fixed amount of energy for essential components such as hydraulics, the control panel, the spindle etc. When the “Run-time” begins, additional components activate such as feed axes and the coolant pump (in the case of wet machining), consuming constant energy. Finally, the “Production” mode involves metal removal, where energy consumption varies depending on the cutting conditions.

$$P_{total}=P_o+K·\left(MaterialRemovalRate\right)$$

(2)

where $P_{total}$ and $P_o$ are the total power consumption and idle power consumption of the machine tool in kW, respectively, and $k$ represents the constant in $\mathrm{k}\mathrm{J}/\mathrm{c}{\mathrm{m}}^3$.

Another approach looks at machine tool states [20], as shown in Equation (3).

$$E_{total}=E_{basic}+E_{cutting}$$

(3)

where $E_{basic}$ and $E_{cutting}$ represent the energy consumption corresponding to the basic state and cutting state of the machine tool respectively. The “Basic” state refers to when minimal components (e.g., computer unit, lightening, cooling fans, lubrication, etc.) are operational to keep the machine ready. During the “Cutting” state, similar to the “Production” mode, energy is used for chip removal, with the amount depending on the cutting parameters. Further classifications include the “Ready” state by Balogun and Mativenga [21], as shown in Equation (4), which bridges the gap between basic and cutting states, representing the energy used to position the workpiece and tool before cutting commences. In the ready state, the spindle achieves the target speed and the tool is positioned for cutting.

$$E_{total}=E_{basic}+E_{ready}+E_{cutting}$$

(4)

where $E_{ready}$ represents the energy consumption corresponding to the ready state of the machine tool.

Similarly, Schudeleit et al. [22] propose “Standby”, “Ready” and “Processing” states for calculating an energy efficiency index, as shown in Equation (5).

$$E_{total}=E_{standby}+E_{ready}+E_{processing}$$

(5)

where $E_{standby}$ and $E_{processing}$ are the energy consumption of a machine tool in the standby and processing states. Standby and processing states are comparable to the basic and cutting states of a machine tool, with the exception being that, in the basic state, the cooling fans and lubrication units are operational and not in the standby state. Figure 4 represents state-based and mode-based classifications of machine tool operation, highlighting the energy consumption of the corresponding machine tool components.

Alternatively, Lv et al. [23] classify total power consumption based on machine tool motions. They argue that the motions of the machine tool during or in support of machining are responsible for energy consumption. They further identify four key types of machine tool motions: basic, auxiliary, air cutting and material removal. In line with this approach, Edem and Mativenga [24] developed a method for predicting total energy consumption by modeling the energy demands of specific CNC machine tool numerical codes, as shown in Equation (6).

$$E_{total}=E_{basic}+E_{tool}+E_{spindle}+E_{feed}+E_{cool}$$

(6)

where $E_{tool}$, $E_{spindle}$, $E_{feed}$ and $E_{cool}$ are the tool change, spindle run, feed and coolant energy consumption, respectively and are modeled using the corresponding numerical codes, e.g., $E_{feed}={(P}_{G01/G02/G03\_feed}+P_{G{00}_{feed}\left(approach\right)}+P_{G00\_feed\left(retract\right)})t_{feed}$, and $P_{G01/G02/G03\_feed}$ $P$ ($G_{00}$, $G_{01}$, $G_{02}$ and $G_{03}$) is the power demand corresponding to the different numerical codes $G_{00}$ (rapid positioning), $G_{01}$ (linear interpolation), $G_{02}$ (circular interpolation clockwise), $G_{03}$ (circular interpolation anti-clockwise). In pursuit of improving machine tool energy consumption models, researchers like Edem and Mativenga [25] have investigated the energy demands of specific components, such as feed axes, with weight of the feed drive and workpiece included, as shown in Equation (7).

$$E_{feed}=P_{basic}·t_{ct}+\left(a·W·v_f+b·W\right)t_c+F_f·v_c·t_c$$

(7)

where $W$ represents the weight of moving components (feed axis, vice, workpiece), $F_f$ represents the feed force, $v_f$ and $v_c$ represent the feed and cutting velocities in (m/min), $t_{ct}$ is the is total machining/cycle time and $t_c$ is the actual cutting time within the cycle. This approach allows for a more accurate evaluation of the total energy consumption. Similarly, significant research efforts have focused on improving the prediction accuracy of machine tool energy consumption, ultimately leading to more energy-efficient machining processes.

Table 1. Overview of energy consumption models for machine tools.

Reference	Model	Remark
Wang et al. [16]	$E_{tooltip}=E_p+E_f+E_k$ where $E_p$, $E_f$ and $E_k$ the energies associated with the primary shear zone, frictional forces and kinetic energy of the chips flow, respectively.	Focused only on tool-tip energy consumption
Dahmus and Gutowski [19]	$P_{total}=P_o+K·\left(MaterialRemovalRate\right)$ $\mathrm{where}{P}_{total}$$\left(\mathrm{total}\right)\mathrm{and}{P}_o$$\left(\mathrm{idle}\right)\mathrm{are}\mathrm{power}\mathrm{consumptions}\mathrm{and}k$ is a constant	One of the primary models and studies which investigated the machine tools energy consumption
Kellens et al. [20]	$E_{total}=E_{basic}+E_{cutting}$ $\mathrm{where}{E}_{total}$$\left(\mathrm{total}\right),E_{basic}$$\left(\mathrm{basic}\mathrm{state}\right)\mathrm{and}{E}_{cutting}$ (cutting state) are energy consumptions	Proposed a machining state-based approach for machine tool energy consumption modeling
Balogun and Mativenga [21]	$E_{total}=E_{basic}+E_{ready}+E_{cutting}$ $\mathrm{where}{E}_{total}$$\left(\mathrm{total}\right),E_{basic}$ (basic state), (ready state) and $E_{cutting}$ (cutting state) are energy consumptions	Introduced a ready state of machine tools for the modeling of the energy consumption
Lv et al. [26]	$P_{total}=P_{standby}+P_{light}+P_{cool}+P_{spindle}+P_{feed}+P_{tool}+P_{cutting}$ $\mathrm{where}{P}_{total}$ (total) and $P_{standby}$ (standby), $P_{light}$$\left(\mathrm{lighting}\right),P_{cool}$ (coolant pump), $P_{spindle}$$\left(\mathrm{spindle}\right),P_{feed}$ $\left(\mathrm{feed}\right)\mathrm{and}{P}_{tool}$$\left(\mathrm{tool}\mathrm{change}\right)\mathrm{and}{P}_{cutting}$ (cutting) are power consumptions	A Therblig-based approach to evaluating and investigating the machine tools’ energy consumption
Liu et al. [27]	$E_{total}={\sum }_{j=1}^{Q_s}{E}_{startup\_j}+{\sum }_{j=1}^{Q_u}{E}_{idle\_j}+{\sum }_{j=1}^{Q_c}{E}_{cutting\_j}$ $\mathrm{where}{E}_{startip}$$\left(\mathrm{startup}\right),E_{idle}$$\left(\mathrm{idle}\right)\mathrm{and}{E}_{cutting}$ (cutting) are energy consumptions	Focused on one of the machine tools’ primary energy-intensive spindle system energy consumption
Kim et al. [28]	$P_{total}=P_{idle}+P_{cool}+P_{spindle}+P_{feed}+P_{cutting}$ $\mathrm{where}{P}_{total}$$\left(\mathrm{total}\right)\mathrm{and}{P}_{idle}$$\left(\mathrm{idle}\right),P_{spindle}$$\left(\mathrm{spindle}\right),P_{feed}$$\left(\mathrm{feed}\right),P_{cool}$$\left(\mathrm{coolant}\mathrm{pump}\right)\mathrm{and}{P}_{cutting}$ (cutting) are power consumptions	Revealed that the actual cutting process itself requires a consistent amount of energy regardless of the specific machine tool. The difference in energy use stems primarily from the power demands during idle states and spindle operation
Schudeleit et al. [22]	$E_{total}=E_{standby}+E_{ready}+E_{processing}$ $\mathrm{where}{E}_{total}$$\left(\mathrm{total}\right),E_{standby}$$\left(\mathrm{standby}\right)E_{ready}$$\left(\mathrm{ready}\right)\mathrm{and}{E}_{processing}$ (cutting) are energy consumptions	Proposed a new metric for rating machine tool designs for energy efficiency, considering individual components and their interaction. This fills a gap in ISO standards for eco-design
Edem and Mativenga [25]	$E_{feed}=P_{basic}·t_{ct}+\left(a·W·v_f+b·W\right)t_c+F_f·v_c·t_c$	Investigated the impact of the feed axis weight and workpiece weight on the total energy consumption of the machine tool
Edem and Mativenga [24]	$E_{total}=E_{basic}+E_{tool}+E_{spindle}+E_{feed}+E_{cool}$ $E_{total}$$\left(\mathrm{total}\right),E_{basic}$$\left(\mathrm{basic}\right),E_{tool}$$\left(\mathrm{tool}\mathrm{change}\right),E_{spindle}$$\left(\mathrm{spindle}\right),E_{feed}$$\left(\mathrm{feed}\right)\mathrm{and}{E}_{cool}$ (coolant pump) energy consumptions	Proposed a Numerical code-based energy modeling for machine tools
Yoon et al. [29]	$P_{total}=P_{basic}+P_{spindle}+P_{stage}+P_{machining}$ $P_{basic}$$\left(\mathrm{basic}\mathrm{power}\right),P_{spindle}$$\left(\mathrm{spindle}\mathrm{rotation}\mathrm{power}\right),P_{stage}$$\left(\mathrm{table}\mathrm{motion}\mathrm{power}\right)\mathrm{and}{P}_{machining}$ (cutting power).	A component-level energy consumption model for machine tools focusing on the feed drive and rotational axes
Feng et al. [30]	$f(x_1,x_2,\dots x_m)=a_0+{\displaystyle \sum _{i=1}^M}{a}_i{x}_i+{\displaystyle \sum _{i=1}^M}{\displaystyle \sum _{j=1}^M}{a}_{ij}{x}_i{x}_j+{\displaystyle \sum _{i=1}^M}{\displaystyle \sum _{j=1}^M}{\displaystyle \sum _{k=1}^M}{a}_{ijk}{x}_i{x}_j{x}_k$ $\mathrm{where}{(x}_1,x_2,\dots x_m)$$\mathrm{are}\mathrm{the}\mathrm{input}\mathrm{variables}’\mathrm{vector}\mathrm{and}(a_0,a_i,a_{ij}{a}_{ijk}\dots )$ represent the coefficients	Workpiece hardness, cutting tool edge, material removal rate and spindle speed were added as influential factors under this data-driven power consumption model
Pan et al. [31]	$E_{total}=\int \left(P_{standby}+P_{unload}+P_{machining}+{P_{additional}}_{loss}\right)dt$ $P_{standby}$$\left(\mathrm{standby}\mathrm{power}\right),P_{unload}$$\left(\mathrm{unload}\mathrm{power}\right),P_{machining}$$\left(\mathrm{cutting}\mathrm{power}\right),\mathrm{and}{P_{additional}}_{loss}$ (additional loss) $t$ (cutting time)	A machine learning (ML) model for machine tool energy consumption prediction, leveraging data to compensate for missing information
Pawanr et al. [32]	$E_{total}=E_{startup}+E_{standby}+E_{acceleration}+E_{rapi{d}_{positioning}}+E_{idle}+E_{tool}+{E_{air}}_{cut}+E_{cool}+E_{machining\_CMRR+}{E}_{machining\_VMRR}$ $E_{startup}$$\left(\mathrm{startup}\right),E_{standby}$$\left(\mathrm{standby}\right),E_{acceleration}$$\left(\mathrm{spindle}\mathrm{acceleration}\right),E_{rapi{d}_{positioning}}$$\left(\mathrm{rapid}\mathrm{positioning}\mathrm{of}\mathrm{tool}\right)E_{idle}$$\left(\mathrm{idle}\right),E_{tool}$$\left(\mathrm{tool}\mathrm{change}\right),{E_{air}}_{cut}$$\left(\mathrm{air}\mathrm{cut}\right),E_{cool}$$\left(\mathrm{coolant}\mathrm{pump}\right),E_{machining\_CMRR+}$$\left(\mathrm{constant}\mathrm{MRR}\mathrm{machining}\right),E_{machining\_VMRR}$(variable MRR machining) energy consumptions	Proposed a new model for machine tool energy consumption considering both constant and variable material removal rates
Brillinger et al. [33]	$E=k_{c1.1}.h^{-m_c}·∆V$ $\mathrm{where}{k}_{c1.1}$$\left(\mathrm{specific}\mathrm{cutting}\mathrm{force}\right),h$$\left(\mathrm{chip}\mathrm{thickness}\right),m_c$$\left(\mathrm{cutting}\mathrm{force}\mathrm{exponent}\right)\mathrm{and}∆V$ (removed volume of material)	Proposed prediction technique for energy consumption during the design phase, allowing designers to make eco-friendly choices upfront.

presents the development of machine tool energy consumption models based on selected significant studies.

The preceding discussion and Table 1 highlight that the progression in machine tool energy modeling reflects the advancements in our understanding of machining processes. Early studies were mainly focused on the energy consumed at the tooltip, which is just a fraction of the total energy profile of a machine tool, providing a limited view of the energy consumption of machining processes. Contemporary models have evolved to encompass a variety of factors such as the startup, standby, spindle acceleration and idle states energy consumptions. Moreover, machine learning approaches are being applied to predict the energy consumption of machining processes. This holistic approach allows for a more accurate assessment of a machine’s environmental impact and operational costs, leading to more sustainable manufacturing practices.

The following section elaborated on the various strategies/advancements made by researchers aimed at enhancing the energy efficiency of machine tools.

4. Strategies for Energy-Efficient Machining

Generally, there are two approaches to enhancing the energy efficiency of machine tools. One involves designing machine tools that are energy-efficient and use materials efficiently. The other focuses on optimizing the machining process itself to save energy. Designing energy-efficient machine tools involves using advanced materials and technologies that reduce energy consumption during idle and active states. This can include the integration of high-efficiency motors, the use of lightweight materials to reduce inertia and the implementation of smart control systems that minimize energy waste. On the other hand, optimizing the machining process focuses on the operational aspects, such as selecting the most efficient cutting parameters and tool paths and minimizing non-productive times.

The pursuit of energy efficiency in machining processes is a multifaceted endeavor, encompassing a range of strategies aimed at reducing the energy consumption and enhancing the sustainability of manufacturing operations. Energy-efficient machine tools are designed to consume less power while maintaining performance, often incorporating advanced materials and control systems. Optimization techniques, such as process parameter and tool path optimization, seek to minimize energy use without compromising the quality of the machined product. Selecting the appropriate machine tool for a specific operation can also significantly impact energy consumption, as it ensures that the process is carried out as efficiently as possible. Accurate energy modeling, including methods like therblig analysis, bond diagrams and machine tool component level modeling, provides a detailed understanding of energy usage patterns and identifies potential areas for improvement [26,34]. The inclusion of transient state energy consumption in these models offers a more comprehensive view, capturing the dynamic nature of machining processes. Furthermore, the application of artificial intelligence in modeling and optimization harnesses the power of data analytics and machine learning to predict and improve energy efficiency, leading to smarter, more adaptive manufacturing systems. These techniques can be categorized into different groups, as discussed in the forthcoming sections.

4.1. Energy-Efficient Design of Machine Tools

Energy usage during the machining process is greatly influenced by the design of machine tools. Incorporating energy consumption considerations into the design phase of machine tools presents a significant opportunity for optimizing their operational efficiency and minimizing their lifetime energy footprint. Machine design itself can be optimized for efficiency. This involves incorporating components like high-efficiency motors, minimizing friction in moving parts, light weight design and strategically designing machine elements to reduce overall energy use. The machine tool industry is undergoing a significant transformation, driven by the imperative of energy efficiency. This shift is not just a response to environmental concerns but also a strategic move to reduce operational costs and comply with tightening regulations. The case study by DMG Mori Co., Ltd. highlights the potential for substantial energy reductions and associated CO₂ emissions savings using modern machinery. The comparison in their case study is of a new machine tool with an older model, resulting in a 45% reduction in energy consumption [35]. A significant portion of energy consumption comes from the moving components of machine tools, and therefore, structure optimization in machine tool design can be an influential energy-saving strategy. Ji et al. [36] proposed a new method for optimizing the structure of moving parts in machine tools for lower energy consumption while considering the static strength and vibration resistance during optimization. By adopting this approach to a hobbing machine tool slide, they designed a light weighted structure and achieved a 3.22% energy reduction while maintaining performance. Wang et al. [37] demonstrate the framework’s effectiveness regarding a four-axis machining center, achieving weight reduction while maintaining stiffness. They combine knowledge-based design with multi-stage optimization and integrate modeling and simulation software and achieve a weight reduction of 8.14% with an increased stiffness of 5.59%. Li et al. [38] proposed a new method that combines optimizing the design of moving components and the control system of a machine tool feed system to minimize energy consumption while maintaining machining accuracy. The results show a significant reduction in energy use (over 10%) and improved control performance with the new design. Triebe et al. [39] employed a genetic algorithm to design lightweight machine tool components, specifically focusing on the slide table. They optimized the structure, particularly the cross-sectional design of the beam of the table as a sandwich panel, prioritizing both weight reduction and maintaining stiffness. Testing confirms that the lightweight designs achieve comparable strength to that of traditional tables. Some researchers used carbon fiber to develop lightweight machine tool components without sacrificing mechanical stiffness. Hu et al. [40] explored lightweighting the machine tool worktable using Carbon Fiber Reinforced Polymer circular tubes with a straight and honeycomb structure, aiming to enhance the processing efficiency for high-precision aerospace parts. Tests and simulations show that the honeycomb arrangement of circular tubes reduces the weight by 45.05% compared to traditional designs.

The spindle is another critical moving component in machine tools, and optimizing its design is crucial for conserving energy in machine tools. The spindle system is recognized as one of the most significant energy consumers in these machines, making its optimization a highly effective strategy for energy conservation. Yi et al. [41] focused the design improvement of one of the major energy-intensive components of a machine tool, i.e., the spindle system. Spindle motors and transmission system parameters were integrated to reduce the specific energy consumption and structure volume. A trade-off between energy efficiency and compact design was accomplished by employing a multi-objective improved teaching-learning-based optimization algorithm. Lv et al. [42] proposed a new spindle design method that considers both energy use and performance during the design stage. They utilized a comprehensive objective function and a biogeography-based optimization algorithm to achieve significant energy reductions while maintaining or improving the spindle performance. This method offers valuable guidance for future spindle optimization efforts. This review of the relevant literature underlines the significant potential of energy-efficient machine tool design in reducing the energy consumption of machining processes.

4.2. Optimization of the Machining Process

Optimizing current machining processes offers a more immediate and cost-effective solution compared to redesigning machines. This approach allows for easier implementation on existing manufacturing units. Optimizing process (cutting) parameters like cutting speeds, feeds and depths of cut during machining can significantly impact energy consumption. Additionally, selecting efficient tool paths that minimize unnecessary movements can further reduce the energy used by the machine. Choosing the right machine tool size and capabilities for the specific machining operation is also crucial. By avoiding oversized machines, manufacturers can prevent unnecessary energy use. Finally, integrating energy consumption considerations during the design phase allows process designers to make choices that promote energy efficiency from the very beginning. This multi-faceted approach ensures that machine tools operate with optimal efficiency throughout their lifecycle.

4.2.1. Optimization of Cutting Parameters

A critical factor influencing the energy consumption of machining processes is the selection of cutting parameters. Optimizing these parameters is a practical approach to achieving energy efficiency in the existing machine tools of production lines, requiring minimal resources and a relative ease of implementation. Existing research supports the efficacy of cutting parameter optimization as a viable strategy for significantly improving machining efficiency [11]. Notably, studies have demonstrated that the cautious selection of these parameters can lead to reductions in energy consumption by up to 40% [43,44]. As a result, parameter optimization for energy-efficient machining has emerged as a popular study area. However, prioritizing/optimizing cutting parameters only for energy efficiency performance in machining processes should not be at the cost of sacrificing other crucial indicators such as productivity and product quality. Multi-objective optimization techniques offer a practical approach for identifying the optimal cutting parameters when multiple performance measures require simultaneous improvement. While traditional multi-objective optimization techniques like Grey Relational Analysis (GRA), Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) and Response Surface Methodology (RSM) have been successfully employed, recent research in machining demonstrates a significant shift towards utilizing advanced methods like genetic algorithms (GA), particle swarm optimization (PSO) and deep learning [45].

Bagaber and Yusoff [46] compared RSM and Nondominated Sorting Genetic Algorithm II (NSGA II) for optimization, demonstrating that NSGA II achieves significantly better results in reducing energy and costs while maintaining quality. Bousnina et al. [47] employed a combination of PSO, artificial neural network (ANN) and GA to optimize machining parameters for features on 2017A alloy. Their approach demonstrated significant impacts on energy consumption, machining costs and surface quality based on machining strategies and sequence planning. Notably, they reported substantial reductions of 78.27% and 44.57% in energy consumption and machining cost, respectively, alongside a 39.77% improvement in surface quality. Wu et al. [45] introduced a novel multi-objective optimization approach for machining processes. Their method utilizes a deep learning-powered genetic algorithm to predict key performance indicators: environmental impact, energy efficiency and surface quality. These predicted values are then incorporated as surrogate objective functions within the optimization process. They employed TOPSIS to identify the optimal solution from the generated Pareto front, ensuring a balance between these competing objectives.

4.2.2. Toolpath Optimization

Toolpaths refer to the routes or trajectories that a cutting tool follows as it moves through the material during a machining operation [48]; some exemplary tool paths for a milling operation are shown in Figure 5. Each toolpath involves cuts in different directions and may vary in energy consumption when machining the same working area. The objective is to choose the toolpath that minimizes energy consumption during the machining process.

These paths are crucial because the selection of an optimal tool path is critical for enhancing energy efficiency in machining processes.

Recent studies have shown that by optimizing tool paths, energy consumption can be significantly reduced, sometimes by as much as 50% [49]. Luan et al. [49] investigated the impact of toolpath optimization on energy consumption in milling. An improved energy model incorporating toolpath effects was developed and validated experimentally. The results show significant energy reduction of around 50% through optimized toolpaths, achieved by balancing machining time, cutting power and surface integrity. Feng et al. [50] focused on reducing energy consumption in hole machining through the simultaneous optimization of tool paths and cutting parameters. An integrated approach is proposed to achieve greater energy savings compared to optimizing these factors separately. An energy model for hole machining was developed, incorporating energy, the tool path and cutting parameters, and further optimized for an energy-efficient toolpath. Recognizing air cutting’s high energy consumption, toolpath optimization prioritizes minimizing the travel distance via an ant colony algorithm. Subsequently, leveraging this optimized path, a multi-objective optimization model using NSGA-II for single-hole cutting parameters is established, and a 57.7% reduction in energy consumption and a 66.4% decrease in the processing time are achieved compared to an unoptimized path.

In multi-axis machining, toolpath optimization becomes increasingly significant. Lu et al.’s research [51] addressed energy efficiency in five-axis flank milling, a complex machining process known for high electricity consumption. The proposed solution utilizes a novel meta-reinforcement learning approach with the Soft Actor-Critic (MSAC) framework to optimize toolpaths. This method dynamically adjusts the feed angle, cutting strip width and path length while considering a maximum scallop height limit. The results demonstrate significant reductions in both energy consumption (69.96%) and processing time (68.44%) compared to traditional methods. Gao et al. [52] presents a novel approach to optimizing a toolpath for energy efficiency in a multi-axis end milling process. They focused on maximizing the cutting strip width at each cutter contact point by adjusting the tool orientation. The cutter contact point details and discretized contact regions for optimal cutter path planning are shown in Figure 6. In the diagram, D_t represents the cutting stripe width, f represents the feed direction, s represents the nominal surface and s′ represents the tolerance surface.

However, excessive driving and cutting energy consumption can counteract these gains. To address this, they analyzed both toolpath geometry features and overall energy consumption (cutting and driving), developed a discrete energy consumption path model to identify the most energy-efficient tool orientation sequence for a given toolpath and achieved energy savings of 10%.

Similarly, in turning centers, although operations may seem straightforward, careful consideration of the toolpath (sequencing of machining steps) is essential to ensuring the accurate machining of all features and energy efficiency, especially when multiple features need to be machined, as investigated by Trifunović et al. [53].

These findings underline the equal importance of the selection of the right cutting parameters and tool path to achieve high machining efficiency and sustainability in manufacturing.

4.2.3. Optimization/Elimination of Non-Cutting Activities

The significance of energy consumption for the non-cutting status of machine tools has received limited attention but is a critical area of research, especially considering its substantial share in the total energy usage of machine tool. The non-cutting activities (NCA), which include periods when the machine is idle, on standby or performing auxiliary functions, can indeed account for a high percentage of energy consumption, sometimes as much as 80% [54]. Optimizing these NCAs presents a valuable opportunity for energy savings. For instance, recent studies [54,55] have highlighted the importance of understanding the power characteristics of machine tools during the NCA to optimize and make machining processes more energy-efficient. For instance, Luan et al. [55] recommended prioritizing spindle speed adjustments within the permissible range of the currently engaged transmission chain. This strategy minimizes the occurrence of power spikes associated with frequent transmission changes.

Researchers [26,56] have leveraged Therblig-based value stream mapping to conduct micro-analyses of machining processes. This approach enhances transparency regarding energy demands throughout the process, particularly by pinpointing non-cutting activities that contribute to energy consumption. By breaking down the machining process into discrete actions, or therbligs, and mapping them, researchers can analyze and optimize energy consumption at a micro level. This method not only enhances the transparency of energy demand but also identifies non-value-added activities where energy waste can be reduced.

Additionally, the sequencing of machining features has been shown to affect non-cutting energy consumption, suggesting that the strategic planning of non-cutting operations can lead to significant energy reductions. Hu et al. [57] investigated the substantial non-cutting energy consumption inherent in machining processes, including tool changes and spindle speed adjustments. The study investigated the impact of the feature processing sequence (FPS) on this energy expenditure. They developed an optimization method utilizing ant colony optimization to identify the optimal FPS that minimizes non-cutting energy use and demonstrated significant reductions in non-cutting energy consumption of 8.70% and 30.42% in two case studies involving 12 and 15 features, respectively. Feng et al. [58] focused on optimizing energy efficiency by sequencing features and adjusting machining parameters to reduce both cutting and non-cutting energy consumption using a genetic algorithm.

These insights underscore the potential for substantial energy savings in the manufacturing industry by focusing on the often-overlooked non-cutting phases of machine tool operation.

Table 2. Energy-saving potential of different optimization approaches.

Optimization Approach	Energy Consumption Reduction Potential
Optimization of cutting parameters	40% [43,44]
Toolpath optimization	50% [49,50]
Optimization/Elimination of non-cutting activities	30% [57]

summarizes the identified energy-saving potential of different optimization approaches from some of the important studies considered in this review.

4.3. Application of Artificial Intelligence

Artificial intelligence (AI) has been rapidly evolving in recent years, finding applications across a wide range of fields and introducing advanced modeling and optimization applications, e.g., from revolutionizing healthcare with disease diagnosis and drug discovery to optimizing traffic flow in transportation and personalizing recommendations in retail. Machine learning, a powerful tool within AI, allows computers to learn from data without explicit programming and has revolutionized numerous fields with its recent advancements, offering robust solutions to complex problems [59]. The machining processes exhibit inherent variability due to uncertainties within the machine itself and its operating environment. As a result, traditional physics-based models, while providing a valuable foundation for understanding machining principles, often struggle to deliver accurate performance predictions due to their inherent reliance on simplified assumptions. This necessitates the exploration of more robust methodologies for modeling machining performance [60]. The complex and stochastic nature of machining processes further complicates the estimation of physics-based model coefficients. As a result, AI techniques such as neural networks, fuzzy logic and genetic algorithms and many more have gained importance in engineering applications [61]. These techniques are favored for their reliable predictive capabilities and their adeptness at handling the inherent complexities of machining processes. They excel in capturing nonlinear relationships between input and output parameters, thereby enhancing both modeling accuracy and optimization efficiency.

4.3.1. Modeling Machining Energy with AI

Garg et al. [61] pointed out the limitations in conventional modeling methods like RSM, Grey relational analysis and Taguchi. To address these shortcomings, they employed the advanced AI techniques Genetic programming (GP), Multi-adoptive regression splines and Support Vector Regression (SVR) to develop predictive models for tool life and power consumption in turning processes. The models were evaluated using various statistical metrics, demonstrating promising results. In another study, Garg et al. [62] introduced a novel complexity-based multi-gene genetic programming (MGGP) approach incorporating orthogonal basis functions to predict the energy consumption of a milling process. They compared this approach with standardized MGGP to model energy consumption in the milling process. Through sensitivity and parametric analyses, they uncovered hidden relations between energy consumption and cutting parameters. The developed model facilitates the identification of optimal input configurations aimed at conserving energy during milling operations. Iqbal et al. [63] applied fuzzy modeling to enhance sustainability in metal cutting. They analyzed the trade-off between energy consumption, tool life and productivity, using a fuzzy rule-based system with optimization and prediction modules. This system recommends optimal cutting parameters and forecasts process outcomes based on chosen settings. Nguyen [64] utilized the adaptive neuro-fuzzy inference system (ANFIS) to model energy consumption and surface roughness in hard turning. Through optimization with simulated annealing (SA), significant reductions of 50% in energy consumption and 20% in surface roughness were achieved, alongside a 33% increase in the machining rate.

MTConnect [65], a communication standard, enables the seamless exchange of information between machines with standardized functions, facilitating the monitoring of the performance and analysis of production data. This real-time data acquisition is crucial for developing accurate energy models that reflect the actual operational conditions of machining tools. By leveraging real-time data, manufacturers can optimize their energy consumption patterns, leading to more sustainable and efficient production processes. Bhinge et al. [60] proposed a data-driven approach to predicting the energy use in machine tools. Their model leverages real-world sensor data (collected using MTConnect) and machine learning (Gaussian Process Regression) to build a more accurate and adaptable model for predicting energy consumption for machining a generic part. Furthermore, the study demonstrates the model’s practical application in process planning. By considering various toolpaths and materials, the model predicted the energy consumption and identifies the most energy-efficient option. The use of real-time data in energy modeling aligns with contemporary approaches that combine data measurement with predictive models to enhance energy efficiency. This shift from experimental to real-time data not only improves the precision of energy models but also supports the implementation of intelligent energy management systems that can adapt to changing conditions and drive sustainable energy solutions.

Machine learning demonstrates its capability to predict energy consumption even in scenarios where datasets are incomplete, which can occur when data are sourced from machining cluster centers, and the data collected from workshops may be incomplete for various reasons. Pan et al.’s research [31] introduces a framework designed to tackle this challenge. They employed data-driven ML approaches, generative adversarial imputation networks (GAIN) and gene expression programming (GEP), to model machine tool energy consumption with missing datasets. It was observed that the proposed framework can efficiently predict energy consumption within an acceptable error range, even when 30% of the data are missing. This underscores machine learning’s potential to handle real-world scenarios characterized by incomplete data.

By utilizing high-frequency energy data, with machine learning algorithms, the researchers were able to successfully perform the energy consumption prediction of parts during their design phase. Building on recent advancements in Machine Learning (ML), researchers have proposed a new method for predicting a part’s energy consumption during machining at the design stage. Brillinger et al. [66] employed high-frequency energy data to predict energy consumption during the design phase of a part using ML techniques (Decision Tree, Random Forest and boosted Random Forests). These models were then validated with actual part machining energy consumption data and show promising results. In another study, by harnessing the data available in the literature, Brillinger et al. [33] developed a predictive model that estimates the energy consumption of parts based on their geometry and material. This approach is particularly beneficial during the design phase, allowing for energy-efficient process planning.

The reviewed literature demonstrates a significant advancement in leveraging machine learning (ML) for modeling and optimizing energy consumption in machining processes. ML’s robust predictive capabilities and adeptness at handling process complexities enhance both model accuracy and optimization efficiency.

4.3.2. AI Optimization of Machining Energy Consumption

Within the domain of machining, advancements in ML are creating a paradigm shift towards optimized machining processes. In recent years, the introduction of machine learning optimization approaches to machining processes such as milling, turning and drilling has provided substantial prospects for improving operating efficiency and product quality. The exponential growth of data and the increasing complexity of models make ML particularly adept at analyzing this information to identify machining parameters with the most desirable outcomes. In the context of energy efficiency, this translates to significant potential for optimization. The integration of machine learning (ML) into machining process optimization is indeed revolutionizing industries by enhancing decision-making and fostering new opportunities for innovation. The literature demonstrates the use of several machine learning approaches for optimization to improve the energy efficiency of machining processes with other necessary performance characteristics like product quality. Techniques like NSGA-II (Non-dominated Sorting Genetic Algorithm II) [67], PSO (Particle Swarm Optimization) [68], Ant Colony Optimization [69], Simulated Annealing (SA) [67] and reinforced learning [70] are widely employed for their robustness in handling complex optimization problems.

NSGA-II is a prominent multi-objective evolutionary algorithm using non-dominated sorting for pareto fronts, genetic operations for solution evolution and a sharing parameter to maintain diversity. It balances exploration and exploitation while avoiding premature convergence, making it widely applied in diverse optimization problems [71]. To tackle high energy use in grinding, Jiang et al. [67] optimized the three key performance aspects, efficiency, energy and cost, by utilizing NSGA-II; the model finds optimal grinding settings that lead to significant improvements: a 16% reduction in grinding time, a 22% increase in energy efficiency and a 16% decrease in cost.

Another ML technique, PSO, is inspired by the social behavior of birds and fish, and it is known for its simplicity and effectiveness in converging to a solution [72]. In multi-pass face milling, Li et al. [72] proposed a novel approach to optimizing both energy efficiency and production costs. They developed a multi-objective model solved by adoptive multi-objective PSO. This method identifies optimal cutting parameters and pass counts, leading to significant energy savings without compromising production cost effectiveness.

Ant Colony Optimization, which mimics the foraging behavior of ants, is another powerful technique for finding optimal paths through graphs, making it suitable for machining processes optimization. Wang et al.’s research [69] proposed a novel system for the energy-efficient process planning of prismatic parts. They used STEP-NC, a common data format, and Ant Colony Optimization to find the optimal settings for energy-efficient machining. They identified optimal cutting tools, operation strategies and cutting parameters specific to the part being machined. Notably, the generated plan, which is compatible with the standard STEP-NC [73] code, ensured seamless integration with existing machining infrastructure. They also showed the successful application of the system on a test part with typical manufacturing features, validating its potential for optimizing machining processes for energy efficiency while maintaining compatibility with current manufacturing practices. Similarly, Simulated Annealing (SA) is a probabilistic technique that searches for a near-optimal solution by emulating the annealing process of metals. These methods have been successfully applied in various fields, from manufacturing to scheduling, demonstrating their versatility and capacity to tackle diverse optimization challenges.

The integration of modeling ML techniques with optimization algorithms has led to the development of hybrid approaches that enhance the process of finding global optima in complex problems. These hybrid models leverage the predictive power of ML to inform and guide the optimization process, resulting in more efficient and effective solutions. Several authors employed hybrid approaches to successfully improve the machining performances, e.g., adoptive ANN-NSGA-II [74], Radial Basis Function ANN (RBFNN)-Multi-Objective PSO (MOPSO) [68], SA-quantum behaved PSO (QPSO) [75] and ANFIS-SA [64] hybrid algorithms. Wang et al. [74] proposed a hybrid adaptive-ANN-NSGA-II method for optimal grinding parameters, focusing on energy management and machining performance. They modeled the surface quality, machining time and power consumption with ANN and optimized using NSGA-II, achieving significant results: the energy efficiency improved by 89.52% and the machining time was reduced by 174.36% while maintaining product quality. Xi et al. [68] introduced a RBFNN-MOPSO approach for parameter optimization. The hybrid algorithm predicts key cutting parameters and validates through experimental design. The results from an application case demonstrate improvements: 2.65% lower energy consumption, 21.44% higher efficiency and 28.50% lower combined bending moment, showcasing the method’s effectiveness in titanium alloy milling optimization and energy efficiency. Chen et al. [75] proposed an integrated model for optimizing process planning and cutting parameters to reduce production time and energy footprint in machining. They employed a hybrid SA-QPSO algorithm. The study emphasized the trade-off between the energy footprint and production time, effectively balanced by the proposed optimization approach. Nguyen [64] performed the optimization of hard turning parameters for the machining rate, energy consumption and surface roughness using hybrid ANFIS and adoptive SA. ANFIS-SA achieved significant improvements: 50% lower energy consumption, a 20% smoother finish and a 33% faster machining rate, which were greater compared to those of traditional methods of RSM-desirability analysis.

Furthermore, the rise of Machine Learning (ML) introduces powerful new techniques, e.g., Deep Learning and Reinforcement Learning. Deep Learning utilizes even more complex architectures for superior pattern recognition and feature extraction [76]. However, it demands even larger datasets and computational resources compared to conventional AI techniques. Reinforcement Learning takes a different approach, where an agent interacts with an environment, learning through trial and error. This makes it ideal for dynamic environments or situations where the optimal solution is unclear [77].

In conclusion, the optimal AI technique depends on the specific machining problem and available resources. For complex problems with abundant data, Deep Learning might be the best choice. However, for simpler problems or limited data, GA, ANN or Fuzzy Logic could be more suitable. Hybrid approaches can offer a well-rounded solution in certain scenarios. By understanding the strengths and weaknesses of each technique, manufacturers can model and optimize their machining processes for efficiency and sustainability.

5. Conclusions and Future Work

This paper has presented a review on recent advancements in improving the energy efficiency of machining processes to achieve sustainability in manufacturing. This review leads to the following conclusions and recommendations for future research:

-

Machine tools are one of the primary energy consumers in machining processes. Several key areas of significance include:

○

The accurate modeling of machine tool energy consumption is crucial for the precise estimation of overall machining process energy consumption.

○

Early studies focused primarily on energy consumption at the tool tip, providing a limited view. Contemporary models now consider factors such as startup, standby, spindle acceleration and idle state energy consumption, providing a comprehensive view of machining energy profiles.

-

There is still potential for improvement in machine tool energy consumption modeling, particularly in the following areas:

○

Limited studies have incorporated transient state energy consumption into their models and require further investigation.

○

Additional load losses, which can significantly impact energy estimation accuracy, have not been extensively studied.

○

Generic energy consumption models are lacking, with most models tailored to specific machine tools and materials. Exploring and developing more generic models is necessary for the practical implementation of energy-efficient strategies.

○

Energy efficiency indexing or rating systems similar to those used for other energy consumers are needed to standardize energy efficiency assessments in machining processes.

○

Non-cutting activities in machining processes, such as air cut, tool path, tool change and spindle rotation speed adjustments, contribute significantly to the total energy consumption of machine tools. By optimizing these non-cutting activities, particularly through the sequencing of machining features and the adjustment of spindle rotation speeds, considerable energy savings can be achieved.

-

Energy-efficient design of machine tools

○

Energy usage during the machining process is greatly influenced by the design of machine tools. Machine design itself can be optimized for efficiency.

○

This includes incorporating high-efficiency motors (with higher-power factors) and employing lightweight designs, such as optimizing the structure of moving components like feed axes (e.g., using honeycomb structures) and utilizing lightweight materials such as Carbon Fiber Reinforced Polymer (CFRP).

○

Lightweight designs of moving components of machine tools have great potential to minimize the non-cutting activities’ energy consumption.

○

The spindle is another critical moving component in machine tools, and optimizing its design is crucial for conserving energy in machine tools.

Due to the substantial number of existing machine tools in operation, these strategies necessitate robust economic considerations for technological advancement and can only be feasibly implemented through the replacement of current production lines.

-

Optimization of the machining process

○

A practical approach to enhancing energy efficiency in existing machine tools and production lines, requiring minimal resources and relatively straightforward implementation. The optimization has three major areas/approaches:

▪

Optimization of cutting parameters,

▪

Optimization of the feature sequence,

▪

Tool path optimization

▪

Optimization of the non-cutting activities of the machine tool

-

Significant advancements have been seen in the application of machine learning approaches to model and optimize the energy consumption of machining processes. Machine learning techniques have shown their reliable predictive capabilities and their adeptness at handling the inherent complexities of machining processes, thereby enhancing both modeling accuracy and optimization efficiency.

○

This real-time data acquisition is crucial for developing accurate energy models that reflect the actual operational conditions of machining tools. By utilizing high-frequency energy data, the researchers were able to predict the energy consumption of a part during its design phase using machine learning algorithms

○

Hybrid models leverage the predictive power of ML to inform and guide the optimization process, resulting in more efficient and effective solutions.

In conclusion, employing strategies such as energy-efficient design, optimizing machining processes and integrating AI techniques hold significant potential to reduce energy consumption, enhance energy efficiency and promote sustainability in machining processes.