**Contents**


## **About the Editors**

**Mariusz Deja** is an Associate Professor and the Head of the Department of Manufacturing and Production Engineering at the Faculty of Mechanical Engineering and Ship Technology, Gdansk ´ University of Technology, Poland. His research interests are abrasive finishing processes, advanced CNC programming, computer-aided process planning using feature-based modeling, as well as additive technologies. He is an author of more than 100 scientific papers in journals, conference proceedings, and book chapters showing the current research in the area of advanced manufacturing processes.

**Angelos P. Markopoulos** is an Assistant Professor in the Laboratory of Manufacturing Technology at the School of Mechanical Engineering, National Technical University of Athens, Greece. His research includes topics such as precision and ultraprecision conventional and nonconventional machining processes, with a special interest in advanced manufacturing and Industry 4.0. Furthermore, he is an expert in manufacturing technology modeling and simulation, including the finite element method, artificial intelligence, and molecular dynamics. He is the author of more than 120 papers in journals, conferences, and book chapters on the abovementioned areas and a member of the editorial boards of several international journals.

## **Preface to "Advances and Trends in Non-conventional, Abrasive and Precision Machining"**

The modern, highly competitive industrial environment demands machining and production processes that result in exceptional quality and precision. The general trend to design and manufacture more complicated mechanical components, along with the rapidly forward-moving material science, raises the need to incorporate and develop new machining techniques in the manufacturing process. Nonconventional machining processes differ from conventional ones, as they utilize alternative types of energy, such as thermal, electrical, and chemical energy, to form or to remove material. Commonly, the energy source has high power density, while the process features prodigious accuracy and the capability to produce and handle demanding shapes and geometries. Examples of nonconventional machining processes are electrical discharge machining (EDM), electrochemical machining (ECM), laser processing, and laser-assisted machining. Abrasive processes such as grinding, lapping, honing, polishing, and superfinishing are constantly developing and allow for obtaining a fine surface finish along with high efficiency. There is an increased scientific and commercial interest in the in-depth understanding and further development of the aforementioned nonconventional and precision machining processes. Research is moving forward through experimental studies, as well in the field of modeling and simulation, exploiting the increased available computational power. Multiphysics and multidisciplinary and multiscale modeling are powerful tools in the effort to optimize existing nonconventional precision machining processes, as well to develop novel ones. As their wider use by the industry swiftly grows, research has to be focused on them, not only due to the academic and scientific interest, but also for the possible financial gain. In light of these aspects, this book contains recent advances and technologies in the aforementioned fields, indicating the future trends for nonconventional precision machining processes. More specifically, a work where brushing with bonded abrasives as the finishing process for ceramics is presented [1]. The processing of zirconium dioxide workpieces with brushing tools of polycrystalline diamond-bonded grains is considered. The goal of the investigation is the reduction of grinding-related surface defects, the preservation of surface roughness and workpiece form, and the evaluation of tool wear in the case of brushing ceramic materials. It was found, by microscopical and surface topography measurements, that the brushing velocity and the grain size play the most significant role. Considering that the material removal mechanisms of abrasive brushing ceramics are largely unknown, this work is one of the few to deal with the specific topic. In the second work [2], a honing cell incorporating a thermographic camera, a sound intensity meter, and software for collecting and analyzing data received during the process on a CNC honing machine is proposed. With the aforementioned arrangement, images from the thermographic camera are analyzed online and the level of sound intensity obtained during honing is continuously monitored, with the purpose of online control of the process and its optimization. For further reducing the temperature of the workpiece due to its interaction with the tool and the subsequent deformations, the machining cell may have an automatic selection of the grain trajectory shape, with a specified value of the curvature radii of the abrasive grain trajectories, according to the wall thickness of the honed workpiece. With the proposed scheme, it is possible to increase the efficiency of the process by about 20-fold. The next work [3] also pertains to honing and more specifically to the possibility of employing wavelet analysis in order to evaluate the changes in the geometrical structure of a surface arising when honing with whetstones of variable granularity. The basics of the wavelet analysis and the differences between filtering with standardized filters, Fourier analysis, and the analysis of the results obtained when measuring the surface roughness with other wavelets are described. As a case study, the honing of a four-cylinder combustion engine was used, where roughness measurements of 3D spatial structures of the machined liners were carried out. The result of the work is the creation of basic recommendations for the selection of wavelets when assessing honed surfaces with different degrees of regularity of the traces that were generated on them. The effect of burnishing on heavily loaded structural elements operating in a corrosive environment is the subject of [4]. This research presents a fatigue resistance test performed on elements operating in seawater, namely a ship propeller, where various processing parameters of burnishing applied on samples are compared to specimens with a ground surface. The results indicate a 30% higher fatigue strength in a seawater environment of the burnished specimens. A device that allows for simultaneous turning and shaft burnishing with high slenderness is also presented. This device can be connected to the computerized numerical control system and automatic processes executed according to the machining program, with the aim of reducing the number of operations and cost of the process. Single-sided lapping is considered in [5], as it is one of the most effective planarization technologies. The process has relatively complex kinematics, and prediction of the tool wear is critical for product quality control. To determine the profile wear of the lapping plate, a computer model that simulates abrasive grains' trajectories in MATLAB is included in this work. Additionally, a data-driven technique was investigated to indicate the relationship between the tool wear uniformity and lapping parameters. The next two papers pertain to wire electrical discharge machining (WEDM) and die-sinking electrical discharge machining (EDM). The former [6] is a precise and efficient nonconventional manufacturing solution in various industrial applications, mostly involving the use of hard-to-machine materials such as, among others, the Inconel super alloys. The related study focuses on exploring the effect of selected control parameters, including pulse duration, pulse-off time, and the dielectric flow pressure on the WEDM process performance characteristics of Inconel 617 material. Parameters such as the volumetric material removal rate, dimensional accuracy of cutting, and surface roughness are considered in an experimental work that was carried out with the Box–Behnken design scheme and analyzed through the response surface methodology analysis of variance (ANOVA) tests. The latter study [7] presents an experimental investigation of the EDM of the aluminum alloy Al5052. A full-scale experimental work was carried out, with the pulse current and pulse-on time being the varying machining parameters. Then, polishing and etching of the perpendicular plane of the machined surfaces was performed in order to observe and measure the machined surfaces. Through analysis of variance (ANOVA), conclusions were drawn about the influence of machining conditions on the EDM performance, with consideration of the material removal rate, the surface roughness, the average white layer thickness, and the heat-affected zone microhardness. In the last paper [8], an analysis of the surface texture of turned parts with length/diameter ratios of 6 and 12 and various rigidity values is presented. The study pertains to samples of S355JR steel and AISI 304 stainless steel, with a detailed analysis of 2D surface profiles, using a large number of parameters that allowed significant differences in the surface microgeometry to be distinguished. The obtained results indicated significantly better roughness and waviness values of the AISI 304 steel surface in terms of its size, periodicity, and regularity; the turning process of AISI 304 shafts with low rigidity allows a better-quality texture to be achieved and has a positive effect on the general properties of a workpiece. Furthermore, it was concluded that the shafts with an L/D ratio of 12 had worse surfaces in the first two sections due to lower rigidity, while the results close to the three-jaw chuck, regardless of the L/D ratio and material type, demonstrated similar waviness and roughness parameters and profiles. The guest editors of this Special Issue would like to thank the authors for their valuable and high-quality work submitted, the reviewers for their efforts and time spent in order to improve the submissions, and the publisher for their excellent work and cooperation.

> **Mariusz Deja, Angelos P. Markopoulos** *Editors*

## *Article* **Surface Finishing of Zirconium Dioxide with Abrasive Brushing Tools**

**Eckart Uhlmann 1,2 and Anton Hoyer 1,\***


Received: 18 November 2020; Accepted: 13 December 2020; Published: 21 December 2020

**Abstract:** Brushing with bonded abrasives is a finishing process which can be used for the surface improvement of various materials. Since the machining mechanisms of abrasive brushing processes are still largely unknown and little predating research was done on brushing ceramic workpieces, within the scope of this work technological investigations were carried out on planar workpieces of MgO-PSZ (zirconium dioxide, ZrO2) using brushing tools with bonded grains of polycrystalline diamond. The primary goal was the reduction of grinding-related surface defects under the preservation of surface roughness valleys and workpiece form. Based on microscopy and topography measurements, the grain size sg and the brushing velocity vb were found to have a considerable influence on the processing result. Furthermore, excessive tool wear was observed while brushing ceramics.

**Keywords:** abrasive brushing; finishing; fine machining; grinding; ceramics; MgO-PSZ; ZrO2

#### **1. Introduction**

In order to meet increasing resource and productivity demands, modern technology requires the development of sustainable and responsible manufacturing processes. This leads to higher expectations in terms of component performance as well as durability [1]. Both are especially impacted by surface friction, which decreases the performance through energy loss and the durability through material wear. Friction between surface pairings is mainly influenced by their respective surface roughness, with high roughness leading to high friction, heat generation, and wear. However, the absence of surface roughness may also lead to a loss of retaining volumes for lubricant fluids. Therefore, surface finishing technologies no longer target total roughness reduction, but instead the ability to manufacture specialized surface textures for given tasks, even partially maintaining topography features. This enables, for example, the utilization of existing roughness valleys for lubricant retainment while only removing roughness peaks to further decrease surface friction.

Exemplary applications which require low surface roughness and high durability under frequent and selectively large pressures are artificial dentures or hip joints made of ceramic materials such as zirconium dioxide (ZrO2) [2,3]. Ceramics are distinguished by high hardness, heat and wear resistance, and bio-compatibility [4]. Nonetheless, their generally low heat conductivity, high brittleness, and predisposition towards fracture formation make the machining of ceramics particularly challenging [5]. After the sintering, ceramics are typically brought into final shape by microcutting or microgrinding processes, both of which may lead to local surface defects and high surface roughness. Despite the relatively ductile machining characteristics of ZrO2 compared to those of other types of ceramics, Fook and Riemer observed "brittle intercrystalline breakouts" as well as

ductile deformation bulges while describing ground ZrO2 surfaces [6]. These features likely affect the surface roughness adversely.

One machining process, which may be used for the finishing of components with hard surfaces, is brushing with bonded abrasives, making use of circular abrasive brushing tools (Figure 1). During tool production, abrasive grains are bonded in a filamentary polymer matrix by an extrusion process [7]. Common grain types are silicon carbide (SiC) and the softer aluminum oxide (Al2O3), although the finishing of ceramic materials suggests the use of harder grains such as polycrystalline diamond (PCD). As a polymer matrix, polyamide (PA) is most popular due to its high abrasion resistance, strength, as well as chemical inertness. Its thermal stability can further be increased by augmentation with additives [8]. The extruded abrasive filaments are cut to length and then attached to a circular brush body by either bonding or stuffing. However, largest filament quantities can be achieved by casting an epoxy resin brush body into a mold after arranging the filaments in a circular fashion. The thereby achieved high denseness of abrasive filaments allows for a high tool stiffness, increasing the productivity [7].

**Figure 1.** Layout of a circular abrasive brush tool [9].

Among the potentials of brushing with bonded abrasives are low process forces and temperatures as well as the utilization of existing machine systems, such as industrial robots, grinding machines, or mills [1,7]. Furthermore, the high flexibility of the abrasive filaments causes them to deflect during contact with a workpiece, thus assuming its shape and compensating minor inaccuracies regarding the geometry of workpiece, tool, and path [10]. Tool wear leads to a slow but successive shortening of the abrasive filaments [8,11,12], until the filaments become too short and therefore too stiff in order to yield consistent processing results. In addition, brushing tools typically require an initial conditioning process to achieve approximately equal filament lengths, which may be time- and cost-intensive, depending on the characteristics of the brushing tool and the requirements of the brushing process.

While abrasive brushing is widely used in industries, process designs are generally based on empirical values. This is mainly attributed to insufficient knowledge of the complex movement behavior of the flexible filaments, complicating the prediction of process characteristics, processing results, and tool wear. Thus, the technological investigations discussed in this article are directed at the gain of knowledge of brushing ceramics with bonded abrasives, specifically the characteristics of brushed surfaces in regard to relevant tool specification and process parameters.

#### **2. Materials and Methods**

To gain a basic understanding of the mechanisms being effective while processing ceramic surfaces with abrasive brushing tools, experiments were conducted on rectangular ZrO2 workpieces of the type Frialit FZM, manufactured and ground to dimensions of 200 <sup>×</sup> 200 <sup>×</sup> 20 mm<sup>3</sup> by FRIATEC AG, Mannheim, Germany. The sintered ZrO2 matrix was partially stabilized (PSZ) with magnesium oxide (MgO) to strengthen the material by retaining cubic fluorite crystal structures—usually only present at temperatures T of >2300 ◦C—and thus prevented monoclinic crystal formation, which would lessen

the fracture toughness [4]. Through a conventional plane grinding process, the surface roughness was then reduced to an arithmetic mean deviation of the roughness profile of Ra = 1.1 μm (Figure 2), with all roughness measurements taken orthogonally to the grinding direction.

**Figure 2.** Workpiece properties.

The circular brushing tools were manufactured by C. Hilzinger-Thum Gmbh, Sindelfingen, Germany (Figure 3d). The tools featured an epoxy resin brush body with an outer diameter of Db = 360 mm, an inner diameter of Di = 203.2 mm, and a width of Bb = 20 mm, holding abrasive filaments with PCD grains bonded in a PA 6.12 matrix. Varied specification parameters include the grain size sg and the filament diameter df. While the filament length lf also impacts the processing results, too short filaments are prone to take thermal damage due to the high stiffness and the low heat conductivity of ceramic workpieces, as the filament plastic matrix has a relatively low melting temperature of Tm ≈ 150 ◦C [8,13] and requires heat losses to be absorbed almost entirely by the workpiece. The accidental melting of the abrasive filaments can be compensated by the use of cooling lubricant [13]. However, PA 6.12 has a high tendency to absorb liquids, which decreases the filament stiffness and influences the consistency of processing results [9]. Therefore, no cooling lubricant was used during the technological investigations and a filament length lf = 40 mm was chosen for all experiments, although further research needs to be done on both cooling lubricants and appropriate filament lengths lf for the surface finishing of ZrO2.

The technological investigations were carried out on a plane and profile grinding machine of the type Profimat MT 408 HTS, manufactured by Blohm Jung GmbH, Hamburg, Germany (Figure 3a). It was equipped with three linear axes, the encoders of which allowed for a positioning accuracy of approximately 1 μm. The spindle had a maximum performance of Pmax = 45 kW and a maximum rotational speed of nmax = 11,000 min−1. All workpieces were brushed in parallel to the direction of the preceding grinding process.

Before conducting the technological investigations, the machine tool was used to condition the brushing tools by shortening the filaments to approximately equal lengths lf. As shearing off protruding filament tips with a sharp edge is a fast, yet imprecise method, a grindstone was used instead. After the radial runout error ΔDb of the brushing tool fell below 0.4 mm, 0.11% of the brushing tool diameter Db, an initialization process was repeatedly run with a brushing velocity of vb = 20 m/s, a feed velocity of vft = 200 mm/min, and a penetration depth of ae = 1 mm, hereinafter referred to as standard process parameters, until the reduction of the surface roughness did not significantly change any more, indicating quasi-static material removal behavior, which was required for consistent and comparable processing results. Subsequent experiments were conducted to determine the influence of the process

parameters brushing velocity vb, feed velocity vft, and penetration depth ae on the workpiece surface roughness (Table 1), as well as the workpiece form deviation, especially after successive brushing cycles.

**Figure 3.** Experimental equipment: (**a**) profile grinding machine; (**b**) tactile surface measurement device; (**c**) tactile roughness measurement device; (**d**) exemplary circular brush used during technological investigations; (**e**) scanning electron microscope; (**f**) light microscope.

**Table 1.** Variation of brushing tool specification and process parameters during the technological investigations.


\* Standard tool specification or process parameter.

Before and after individual brushing cycles, the workpiece profile depth h was measured orthogonally to the brushing direction with a laser triangulator of the type optoNCDT ILD2300-2 by the manufacturer Micro-Epsilon Messtechnik Gmbh & Co. KG, Ortenburg, Germany. The sensor had a measurement range of 2 mm and was operated at a measurement frequency of fm = 1.5 kHz and a measurement feed velocity of vm = 100 mm/min occurring orthogonally to the brushing direction. The vertical resolution of the sensor was denoted by the manufacturer as 0.03 μm at a reference measurement frequency of fm = 20 kHz and should analogously be smaller for lower measurement frequencies fm. The manual fitting and subtraction of the pre-brushing profile h0 from the post-brushing profile h1 provided information on the material removal depth hr. The workpiece surface roughness was determined with a tactile instrument of the type Surftest SJ-210, manufactured by Mitutoyo Corporation, Sakado, Japan (Figure 3c). Furthermore, the workpiece topography was measured with a tactile surface measurement device of the type Nanoscan 855, manufactured by Hommel-Etamic GmbH, Villingen-Schwenningen, Germany (Figure 3b). Additionally, light microscopies were made with a microscope of the type VHX 5000, manufactured by Keyence Deutschland Gmbh, Neu-Isenburg, Germany (Figure 3f). The images were taken with a 500× magnification, with a high dynamic range (HDR) enabled, and in 3D mode in order to increase the depth of focus. Scanning electron microscopy (SEM) images were obtained with a microscope of type LEO 1455VP, manufactured by Carl Zeiss

AG, Oberkochen, Germany (Figure 3e). For the evaluation of the material removal depth hr and the workpiece topography, the software MATLAB R2019b by The Mathworks INC., Natick, MA, USA was used.

#### **3. Results**

#### *3.1. Ground Surface Characteristics*

Characterizing the surface of the ground workpieces on the basis of SEM images (Figure 4), two prominent features became apparent: Firstly, the unidirectional grooves of the preceding plane grinding process stood out, discernable as parallel lines from left to right, and the breadth and occurrence of which depends predominantly on grain size and grain distribution in the grinding tool. Secondly, the plane ground surface was scarred with unevenly spread pockmarks of various sizes and shapes. Their approximate depths could not be determined based on SEM images but are suspected to be larger than those of the unidirectional grinding grooves. Equally unclear is whether the pockmarks were caused solely by the plane grinding process or originated from the workpiece composition and sintering. The comparison with workpieces from the same batch, which were lapped instead of plane ground, showed similar pockmarks and therefore suggests that they might be induced prior to the grinding process and then exposed by it, likely due to inhomogeneous crystal formation.

**Figure 4.** SEM images of the plane ground ZrO2 surface.

The characteristic unidirectional grinding grooves and nonuniform pockmarks can also be observed on the light microscopy images (Figure 5a). The comparison with a gently brushed surface showed that both features can be partially removed by the abrasive filaments and replaced with multidirectional brushing grooves, which were thinner and shallower than the initial grinding grooves (Figure 5b).

As surface defects such as grooves and pockmarks are prone to form fractures and thus weaken the material, especially in brittle materials such as ceramics [4], their removal can increase the component durability. Considering medical engineering as the most important field of application for ZrO2, pockmarks and other pitted blemishes in artificial dentures may lead to perpetual deposits of food debris and hence should be avoided. Similarly, rough surfaces possess a larger surface area and therefore increase bacteria accumulation [14].

**Figure 5.** Light microscopy images of ZrO2 with characteristic surface defects: (**a**) ground surface; (**b**) brushed with abrasive filaments with a filament diameter of df = 0.6 mm.

#### *3.2. Tool Specification*

Tool specification parameters, which influence the processing result, are mainly the grain size sg, the filament diameter df, and the filament length lf. While large grain sizes sg generally lead to higher material removal rates, lower surface roughness can be achieved with small grain sizes sg [12,15]. Therefore, the processing results for three different grain sizes sg were compared: sg = 320 mesh (29.2 μm), sg = 240 mesh (44.5 μm), and sg = 80 mesh (185 μm) [16,17] (Figure 6). Operated at standard process parameters, after a number of brushing cycles Nb = 3, the processing results of the tools with relatively fine grain sizes of sg = 320 mesh and sg = 240 mesh appeared similar, with unidirectional grinding grooves replaced to some extent by multidirectional brushing grooves and pockmarks being exposed rather than removed. However, the comparison with a coarse grain size of sg = 80 mesh showed that all unidirectional grinding grooves and most pockmarks were replaced by multidirectional brushing grooves, which were wider and seemed deeper than the brushing grooves created with smaller grain sizes sg.

As light microscopy images are only suitable for a qualitative analysis of the processing result, the workpiece topography was furthermore measured with a tactile instrument before and after brushing. Due to the assumed axial symmetry of the brushed profiles, the measurement tip was placed in a reproducible position near the center of each brushed profile and moved outwards in axial direction xm. The measured surface comprised a projected area with a length in the axial direction of xm = 10 mm and a width in the feed direction of ym = 4 mm. This led to a total of 41 roughness profiles per brushed surface and a 0.1 mm distance between single roughness profiles.

Considering the topographies before and after brushing with a grain size of sg = 320 mesh and standard process parameters (Figure 7), a preservation of the workpiece waviness can be observed alongside with an alteration of the surface roughness, characterized by the retaining of roughness valleys and the removal of roughness peaks. This suggested that successive abrasive brushing with fine grains subjected to a force-controlled principle allowed for the surface finishing of ZrO2 without the formation of entirely new surfaces, as would be the case for similar finishing processes such as fine grinding or lapping, subjected to position-controlled principles.

**Figure 6.** Light microscopy images of ZrO2 under the variation of the grain size sg: (**a**) ground surface; (**b**) brushed with a grain size of sg = 320 mesh; (**c**) brushed with a grain size of sg = 240 mesh; (**d**) brushed with a grain size of sg = 80 mesh.

**Figure 7.** Topography of ZrO2, brushed with a grain size of sg = 320 mesh.

The comparison between the topography of a workpiece brushed with fine grains with a size of sg = 320 mesh (Figure 7) and the one which was also brushed with standard process parameters but with coarse grains with a size of sg = 80 mesh (Figure 8) suggested that the material removal mechanisms strongly depended on the grain size sg.

**Figure 8.** Topography of ZrO2, brushed with a grain size of sg = 80 mesh.

Whereas fine grains altered the workpiece topography only slightly, coarse grains led to a large form deviation, exceeding the total height of the initial roughness profile and forming a new topography with arbitrary-form peaks and valleys. Depending on the application, this might affect the workpiece quality and functionality more than the presence of pockmarks, which remained yet to be detected based on tactile workpiece topography measurements.

Apart from the grain size sg, tool specification parameters worth investigating during future research are the filament diameter df and the filament length lf, both of which have an effect on the contact pressure pc (i.e., short and thick filaments leading to high contact normal forces Fn), thus impacting the grain penetration depth and consequently the processing result [7,9]. Additional tool specification parameters with minor influence on the processing result are the grain mass percentage cg and the type of plastic matrix used for the abrasive filaments [9]. Based on prior experimental results, finishing ceramic workpieces with grain types softer than PCD is possible but might cause excessive tool wear due to the similar hardness of abrasive grains and workpiece material.

#### *3.3. Process Parameters*

Apart from the grain size sg, the process parameters brushing velocity vb, feed velocity vft, penetration depth ae, and number of brushing cycles Nb influenced the processing result, the productivity, and the tool wear. The processing parameter mostly responsible for the number of contacts between the workpiece and the cutting edges of the abrasive grains was the brushing velocity vb, measured at the filament tip, being the outermost point of the circular brush tool. Its variation showed a significant impact on the removal of the unidirectional grinding grooves and the pockmarks (Figure 9). Whereas a brushing velocity of vb = 10 m/s left most of the initial topography unchanged even after a number of brushing cycles of Nb = 3 (Figure 9b), most unidirectional grinding grooves and some pockmarks were removed with a brushing velocity of vb = 20 m/s (Figure 9c), while all grinding grooves and most pockmarks were replaced by multidirectional brushing grooves with a brushing velocity of vb = 30 m/s (Figure 9d).

**Figure 9.** Light microscopy images of ZrO2 under the variation of the brushing velocity vb: (**a**) ground surface; (**b**) brushed with a brushing velocity of vb = 10 m/s; (**c**) brushed with a brushing velocity of vb = 20 m/s; (**d**) brushed with a brushing velocity of vb = 30 m/s.

The light microscopy images indicate that the pockmarks were reduced in size as more workpiece material was removed, suggesting that the phenomenon only occurred close to the workpiece surface. This contributes to the assumption that the surface defects were caused by the initial grinding treatment.

Examining the workpiece surface after a brushing process with a brushing velocity of vb = 30 m/s on the basis of SEM images (Figure 10), the multidirectional brushing grooves were less conspicuous than in the corresponding light microscopy images, while the complex shape of the pockmarks as well as their relatively large depths were most salient. Additionally, the edges around the pockmarks appeared rounded, although their rough inward texture depicted the brittleness of the workpiece material.

**Figure 10.** SEM images of ZrO2 surface, brushed with a brushing velocity of vb = 30 m/s.

Comparing the workpiece topographies before and after a high velocity brushing process (Figure 11), the workpiece form and waviness remained nearly the same, although all previously produced roughness features were removed, indicating the fabrication of an entirely new surface as well as a material removal depth hr of >9.9 μm, the total height of the initial roughness profile Rt (Figure 2). The corresponding total height of the roughness profile after Nb = 3 brushing cycles was Rt = 1.9 μm, which amounted to a percental roughness reduction of ΔRt = 80.7%, approximately the same as the percental peak height reduction of ΔRpk = 84.2% and the percental valley depth reduction of ΔRvk = 82.3%. This means that roughness peaks and valleys were equally reduced.

**Figure 11.** Topography of ZrO2, brushed with a brushing velocity of vb = 30 m/s.

In comparison, the workpiece topography after a brushing process with a brushing velocity of vb = 10 m/s shows that only the roughness peaks were partially removed while the roughness valleys and the workpiece form remained mostly intact (Figure 12). With an initial total height of the roughness profile of Rt = 7.8 μm, the resulting total height after Nb = 3 brushing cycles was Rt = 5.5 μm, amounting to a percental roughness reduction of only ΔRt = 29.7%. The considerable difference between the percental peak height reduction of ΔRpk = 55.3% and the percental valley depth reduction of ΔRvk = 32.3% supported the visual observation that roughness peaks were removed more efficiently than roughness valleys. This can be attributed to the fewer and less impactful contacts between abrasive filaments and workpiece compared to processes with high brushing velocities vb (Figures 7 and 11).

**Figure 12.** Topography of ZrO2, brushed with a brushing velocity of vb = 10 m/s.

In addition, the technological investigations conducted for this work confirmed an independence of the processing result from the feed velocity vft, meaning that multiple brushing cycles Nb at high feed velocities vft amounted to similar results to few brushing cycles Nb at low feed velocities vft, as long as the overall contact duration between tool and workpiece remained the same. Nonetheless, the low heat conductivity of ZrO2 might make brushing at high feed velocities vft more feasible and reduce tool wear due to a better distributed heat flow and more cooling time between brushing cycles, which prevents the plastic matrix of the abrasive filaments from melting and leaving residue on the workpiece, avoiding additional cleaning processes.

Further technological investigations showed no apparent qualitative differences between the penetration depths of ae = 1 mm, ae = 2 mm, and ae = 3 mm in regard to the processing result. However, contemporary research on steel workpieces (16MnCr5) indicates that large penetration depths ae slightly increase the material removal rate, the reduction of the surface roughness, and the tool wear due to increased contact forces [7,9].

The variation of the number of brushing cycles Nb supports the established theory that the surface roughness is reduced successively and regressively, until a lower roughness limit is reached—its value depending mainly on the tool specification—after which consecutive brushing cycles Nb always yield new topographies as opposed to merely reducing certain roughness features such as roughness peaks [9].

#### *3.4. Processing Result Deviations*

Since most aggressive brushing processes, specifically those with a large grain size sg and a high brushing velocity vb, do not only alter the surface roughness, but may also lead to a potentially unwanted change of the workpiece form, a more detailed analysis of the proceeding shape deviation is required. Therefore, the measurement length of the workpiece topography was increased to sm = 50 mm in order to capture the entire width of the brushed profile (Figure 13), as it was not homogenous due to the diverse interactions between abrasive filaments, depending highly on their positions on the brush body. The technological investigations showed that besides large grain sizes sg and high brushing velocities vb, a large number of brushing cycles Nb can also lead to a form deviation after the initial roughness features were completely removed. This resulted in a characteristic "W"-shape, the depth

of which was successively increased with each additional brushing cycle. Likewise, noteworthy is that the width of the "W"-shape amounted to Δxm = 27.8 mm, more than the nominal tool width of Bb = 20 mm, as the abrasive filaments were parted in the middle and then deflected towards both sides in the axial direction (measurement direction xm), dodging the high pressure caused by filament interactions and taking the path of least resistance, meaning the axial side obstructed by fewer neighbor filaments. Although a sharp transition from the brushed profile to the unprocessed workpiece surface can be recognized, a small number of protruding abrasive filaments appeared to have been in contact with the unprocessed surface, as the roughness beyond the transition edge was slightly reduced.

**Figure 13.** Topography of ZrO2, measured across the width of an overbrushed profile.

In addition to the workpiece topography, laser distance measurements were made between brushing cycles in order to monitor the workpiece form deviation over time, characterized by the material removal depth hr across the measurement width sm (Figure 14). The hypothesis that the crest of the "W"-shape was formed by the parting of the abrasive filaments as opposed to being predetermined by the initial workpiece waviness, as seen in Figure 13, was supported by the fact that it was first observed after several brushing cycles, whereas the material removal depth hr after Nb = 1 brushing cycles resembled a crestless "U"-shape instead. Moreover, the "W"-shapes of most brushing processes showed an asymmetry towards one axial side with less material being removed on the opposite side, presumably resulting from a minor unevenness or inclination of the brushing tool profile and/or the workpiece. As the formation of a "W"-shape was also observed on the brushing tool profile, mutually reinforcing form deviation mechanisms between tool and workpiece were most likely to occur. The technological investigations furthermore suggested that the workpiece material removed per brushing cycle followed a degressive trend, implying that rough surfaces were processed more efficiently than smooth surfaces and large numbers of brushing cycles Nb led to unproductive machining.

Apart from the aforementioned profile deviation, brushing tools are prone to lose sharpness over time due to an adjustment of the initially cylindrical abrasive filament tips to the workpiece surface, gradually transforming a sharp point contact into a flat, elliptical contact geometry [11]. Although the interrelations between wear and machining characteristics were not yet scientifically understood, first conclusions can be indirectly drawn from the processing result. One such correlation was the

percental reduction of the workpiece roughness, specified by the arithmetic mean deviation of the roughness profile Ra, measured over the entire period of application tu of the brushing tool (Figure 15).

**Figure 14.** Material removal depth hr, measured across the width of an overbrushed profile.

**Figure 15.** Deviating roughness reduction caused by the wear of the brushing tool.

After a brief initialization period, the roughness reduction continuously decreased. For the purpose of productivity, the brushing tool either needed to be expensively reconditioned or at moderate cost turned over to utilize the sharp sides of the abrasive filament tips as opposed to the flattened sides. During the technological investigations, these methods resulted in percental roughness reductions between ΔRa = 28.3% and ΔRa = 80.6% after single brushing cycles with standard process parameters, using one brushing tool with a grain size of sg = 320 mesh. However, depending on the application, a sudden increase of the material removal rate may be undesirable, as it might cause further form deviation of the workpiece instead of merely roughness reduction. In this case, frequently turning over the brushing tool or changing the rotational direction is advised in order to maintain a consistent level of productivity, represented by the material removal rate.

#### **4. Summary and Discussion**

Brushing with bonded abrasives is a finishing process which can be used to enhance the quality of technical surfaces, primarily by decreasing the surface roughness and removing near-surface defects. This could considerably improve the machining results of ceramic materials with defects such as

grooves or pockmarks caused by grinding processes, although the finishing of sintered ceramics without preceding grinding treatments is also conceivable in order to decrease the substantial machining costs [4].

Within the scope of this work, the improvement of the surface quality was demonstrated on the basis of MgO-PSZ (ZrO2) finished with circular brushing tools, the abrasive filaments of which contained PCD. The findings showed that abrasive brushing can furthermore be utilized to create specialized surfaces by only removing roughness peaks while leaving roughness valleys intact. Abrasive filaments with small grain sizes sg proved appropriate for this task, as large grain sizes sg rapidly led to form deviations larger than the total height of the roughness profile Rt and the resulting coarse brushing grooves negatively affected the minimum achievable roughness. Since small grain sizes sg usually require multiple brushing cycles due to the low material removal rate, performing one brushing cycle with coarse grains and one with fine grains might increase productivity, which remains to be confirmed by further experiments. In addition, pending are the influences of filament diameter df and filament length lf on the processing result while brushing ceramics. First experiments suggested that abrasive filaments shorter than lf = 40 mm are prone to take permanent thermal damage due to the low heat conductivity of ZrO2, which in industrial applications could be compensated by adding cooling lubricants.

Additionally, the brushing velocity vb was confirmed to have a significant impact on the machining characteristics, as a high brushing velocity of vb = 30 m/s proved advantageous for the reduction of near-surface defects and yielded topographies with equally reduced roughness peaks and valleys, indicated by a percental peak height reduction of ΔRpk = 84.2% and a percental valley depth reduction of ΔRvk = 82.3%. Nevertheless, the complete removal of pre-existing roughness features might be undesirable and also cause excessive tool wear due to increased heat input. In comparison, a low brushing velocity of vb = 10 m/s yielded a percental peak height reduction of ΔRpk = 55.3% and a percental valley depth reduction of ΔRvk = 32.3%, and their difference indicated that roughness peaks were removed more efficiently than roughness valleys.

Generally, workpiece form deviations and tool wear developed more rapidly than during comparable technological investigations priorly carried out on steel 16MnCr5 workpieces [9]. Even with fine grains and medium brushing velocities vb, after a small number of brushing cycles Nb, the forming of a "W"-shaped workpiece profile was observed, caused by the quasi-symmetrical deflection of the abrasive filaments attached to the lateral area of the cylindrical brush body. However, this might be of no consequence, if pot-shaped brushing tools were used instead, which comprised abrasive filaments attached to the base area of the cylindrical brush body, therefore yielding different kinematic relations as well as circular brushing groove patterns on the workpiece surface.

Tool wear was measured based on the consistency of processing results, specifically the arithmetic mean deviation of the roughness profile Ra, and could be compensated by frequent turning of the brushing tool or alternating of the rotational direction. A wear study based on the change in form of single abrasive filament tips remains yet to be conducted.

**Author Contributions:** Conceptualization, A.H. and E.U.; methodology, E.U.; software, A.H.; validation, A.H.; formal analysis, A.H.; investigation, A.H.; resources, A.H.; data curation, A.H.; writing of original draft preparation, A.H.; writing of review and editing, E.U.; visualization, A.H.; supervision, E.U.; project coordination, E.U.; funding acquisition, E.U. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Deutsche Forschungsgemeinschaft (DFG) within the scope of the project "Analyse des Zerspan- und Verschleißverhaltens beim Bürstspanen mit abrasivem Medium sprödharter Werkstoffe" (project number: 392312434).

**Acknowledgments:** The authors kindly thank the funder for their support.

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

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **The Proposition of an Automated Honing Cell with Advanced Monitoring**

#### **Adam Barylski and Piotr Sender \***

Department of Manufacturing and Production Engineering, Gda ´nsk University of Technology, 80-233 Gda ´nsk, Poland; abarylsk@pg.edu.pl

**\*** Correspondence: piotr.sender@pg.edu.pl; Tel.: +48-662-082-579

Received: 3 September 2020; Accepted: 22 October 2020; Published: 28 October 2020

**Abstract:** Honing of holes allows for small shape deviation and a low value of a roughness profile parameter, e.g., Ra parameter. The honing process heats the workpiece and raises its temperature. The increase in temperature causes thermal deformations of the honed holes. The article proposes the construction of a honing cell, containing in addition to CNC honing machine: thermographic camera, sound intensity meter, and software for collecting and analyzing data received during machining. It was proposed that the level of sound intensity obtained during honing could be monitored continuously and that the images from a thermographic camera could be analyzed on-line. These analyses would be aimed at supervising honing along with the on-line correction of machining parameters. In addition to the oil cooler, the machining cell may have an automatic selection of the grain trajectory shape, with specified value of the radii of curvature of the abrasive grain trajectories, according to the wall thickness of the honed workpiece, which will result in reducing the temperature generated during honing. Automated honing cell can mostly increase honing process efficiency. Simulations in FlexSim showed the possibility of increasing the efficiency of the honing process more than 20 times.

**Keywords:** honing of holes; kinematics of honing; automation of honing; abrasive grain trajectories; FlexSim; thermography

#### **1. Introduction**

The honing process heats the workpiece and raises its temperature. Specialist literature dealing with the honing process of cylindrical surfaces draws attention to the issue of the temperature increase during machining [1–11], which causes the thermal deformation of the honed hole. The temperature increase is also a very important issue during the honing process of flat surfaces by grinding with lapping kinematics as shown in [12]. Another interesting issue, in addition to the coexisting issue, i.e., surface texturing [13], is the use of new honing methods [14–17] and honing equipment used during machining [18,19] to improve the quality of the holes obtained [20–24].

The process of honing is influenced by the selection of proper machining parameters [1,25–30], which directly affects the obtained cylindricity deviations of the honed holes [9,10,31–35]. Much attention in the literature has also been devoted to examining the possibility of creating a new honing equipment and machines [18] as well as machining strategy, during which a different shape of oil channels cross-hatch could be obtained [12,36–46] which influences the further development of honing methods [47–52].

Irene Buj-Corral [2] described the methodology of selecting the appropriate density of the abrasive whetstone based on the analysis of the acoustic signal recorded in honing process and checked that roughness of honed cylinders increases mainly with abrasive grain size, followed by honing head pressure. She researched that tangential speed influences roughness slightly [53]. Similarly, Chavan [38] describes the effect of the pressure of the abrasive whetstone and the honing speed on the obtained roughness profile parameters. Deepak [27] dealt with in detail the influence of honing parameters on the obtained parameters of the roughness profile of honed surface. He stated that the rotational speed of the head had the greatest influence on the *Rz*, *Rk*, and *Rvk* parameters, while the linear speed of the jump had the greatest influence on the *Rpk* parameter. Barylski [1] indicates that the selection of parameters affects the size of the honed material and affects the amount of temperature generated in the workpiece.

The latest literature describes the use of the neural network method to supervise the honing process [54], but the analysis of parameters leads to obtaining the desired roughness profile parameters. In this article, it is proposed that the neural network completely diagnose the process of honing, including the analysis of the sound signal obtained during honing and the analysis of thermograms recorded during honing, which is a novelty in proposing the direction of the honing process and its automation.

#### **2. Kinematics of Honing Process**

Goeldel [41,55] proposed to use new shapes of grain paths for honing of cylindrical holes. Bujukli [56] describes the effect of the honing head stroke length on the improvement of the cylindrical shape deviation of the honed hole.

Figure 1 shows the cross-section of the honing head and the speeds occurring in the machining system: *Vax*—axial linear speed of the honing head [m/min], *Vaz*—peripheral speed of the honing head [m/min], *Vexp*—infeed speed of abrasive whetstone [μm/min].

**Figure 1.** Honing head with main honing speeds; 1—honed cylinder liner, 2—abrasive whetstone, 3—expanding mandrel, 4—pressure and machined diameter measurement, 5—temperature measurement, 6—vision system.

The honing head has a vision system (item 6), which records the real-time image of the honed surface, introducing data to the neural network that verifies the number of oil channels produced in such a way that the continuity for the flowing oil is broken. The gauge of the pressure force and the diameter of the machined hole (item 4) and the temperature gauge (item 5) enable the supervision of the honing process in real time.

The basic formulas that define the honing kinematics (Figure 1) are:


$$V\_{\mathfrak{c}} = \sqrt{V\_{ax}^2 + V\_{az}^2} \tag{1}$$


$$V\_{\text{ax}} = 2L\mathfrak{n}\_{\text{ax}} \tag{2}$$

where: *L* is the length of the head stroke in the axial direction, in reciprocating motion [m]; *n*α*<sup>x</sup>* is the head stroke frequency in reciprocating motion [1/min]; *Vaz* is the peripheral speed of the head [m/min]

$$V\_{az} = \frac{\pi dn}{1000} \tag{3}$$

where: *n* is the head rotation speed [min<sup>−</sup>1]; *d* is the diameter of the honed hole [mm]; And

$$
tg a = \frac{V\_{ax}}{V\_{az}}\tag{4}$$

where: α is the honing angle [◦].

The honing angle is a parameter influencing the oil consumption and the amount of toxic compounds emitted to the environment during the operation of internal combustion engines [57]. It also affects the coefficient of friction of the piston rings against the cylinder surface [38,42,58,59], which has a direct impact on engine power losses.

The honing angle depends on the mutual relation of the rotational speed and the speed of the axial linear stroke of the head. A higher value of the rotational speed with a lower value of the honing head stroke speed makes the angle closer to the horizontal direction, a lower value of the rotational speed with a lower value of the head stroke speed makes the angle closer to the vertical direction.

Figure 2 shows an exemplary trajectory of the abrasive grain (item 1) obtained during honing on the developed surface of the machined hole (item 2). The advantage of the honing cell is the automatic selection of the shape of the abrasive grain trajectory, with different values of the grain trajectory curvature, which directly affects the amount of temperature generated in the workpiece.

There are a many combinations possible in the honing process of oil channel angle. This diversity is presented in Table 1. Various honing researchers and different engine manufacturers suggest the use of different honing angles, but this article proposes the use of honing with variable kinematics, which enables the creation of a scratch grid in the form of curvilinear oil channels.

Honing with traditional kinematics is carried out without a change in the machining parameters. Figure 3 shows traditional honing process, where rotational speed of honing had has a constant value during process. This honing method allows to create cross-hatch shape of abrasive grain trajectories. The rotational speed of the honing head at the beginning of the cycle increases from zero to a constant value (Figure 3a), the feed value increases from zero to a constant value (Figure 3b). In honing with constant kinematics, the stroke length increases linearly (Figure 3c), while the grain path length increases according to the curve shown in Figure 3d.

Honing with variable kinematics is carried out with a change in the machining parameters. Figure 4 shows non-traditional honing process, where rotational speed of honing had has a different rotation speed in one single cycle of honing process. This honing method allows to create different shape of abrasive grain trajectories. The rotational speed of the honing head at the beginning of the cycle increases from zero and can still increase or decrease in turn (Figure 4a), the feed value increases

from zero and can increase and decrease in turn, etc., the return of the speed vector can change to the opposite in any way (Figure 4b). In honing with variable kinematics, the stroke length increases not linearly but curvilinear and according to the rotational speed and stroke speed value (Figure 4c), while the grain path length increases according to the curve shown in Figure 4d.

**Figure 2.** Possible directions of movement of honing head in honing of cylindrical holes: 1—abrasive grain trajectories, 2—expanding of honed surface, 3—honed cylinder liners with honing head, 4—additional honing head oscillation motion in vertical direction, 5—additional honing head oscillation motion in horizontal direction, 6—additional oscillation motion of honing head rotation direction.


**Table 1.** The honing angles discussed in the literature.

Table 1 shows the honing angles that are discussed in the specialist literature.

**Figure 3.** Traditional kinematics of honing process, (**a**) constant value of honing speed, (**b**) constant linear speed of honing head, (**c**) linear value of honing head, (**d**) length of abrasive grain path in traditional honing process; 1—an example of abrasive grain path.

Because of the disordered distribution of the grain in the abrasive stone, the grain fracture planes are at any angle to the grain direction, determined by the vector of the resultant honing speed *Vc*. During honing, the direction of the forces acting on the grain changes. Some of the grains are too weak in a given plane to transfer cutting forces, so that new cutting edges are constantly created during the honing process. Changing the direction of grain work also prevents the deposition of the processed material particles on the grain working surfaces; because of this phenomenon the change of the direction of the velocity vector *Vc*, i.e., variable honing kinematics, is an important parameter influencing the course of the process [5–8] and reduces the friction coefficient in the piston-cylinder assembly [80,82,84,85]. The value of normal acceleration *an* is responsible for the change of the vector *Vc* direction and the trajectory curvature.

**Figure 4.** Variable kinematics of honing process—an example of abrasive grain trajectory obtained during honing with variable linear and tangential speed. Components of the cylindrical honing process: 1—an example of abrasive grain path received in variable kinematics of honing, 2—comparative abrasive grain path obtained in traditional honing.

#### **3. Methods**

The article proposes the construction of a honing cell, containing in addition to CNC honing machine: thermographic camera, sound intensity meter, and software for collecting and analyzing data received during machining. It was verified that the main factor hindering the serial honing treatment is the deformation of the shape of the honed hole, caused by the heating of the workpiece due to the friction of the abrasive stone against the honed surface, which also causes a change in the diameter of the hole. The prevention of this phenomenon consists in controlling the temperature to which the honed workpiece is heated during machining, and in carrying out the treatment in such a way that the amount of temperature increase is reduced due to the selection of proper honing parameters, depending on the shape and *CCR* (the curve curvature radius) of the abrasive grain trajectories [5].

Honing cell principle of operation: 1—numerical simulation of deformations, stresses, and heat flow, 2—programming of honing head movements with the selection of the appropriate shape of the abrasive grain path adjusted to the thickness of the section or sections of the honed workpieces, 3—supervising of honing during the process, measuring of diameter and cylindricity of honed hole, measuring of sound signal, 4—correcting of actual machining parameters.

#### *3.1. The Equipment of the Honing Cell*

Figure 5 shows a CNC vertical milling center equipped with the honing instrumentation, a thermographic camera, surface roughness measure gauge, and sound intensity meter. Each of the three shown laptops analyzed different signals and their output values obtained during the honing process.

The key issue of the proposed automated honing cell is to be able to supervise the honing process by analyzing the audio signal and by analyzing the images obtained from the thermographic camera during the process. The task of supervision of honing process should be to generate abrasive grain paths with different shape of grain trajectories and with different radii of curvature, depending on the data obtained from the analysis of the acoustic signal and from the analysis of the thermogram of the workpiece.

**Figure 5.** Haas VF-3SS milling machine with honing equipment: 1—thermal imaging camera, 2—software of thermal imaging camera, 3—Mitutoyo SJ-210 roughness meter, 4—software of Mitutoyo SJ-210 roughness meter, 5—sound intensity meter, 6—vibration meter, 7—air nozzle.

#### *3.2. Numerical Simulations of Honing Process*

The honing cell should verify the influence of honing parameters on the size of the temperature increase, which is related to thermal deformation of the honed hole. The measurement should be carried out before machining, by performing a numerical simulation of honing.

Figure 6 shows the image recorded during the numerical simulation, showing the heat flux flow through the honed cylinder liner. The occurrence of different values of heat flux is clearly noticeable, depending on the thickness of the section of honed workpiece. The differences affect the non-uniform cylindrical deformation, more information is included in [5–8]. The simulation will provide information about the size of deformation for various possible variants of machining parameters, e.g., the amount of pressure of the abrasive whetstone on the workpiece.

The different temperature value of the honed workpieces causes the occurrence of different values of thermal stress in different places of the workpiece and affects the deformation of the cylinder shape, which is an undesirable phenomenon.

The honing cell should verify the influence of honing parameters on the amount of stresses and deformations occurring during honing in the workpiece. The measurement should be carried out before machining, by performing a numerical simulation of honing (Figure 7).

#### *3.3. Programming of the Grain Trajectories in Non-Conventional Way*

It was verified in [5] that the shape of the abrasive grain trajectories obtained in the honing process influenced the size of the temperature rise in the honed workpieces. In addition, abrasive grain trajectories can be generated using mathematical functions such as sin(x), cos(x) etc., which would allow the creation of a path of any shape, also curves with different radii of curvature.

A very interesting issue is the problem of programming the path of the abrasive grain, different one than the traditional path resultant from the rotational and linear speeds of the honing head, which is so far used in most honing machines manufactured by leading manufacturers. In the proposed approach, a curvilinear path can be selected and generated depending on the cross-sections of the honed workpiece. It is defined in the form of a mathematical formula that defines a curvilinear path of various shapes with variable shape parameters (radius of curvature, amplitude size, frequency of change of direction). This kind of non-conventional programming can be realized in the CAD/CAM system, and previously planned in *CCR* (curve curvature radius) module.

Figure 2 shows the window view from the CAD/CAM Alphacam software, where it is possible to define a curvilinear path by adding a circular vibration to the path of varying magnitude and frequency of change of abrasive grain move direction.

Changing the shape of the abrasive grain path from a circular path to a sinusoidal path. Most basic form of sine wave describing the time function t is:

$$\mathbf{y} = \mathbf{A}\sin(\omega\mathbf{t} + \mathbf{f})\tag{5}$$

where: A—amplitude, ω—pulsation in radians per second (closely related to the frequency in hertz), f—phase shift (if the phase is different from zero, the function graph looks shifted in time by 0 s).

**Figure 7.** Deformation of the honed workpieces—different value in different places. (**a**) view of the window from the simulator with the effect of the deformations obtained, (**b**) view of the window from the simulator with the presentation of non-linear deformation results of the honed workpiece.

The curves describing the trajectory of the abrasive grain may have various shapes, characterized by a different curvature of the grain trajectory [5]. The grain trajectory may take any shape, while performing which the honing head may more or less frequently change the direction of the axial movement (Figure 8). The quality of the hole made depends on the accuracy of the honing head movement [89,90].

**Figure 8.** Examples of different grain trajectory shapes, shown on the developed surface of machined hole, with different oscillation frequency, with different path length for the same length on horizontal direction of the treated surface. The function marked with digit 1 has one change of the head movement direction in the lower and upper turning point. The function marked with digit 2 has one change in the head movement direction in the lower turning point and two changes in the direction of the upper turning point.

Figure 9 shows the modification of the cylindrical path, which consists in changing the shape into a sinusoidal shape.

**Figure 9.** Programming in non-conventional way by adding oscillation frequency and amplitude's high to circular path generated in CAD/CAM system; 1—entering parameters, 2—circular path, 3—the resulting sinusoidal trajectory.

Figure 10 shows how to modify the shape of the machining path (item 2 from Figure 9) to a zig-zag shape path or a sinusoidal path (item 3 from Figure 9).


**Figure 10.** The method of modifying the shape of the machining path in CAD/CAM system Alphacam.

#### *3.4. Setting of Honing Parameters*

Automated honing cell should automatically setup the needing abrasive grain trajectories shape, as shown on Figure 9. Construction of honing head should allow for two-way steering of direction of abrasive whetstone and honing head movement (Figure 11) and should have possibilities to receive a machined surface images during process.

**Figure 11.** Idea of honing process with abilities to control the abrasive whetstone and honing head movements in both direction: 1—honing head body; 2—expanding pin for abrasive whetstones; 3—abrasive whetstone; 4—possible TWO-WAY direction of movement control; 5—automatic vision system.

An important task is to use appropriate honing parameters that influence the course of the process. Incorrect honing parameters result in an excessive temperature increase and may cause the whetstone wear out in a very short time. Figure 12 shows the whetstone: 1—new, 2—after machining with incorrect pressure at the beginning of honing process, 3—worn out in a very short time.

**Figure 12.** Machining tool: 1—new whetstone; 2—damaged whetstone; 3—worn whetstone.

It is advantageous to set machining parameters influencing uniform wear of the machining tool.

#### *3.5. Supervision of Honing during Process*

A very important issue for the automatic honing cell, in addition to the selection of machining parameters after the initial numerical simulation of the honing process, is to verify, supervise, and correct the obtained machining results in real time of honing process.

#### 3.5.1. Supervising of Surface Textures

Figure 13 shows the surface obtained after honing with a variable value of the linear feed of the honing head. Variable honing kinematics, unlike traditional kinematics, enables the creation of oil scratches with new, and never used earlier, shapes of abrasive grain trajectories. The literature clearly shows the advantages of honing with variable kinematics, which reduces the wear of the machining tool and with surfaces with lower roughness profile parameters, e.g., *Ra* and *Rpk*.

**Figure 13.** Obtained texture of the honed surface for variable stroke speed of honing head in the range of 1000–3000 mm/min—average value of tangent angle to the grain trajectory of 14◦; 1—sample abrasive grain path; 2—tangent line to the abrasive grain path.

Figures 14 and 15 shows examples of surfaces obtained after honing, on which, in addition to measuring the deviation in the shape of roundness and cylindricity and the parameters of the roughness profile, the quality of received texture of machined hole is checked.

**Figure 14.** Example of a honed surface without texture defects, 1—probe tip.

**Figure 15.** An example of a honed surface with texture defects: 1—scratch; 2—point flaw. Figure 14 shows the surface after honing, without texture defects.

Figure 15 shows the surface after honing, with texture defects in the form of: 1—scratches, 2—point heterogeneity. The quality of the obtained surface is determined by the homogeneity of the dimensions, shape, and texture of the obtained surface after honing process.

Figure 16 shows the verification of the obtained oil channel pattern in a schematic manner. The verification consists in checking whether the obtained oil channels are continuous, or whether they are clogged with fragments of the honed workpiece's material. The individual layers of neural networks check whether the image fragment shows an oil channel shape is a line, whether it is a curve, whether the shape is broken, or the break is caused by the intersection of oil channels or the presence of the workpiece material in the oil channel. In the event that the verification would confirm a significant number of oil channel breaks, the shape of the grain path should change, and the treatment should ensure the minimum number of places with a break for oil flow.

**Figure 16.** Verification of the shape and it's continuity of oil channels using a neural network. The numbers indicate the stages of the subsequent stages of surface texture verification.

#### 3.5.2. Analysis of Image of Honed Workpieces in Matlab's Image Processing Toolbox

Figure 17 shows a cone representing the HSV color description method. Each color has its own shade, brightness, and value by which the color can be defined.

At the beginning of the honing process, the workpiece temperature is equal to the ambient temperature. During processing, the temperature of the workpiece increases, which causes thermal deformations of the honed hole. The temperature rise of the item can be observed on-line through the Matlab Image Processing Toolbox module.

#### *3.6. Correcting of Honing Parameters*

The honing cell should include a CNC honing machine, a thermal infrared camera, a microphone, and modules of e.g., Matlab's software for on-line image and sound level spectrum analysis, an oil cooler and a *CCR* (curve curvature radius) module—the matching module of the shape of the abrasive grain trajectories received on developed surface of the machined workpieces to a certain thickness of the cross section of the honed workpiece.

Figure 18 shows possible actions to be performed during the honing process by automated honing cell: (a) FEM mesh overlay and working pressure setting, (b) determination of thermal conditions, (c) deformation, sound level and image analysis, (d) determination of honing parameters, (e) honing with variable kinematics setting with *CCR* module, (f) receiving variable shape of abrasive grain trajectories optimized to manufacturing conditions.

**Figure 17.** Cone showing the color description method named HSV.

**Figure 18.** Schematic diagram of honing algorithm: (**a**) FEM mesh overlay and working pressure setting, (**b**) determination of thermal conditions, (**c**) deformation, sound level and image analysis, (**d**) determination of honing parameters, (**e**) honing with variable kinematics setting with *CCR* module, (**f**) receiving variable shape of abrasive grain trajectories optimized to manufacturing conditions.

Figure 19 shows schematically the process parameters verified on-line during the treatment. Depending on the values of the obtained parameters, the values of the machining parameters would be corrected automatically.

**Figure 19.** Schematic diagram of automated honing process.

The main factor influencing the differentiation of efficiency of manufacturing is the lack of the need to multiple cool the honed workpieces before the end of the honing process, and lack of the multiple measurements of the obtained diameter and shape deviation of honed hole.

#### **4. Results**

Performing numerical simulations makes it easier to plan the machining process. Owing to the simulation, we can find out the size of deformations and stresses occurring during honing, which allows us to decide on the selection of the right machining parameters.

Owing to the sound signal level analysis and thermal image of honed workpieces analysis one can establish the needed parameters of the honing process.

#### *4.1. Stresses and Deformations in Machined Workpieces*

In Figure 20 digit 1 indicates the location of the stress measurement, that the honing cell could monitor online. A honing head with the ability to measure the amount of pressure of the whetstone on the treated surface would allow for the verification of the shape deviation of the honed hole.

**Figure 20.** Results of numerical analysis of honing process of cylinder linear with different thickness of cross-section; (**a**) simulation result of the entire assembly, (**b**) assembly simulation result without cylinder linear, (**c**) assembly simulation result without cylinder linear and without honing head body, (**d**) simulation of stresses obtained during honing process.

Figure 21 shows the numerical inhomogeneous simulation values of the deformation of the honed hole obtained during honing of thin-walled workpiece with a variable wall thickness, which shows the actual manufacturing difficulties of this type of workpieces.

**Figure 21.** Numerical simulation—deformation of a cylinder with a variable wall thickness: 1—the greatest cylindrical deformation value of honed workpiece; 2—the smallest cylindrical deformation value of honed workpiece.

#### *4.2. Analysis of Sound Signal Level, Received During Honing Process, in Matlab's Audio Toolbox*

Figure 22 shows a graph of the intensity of the sound signal, which was obtained during honing with varying kinematics. Measured sound was introduced into the Matlab Audio Toolbox. Distance from peak to peak in horizontal direction, in the diagram shown in Figure 22, indicates the time of the honing cycle (movement of the striking head up and down). Similarly, to Buj-Corral I. [2], it has been verified that the choice of honing parameters is reflected by the amount of sound signal emission (Figure 22), which means that the acoustic signal analysis is a good tool for verifying the honing process. Figure 23 shows the test stand with the equipment for measuring the sound intensity level.

**Figure 22.** Matlab Audio Labeler—analysis of sound level vs honing process time conducted on CNC vertical milling machine Haas VF-3SS. Honing with variable kinematics condition, with different value of honing head stroke speed; 1—shorter honing cycle time, 2—longer honing cycle time, 3—honed workpiece, 4—honing equipment.

Figure 24 shows the sound pressure level for different mean values of the variable stroke feed of the honing head. Figure 24 shows that the lower value of the sound intensity level is obtained for the mean value range of the variable stroke feed of the honing head.

**Figure 23.** Test stand: 1 equipment for measuring of the sound intensity level.

**Figure 24.** Sound intensity level depending on the average value of the variable head feed, obtained during honing with variable kinematics.

A lower sound level [dB] value means the occurrence of lower cutting forces, associated with the occurrence of lower cutting forces for the middle range of the applied feed.

#### *4.3. Analysis of Image of Honed Workpieces in Matlab's Image Processing Toolbox*

Figures 25 and 26 show the temperatures of the processed object obtained during honing. Using the HSV description method, it was proposed to analyze the temperature distribution over the surface of the entire object. Instead of checking the highest temperature to which the honed object has heated up, an analysis of the uniformity of the temperature of the entire object was proposed. As shown in Figure 25, the temperatures analyzed in the Matlab Image Processing Toolbox module (value and degree of heating of the object) are verified using the Hue, Value, and Saturation parameter.

**Figure 25.** Matlab Image Colour Thresholder—analysis of thermogram of honed workpiece on CNC vertical milling machine Haas VF-3SS (earlier stage of honing (than on Figure 26). H—hue, S—Saturation, and V—Value (HSV).

It can be clearly noticed that in Figure 26, i.e., on the workpiece after processing (heated from honing), the H parameters shift to the right toward the red shade, which means that the workpiece has heated up (the recorded thermogram shows a higher temperature than at the beginning of the treatment). The value and saturation of the color, the *S* and *V* parameters, respectively, also shift to the right, which suggests an increase in the temperature value on a larger surface of the object to be polished.

The task of the automatic cell is to monitor the degree of heating of the object, while the *H* parameter takes the appropriate color value, the *S* and *V* parameters respectively will confirm that the object has been heated evenly over the entire surface.

As shown in Figure 26 the hotter the temperature of honed workpiece, the more bright red color. Figure 26 also shows the shifting of the amount of the measured light color to the right, i.e., toward the lighter colors, by means of arrows.

**Figure 26.** Matlab Image Colour Thresholder—analysis of thermogram of honed part on CNC vertical milling machines Haas VF-3SS (later stage of honing process than shown on Figure 25). H—hue, S—Saturation, and V—Value (HSV).

The analysis of the thermogram, in contrast to the analysis of only the maximum temperature of the workpiece, can provide information about the local temperature increase of the honed workpiece during machining, which provides comprehensive knowledge about the temperature rise in the workpiece for different cross-section thicknesses and for different places on the honed hole.

Figure 27 shows a window view from the FlexSim 2020 simulator in which a simulation of honing on an automated cell was prepared and compared to conventional honing. The automatic cell enables more than 20 times faster production than in the conventional way.

**Figure 27.** Window from the FlexSim 2020 simulator—comparison of production on a conventional station and on an automatic cell; 1—automated honing cell (efficiency of automated honing cell: 11.6 workpieces/h), 2—conventional honing cell (efficiency of traditional honing: 0.5 workpiece/h). 3 and 4—FlexSim 2020 simulation's algorithms of honing process in Process Flow.

#### **5. Discussion**

Building a test stand with all components listed in the article will allow to produce workpieces of complex shapes with different wall thickness and required specifications. The proposed honing cell consists of CNC honing center with honing equipment, thermographic camera, profilographometer, sound intensity measuring device, software for a thermographic camera, software for a roughness gauge, software for a sound level meter, cooling nozzle, measuring instruments—a diaphragm gauge, and computers enabling simultaneous observation of changes of data for individual instruments could enable the automation of honing process and the reduction of manufacturing time.

When changing the machining parameters and when switching the honing process on and off, individual devices had to be turned on, set, turned off, and properly turned on and off for collecting data in a dedicated software. After setting up the tooling, it was necessary to upload the appropriate program to the CNC machine, turn on individual devices and their dedicated software.

A very advantageous solution would be to create a honing cell with all listed elements implemented in honing machine and used simultaneously during honing.

The automatic honing cell could monitor on-line the temperature of the honed workpiece, the pressure of the whetstones obtained during machining, the sound level of honing process, and depending on the honing process conditions, it could automatically, using the *CCR* module, create an abrasive grain path shape with an appropriate radius of curve curvature to improve machining conditions. An important conclusion is the possibility of about a 20-fold increase in the efficiency of serial production of honing thin-walled objects

#### **6. Conclusions**

Performing a simulation of machining before the honing process enables the selection of machining parameters before the process begins, it would be particularly valuable to program the shape of the grain trajectory with a specific value of the curve radius adjusted to the honed workpieces.

The measurement of the sound intensity in the honing process is a valuable source of information because the value of the sound level is directly related to the processing conditions, which is easily noticeable in the way that the lower sound level corresponds to the lower cutting forces occurring at a given processing time.

The temperature increase of the honed workpieces can be monitored during the honing process, the advantage of the HSV method is the fact that we analyze the temperature distribution over the entire smoothed surface, and not only collect information about the maximum temperature obtained, this method has many advantages, mainly that it can control the temperature distribution in such a way that the honing process heats the workpiece uniformly.

Supervision and control of the honing process enables about twenty-fold reduction of the time needed to perform honing machining of thin-walled objects and of variable thicknesses of workpieces, because of the elimination of the need to divide the processing into several stages, followed by cooling to the ambient temperature, in which diameter of the honed hole can be measured.

The analysis of the surface texture obtained during honing process, with the use of the neural networks, allows for quick verification whether the obtained surface has a uniform texture shape of the obtained oil channels or whether additional corrective machining passes of honing head are required.

**Author Contributions:** Conceptualization, P.S.; methodology, P.S.; software, P.S.; validation, A.B.; formal analysis, A.B.; investigation, P.S.; resources, P.S.; data curation, P.S.; writing—original draft preparation, P.S.; writing—review and editing, A.B.; visualization, P.S.; supervision, A.B.; project administration, A.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors thank for co-financing the Mazovia/0127/19 project by the National Center for Research and Development.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
