A Review on Metallic Drilling Burrs: Geometry, Formation, and Effect on the Mechanical Strength of Metallic Assemblies

A burr is generally defined as any unwanted residual material resulting from a manufacturing operation that lies outside the ideal geometrical shape of the workpiece [1]. All machining processes such as milling, drilling, turning, or grinding produce burrs on the free edges. Of special importance is the burr formed in the drilling process. The reason for this is that drilling is the most frequent machining process, accounting for around 35% of all machining operations [2,3]. In the aeronautical industry, around 300,000 holes are drilled in the structure of a fighter jet, and between 1.5 and 3 million for a commercial aircraft [4]. The standard assembly process used in the aerospace industry for mechanically fastened joints is illustrated in Fig. 1. An alternative to the traditional drilling and assembly process is the so-called One-Shot or One-Way-Assembly. This designates the process (often machine-based) where fastener holes are drilled upon assembly without the joint being separated, cleaned, and holes deburred after drilling, as would be the norm under a more traditional process [5]. The advantages of One-Way-Assembly involve the reduction of cycle time and cost associated with the deburring operation. The commercial aircraft industry acknowledges these advantages (for example, Refs. [5,6] from Boeing, Ref. [7] from Airbus). In the military aircraft industry, it is also recognized that it should be the process of choice for the mass production of the next generation of fighter aircraft [8]. The One-Way-Assembly strategy poses however an industrial challenge. By not separating the drilled parts, there is a high risk of interface burrs entrapment. The nonremoval of these burrs has an experimentally demonstrated detrimental effect on the static and fatigue strength of the assembled parts [5,9,10]. To reduce the risk of burr formation, a better control of the clamping conditions to limit the gap between the parts is required [6]. Nondestructive inspection methods capable of assessing the existence of small-size contamination on nonaccessible surfaces are also necessary for quality assurance. Furthermore, an engineering capability to assess the impact on the mechanical strength of structures affected by the presence of these types of defects is still lacking today.

There is an extensive literature on burrs. The earliest systematic research was carried out in the 1970s, with Gillespie and Blotter's [11] innovative work on the qualitative description and mechanical modeling of burr formation for different manufacturing processes. The first geometrical classifications and descriptions of burrs [12], as well as the first systematic studies on deburring [13–17], were also carried out at this time. Between the years 1980 and 2000, a significant number of analytical models of burr formation were developed, leading to a good understanding of the basic underlying mechanisms. Examples can be found in the work by Ko and Dornfeld [18] and Chern [19]. The aim of these studies was to understand the influence of manufacturing parameters to optimize the process and minimize burr formation. From 2000 to the present, numerous theoretical and experimental studies have been published. The first studies on drilling burr formation for stacked materials belong to this period. Again, the focus of these publications was on the optimization of manufacturing parameters to reduce the burr size. There is a significant gap in the literature regarding the assessment of the effect of burrs on the mechanical behavior of structures. It is generally assumed that the effect is detrimental, both for static and fatigue. Few publications have attempted to quantify it or to physically explain the reason for the negative impact.

In this article, a synthesis of existing literature on drilling burrs is presented. It covers relevant studies performed from the 1970s until the year 2024. It deals with burr description, formation mechanics, and modeling and with burr minimization and deburring techniques, as was done in previous review works [20–22]. In addition, the relevant studies regarding the effect of burrs on the static and fatigue strength of mechanically fastened joints are synthesized. A systematic extraction and analysis of literature static and fatigue test data has been performed. In the following, a geometrical description of drilling burrs and of the methods that are used for their experimental and industrial characterization is made in Sec. 2. The mechanisms of drilling burr formation are reviewed in Sec. 3. A synthesis of theoretical models is then given in Sec. 4. The influence of relevant parameters on burr size is reviewed in Sec. 5. The different burr control strategies and the most common deburring techniques are examined in Sec. 6. The limited literature on the effects of burrs on the static and fatigue strength of mechanical assemblies is reviewed in Sec. 7. In the last section, a discussion with conclusions is proposed.

There is no unique definition of drilling burrs or a universally accepted standard on how they should be described. The ISO 13715 standard [1], which is dedicated to edge quality, defines a burr as rough residual material outside the ideal geometrical shape of an external edge, resulting from a machining or forming process. This definition applies to burrs from all manufacturing operations. This standard also gives a geometrical definition of burr size, denoted in Fig. 2. An attempt to standardize definitions related to edge quality can be found in Ref. [23]. The definition from the ISO 13715 has been widely adopted [5,11,24–28]. It is often complemented by the idea that a burr is generated by a plastic flow of material [18,29–33]. Thus, the following definition is proposed for drilling burrs: rough residual material outside the ideal hole edge, formed as a result of plastic flow from the drilling operation.

Note also that the term burr is sometimes used for drilling defects in composite materials. Any measurable excess of material left on the product is called a burr [34]. Composite burrs will not be covered hereafter.

Classifying drilling burrs helps to understand the formation mechanisms, to select a suitable deburring technique, and to assess the potential effects if burrs are not removed. Three main criteria can be found in the literature to classify burrs: (i) location within the workpiece, (ii) morphology, and (iii) formation mechanism. A synthesis of the existing classifications is given in Table 1.

Regarding drilling burr location, entry and exit burrs are distinguished [20]. Typically exit burrs are larger, and therefore more difficult to remove. The formation mechanisms are different for entry and exit burrs [15]. Also, when drilling simultaneously more than one plate, burrs may develop at the interface, at both mating surfaces [35]. Independently of the location, drilling burr morphology is diverse and often complex and irregular. Multiple classifications regarding morphology exist, which focus on shape, uniformity, and the presence of a cap attached to the hole. The classification of burrs according to the formation mechanism was first proposed by Gillespie and Blotter [11]. Three main mechanisms are identified, which are reviewed in detail in Sec. 3. This classification is general to all types of machining. Different processes promote one type of burr or another. In practice, more than one mechanism is involved in drilling burr formation at the same time [11].

Drilling burr shapes are often complex and irregular. A simplified geometrical idealization is necessary for qualitative and quantitative analysis. Drilling burr geometry can be described by a small number of simple measurable parameters. Schäfer's parameterization of a typical burr profile [20] is given in Fig. 3. A profile corresponds to the projection of the burr onto a radial plane. Four parameters are used: burr height h, burr thickness t, burr root radius r_r, and burr tip thickness th. There are some experimental studies that make direct use of this parameterization [46–48]. However, it is difficult to obtain experimentally the four parameters with sufficient accuracy. That is why only two are often used: the height and the thickness. Moreover, in many cases, only the burr height is retained.

There is no universal definition of burr height and thickness. In fact, most studies that report values for burr height and/or thickness do not formally define these two parameters. According to Aurich et al. [20], burr height is defined as the distance between the nominal edge of the hole and the highest point in the cross-sectional area. This definition simplifies the measurement procedure. It is the one applied in practice in most of the studies that can be found in the literature. Aurich et al. [20] define thickness based on Schafer's idealization as the width parallel to the burr root at a distance equal to the radius of the tangent circle, as measured in the cross-section. The experimental measurement of this thickness is not straightforward. Most studies use root thickness as burr thickness.

There is circumferential variability of the burr profile. This means that burr height and thickness depend on the radial plane in which the profile is extracted. Only a limited number of studies deal with the statistics of drilling burr geometrical variability [49,50]. Most studies report a single value for burr height and thickness, which is computed as the maximum [25,36,51–55], the average [28,33,37,38,41,42,44,56–70], or the median [12] of a small number of measurements along the circumference of the burr. Four measurements separated by 90 deg are often taken.

To perform a geometrical description of a drilling burr, experimental measurements must be carried out. A wide variety of techniques are available [20]. A synthesis of the main technologies is given in Table 2. This synthesis is not intended to be exhaustive. For more details on burr measurement technologies, see Refs. [20,97]. A choice of technique must be made according to the desired level of accuracy, the parameters needed to be measured, the type and size of the burr, and its location and accessibility. Mechanical systems, such as dial indicators and profilometers, are the most commonly used technologies, due to their simplicity. They are applicable both in industrial and research contexts. Metallographic cuts and microscope observations are also a common choice for research purposes, due to their accuracy.

The analysis of the physics underlying drilling burr formation is derived from the studies on the mechanics of cutting and chip formation. On the basis of an extensive experimental study involving different machining processes, Gillespie and Blotter [11] isolate three basic mechanisms of burr formation: (i) lateral extrusion, (ii) plastic bending, and (iii) tearing.

The first mechanism is the lateral extrusion of the workpiece material. Indeed, the contact between the tool and the workpiece generates local compression of the material, which causes bulging on the sides of the tool as a consequence of the Poisson effect. Local compression can be due to pressure in effective cutting edge radius, pressure on the tool’s lateral surface, or an initial indentation of the tool [15]. As part of the material undergoes plastic deformation under these mechanical actions, to comply with the plastic incompressibility constraint, part of the material will continue to be pushed outwards and will not retract when the tool is removed, resulting in a burr. Such burrs, due to the importance of the Poisson effect in their formation, are called Poisson burrs. The second mechanism of burr formation is the plastic bending of the machining chip. In this case, when less energy is required to bend the chip than to shear it, the chip will undergo a state of plastic bending that will result in a rotation of material. This results in a so-called rollover burr. The third mechanism of burr formation is the tearing of material, due to plastic flow in shear. Two failure modes are distinguished, bending and pure shearing [11]. This is similar to the burr produced in the punching process [98]. This mechanism results in a tear burr. Gillespie and Blotter identify three additional mechanisms that lead to an excess of material, which according to certain definitions [1] could be burrs. Thus, redeposition of material in casting, incomplete cut-off, or flow of material in cracks in molding processes is identified as burrs. These types of burrs are not considered hereafter.

The three mechanisms identified by Gillespie and Blotter are helpful in understanding the physics behind burr formation. They are general to all manufacturing processes. A classification according to these formation mechanisms is presented in Table 1. The distinction of the three mechanisms has helped in the formulation of predictive models of burr size for different manufacturing processes, such as milling [11,99], drilling [15], turning, and grinding [11]. It was hypothesized that for each manufacturing process, only one mechanism is actively contributing. This approach is used in the following to explain entry burrs formation. However, isolating a single mechanism is not entirely valid, as more than one mechanism can often act simultaneously. Studies on exit and interface burrs formation focus more on the identification of a sequence of formation rather than on the isolation of a physical mechanism.

The study and modeling of burr formation are experimentally supported by observation, either during or after drilling. In-process methods include the use of high-speed photography [42], the measurement of temperature fields with infrared cameras [100], and drilling force/torque monitoring [85]. Out-of-process observation techniques include metallographic cuts [62,101]. Metallographic cuts along radial planes allow microstructural observations to be carried out. Analysis of grain size and orientation and microhardness measurements are often performed [62,71].

The formation of entry burrs during drilling has received limited attention in the relevant literature. The most likely reason why most studies focus on exit burrs is that they are typically larger, and therefore more difficult to remove. Two mechanisms have been identified to produce entry burrs: (i) lateral extrusion of material and (ii) vertical tearing due to upward chip motion. These two mechanisms have been illustrated in Fig. 4. The mechanism by lateral extrusion acts at the drill margins at the point where the tool first fully penetrates the workpiece. The importance of this mechanism becomes greater the more worn the flanks of the tool are [15]. The tearing mechanism acts from the beginning of drilling [11,15,102]. Indeed, the chip starts to form as soon as the cutting lips enter the workpiece. When moving upwards at the flanks of the drill, the chip is still attached to the edge of the hole. The drill geometry forces the chip upwards. This generates a state of vertical shear on the material in the vicinity of the hole edge, which will induce an upward plastic deformation. Finally, when the chip is torn away from the rest of the workpiece, it leaves a residual material which is the entry burr.

Unlike the entry burr, the formation of exit burrs has been extensively documented in the relevant literature. The sequence of cutting and burr formation is clearly identified [103,104]. It is illustrated in Fig. 5. The process has been divided into six stages.

The first stage (Fig. 5(a)) is steady-state cutting. This stage starts shortly after the drill fully penetrates the material. It is a complex cutting process, with variable local conditions at different points of the tool. Typically, the chisel edge action is identified as indentation, with workpiece material being pushed out to the cutting lips by the Poisson effect. Cutting at the extremities of the chisel edge and at the flanks resembles locally oblique cutting with varying cutting angles, speeds, and forces [105]. During steady-state cutting, a zone of elastic compression exists ahead and below the drill bit point, which extends to the exit surface over the entire workpiece. In the area close to the cutting edge plasticity conditions are attained.

The second stage of the burr formation process (Fig. 5(b)) is called initiation and begins when plasticity is attained at the bottom surface. An observable bulge develops at this surface, which grows as the tool advances. This plastic bulging can be seen in tests such as those performed by Patil et al. [42].

In the third stage (Fig. 5(c)), called the development stage, the zone affected by vertical compression and shear moves toward the surface of the hole, while the plastic deformation or plastic bulge at the exit surface enlarges.

The fourth and final stage (Fig. 5(d)) consists of the final formation of the burr on the exit surface. Three sub-stages are identified in the literature [4,42,62,106–108]. First, under the action of the thrust force, failure of the material below the drill bit point occurs, resulting in drill bit breakthrough through the bottom surface (Fig. 5(d.1)). This results in the formation of a cap, which may either detach immediately or become attached at some point on the circumference of the remaining material. Drill breakthrough is followed by plastic bending of the remaining material, resulting in a rotation around a plastic hinge located in the vicinity of the hole edge at the exit surface (Fig. 5(d.2)). The rotation of the material is accommodated by a plastic elongation of the material near the hole edge, in a shear mechanism similar to that described for entry burrs [11,102]. Third and finally, it is possible that a final fracture may occur at drill edges, with the remaining material being the burr (Fig. 5(d.3)). This fracture may or may not occur. If it does not occur, it will promote the formation of a more irregular, crown-like burr [37].

The formation of a drilling exit burr is a complex mechanical process, in which the main driver is the thrust force due to the interaction between the drill bit and the workpiece. Increased ductility, which is also favored by the local high temperatures due to the heat released by friction and plastic deformation, promotes the formation of larger and more irregular burrs. A metallographic cross-section showing a significant microstructural difference between the normal workpiece material and the burr material has been presented by Sofronas [62]. A smaller grain size can be observed in and close to the burr, due to strain hardening. Indeed, microhardness measurements performed on this burr show that the burr material has undergone significant plastic strain hardening [62]. A similar study can be found in Ref. [71]. In Sofronas's [62] study, a different grain orientation can be observed around the burr, which may be a source of different local mechanical properties. Shiozaki et al. [98] show that the residual stress state is complex in this region.

Note that in industrial applications, multistage drilling is a common practice, which consists of successive enlargements of the hole up to the nominal diameter [109], with a final reaming [15]. This generally results in the partial removal of drilling burrs and in the modification of the local material conditions described above.

Burr formation at the interface of two plates involves the development of one exit and one entry burr. These burrs appear whenever at least one of the stacks of the drilled element is metallic. Interface burrs have received attention in the literature, especially for metallic stacks [6,9,51,72,73,110–112]. Composite-metal hybrid stacks have received less attention in this regard, as they are less common [29,74,102,113].

The mechanisms behind interface burrs formation are essentially the same as those of free surface burrs [102]. The fundamental difference lies in the restriction to development due to a nearby surface, which favors the lateral bulging of material [72,102]. It is widely recognized that the most important factor in interface burrs formation is the gap between the surfaces [6,9,51,53,75,110–112]. Ideally, the distance between the mating surfaces to be drilled should be zero at all points. In reality, however, a gap always exists. Sometimes there is a predrilling gap in the area of the hole. This is due to the fact that perfect positioning of the parts is impossible, the parts are not perfectly parallel, the clamps are far from the drilling area, and/or the clamping force is insufficient [6,33,60,114]. Moreover, this natural gap, which can be negligible, is aggravated by the separation of the plates due to the thrust force during the drilling process [75]. Three successive stages can be distinguished in interface burrs formation. They are linked to the development of a gap between the mating surfaces. These stages are illustrated in Fig. 6. An initial gap has been assumed to exist between the plates.

The first stage (Fig. 6(a)) is the steady-state cutting of the upper plate. The thrust force produces the elastic bending of the two surfaces. The second stage starts when the drill breaks through the bottom of the first surface (Fig. 6(b)). At this point, the first plate experiences a spring back. At the same time, the thrust force applied on the second plate pushes the plates apart. This further increases the gap between the plates, leaving more space for burr development. There is also a risk of chip entrapment between the plates [75]. The third stage (Fig. 6(c)) consists of the formation of an exit burr on the bottom surface of the first plate. This happens as described above for free surface burrs, but under the restriction due to the presence of a backup plate. At this same point, an entry burr on the second plate may form, also under the restriction due to the presence of the other surface. It is also possible that the exit burr generated on the first plate penetrates into the material of the backup plate, as illustrated in Fig. 7. This behavior becomes more likely when the second plate has a small transverse strength compared to the first plate. It is the case of metallic-composite drilling [102]. As the drill advances into the second plate, the gap created due to the thrust force is reduced, and plates tend to spring back to the original position, generating a crushing force on the burrs. The final geometry of the burrs, as well as the local mechanical properties, will be affected by both the developing process and the final crushing.

The development of burr formation models, with qualitative and quantitative predictive capabilities, is of direct industrial and research interest. Burr formation models provide a better understanding of the physics of burr formation. Indeed, an experimental study is sometimes insufficient or even impossible. Representative models also allow to evaluate the influence of parameters without the need for long and costly experimental campaigns. Furthermore, being able to predict burr size or the postdrilling state of the material allows the development of burr control strategies for burr minimization and gives an idea of the potential effects of not removing them.

Drilling burr formation is a complex process, which raises numerous modeling difficulties [115]. The modeling complexity comes simultaneously from different sources. First, it is a problem with large displacements and rotations, involving geometrical nonlinearities. Second, the workpiece material experiences large deformations and strain rates, as well as high temperatures that affect its mechanical behavior. This viscoelastoplastic behavior with thermal effects is further complexified by damage initiation and propagation phenomena leading to chip and burr development. Third, the contact conditions between the tool and workpiece are complex, with small-scale geometrical details such as tool wear or material adhesion [116] having a strong influence. The interaction of the already generated chip with the rest of the workpiece can be equally important in some cases. Fourth, the use of lubrication affecting the contact condition and heat transfer process adds significant complexity. Finally, the boundary conditions of the drilling process, such as the clamping force or the distance to the clamping supports, can have a decisive influence on the burr formation process and must be properly modeled.

Different modeling approaches can be found in the literature [20]. The most common one is that of mechanistic models. They aim to represent the physics of drilling burr formation, more or less exhaustively. Two different types of mechanistic models can be distinguished. On the one hand, there are analytical models. In this review, analytical models are defined as those that are derived from a simplified analytical formulation, based on the mechanics of materials, of the burr formation processes. These models, although named analytical often require the application of some numerical resolution techniques. On the other hand, finite element models (FEM) have also been applied to the study of drilling burr formation [97]. These models make use of explicit schemes [117]. An alternative to the mechanistic approach is data-driven modeling. It consists in exploiting a database of experimental results and range from linear regressions to complex machine learning models. Data-driven models are blind in the sense that they do not capture the physical phenomenology of burr formation and are highly dependent on the dataset used in their calibration. This article does not further elaborate on data-driven models. For the reader's interest, examples can be found in Refs. [4,24,26,28,42,56,58,65,66,68,69,76,86–88,106], from which the experimental trends observed in the data used for their calibration are synthesized in Sec. 5. The following is a review of mechanistic burr formation models.

There is very limited reported experience in analytical modeling of drilling entry burr formation. Gillespie [15] establishes a model for Poisson burrs but does not apply it to the prediction of drilling entry burrs. Abdelhafeez et al. [102], for whom the formation mechanism is pure shear in the vertical direction due to chip upward motion, developed a simplified model based on an energy balance approach. Based on a simplified geometrical description and assumptions about the viscoelastoplastic behavior, the point at which the strain energy is equal to a shear failure energy is determined. The approach proposed by Abdelhafeez et al. is illustrated in Fig. 8. The model has been validated against experimental results for aluminum and titanium alloys and for different drilling conditions. The predicted burr heights and thicknesses showed relative errors below 20% with respect to experimental measurements. Publications on numerical simulations of entry burrs are scarce. Choi et al. [9] develop a Lagrangian FEM model in which the formation of entry burrs is observed. Results are however not validated against experimental measurements, and neither a detailed study of the formation mechanisms is made. Similarly, Efstathiou et al. [36] present an arbitrary Lagrangian Eulerian (ALE) model in which the formation of entry burrs is observed. Abdelhafeez [71] develops a coupled Eulerian Lagrangian (CEL) model for burr prediction in aluminum and titanium. The model shows the formation of entry burrs for various feed and cutting speed conditions and for three different materials, two aluminum and one titanium. The entry burr height prediction is validated against experimental results, with relative errors between 6% and 17%.

The modeling of exit burrs has received extensive attention in the literature. Mechanistic models based on analytical formulation discriminate between the successive stages that are illustrated in Fig. 5. The first three stages, steady-state cutting, initiation, and development, are usually addressed together. Models focus on predicting the thrust force at these stages as a function of drilling conditions and tool geometry. The thrust force is considered to be the main driver of burr formation and is used in the modeling of later stages. For the calculation of this force, the tool tip is usually divided into three different regions, as in Fig. 9. The chisel edge comprises two of these regions. In region 1, near the center of the chisel edge, the mechanical action of the drill bit on the workpiece consists essentially of indentation [106,107]. In regions 2 and 3, which are sometimes treated together, cutting is performed under oblique cutting conditions, i.e., with the cutting edge not perpendicular to the relative speed between the cutting edge and the workpiece. It is common practice to treat such cutting conditions as a superposition of local orthogonal cutting with varying cutting angles and speeds [118,119]. Sofronas [62], who developed the first analytical model of exit burr formation, assumes that the entire thrust force is due to action in regions 2 and 3. The force is calculated by assuming the same orthogonal cutting conditions along all cutting lips, with the same dynamic cutting angles, making use of previous work done by Williams [120]. Kim and Dornfeld [119] make the same assumption, although the cutting lips are discretized to account for varying cutting conditions. Local cutting forces are calculated on the basis of Merchant's orthogonal cutting theory [121]. Zai et al. [122] apply a similar model for the study of micro-drilling. Rana et al. [108] also model the thermal effects due to local heating, causing the flow stress of the material to decrease with increasing temperature. This is used to correct the calculated thrust force. Mandra et al. [89] include the effect of tool wear by adding to the thrust force due to cutting a contribution in the form of a plowing force. On the other hand, Lauderbaugh [106], Saunders and Mauch [107], and Li et al. [4] include an indentation force model for region 1 in the central area of the chisel edge. Thermal effects are also taken into account by solving a 1D heat conduction problem to locally modify the mechanical strength of the material by temperature effects.

The thrust force during steady-state cutting is the essential input for the study of burr formation stages. For the study of the three latter stages, two types of approaches have been reported. The first approach calculates the position of the tool at the moment of chisel edge breakthrough, to determine a geometry to which a rotation with deformation can later be applied. This approach is illustrated in Fig. 10. This approach is followed by Sofronas [62], who determines the chisel edge breakthrough point as the position at which the mechanical power supplied by the tool is equal to the power required for the shear failure of the material under the drill bit point. Once the initiation point has been determined, the final geometry of the burr is calculated using the incompressibility hypothesis of the remaining material. In particular, the maximum height and the thickness at the base of the final burr are determined. A similar approach is followed by Lauderbaugh [106] and Saunders and Mauch [107]. Their approach consists of calculating the stresses with a variable thickness plate model from Ref. [123] until stresses are higher than the ultimate tensile strength of the material. The variable thickness plate represents the material under the tip of the drill bit, subjected to the loads due to the thrust force. Thermal effects are also taken into account in the model by modeling the heat released in the process and a 1D modeling of heat conduction, as well as the effect of temperature on the ultimate tensile strength of the material. Once the position of the tool at the breakthrough point is determined, the final geometry of the burr is also determined using the same plate bending model. A possible second fracture is also taken into account in the model. The second type of approach for the final stages of burr formation is based on the principle of energy balance. The breakthrough of the tool is not modeled. The underlying assumption of these models is that the mechanical power during the entire burr-forming process is constant. This work includes the work due to thrust force and the work due to deformation leading to burr formation. According to Kim and Dornfeld [119], the thrust force work is determined from the steady-state phase, while the deformation work comprises the work due to the elongation of the material as well as to the plastic bending associated with the rotation of the material at the hole edge. These energies are expressed as functions of the final burr geometry. The conservation of power equation allows to obtain explicit expressions for the height and thickness of the burr. The same approach is followed by Li et al. [4], Mandra et al. [89], and Zai et al. [122], with slight variations in the calculation of the energies to include the effects of vibratory drilling and tool wear. Segonds et al. [124] provide an analytical model to establish the burr type from the tool geometry and drilling conditions. It does not, however, predict the height and thickness of the final burr. This model is based on a slip planes theory and the application of the equilibrium equation on a sheared volume.

Most of the analytical models mentioned above have been validated against experimental results. However, a quantification of model accuracy is often not available. In this case, when experimental data and model predictions are provided graphically, an estimation made by the authors of the root mean square error is given. Kim and Dornfeld [119] present a validation of the effect of feed rate on burr height and thickness, for different diameters and two different steels. An average prediction error of 40% for burr height and 45% for burr thickness is obtained. Li et al. [4] validate their model against aluminum tests, with an average discrepancy of 11% on the effect of feed on burr height. Lauderbaugh [106] and Saunders and Mauch [107] both study the feed rate effect for aluminum, with average prediction errors below 45%. The thermal aspects of their models are validated using a finite element model [125]. Mandra et al. [89] validate their model with tool wear effects by means of experimental observations, with an average error of 19%.

Regarding numerical simulation of exit burr formation, Guo and Dornfeld [103] develop a Lagrangian model with variable ductile damage initiation and evolution criteria, adapted to the different stages of burr formation. This is equivalent to the work done by Park and Dornfeld [126,127] for 2D orthogonal cutting. Min et al. [104] also present a Lagrangian model which qualitatively shows good agreement with experimental observations on the exit surface. Min et al. [27] also present a model capable of predicting the thrust force of the drilling process, with an average error of 60% for different feed rate values. The prediction is shown to be close to experimental results during steady-state cutting, but not during the burr formation stage. Isbilir and Ghassemieh [128] establish another Lagrangian model for the simulation of burr formation in titanium. The model has prediction errors below 20% for thrust and torque measurements. However, burr height is underestimated between 50% and 75%, due to the lack of thermal modeling. Abdelhafeez et al. [101] develop another Lagrangian model of drilling in aluminum and titanium. Thrust force and torque are predicted with 8 and 56% error, respectively. Finally, Gajrani et al. [26] also present a Lagrangian model for the prediction of burr height and cutting forces and torque for titanium, with a 6.5% discrepancy between the predicted thrust force and experimental measures. Non-Lagrangian models have also been reported in the literature. Efstathiou et al. [36] develop an ALE model for burr formation in aluminum. The model is validated qualitatively and quantitatively by observation and optical measurement of the drilling exit surface. Abdelhafeez et al. [101] present a CEL model. Abdelhafeez et al. provide an error value for thrust force and torque of 19% and 22%, respectively. Burr height is predicted with an error between 9% and 17%.

Most studies on interface burrs modeling focus on determining the gap created between two plates during the drilling process, and the influence of process parameters such as clamping force or clamping distance to the hole. Jie [53] presents a model based on beam theory to determine the gap due to the action of the thrust force under the constraint of clamping pressure at a certain distance from the hole. Hu et al. [111] develop a similar model, replacing the beam with an axisymmetric plate model. Tian et al. [114] and Yin et al. [64] do the same for a rectangular plate. These models are able to quantitatively predict the gap developed between the two plates, but the relationship between interlayer gap and interface burr size is not rigorously established. Hu et al. [111] show a 35% error between gap prediction and experimental measures of interlayer burr height. These results are shown in Fig. 11, where it can also be seen that the exit burr is significantly larger than the entrance burr at the interface. It is then demonstrated that assuming that the interlayer burr size is equal to the total burr height at the interface is a reasonable approach.

There are also analytical models capable of predicting the interface burr in a straightforward way, without going through an interlayer gap calculation. In this regard, Hellstern [72] presents a modification of the Sofronas [62] model to account for the resistance to the thrust of the backup plate. Similarly, Abdelhafeez et al. [102] apply their entry burr model to the prediction of interface burrs in hybrid assembly drilling. The presence of a composite plate is taken into account by adding a strain energy representative of its resistance to transverse deformation.

Regarding numerical models, Choi and Min [110] develop a Lagrangian model to predict the gap evolution during the drilling process of two metal plates. Choi et al. [9] present another Lagrangian model with the same objective. The thrust force predicted by the model is experimentally validated. The model allows for a gross prediction of interlayer burr height and thickness.

From a manufacturing engineering perspective, it is necessary to identify all parameters that are relevant to the generation of burrs. The main concern is to determine how these parameters affect burr size. The parameters affecting burr formation are practically the same as those identified for hole quality by Aamir et al. [116]. The diagram in Fig. 12 compiles the various parameters with potential effects on burr formation. The parameters have been arranged into three categories to facilitate their systematic study: (i) tool, (ii) workpiece, and (iii) process parameters.

This category comprises all factors concerning tool geometry, material, coating, and wear. The effect of drill bit geometry on burr formation has been extensively documented in the literature. Influential parameters regarding tool geometry are the diameter, the point angle, the helix angle, the lip relief angle, and the point geometry. Regarding diameter, studies show that an increase in hole size typically leads to an increase in burr size, due to a scale effect [25,106,129]. Other studies [28] show, on the contrary, that a nonmonotonic trend exists. Dix et al. [25] further show the impact of diameter on burr morphology, stating that small diameters favor the formation of irregular, crown-like burrs and that with increasing diameter the burrs tend to become more uniform. Concerning drill point angle, it is widely accepted that a large point angle helps to reduce burr size [43,44,52,69,106,130]. The greater the point angle, the smaller the thickness of remaining material at burr initiation, and the less concentrated the thrust force will be in the central zone of the drill bit. This promotes a first breakthrough of the drill away from the center of the hole, resulting in smaller burrs than if the fracture is limited only to the chisel edge zone. For certain diameter values, however, a nonmonotonic trend has been reported [26,67,86]. A negative effect of increasing point angle has also been documented [3]. The effect of helix angle has received less attention in experimental studies. In this regard, conclusions have been drawn both concerning the negative effect of increasing the helix angle [44] and its positive effect [62,130]. Regarding lip relief angle, there are no clear conclusions on its effect [44]. Drill point geometry also has an impact on burr formation. In this regard, the advantages of the split point [43,44], the step drill [32,90,131], or the chamfered drill [26] over traditional helical point drills have been demonstrated. A schematic representation of these drill point geometries can be seen in Fig. 13. A split-point geometry can reduce the thrust force compared to a traditional helical point [43,44], resulting in smaller burrs. The use of step drills can be seen as a multistage drilling process, with successive hole enlargements that are produced by the different stages of the tool [109]. The last drill stage is the one producing the burr. Since the volume of material to be removed by this stage is smaller with respect to traditional twist drills, the thrust force is reduced, resulting in smaller burrs. Chamfered drills act by favoring the cutting action in the periphery of the hole, therefore helping to produce clean edges with small burrs [26]. Regarding tool wear, experimental conclusions are unanimous and point to the negative effect of wear on burr formation [74,77,78,89,100,113,132–134]. Tool wear is typically measured in terms of the number of holes drilled with the same tool, or geometrically as flank wear. Indeed, tool wear causes a decrease in the cutting capabilities of the tool, which translates directly into an increase in the thrust force. As thrust force is the main driver of the exit burr formation process, increasing tool wear results in the formation of larger burrs.

This category involves the material and geometrical characteristics of the workpiece. As for the material, it is known that those features that foster plastic deformation result in the development of larger burrs. In this regard, higher ductility favors burr formation [4,129]. Strain hardening characteristics and the strain rate and thermal sensitivities of plastic behavior also play an important role. Regarding thermal behavior, the thermal properties of the material, such as specific heat or thermal conductivity, influence the development of elevated local temperatures that affect the mechanical properties and influence burr development. For example, titanium alloys are prone, due to their low thermal conductivity and specific heat, to experience high local heating during manufacturing processes [135]. This promotes an increase in local ductility, and therefore the formation of larger burrs. Fracture behavior may also play a role. For example, higher fracture strain facilitates the development of larger burrs [119]. With regard to the geometry of the workpiece, aspects such as the thickness or the angles between exit surfaces and machining directions can be of importance [136]. For drilling multistack parts made of different materials, the position of each material, or stack sequence, can be relevant [51,53].

The third and final category of relevant parameters comprises those defining the drilling process. In this regard, it is widely recognized in the literature that increasing the feed rate promotes the development of larger burrs [4,26,39,50,62,69,75,79,86,90,106,119,129,131,137–140]. Indeed, increasing the feed rate implies an increase in the thrust force [64]. Since thrust force is the main driver of burr development, increasing the feed rate promotes burr formation. There are studies where a nonmonotonic trend is shown [44,46,87], or even where increasing the feed reduces burr size [42,80,113], but these are relatively rare. Similar conclusions can be drawn regarding cutting speed or spindle speed. Indeed, increasing the cutting speed tends to favor the development of larger burrs [36,39,42,44,69,78,80,90,129,139,140]. Increasing the cutting speed favors the development of higher local temperatures caused by friction between the tool and the workpiece. The thermal increase makes the workpiece material more ductile, thus favoring the formation of burrs. There are also experimental studies that show the positive effect of increasing the cutting speed [33,60] or its low influence [46], as well as a nonmonotonic trend [87,141]. Another process factor that influences burr formation is the lubrication and the parameters that characterize it. Indeed, it has been experimentally demonstrated that under wet-cutting conditions, burr development is lower than under dry-cutting conditions [44,58,78,142]. Lubrication mitigates local temperature rises by allowing heat to evacuate. As the temperature increases are not so significant under lubricated conditions, the increase in ductility is limited, which makes it more difficult for burrs to form. Other aspects can have an effect on drilling burr development. Strategies such as vibratory drilling [4,28], multishot drilling [109], or the use of variable feed [143] have been shown to be effective in reducing burr size. Others such as pecking [113] have been shown to have a negative effect by promoting the development of burrs. Finally, for multistack drilling of the same or different materials, there are several process parameters that affect the formation interface burrs. These are parameters that have an effect on the development of the interlayer gap, which, as has been demonstrated, is the main driver for the formation of interface burrs. In this sense, the higher the clamping force that is used, the greater the obstacle to interlayer gap development and the larger the burrs obtained [6,33,51,53,60,64,72,75,111,114]. An example of this can be seen in Fig. 14. The clamping strategy [6] and the distance between the clamping position and the hole also play a role. Indeed, the greater the distance, the less restriction to interlayer gap development and the larger the interface burrs that are likely to form [73].

A summary of the effect of the quantitative parameters discussed in this section is given in Table 3. Parameters involving multistack drilling were not included.

The literature contains contradictory conclusions regarding the effect of some of the relevant parameters. This is due to the fact that most of the studies draw conclusions from a limited number of machining experiments, typically for a single material and for a limited range of values and combinations of tool geometry and process conditions. Given the complexity of the burr formation process, conclusions drawn from small test campaigns are difficult to generalize. Nevertheless, several general conclusions can be derived from the literature survey and from the study of burr formation mechanisms. Indeed, burr formation involves large plastic strains under high strain rates and temperatures. Therefore, the higher the baseline ductility of the material, the larger the burrs are likely to be. All factors that increase ductility will result in easier formation of larger burrs. In this sense, the higher the temperature, the more ductile the material becomes, and the larger the burr is likely to be. Conditions favoring local heating, such as the absence of lubrication, high contact forces due to tool wear or high cutting speeds, and high strain rates leading to high adiabatic heating due to irreversible plastic deformation or low heat conductivity and low specific heat, promote the formation of burrs. Regarding process parameters and tool geometry, the higher the thrust force, and the more concentrated near the chisel edge due to small point angles, the higher the probability that large irregular burrs will develop. For interface burrs, all of the above remains true. Furthermore, it should be noted that clamping conditions are the main driver for interlayer gap formation during the drilling process, and therefore also for interface burr formation. Worse clamping conditions, i.e., smaller clamping force, bigger predrilling gap, or bigger clamping distance lead to bigger burr and potentially to other types of interface pollution. Furthermore, the lower the resistance of the backup material to transverse deformation, the more easily interface burrs will develop.

Deburring comprises all postmachining techniques through which the total or partial removal of burrs is carried out. A wide variety of deburring processes is available for industrial use [20]. No deburring process is however universally applicable. In fact, most deburring processes and tools are only applicable to certain workpiece geometries and cannot be used on a general basis. Among the available techniques, a choice must therefore be made. The choice of deburring technique must be made after an assessment of burr size and location and of the volume, geometry, material, and criticality of the affected part. This assessment must lead to an estimation of the potential impacts of the deburring process on workpiece quality [20]. The cost of and the potential automatization play also a key role in the choice of the deburring technique. There is an extensive literature on deburring processes. The first systematic studies were carried out by Gillespie [13,16,17]. Gillespie proposes a classification of deburring processes into four categories: (i) mechanical, (ii) thermal, (iii) chemical, and (iv) electrical. A compilation of some of the most important deburring technologies following this classification is given in Fig. 15.

Mechanical methods remove burrs by mechanical abrasion or cutting between the deburring tool and the burr material. Examples of techniques belonging to this category are abrasive jet deburring, vibratory tumbling, or hand deburring [17,145]. Among the mechanical deburring techniques, hand deburring stands out as the one most commonly used [146]. Thermal methods remove burrs by high local heating. Flame deburring, thermal shock deburring, or laser deburring [20] belong to this category. Chemical techniques use a chemical agent to eliminate the burrs. Chemical barrel finishing and chlorine gas deburring [17] belong to this group. Electric methods apply a current or discharge to the burr to remove it. They often require the use of a chemical agent. Examples are electrochemical deburring and electrochemical polishing. For extensive reviews on deburring techniques, see Refs. [17,145–147].

Deburring is a costly activity, which can account for up to 30% of manufacturing costs [16,17]. More moderate estimations situate the cost of deburring between 5% and 20% [148–150], with costs scaling with the size of the burr and the geometrical complexity of the workpiece [148]. For multistacks drilling, cleaning and deburring of the interface are nonvalue-added tasks that require significant processing time and whose elimination would allow a nonnegligible reduction of costs. More details on the economic impact of deburring can be found in Ref. [146]. Deburring is therefore an activity that, from an industrial point of view, should be avoided. If this is not possible, the volume of burr material to be removed should be kept as small as possible. This is the aim of burr control strategies, which, based on a knowledge of the parameters that are relevant to the generation of burrs, allow postmachining deburring needs to be reduced or even eliminated.

Burr control strategies comprise all techniques aimed at minimizing the burr resulting from a machining process. The development of burr control strategies is based on the knowledge of burr formation mechanics and on the estimation of the influence of the different relevant parameters. Burr control and minimization strategies can be classified into three main groups: (i) workpiece design optimization, (ii) drill selection, and (iii) drilling parameters optimization.

The first group comprises those strategies that involve proper material selection or part geometry design. These are typically driven by part functionality constraints or structural considerations, and there is therefore not significant flexibility to make modifications for burr minimization [22]. In the case of multistack drilling, the selection of a suitable drilling order can significantly help to reduce burrs [6,51].

A proper selection of drill bits can help to minimize burrs. Material selection and tool coating should be made according to manufacturing criteria that take into account the workpiece material and machining conditions. For the case of drilling, different strategies involving tool geometry have been reported as potentially suitable for burr minimization. Ko and Lee [43] develop a split-point drill bit, which helps to avoid the formation of drilling caps and thus contributes to reducing the size and irregularity of burrs. Franczyk et al. [79] present a chamfered drill that proves experimentally advantageous over traditional tools. Gajrani et al. [26] also demonstrate that a chamfered drilled with slits contributes to the reduction of cutting forces and burr size. Another geometry that has been shown to be effective in minimizing burr size is that of step drills. These are drill bits with multiple axial steps of increasing diameter. Jia et al. [90] demonstrate that the step drill configuration helps to reduce burr size by redistributing the thrust force more effectively. Ko and Chang [32] and Ko et al. [131] also conclude on the advantages of the step drill and conduct studies on the influence of step size and angle. A different strategy, which also involves modifying tool geometry, allows to perform a deburring operation at the same time as drilling. Properly speaking, this is a deburring operation, but as it is performed in-process and not as a postprocess operation, it has been included here as a burr control strategy. For drilling operations, Kim et al. [136] present a drill bit geometry with a retractable cutter in the shank area, which allows for the removal of the entry and exit burrs during the drilling operation. Franczyk and Zębala [81] propose a similar tool geometry with in-process deburring capability. An analogous strategy has been developed by Kubota et al. [151].

Extensive literature is available on the optimization of drilling parameters to minimize burrs. Several approaches are possible. A first approach consists of the use of control charts, obtained from a sufficiently large number of experimental data. These control charts allow the selection of optimal parameters as long as the machining conditions do not deviate too far from those used in the creation of the chart. Examples of burr control charts can be found in Refs. [36,38,57,104]. Another approach to process parameter selection is to apply numerical optimization algorithms on objective functions such as burr size or deburring cost. The objective functions relate the process parameters, and sometimes also the tool geometry or drilling strategy, to the burr size or deburring cost by means of analytical models or reduced models such as response surface models (RSMs) or artificial neural networks (ANNs). Examples of application of optimization techniques for drilling processes aiming to minimize burr size can be found in Refs. [28,30,36,51,66,67,76,86,88,152] and for deburring cost optimization in Ref. [109]. The most important limitation of the application of optimization techniques lies in the difficulty of defining an objective function that is fast to evaluate and sufficiently realistic, capable of taking into account the interaction effects between different parameters. RSMs or ANNs are calibrated from tests and are hardly applicable outside the conditions that were used to obtain the calibration database, which makes the optimization results difficult to generalize. A final approach involving the manufacturing process is the use of process modifications that hinder the formation of burrs. Li et al. [4] propose the use of low-frequency vibration-assisted drilling and ultrasonic-assisted drilling. The goal of this technique is to reduce the thrust force, the main driver of drilling exit burr formation. Lin and Shyu [143], on the other hand, propose a slow sinusoidal feed variation over the workpiece thickness to be drilled. This can be seen as very low-frequency vibratory drilling. The idea of this method is to reduce the thrust force during tool entry and exit. A similar strategy is proposed by Pande and Relekar [28]. Another technique is tested by Mondal et al. [76], who show that the use of a backup material is useful for burr size reduction.

In certain situations, it is possible that burrs are not removed and remain on the manufactured parts by the time they are brought into service. If the burr is accessible, and given that deburring automation is not widespread, it may be decided not to remove the burrs to reduce manufacturing time and cost. This is not common, given the many negative effects that not removing burrs may have. However, it is possible that burrs are not removed if they are not easily accessible, as in One-Way-Assembly processes [6,8]. The nonremoval of burrs can have several negative consequences. The presence of burrs can hinder the correct positioning of parts during assembly, resulting, for instance, in hole misalignment and nonperpendicularity [153]. It can also lead to unwanted interference or jamming between parts. The presence of burrs can also damage protective coatings, reducing their effectiveness and promoting corrosion problems [22]. Finally, burrs can be indicative of the presence of other types of defects, such as subsurface cracks [20,132], excessive surface roughness [116], or interface pollution [52,75]. The presence of burrs has an experimentally demonstrated detrimental effect on the mechanical strength of mechanically fastened assemblies, both in static [54] and fatigue [154].

From a manufacturing perspective, burrs can also be problematic for the machining process itself, for the subsequent heat and surface treatment, and for the assembly and disassembly operations. The presence of burrs can lead to accelerated wear of cutting tools, resulting in reduced tool life [75]. The detachment of hard particles coming from burr material may also erode the surfaces in contact [53], triggering interface fretting wear and in certain cases electrical problems and short circuits [155]. The application of protective coatings that are electrically deposited can also be hindered, such as for powder coating or anodizing, as they interfere with the electric field created during this type of treatment. Special care should also be taken during heat treatments, as the presence of burrs can result in high local stress concentrations during cooling leading to cracking [82]. Burrs also pose a hazard to workers during the assembly and disassembly of parts, as they are sharp edges that can be cut. Burrs and their potential detachment can also pose a risk to the functionality of other components. For instance, if absorbed in a fluid flow the burr will increase the head loss of the tube [82] and potentially cause a serious system failure [20]. They can also have a negative aesthetic impact on the manufactured parts.

The number of available experimental studies concerning the effect of drilling burrs on the mechanical strength of assemblies is small. On the one hand, this can be justified by the fact that burrs are systematically removed for most situations of industrial interest. On the other hand, mechanical testing involving burrs is complex. The complexity is mainly due to the need to obtain geometrically and materially reproducible burrs, adapted to test conditions. A burr height and a uniform shape are targeted. Most publications do not report the process that is used to generate the burrs, or even neglect the influence that the process can have on test results. A complete geometrical study of the burr is often lacking. For the testing of assembly conditions, access to burrs is limited. For example, in fatigue testing of assemblies, if the test is not interrupted and the specimens are disassembled, the evolution of the burr is not observable. Studies report burr characteristics before assembly and do not pay attention to the evolution of burrs during testing.

Effects of burrs on the installation of fasteners have been reported. Dols [156] observes during fastening that a burr can either get crushed or entrapped between the fastener shank and the hole surface. A measurable impact on the relationship between torque-number of turns is reported. For the correct installation of a fastener, it is also necessary to ensure that the preload level is adequate. Zwerneman [157] shows that the effect of burrs on the fastener preload depends on the installation system. Experimental measurements for iso-torque installation can be seen in Fig. 16. For burrs of 2.5 mm, an average decrease of 25% is observed. For iso-torque and turn-of-nut installation, it is shown that bolt preload decreases with burr size. Mackiney [54] presents test results measuring the number of nut turns required to achieve a proper target bolt tension. Results show that the number of required turns increases with burr size. This result is complementary to Zwerneman's study. The Research Council on Structural Connections (RCSC) reports these results and states that for large burrs, greater than 1/16 in., more rotation is needed to achieve specified pretension if the turn-of-nut method is used. Burrs can also affect the relaxation of the preload with time. Chiza et al. [158] and Annan et al. [159] show that the relaxation with time of the preload is higher in the case of nondeburred specimens. Incorrect installation or rapid loss of preload can have a significant impact on fatigue behavior, being a potential source of a knockdown effect [160].

Regarding the static strength of assemblies, only two aspects have been studied: the effect on the (i) slip coefficient and on the (ii) ultimate shear load.

The slip coefficient concerns load transfer by friction. Chiza et al. [158] report a slight increase of about 10% in the static friction coefficient between the faying plates if the burrs are not removed. Polyzois [161] and the RCSC [162], based on the work of Mackiney [54] and Zwerneman [157] for burr sizes up to about 2.5 mm, report a nonmonotonic trend. Experimental results can be seen in Fig. 17. For burrs below 500 µm, increasing the height increases the coefficient of friction, whereas for larger burrs the effect is the opposite. However, the variations are small, lower than the experimental scatter. For the tested configuration, it can be therefore concluded that the effect on the friction characteristics is negligible.

Regarding the ultimate shear capacity of joints with burrs, Zwerneman [157] presents test results up to failure for double lap shear specimens. The experimental data in Fig. 18 show a small knockdown effect that scales up with burr size. No information is given regarding the mode of failure. For burrs of 3 mm, a reduction of around 10% has been observed. This reduction is smaller than the observed scatter. For the tested configuration, it can therefore be concluded that the effect of burrs on the ultimate shear strength is negligible.

The effect of burrs on the static strength of joints is expected to be small. Even for large burrs of several millimeters, the results above demonstrate that the friction and ultimate capacity of the joint remains unchanged. More experimental data is however considered necessary to consolidate this conclusion.

The impact of drilling burrs on the fatigue strength of mechanical assemblies has been reported to be significant. Barter et al. [163] show cases of fatigue failures in aircraft structures initiated at burrs. A synthesis of experimental test conditions concerning the effect of burrs can be seen in Table 4. Results are available for different configurations and materials. A schematic representation of the test specimens is shown in Fig. 19.

Abdelhafeez et al. [154] present fatigue test results for open-hole specimens in aluminum 2024-T351 and 7010-T7451 and for titanium Ti-6Al-4V α–β annealed. Specimens without burrs and specimens with burrs on the hole edges, both at the entry and exit, are tested. Burr height and thickness are measured with a 3D microscope. Experimental results presented by Abdelhafeez et al. show a knockdown effect of the presence of burrs. It is assumed that the burr acts as a notch that locally concentrates stresses and promotes crack initiation. Under this hypothesis, a good correlation of this knockdown with total burr volume is found. Total burr volume is calculated assuming a triangular burr section with constant height and thickness around the hole. Micrographic analyses of the fracture surfaces are also provided, which show that the burr acts as a preferential site of crack initiation. Experimental results are shown in Fig. 20.

Nix et al. [164] report fatigue tests for open-hole specimens in 7010-T76 aluminum. A significant detrimental effect of burrs is demonstrated. The experimental results can be seen in Fig. 21. A distinction is made between specimens where postdrilling reaming has been applied and specimens where it has not. The most detrimental effect corresponds to specimens where reaming has not been applied. No explanation is given regarding this.

Nishimura [165] carried out another study on open-hole specimens. Burrs are machined by milling around the hole of the specimen. For fixed diameter, it is shown that, in the presence of burrs, there is a detrimental effect on the fatigue strength. This is explained again by arguing that the burr acts as a stress concentrator. Experimental results demonstrate the detrimental effect is greater at low-stress levels than at high-stress levels. This trend has also been observed in Refs. [166–168]. Nishimura also studies the size effect resulting from varying the hole diameter at a given stress level. In this case, it is observed that the smaller the diameter, the greater the detrimental effect. This is explained by arguing that reducing the diameter brings the burr closer to the center of the hole, increasing the intensity of stress concentration.

Shiozaki et al. [98] analyze the detrimental effect of the presence of burrs resulting from a punching process. A study is made of residual stresses at the edge of the hole, together with experimental measurements with X-ray diffraction. Areas with tensile residual stresses, which favor fatigue crack initiation, are identified. Fatigue results are presented comparing as-punched configurations and configurations to which a residual stress relaxation heat treatment has been applied. Two different levels of residual stresses are compared, corresponding to different values of clearance between the punch and die. The results presented in Fig. 22 show that the higher the residual stresses, the lower the fatigue strength. This shows that the detrimental effect indeed comes from the existence of positive residual stress zones. It is also observed that the detrimental effect is greater at low stresses. These results show the potential importance of residual stresses in the effect of burrs on fatigue.

Lanciotti and Polese [166] present a study for dogbone drilled specimens. The test results can be seen in Fig. 23. Deburred specimens where drilled and reamed individually. Nondeburred specimens where drilled and reamed as a stack of 10 specimens. For series B, a 0.15-mm thick plastic foil was also inserted between the specimens. It is observed that burrs have a detrimental effect, the greater the lower the applied stress. No data is provided on the geometry of the burr, nor is there any explanation for the observed knock-down.

Lorenzo et al. [167] present the results of fatigue tests on tension-compression circular coupons in casting magnesium. Burrs appear as a result of the casting process. The tests show a detrimental effect between the deburred and nondeburred configurations. These burrs are not representative of a drilling process. Matsunaga et al. [168] also present fatigue tests with circular tension-compression specimens, but this time with a hole drilled on the cylindrical surface. Burrs are machined by turning, in a similar way as Nishimura did by milling in Ref. [165]. The burrs are shown to have a detrimental effect, especially at low stresses.

Stalley [7] concludes that burrs have no effect on fatigue behavior in open-hole configurations for aluminum from the 2000 series. No geometrical characterization of the burrs or fatigue test results is provided. Stalley [7], without presenting experimental results, also argues that a knockdown exists in high-load transfer (HLT) configurations for 2000 series aluminum, but that for the 7000 series, there is no effect.

For low-load transfer (LLT) specimens, Jochum et al. [5] argue that the detrimental effect on fatigue can be significant, with drops of around 40% in fatigue life for titanium. A summary of their study can be seen in Fig. 24. For HLT specimens, Jochum et al. [5] state that the detrimental effect of burrs at the interface is between 60% and 70% of fatigue life reduction. The impact is therefore higher than in the LLT specimen.

Vallellano et al. [169] present their results with aluminum–CFRP hybrid HLT specimens, showing in this case that the effect of the burr is beneficial, increasing the fatigue life. This is explained by arguing that the burr has served to fix the interlayer, reducing the damaging effect of fretting. The experimental results can be seen in Fig. 25.

Koster et al. [83], on the other hand, conclude that there is no detrimental effect for LLT specimens, but only performs tests with burrs at the nut side, not at the interface. Vallellano et al. [169] reach the same conclusion for LLT and medium-load transfer (MLT) aluminum–CFRP hybrid specimens. The results of the latter study can be seen in Fig. 25.

While there is broad qualitative agreement on the negative impact of burrs on fatigue strength, the reasons explaining the fatigue initiation mechanism in the presence of burrs remain unclear. The effect of burrs may be due to different reasons [170]. The most common and simplest approach is to consider that the burr acts as a stress concentrator due to its sharp geometry. This would easily explain the knockdown that exists even in simple configurations such as open-hole or dogbone specimens. For assembly conditions, the burr can be seen as a preferential load-passage zone, thus concentrating stresses and favoring the initiation of fatigue cracks. Other potential explanations for the detrimental effect of burrs are the existence of local tensile residual stresses as a result of the burr formation process. This explanation is employed by Shiozaki et al. [98] for punching burrs. As an alternative to the above stress-based explanations, it has been proposed that burr material is a damaged material that may contain micro-cracks [132,171], reducing the experimentally observed fatigue life to a propagation phase from microscopic cracks. For assembly conditions, there are other potential sources explaining the knock-down due to burrs. On the one hand, it is possible that during assembly or cycling, burrs may detach and form hard particles at the interface, causing fretting fatigue [53]. Even if not detached, the burr may indent, or potentially enter the other material or the hole, with effects that are difficult to anticipate. Furthermore, it is possible that the installation conditions when burrs are present at the interface may not be correct, or may degrade excessively during cycling. In particular, it has been shown in the preceding section that the presence of burrs may hinder the correct installation of fasteners, leading to lower-than-desired preload and faster relaxation during cycling.

It is difficult to discriminate among the above reasons which one applies to a particular real-life situation. Moreover, in many situations, it is most likely that several of these explanations are true at the same time. It may also be difficult to discriminate the effect of burrs from potential alterations linked to the formation of the burr, such as high roughness in the burr zone or altered microstructure due to plastic deformation and local heating.

There is an extensive literature on drilling burrs. Most of the available publications address the topic from a manufacturing perspective. Given the high estimated economic impact of deburring tasks, research has been directed toward the improvement of drilling to minimize and control burrs. The attempts to minimize burrs have led to the study of formation mechanisms, as well as to the development of predictive models capable of capturing the influence of the relevant parameters. However, a significant number of knowledge gaps regarding drilling burrs have been identified. First, a standardized and unique method of describing drilling burr geometry is lacking. This knowledge gap is already mentioned in Refs. [20,146]. A formal universal parameterization and harmonized measuring method, adapted to the variety of possible geometries, would be desirable. Nowadays, burr geometry is described by a single value for height, and sometimes also of thickness. A rigorous assessment of the suitability of this parameterization for different purposes is lacking. The development of a homogeneous geometrical parameterization would allow the choice of a robust measurement method, as well as facilitate the development of predictive models capable of assessing the impact of the burr. Ideally, the parameterization should be such that it can be obtained with simple, industrially applicable nondestructive methods.

Second, there are important knowledge gaps in relation to the drilling burr formation process. The entry burr has been poorly studied, with limited information on the effect of manufacturing parameters on its development. Predictive models are lacking. The material evolution in the area close to the entry burr is also unknown. Knowledge gaps regarding exit burr formation have also been identified. Analytical models are rare and their scope is very narrow, not being able to deal with the complex thermomechanical behavior observed experimentally. FEM models are not widespread either, due to their high computational cost. This means that there is no real predictive capability for burr type, burr size, and burr irregularity. There is also no effective ability to predict the postprocess material state, in terms of local properties and residual stress distribution. There are a number of reduced models capable of assessing the impact of different parameters on burr size, but these are limited to conditions close enough to those of model calibration. Moreover, the formation of burrs at the interface in multistack drilling has been little studied. It is known that the important factor affecting bur formation is the gap formed between the stacks, and validated models are available to predict the evolution of this gap during drilling. However, the relationship between the gap and the size of the exit and entry burr formed at the interface is unknown. It is also unknown what happens when the two burrs are pressed against each other after drilling. Other phenomena such as the penetration of burrs into the backup material, or the behavior in hybrid joints, where the order of the layers may be of importance, have received little attention so far. There is another important knowledge gap regarding burr formation in multistage drilling and during final reaming, which are systematically applied in the aerospace industry [15] and in which the last operation is the one that most affects the final burr size.

Third, there are large knowledge gaps regarding the impact of drilling burrs on the mechanical strength, in case they are not removed. Studies on the effect of burrs on the static strength are very limited. The effect on fastener installation conditions, such as potential perpendicularity defects or the risk of a burr penetrating the hole, has not been studied. The impact of burrs in fatigue has been studied more than for static. The existence of a detrimental effect has been demonstrated experimentally, but it has not been possible to explain it qualitatively in a conclusive way. A greater detrimental effect is observed for lower values of stress. Multiple reasons explaining the knockdown have been suggested, but there is not enough experimental information to discriminate between them. The possible existence of a scale effect, in the sense of the impact of hole diameter, is also unclear. The ability to discriminate the effect of burrs on different workpiece materials is also lacking. For assembly conditions, a number of additional questions remain open to this day. There is no ability to predict the burr geometry after fasteners are fixed nor to assess the impact of the load transfer rate to explain the observed differences between high-load transfer and low-load transfer configurations. Interaction with sealant and with various types of interface pollution has also not been studied.

The open questions listed above represent opportunities for applied research. This research is of direct industrial interest, both in terms of reducing manufacturing costs by minimizing the generation of drilling burrs and in terms of developing the capacity to assess their effects. From the authors’ perspective, a research priority should be to enlarge the static and fatigue test databases to allow a deeper knowledge of the effect of burrs on the mechanical strength of metallic and hybrid joints. These tests should be conceived to better understand the detailed failure mechanisms when burrs are present, as well as the influence of burr size. The identification of a limit burr size below which the impact on static and fatigue can be considered negligible must also be a research target. For testing, a method to produce repeatable burrs that are representative of realistic manufacturing conditions must be developed first. Special interest should be devoted to the manufacturing conditions of One-Way-Assembly processes, due to the increasing industrial interest in them. Another research priority should be to improve the understanding of burr formation, especially of the material state around the burr. This should allow to develop predictive capacity to minimize the size of burrs produced during manufacturing, as well as to better understand and anticipate the impact that burrs may have on the static and fatigue strength.

The authors gratefully acknowledge the ANRT (Association Nationale de la Recherche et de la Technologie) and Airbus Operations SAS for financially supporting this study. The authors are members of the working group Manufacturing 21 which gathers 21 French research laboratories. The topics covered are the modeling of the manufacturing process, virtual machining, and the emergence of new manufacturing methods.

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.