1 Introduction

Additive manufacturing (AM) technologies, which offer a great deal of flexibility in object manufacturing and allow the manufacture of very geometrically complex products—which would be impossible to obtain with traditional tools—have an undeniable advantage over conventional manufacturing technologies [1]. Current possibilities to manufacture structures unachievable by subtractive and forming manufacturing technologies, aroused by introduced layerwise manufacturing, are further extended thanks to simultaneous implementation of the functionally graded materials, embedded electronics, as well as other integrated functionalities. All mentioned areas can be applied at once in parts manufacturing, but this approach is very challenging and requires specialized knowledge in decision making, which is the spark for the development of an intelligent additive manufacturing and design (IAMD) concept [2].

Since the 1980s, the Rapid Prototyping trend has evolved into additive manufacturing, and many different developed technologies belong to the AM group. In 2013 the first ISO/ASM standard from the 529XX series was published which introduced the terminology and nomenclature of different techniques [3]. A special place in the classification is for technologies which can process metals and produce free of voids, fully functional, high performance metal parts [4]. A more detailed classification of metal AM techniques divides them according to, e.g., the form of material feedstock (filament, sheet or powder), or the source of fusion energy (ultrasound, electron or laser beam) [5]. Three main groups can be distinguished in this category: directed energy deposition, powder bed fusion and sheet lamination [6].

The application of AM is an economic alternative to single unit production, in which costs related to the preparation for production can be spared and special tools are not necessary to design and manufacture them, which would substantially increase the unit cost [7]. In particular, technologies from the laser-powder bed fusion (L-PBF) group, which allow the manufacture of metal objects by melting metal powders with a laser beam, are widely used in sectors that feature high expenditures, such as medicine, aviation, or aeronautics. The effect of synergy in the manufacture of fully functional products is used in these sectors, thanks to the combination of the merits mentioned above.

The development of the design for additive manufacturing (DfAM) methodology has contributed substantially to the reduction of manufacturing time and costs for an object with an optimised geometry, by reducing the volume of the material, for example, with the introduction of lattice structures [8]. Tofail et al. [7] demonstrated that these sophisticated technologies also have limitations, which at the same time are an obstacle to their implementation in industry. These barriers mainly include the investment in purchasing the equipment and the level of return on this investment, in which replacing conventional production with additive technologies will provide the enterprise with a profit. When analysing the market development of metal additive technologies, a trend of increasing build volumes in machines can be observed [9]. Such a development opens new opportunities for the industry to apply such techniques in the manufacture of elements that are large despite their low mass (e.g., elements of aircraft engines or gas turbines). In the case of production orders for small elements, an AM solution allows for a higher production volume in a single process, which directly translates into increased production efficiency. In the case of building large objects, it is frequently impossible to add additional models to the build volume; therefore, it is necessary to increase productivity using other methods to make the process more economical. The process parameters must be prepared as quickly as possible, resulting in a high build rate, in addition to maintaining satisfactory properties of the finished product.

Numerous researchers have emphasised the complexity of the L-PBF process, the results of which—the quality of the melted material—depends on many parameters. Scientific circles focus on the development of new materials for additive technologies that will expand their capabilities. Various optimisation methodologies are used to shorten the time and reduce the financial expenditure of developing a process window for new materials, such as the assessment of single scan tracks (SSTs), statistical instruments for experiment planning or other advanced methods for multiple-criteria process optimisation [10].

The SST method allows the basic process parameters (linear density of energy, consisting of scanning speed and laser beam power) to be quickly assessed, at the cost of a small amount of powder material, which is frequently very expensive—in small-scale manufacturing and in preliminary research [11]. In the methods for planning experiments (design of experiments) the process parameters are selected, and their values have a significant impact on the qualitative outcome achieved [12]. Yuan et al. [1] have suggested a multiple-criteria tool for optimising the PBF process for thermoplastics sintering to assess the relationships between the process parameters and the many factors of product assessment, e.g., mechanical properties, degree of material degradation and process productivity.

The cost of the AM system is the main cost component of additive technologies, which is estimated to be even 50–75% of the overall cost [7]. Because it is impossible to reduce the cost of purchasing an AM system, the only way to reduce machinery costs is to increase the productivity of manufacturing processes. By reducing the time of machine operation, the cost of manufacturing parts in an AM will be reduced. The gain in process productivity suggested by researchers is mainly limited to increasing the values of surface process parameters, such as scanning speed or distance between scan tracks [11]. Herzog et al. [13] presented an approach that achieves higher process productivity by increasing the scanning speed by two-thirds, resulting in a lower relative density of material, which is compacted next in the HIP process. Sow et al. [14] improved productivity by increasing the diameter of the laser spot, therefore, reducing the number of generated vectors filling the contour. Leicht et al. [15] presented results for a fourfold increase in the thickness of the melted powder layer, improving the productivity and mechanical properties of 316L steel. The growth in productivity was obtained at the cost of worse mechanical properties, although the values consistently exceeded those defined for cold-rolled steel in ASTM A240M-18. De Formanoir et al. [16] described the ‘hull-bulk’ strategy, which consists of manufacturing objects from the popular Ti6Al4V alloy, melting the contours of the manufactured parts every 30-µm layer and melting the hatch every third layer with the use of much higher energy. Thanks to this strategy, a good surface quality was maintained, and the process productivity was improved.

Performed review of the available literature did not result in a reliable and global assessment of the factors that influence process productivity or an analysis of the impact of these factors on process productivity, and hence on manufacturing times, which comprise the majority of the cost of objects manufactured with additive technologies.

This paper presents the relationships between the factors of the L-PBF process and the resulting productivity of an additive process, which have been verified. In particular the authors analysed to what extent a change in process parameters contributes to better process productivity, taking into consideration the effects which can occur in terms of the quality of the melted material.

2 Materials and methods

For the purposes of this paper—verifying the impact of selected factors on the productivity of an L-PBF process—an experiment was planned which consisted in evaluating the manufacture time of test objects for various sets of values of process parameters and using various L-PBF systems equipped with several laser sources to analyse the degree to which they result in higher productivity.

The designed objects had the geometry of the same volume (490 cm3) and the same height along the Z axis (100 mm), but they differed in surface area, according to which they can be described. The first model was a cuboid with dimensions of 70 × 70 × 100 mm, while the second was a thin-walled, closed profile of a square cross section with dimensions of 250 × 250 × 100 mm and a wall 5 mm thick (Fig. 1). Such a choice of geometry allowed us to analyse an additional factor related to idle runs of the laser beam when creating thin-walled geometry. The strategy of scanning models in each manufacturing process took into account only the scanning of the cross-sectional hatch, without contour scanning, and different values of process parameters for the down-skin and up-skin used in the case of regular manufacturing processes to obtain the best objects in terms of quality. For the same reason as for simplification of the scanning strategy, the evaluation performed did not take into account any support structures, but only the manufacture of solid bodies of the designed models.

Fig. 1
figure 1

Geometry of the models on which the studies were carried out. A Basic model, 70 × 70 × 100 mm; B thin-walled model, 250 × 250 × 100 mm, 5 mm wall

Full size image

The initial values of the process parameters used in the evaluation were determined for the processing of 17-4 PH steel and were found in the library of process parameters provided by the system manufacturer (SLM Solutions Group AG, Lubeck, Germany).

The first experiment was on multi-laser systems, allowing for the simultaneous melting of cross sections of manufactured objects on each layer, therefore, shortening the process times. To remove any additional variability factors, the machines were equipped with one, two, and four laser sources, namely, an SLM 280 HL 2.0 in a single and twin system and an SLM 500 system (SLM Solutions Group AG).

The next experiment was designed to assess a drastic change in basic values for 17-4 PH steel (by 100%), only to determine to what extent the change affects productivity while ignoring the quality aspect of the resulting object and its material properties. The experiments were based on a full-factorial plan for four variables at two levels (24). To supplement the study, the actual time of machine operation was measured during a few layers of the manufacturing process.

Experiments were carried out for a laser power of 0 W. The default values for 17-4 PH steel were taken as the basic process parameters for the evaluation, while values increased by 100% were taken for the second, extreme level of factors. Such an approach not only allows us to save material and energy, but also allows us to conduct extremely theoretical processes using abstract parameter values without build failure. The 0 W laser power does not affect the scanning time, because the machine moves the Galvo system the same as in the case of a standard melting. The coded values of the factors used in the experiment are presented in Table 1. Sixteen experiments were carried out on the presented geometry cases, checking the manufacture time on an SLM Solutions 280 HL 2.0.

Table 1 Range of analysed parameters
Full size table

The experiments planned as above yielded data on the time of individual stages of the process: scanning and depositing the next powder layers without wasting material. Because the models have a constant cross section across the entire height, the process was only running during the initial ten layers of the models. Then, the average time taken to deposit powder layers and scan the model cross section on a single layer was calculated. The entire build time is a total of the times for depositing powder layers and scanning cross sections, multiplied by the total number of layers in the model. The analysed data came from files that the machine saved after the process that was completed. Therefore, the analysed times are the actual times of the SLM processes performed with different parameters.

After analysing the machine data, the decision was taken that the basic production process would be split into two aspects. The first one was related to the deposition of a powder layer, which comprises the material collection from a container, movement of the rake, and control actions of the system (testing sensor values and pictures of the layer control system). The second component included scanning of the material with a laser beam. The recorded times were analysed, determining their main effects based on the differences between the process times for two extreme values of the analysed parameter. An example of calculating the effect of factor A (scanning speed) is as follows:

$$\widehat{Effect\,A} = \left| {\overline{A^+ } - \overline{A^- }} \right|,$$

where \(\widehat{Effect\,A}\) is the effect of factor A (scanning speed) (s) on the SLM process time. \(\overline{A^+ }\) is the average build time for all sets in which the scanning speed was on the upper level. \(\overline{A^- }\) is the average build time for all sets in which the scanning speed was on the lower level.

3 Results

3.1 Factors influencing the productivity of the L-PBF process

The time of part manufacturing, depending on the productivity of the technology, is a factor that directly influences the cost of the production process. Interest in and widespread use of metal AM technologies in industrial plants will be enhanced if the process can become cheaper and faster. To this end, the work on permanent improvement and optimisation of the L-PBF process is ongoing.

The complexity of the L-PBF process and the multitude of factors that influence the production time of a finished element are presented in the Ishikawa diagram in Fig. 2.

Fig. 2
figure 2

Ishikawa diagram presenting the factors that influence production time in the L-PBF process

Full size image

The methods for increasing the productivity of the L-PBF process may be divided into a few areas. The first (Machine) is related to the machine and its capabilities; these are factors that depend on the manufacturer, required specification, and budget. They include multi-laser systems, modular machines, systems for monitoring the process, and increased build volume, among others. The next area may be developed in all AM metal systems, because it is related to the process and its parameters (manufacturing process). The next two areas are inter-process actions, comprising all actions related to preparing the machine and material for manufacturing (process preparation) and to performing the required post-processing on the manufactured object (post-processing). In addition, a group of factors related to the operator who performs all process-related actions (user) has been distinguished. When analysing this area in terms of productivity, one can state that it is highly exposed to the human factor and the errors and decreased productivity that can result from the operator’s work. The last area considered in the paper is that related to designing the elements and preparing the build files (part geometry).

3.2 Main manufacturing process and process parameters

Three factors that influence the time taken for the basic manufacturing process are depositing a powder layer, scanning the layer (melting) and other actions related to the machine system and process monitoring, e.g., taking pictures before and after depositing a layer, and after scanning in the system of layer monitoring. Factors that influence the basic manufacturing process are presented in Fig. 3.

Fig. 3
figure 3

Ishikawa diagram presenting the factors that influence the basic manufacturing process

Full size image

A standard user does not have any influence on the operating system work and its delays. It is not recommended to switch off the process monitoring, because it can be used to analyse the process and to detect possible errors, which industrial customers require. Therefore, to increase the process efficiency, we placed the focus on the other two areas: depositing a powder layer and scanning with a laser beam (melting).

3.3 Machine design and the solutions available on the market

The choice of machine for a specific application is crucial. To satisfy the market’s expectations, manufacturers introduce design solutions that can shorten the build time and monitor the process to respond to any issues as quickly as possible.

  • Machine build volume The choice of a smaller machine reduces the cost of purchase. In the case of larger machines, it is possible to produce a few parts in one process, therefore, increasing the productivity by decreasing the number of basic manufacturing processes and inter-process actions. In the literature, it is possible to notice a division of LBPF machines according to specific applications [9]. Small machines, with a maximum build volume of 125 × 125 × 125 mm, are used in research and development work carried out in companies and scientific institutions. Machines of medium size are another group that is used most frequently in research and development projects and short series production. In this case, the build volume size can reach 300 × 300 × 300 mm. The specification ends with the largest machines, intended for industrial manufacturing. In this group, the focus is on process productivity and minimising downtimes; therefore, this category includes machines with the largest build volumes and the most laser units which enable the processing of high powder layers in the process. It is obvious that the price of the machine increases with size and application. Figure 4 presents a comparison of the build volumes of the L-PBF systems of three leading market manufacturers.

  • Multi-laser systems Multi-laser systems can melt the material by operating a few laser beams at the same time. To verify the effect of using such systems, the process time was evaluated for three types of machines, based on the same 3D model and basic parameters for 17-4PH steel provided by the system manufacturer, SLM Solutions Group AG. The SLM Solutions 280 HL single-laser version was considered as the basic machine; then the build rate was compared with the results of two machines by the same manufacturer, but with twin-laser (SLM 280 HL 2.0) and quad-laser (SLM 500) systems. The time of the basic process is the sum of the times for scanning with the laser beam and for depositing a powder layer. Because increasing the number of laser sources only affects the scanning time, only these values were analysed. Figure 5 presents a comparison of scanning times for the solid and thin-walled models between the single and twin systems. It is necessary to mention that if the geometry of the model allows for it, the twin system can work in two ways: the two lasers interact, scanning one model with an overlap zone (red area), or each laser manufactures its object independently allowing to build 2 times more parts in the same way. As the thin-walled model occupies the entire build platform of the SLM 280 machine, it cannot be duplicated in the process; therefore, the second method for twin system operation was analysed only for the geometry of a solid object, and overall processing time per single part (and not an entire process time) was taken for comparison. For a better interpretation of the results, Fig. 6 presents an operation scheme of the twin laser machine. The workspace in a multi-laser machine is divided into work areas for each laser. The lasers share together a 30 mm wide overlap zone. In the scanning strategy, the location of the contact point between lasers changes with each layer.

    As can be seen in Fig. 5, the scanning time for the solid and thin-walled model of the same volume differed by less than 2 h 31 min. Such a difference can be caused by many factors, including the need for the Galvo mechanism controlling the laser beam to make longer movements in the case of a model occupying the entire platform area. Another thing that can affect process time is the increased number of laser on–off delays for the thin-walled part. For the machine with two lasers, the time difference was less pronounced (approx. 55 min), while in this case, the scanning of the thin-walled model was faster. Comparing the single and twin systems, the manufacture of the solid model in the first method with the overlap zone on the twin machine reduced the scanning time by 23 h 43 min 32 s (nearly a 45% reduction), while the second way reduced the total exposure time per part by 26 h 22 min 40 s (a 50% reduction) as compared to the machine with one laser. For the thin-walled model and the machine with two lasers, the time was reduced by 27 h, which roughly equals a 49% reduction in scanning time over the machine with a single laser. The difference in the time reduction between the two models, solid and thin-walled, for the twin machine may have resulted from a different volume of the overlap zone. For small models, the second exposure method—where both lasers perform individual actions independently—was a better solution, allowing the models to be duplicated and arranged on the build platform. This led to a greater reduction in the manufacturing time of a single part and did not require the setting of additional parameters of an overlap zone. However, the even distribution of the scanners splitting the build plate into smaller laser working areas has a positive effect on the quality of the parts. The smaller work area of a single laser minimises the phenomenon of changing the shape of the laser spot along with the distance from the centre of the optical system. Because of this, the process is more stable. There are no large energy density deviations supplied to the powder, and the thickness of the melted track is more uniform. Furthermore, by courtesy of SLM Solutions Group AG, the single and twin systems were compared with the quad-laser SLM 500 machine. The results are presented in Fig. 7.

    The small difference in scanning times between the machines in the twin and quad systems in the case of the thin-walled model resulted from the fact that the model covered most of the build volume of only two lasers, while the other two scanned only small fragments in the overlap zone. Unfortunately, the geometry of the model was too large to duplicate on the building platform, so the analysis was carried out for a single part placed in the centre of the platform. When analysing the results for the solid model, a significant difference can be observed in the scanning time as compared with the systems discussed above. As in the case of the twin system machine, the process may be planned in two ways. Analysing the first way (interacting lasers), the use of a machine with four lasers reduced the time of manufacturing a single part by 35 h as compared to the machine with one laser, and by 11 h 23 min against the machine with two lasers. In the case of the second method, the time needed to manufacture a single part was 39 h 17 min shorter compared to the single-laser machine; compared to the twin machine, it was 15 h 43 min shorter.

  • System of bi-directional layer deposition Originally, the movement of the rake that deposits new powder layers required two passages in opposite directions, that is, a passage that deposits the powder and a return passage to the initial position (Fig. 8a). Return movement, which is an idle movement, unnecessarily extends the time of layer deposition; therefore, a method for layer deposition was developed with single passages, regardless of movement direction, called ‘bidirectional recoating’ (Fig. 8b). The new design of the powder spreading apparatus was patented by SLM Solutions [17], especially to make this invention reliable and uniform the distribution of powder during both passes. Based on data from the process on an SLM Solutions 280 HL 2.0 machine, it was determined that the passage of a rake through the build volume, in one direction with the default speed, took 2 s. The system of bi-directional powder deposition allows the next layer to be scanned after a single rake passage without the need for it to be returned to the base point, as in the original system of powder deposition. The entire break between layer scanning, roughly treated as the layer deposition time, consists not only of the rake passage, but also of the time needed for additional actions of the machine system, i.e., data loading, platform lowering, and inspection of the deposited layer by the monitoring system. This time, in the case of powder deposition at a single passage, was 11 s, while changing the mode to a double passage extended it to 13 s. Removing the need for the rake to make a double passage reduced the deposition time of a powder layer between consecutive scans by 2 s. A small reduction in time, even in the manufacture of an object 10 mm high (applying a layer 30 µm thick), results in the entire process time being reduced by 11 min.

  • Systems for automatic platform cleaning The time required and the costs of the stages related to the preparation and finishing actions in most cases depend on the human factor, i.e., the operator’s experience, the dynamic of the work, and the size of the build itself. Machines featuring a build plate that self-levels before the basic process are available on the market, as are automatic systems for removing the finished product, which eliminates the human factor at these stages. The solution of Additive Industries is one example: it automatically allows the chamber of the metallic powder to be emptied [18]. The machine, having finished the last layer, moves the platform slightly downward and then lowers a special hatch, which tightly closes the powder bed. The powder material is then sucked into machine tanks, resulting in a clean powder build platform, with models built in the process. This solution is much more effective than the standard method of manually directing the powder into the machine chutes using a brush or rake. An additional benefit of such a solution is that the material remains permanently shielded with an inert gas, preventing its oxidation. This solution helps minimise the operator’s contact with the metallic powder, and also increases work safety and comfort.

Fig. 4
figure 4

Chart presenting the build volume in selected L-PBF machines of three leading market manufacturers

Full size image
Fig. 5
figure 5

Comparison of exposure times between SLM single- and twin-laser systems

Full size image
Fig. 6
figure 6

Twin lasers workspace section outline (A) and scanning strategy in the overlap zone (B)

Full size image
Fig. 7
figure 7

Exposure time on a quad-laser SLM 500

Full size image
Fig. 8
figure 8

Schematic presentation of powder layer deposition at a double (a) and single (b) passage

Full size image

3.4 Sensitivity of process parameters to productivity of the L-PBF process

The output values of the estimated manufacturing time of the test models for all manufacturing experiments carried out on cuboidal and thin-walled objects were gathered and analysed in terms of the total process time. 16 experiments were carried out according to the factorial experiment plan for four variables at two levels, as presented in Table 1. Figure 9 presents the main effects (absolute value of the process time differences, recorded and analysed for the extreme values of the parameters tested in the experiment) and the interactions between them, presented on a Pareto chart. The increased height of the melted powder layer (B) had the greatest impact on process productivity for both models. The other parameters, in order of their impact on process time, were increased distance between scanning lines (C), increased scanning speed (A), and second-order interactions between these factors (BC, AB, AC). In Fig. 9, it can be seen that the length of the scanning vector (D) had practically no influence on the build rate of the test models.

Fig. 9
figure 9

Standardized Pareto chart for SLM build time

Full size image

When analysing the Pareto chart, it can be observed that there was a stronger effect from changing the layer height and the distance between scanning lines in the case of the thin-walled model than for the solid model. An opposite result was obtained for the scanning speed, in which case the effect was stronger for the solid-type model. We must point out that both models had the same volume of 490 cm3.

4 Discussion

In Fig. 10, the process times carried out with the basic parameters are compared with those in which changes were introduced to increase the build rate. The samples discussed above (solid and thin-walled) were adopted as test geometry cases. The improvements were the use of multi-laser machines and changes in the main parameters by doubling the values, i.e., changing the layer height, changing the distance between scanning lines, and changing the scanning speed. For twin and quad systems, the time of analysis for thin-walled objects was not given, because it was not possible to place objects in such a way as to utilise the potential of all lasers during the manufacturing process. In the case of the thin-walled model, the layer thickening reduced the process time the most among all the factors. For the solid-type sample, the process carried out on a machine with four lasers turned out to be the fastest. Very small differences in process time (approx. 7 min) were found for the process done on a machine with two lasers and the process using an increased layer height. It should be noted that the process time for the twin machine, where both lasers were interacting, while scanning a single sample with an overlap zone was longer than that of the one with a thicker layer and the process with a longer distance between scanning lines.

Fig. 10
figure 10

Comparison of SLM process time for multi-laser systems and for changes in the main parameters to affect the build rate

Full size image

The height of the layer had a significant impact on the build rate of the L-PBF process. It is obvious that a thicker layer of the model means a smaller number of layers required to scan during the model build process and consequently a significantly shorter time of laser beam exposure and a shorter total time of layer deposition. However, it is necessary to consider that a change in layer thickness requires a new process window to be developed to obtain a satisfactory effect. Because of the increased layer height, the quality of a part can deteriorate, showing reduced precision, increased surface roughness, and especially a lower relative density [19]. The changes in layer thickness during melting of nickel alloy powder were the reason for changing the density of energy (Eq. 1) supplied to melt the powder, while the same scanning parameters were maintained (power, speed and distance between lines) (Eq. 1: energy density [20]):

$${\text{VED}} = \frac{P}{V\, \cdot \,h_d \, \cdot \,t}\,\left[ {\frac{{\text{W}}}{{{\text{mm}}^3 }}} \right],$$

where VED is the energy density, P is the laser power (W), V is the scanning speed (mm/s), hd is the distance between scanning lines (mm), t is the layer height (mm).

A lower layer height resulted in a finer grain in the structure of the material. A higher energy density caused an increase in temperature and a faster cooling, resulting from the increased temperature gradient, which led to refinement of the structure. Moreover, with the increase in the height of the deposited powder layer, the level of porosity grew, directly leading to worse mechanical properties: material hardness, modulus of elasticity, ultimate tensile strength, and yield strength. These observations were also confirmed [21]. An excessive increase in the layer thickness, despite the optimisation of the process in terms of porosity, can also result in lowering the yield strength or ultimate elongation, for example. This effect was presented by Leicht et al. [15], whose aim was to determine the influence of changing the layer thickness from 20 to 80 µm on the resulting mechanical properties of 316L steel. These values were 15% (from 540 to 460 MPa) and 30% lower (from 61% to 44%), respectively, for samples made with an 80-µm layer. However, samples made with a thicker layer exceeded the requirements specified for 316L steel in the standard ASTM A240M-18 [15]. A previously presented paper described the effect of optimising the L-PBF process for layer thicknesses of 30 and 60 µm in the case of material quality and preservation of mechanical properties of low carbon medium manganese steel [22]. The change in processing parameters used to melt a twice-thicker layer with an acceptable level of porosity also requires a doubling of the line energy density (LED) from 0.25 to 0.50 J/mm, respectively, but the change in individual processing parameter values is not so clear-cut and, for example, the laser energy used to melt a 60 µm-thick layer is only 100 W higher than the 250 W used to melt a 30 µm-thick powder layer. In the case of the mechanical properties and the crashworthiness of the transformation induced plasticity (TRIP) steel studied by Pawlak et al., the most important factor affecting these characteristics was the residual austenite content of the phase composition obtained during post-process heat treatment.

In certain cases, full material density or high surface quality is not necessary. The automotive industry now uses a large number of parts manufactured in the classic powder metallurgy technology, with densities of 90–95% [15]. The increase in the thickness of the layer of melted material, especially a few times, also has a negative impact on the accuracy of dimensions [19] and the surface quality of objects manufactured with L-PBF technology, despite ensuring good mechanical properties [23]. The minimum value of surface roughness achievable in the L-PBF process is around a few Ra, which differs substantially from the expectations of the aviation industry, for example. This introduces the need to apply post-processing, which means that the surface quality in the as-built state should not be the key factor in process optimisation, which should instead aim at yielding as high a build rate as possible. Furthermore, the quality of the melted material does not only depend on the values of the process parameters, but many other factors related to the batch of powder material also affect the reproduction quality and geometric accuracy [24].

Similarly, increasing the porosity of products manufactured with L-PBF technology does not necessarily mean giving up their use. Increasing productivity by 24.5% with the application of a scanning speed that was two-thirds higher led [13] to obtain a higher porosity of the material porosity (approx. 5%). Despite this, as a result of additional hot isostatic pressing (HIP) treatment, it was possible to cancel this difference and increase the relative density to a level of 99.8%. An additional effect of the optimisation carried out by these authors consisted of increasing the internal stresses in the processed material by 15.8% with the use of a higher scanning speed.

An increase in process productivity can be achieved by increasing the distance between scanning lines; Dong et al. [25], presented the results of an L-PBF process for 316L steel in which this parameter was changed. The distance between scanning lines increased from 100 to 200 μm, resulting in a reduction in the relative density from 99.9% to 95.4%; an increase in surface roughness was also observed, from 2.68 to 6.8 μm, as well as a decrease in maximum temperature and heat accumulation. One factor that allows the distance between scanning lines to be increased is the higher individual track width that results from supplying a higher energy to the bed or changing the level of beam focus and creating a larger spot diameter.

The methods for increasing process productivity presented in this paper are not an exhaustive list of the discussed issue, a particularly important one from the point of view of industry. Many researchers have also analysed other factors related to the modification of optical systems, such as changing the laser beam focus, therefore, enlarging the laser spot, or applying special scanning strategies for the manufactured objects. As a result of a lower beam concentration and contour hatch scanning with a laser beam 0.5 mm in diameter, significantly fewer splinters were observed and evaporation of the material was almost entirely eliminated—in addition to increased productivity [14].

The application of a scanning strategy called ‘hull-bulk’ or ‘hull core’ is a compromise solution to the resulting decrease in surface quality caused by the thicker layer of melted powder. This strategy consists in scanning the contours of manufactured objects on each deposited layer (e.g., 30-µm thick), whose hatch is melted only every nth layer (e.g., every third layer of melted contours). As a result of this strategy, it is possible to obtain a satisfactory—as high as possible—surface quality of the manufactured object at the cost of maximum reduction of the process time, tripling the time of layer deposition and contour scanning. The profitability of the hull-bulk strategy depends on the ratio of the surface of the object to its volume. This ratio directly translates into the proportion of the overall time of the contour scanning on each layer and hatch. If this proportion does not exceed 5 cm−1 it is possible to increase productivity by at least 25%, but if this ratio exceeds 10 cm−1, the use of the hull-bulk strategy is not recommended [16]. Considering the application of the hull-bulk strategy, it is necessary to consider that the benefits resulting from the use of additive technologies grow with the degree of complication of the manufactured object and thus with a higher surface-to-volume ratio.

The data presented here prove that in each case it is necessary to adapt the method to shorten the build time to the customer’s expectations and assumptions or the needs dictated by a specific product application.

5 Conclusions

The aim of this work was to quantify the extent to which the change of selected process parameters and the use of multi-laser systems increase the productivity of the process, and thus shorten the manufacturing times. For the geometry of the samples analysed in the paper, the use of a machine with two lasers led to a reduction of the process time by nearly 45% when the two interacting lasers were employed within one sample, while the second method, where each laser manufactures its own model separately, the process time was shortened by 50% compared to the machine with one laser. In the case of a thin-walled model, the system with two lasers reduced the scanning time by 49% in comparison with the machine with a single laser. For small models, it is more favourable to arrange them so as to be scanned by each laser independently. This allows for a greater reduction of the process time and also avoids the need to select parameters for an overlap zone. In addition, in the case of sensitive products (e.g., implants) it is recommended to scan them by single laser beams, because this facilitates the certification of the final elements. However, it should be mentioned that the even distribution of the scanners that divide the build plate into smaller laser working areas has a positive effect on the quality of the parts. The smaller work area of a single laser minimises the phenomenon of changing the shape of the laser spot along with the distance from the centre of the optical system. There are no large energy density deviations supplied to the powder, and the thickness of the melted track is more uniform.

Depositing material bi-directionally shortens the deposition time of a single powder layer by 2 s. Referring to this value to the overall process time of the model studied here, consisting of 3336 layers, it can shorten the time by nearly 2 h over a system of unidirectional deposition.

In addition, the paper comprises a quantitative analysis of the influence that the most important parameters of the L-PBF process have on the process productivity. Among the selected parameters, the height of the melted powder layer had the greatest influence on the build time. In the analysis, changing the melted powder layer from 30 to 60 µm resulted in practically the same increase in the build rate as the application of a two-laser system, where both sources were working independently (no overlap zone). In contrast, comparing the case where both lasers were scanning one part with an overlap zone, the thicker layer reduced the process time more. This confirms that there is a possibility to significantly increase the build rate even in cheaper single-laser systems. Changes to this parameter should be considered in industrial applications of L-PBF technology only if detailed reproduction precision and surface quality are not key factors or the built part will be post-processed. In situations where surface quality and geometric precision are crucial factors for the process outcome, it is necessary to consider the use of the ‘hull-bulk’ scanning strategy, which is a compromise between substantially increasing productivity and maintaining high surface quality. If the hull-bulk strategy cannot be applied, e.g., due to the ratio of the manufactured object surface to its volume, then it is necessary when optimising the L-PBF process for higher productivity to focus on developing a process window for the longer distance between scanning lines or on increasing the scanning speed.