Modeling of Feedstock Formability to Optimize Mold Design and Prevent Possible Defects During Metal Injection Molding

Abstract

Metal injection molding (MIM) is a current technology used to produce products with complex shapes. Despite the accumulated worldwide experience in using this technology, manufacturers sometimes fail to launch new products of proper quality. Often, this is because defects can occur at the casting stage, the prevention of which is impossible through the use of standard design and technological solutions of molds design and the experimental selection of technological modes. This study aimed to analyze the causes of such defects in the green part and optimize the mold design to ensure defect-free casting, which was impossible for the manufacturer using standard solutions. The core method used in this study was simulation modeling of the casting process. A hypotheses were selected, and an analysis of the causes of defect occurrence in casting was conducted. Simulation modeling proved that defect occurrence was due to the formation of a free melt jet and subsequent air capture by this jet. Based on modeling, different gating and feeding system designs were analyzed, which made it possible to choose a variant that provided the uniform filling of gating cavities without forming jet flows and defects. The novelty of this study lies in its optimization of the mold design to prevent free melt jetting and similar defects in other MIM products by modeling feedstock molding processes. The presented results can help enrich the knowledge base of the mold design for MIM and ensure defect-free production.

1. Introduction

The development of powder injection molding (PIM) technology and its main technological stages is associated with the use of feedstocks, which are dispersed–filled composites with a thermoplastic matrix that allows for the injection molding of a suspension filled with powder of the required chemical composition during melting. At the same time, the study of such multi-phase structured media is the subject of research on the metallurgy of thixotropic materials, which includes the thixoforming of aluminum alloys and steels. The processing of materials in the thixotropic state utilizes the viscosity reduction effect along with the increasing shear rate of the heterophase medium to cast products of the most complex shapes, decrease the production cycle, and expand the possibilities of the manufacturer in choosing the configuration of the future product [1,2,3].

Today, metal injection molding (MIM) technology is among the industrial technologies used for the high-volume production of shaped products made of metals and their alloys [4,5,6]. MIM technology inherits the technological advantages and disadvantages of combining plastic injection molding and powder metallurgy methods. One such feature of this technology is the relatively high complexity of designing and manufacturing molding tools. Because casting completely repeats the geometry of casting cavities, the technological limitations associated with mold production also limit the designer’s options regarding the geometry of a future product [6,7,8]. The most essential aspect that should be considered when designing molds is the complex rheological behavior of the polymer powder, which is a feedstock [9,10,11]. To choose the design of the gating and feeding system, it is also essential to consider the thermophysical properties of the feedstock, because the nature of the process of filling the mold cavity depends on them significantly [12,13,14]. At the feeding and pre-pressing stages of the green part, the pressure, specific volume, and temperature (PVT) characteristics of the feedstock play a crucial role [15,16,17]. Let us note for the wide range of readers not familiar with MIM technology that the term ‘green part’ refers to the composite powder–polymer blank of a part obtained by casting. All the aforementioned feedstock characteristics change significantly within the technological cycle of green part molding, which leads to specific difficulties in designing the system of running channels and the feeder, cooling, and ejection systems; in choosing the location and orientation of the casting; and in the number of nests in the mold [18,19,20].

Lack of experience in designing technological tooling for feedstock molding and the complex geometric configuration of products can lead to errors in the design of molds for MIM, which increases production and future product costs. During the development of MIM technology by industrial enterprises, it is essential to ensure economic efficiency, which is associated with research, development, and engineering costs. Despite many years of experience in the application of MIM technologies and the accumulated knowledge base of the typical design and technological solutions of molds, the experimental selection of technological modes of the MIM process does not always allow manufacturers to launch new products of the proper quality. This is often because defects occur in castings during the injection molding stage. In particular, one of the possible types of defects includes shrinkage cavities and voids with trapped gas, which are very similar in appearance but have different mechanisms of origin and will differ regarding the way the causes of their appearance are eliminated. Therefore, this study aimed to analyze the causes of such casting defects in the green part and optimize the mold design to ensure defect-free casting, which was impossible for the manufacturer using standard solutions.

2. Literature Review

The most significant advantage of MIM is its ability to produce a large series of complex-shaped parts with little or no additional machining. In mass production, it is essential to ensure consistent quality and defect-free characteristics from part to part. Defects can occur at each technological stage of the MIM process, and most of them have the property of technological heredity; that is, they originate at the initial stage (e.g., when casting a green part), are not eliminated at subsequent technological stages, and are inherited in sintered parts [21,22,23]. Sometimes, a defect formed in the early stages may not manifest until the stages of binder removal or even sintering a final part are reached.

For the defect-free production of MIM products, it is necessary to know the defects typical for each stage and the reasons for their occurrence, assessing the level of influence of this or that technological parameter on the possible defects in the product [24]. The most common defects of the MIM process occurring at the feedstock preparation, binder removal, and sintering stages should be thoroughly studied, as well as their manifestations, causes of occurrence, and possible remedies for each feedstock. Many defects occurring at the feedstock injection molding stage are typical for thermoplastic injection molding and have similar causes of occurrence; therefore, the study of feedstock injection molding at this stage should begin with the study of similar defects in the thermoplastic molding process [25,26,27]. However, within the framework of this work, we limited the scope of the analysis of possible defects, mainly to defects in the volume of castings, without considering possible defects on the surface. As shown below, this completely corresponds to the aim of this study to prevent molding defect formation in the volume of the green part.

The development and growth of bubbles are the most common defects in thermoplastic molding. These defects originate as discontinuities in the solidifying material (vacuum cavities) and take the form of round or oblong cavities inside the product during the sintering stage, thereby minimizing the internal energy of the system according to the laws of thermodynamics [28]. They occur mainly in relatively thick-walled products in the zone of the maximum thickness. Therefore, it is essential to distinguish between void–vacuum cavities and cavities formed by trapped air or moisture, which occur for other reasons. The molded product first receives a hard shell, which thickens faster or slower towards the core, depending on the cooling rate of the mold. However, in the zone with a large wall thickness, the core region remains viscoelastic for a long time. The shell is thick enough to withstand the shrinkage stresses. The mass inside is pulled outside, such that vacuum cavities occur in the still-plastic core of the casting. The causes of this occurrence can be attributed to the technological parameters of casting (low pressure and excessively short holding period, excessively low mold temperature, and high melt temperature) and the design features of the mold, such assmall feed system and nozzle hole cross-sections and injection in a zone of low thickness [29,30,31].

Common defects are gas bubbles or voids near the feeder and away from the sprue, and not only near the maximum wall thickness of the molded product. Bubbles can vary in size and geometry. They are present in thermally sensitive materials at high temperatures. A high molding temperature increases the risk of the polymer material degrading, leading to the formation of gas inclusions. The reasons for these defects include excessively long holding times and low pre-pressing material dosage if the molding cycle is long. Local overheating of the mass in the screw cylinder is also possible. The primary causes of gas voids include an excessively high temperature of the molded material, an excessively long holding time of the melt in the cylinder, and unfavorable screw geometry with excessively high compression [30,31,32].

The appearance of the free jet formation effect negatively affects the quality of the final product. free jet formation. A visible gray or mat-melt jet emerges from the feeder and is poured by the subsequent melt. It may be entirely or partially hidden in a molded product. This is caused by the instability of the polymer melt front in the mold nest, particularly when the melt in molds with large cavities forms a high-velocity free jet inside the mold. After the jet leaves the feed system, the hot melt flows onto the relatively cold surface of the mold and cools such that it cannot homogeneously combine with the rest of the molded material during further filling. In addition to visible defects, heterogeneities, internal stresses, and deformations appear in the cold state, thereby reducing the quality of the products. In most cases, only a suitable change in the position and geometry of the feed system (to round off the transition between the feed system and casting, which involves enlarging the feed system, moving the feed system away from the center of the cross-section, and injecting it in the direction of the wall) can reliably eliminate the formation of a free jet [33]. The technological reasons can be an excessively high injection speed and a high temperature difference between the melt and mold [34,35,36].

A cold plug is a cold melt particle near a feeder that forms traces on the product surface and significantly reduces the mechanical properties of the product. This occurs because of a cooled melt in the area of the machine nozzle or hot channel nozzle due to excessively low melt and hot runner temperatures and unfavorable geometry of the feeder and hot runner nozzle (nozzle cross-section is too small). In the case of feedstock, cork flow can occur in addition to cold plugging. This is the phenomenon of the dynamic glass transition of a highly filled suspension [37,38,39].

During the filling process, a weld line, which is a line that connects the individual flow fronts, may occur when the melt flow separates and reconnects, e.g., when flowing around rods. The larger the streamlined rod, the more clearly the weld line is visible and the greater its influence on the mechanical properties. The smaller the flow-merging angle, the more clearly the weld line becomes visible. This is because of the excessively low injection speed and mold and melt wall temperatures, lack of air exhaust at the merge point, and unsuccessful location of the feed system [40,41,42]. Feeder marks, cracks, low dimensional stability, and warps are examples of defects. Some of these problems are easily removable and their causes have been well studied.

As noted earlier, most of the defects occurring during the molding of MIM products, including the aforementioned ones, are similar to plastic molding defects.

Table 1. Defects occurring at the injection molding stage of MIM preforms.

Type of Defect	Possible Causes	Possible Solutions
Flash	Excessively high molding pressure, poor flatness of the molding split line, excessively large ventilation channel	Use larger tonnage machines, correct mold design, use lower injection rate and molding pressure, optimize switch point
Sticking in the injection mold	Excessively high molding pressure, specifics of the thermal shrinkage, early ejection, irrational mold design, and/or poor mold quality	Use lower injection rates, molding/holding pressures, and mold temperatures; increase cooling time; increase mold slopes; change pusher areas and locations; change binder composition
Caverns	Thermal shrinkage, low density	Increase molding/holding pressure and injection rate, reduce mold temperature, increase feeder area, add ventilation ducts, reduce rate while passing through thick sections
Voids	Captured gas, absorbed moisture	Increase holding pressure, reduce injection rate, increase mold temperature, increase feeder area, move feeder to thicker sections
Burn marks	Superheated binder	Reduce injection rate and feedstock temperatures, increase feeder area, change feeder location
Weld line	Supercooled feedstock	Increase injection rate and mold and feedstock temperature, increase feeder area, add ventilation ducts or overflow ducts near the weld line location, relocate feeder, change part design to avoid flow partition
Flow marks	Cold feedstock in the matrix	Increase injection rate and mold and feedstock temperatures, increase feeder area, change feeder location

systematically summarizes core information on common defects occurring in green parts during the MIM manufacturing of products based on data from various sources [21,25,27]. Table 1 provides the causes and solutions for these typical defects.

One of the most relevant publications to the content of our study is the one by Knöpfle et al. [43], in which the authors determined the influence of the casting process parameters and feeder shape on the quality of the formed rods from the Catamold 17–PH A feedstock from BASF SE. The studied factors were the injection speed, holding pressure, material temperature, feeder geometry (rectangular or round), and the use of a vacuum. The influence of several factors was also studied simultaneously. The influence of factors on the surface roughness, presence of a gas inclusion, shape distortion, and change in the length of the part, as well as on the width of the part in the areas close to the feeder and at the end of the flow path, were assessed. Knöpfle et al. [43] showed that the holding pressure had the greatest effect on the quality of the sample. At the same time, the holding pressure did not affect the surface roughness or size of the gas inclusion. The main factors influencing gas inclusion are the use of a vacuum, changes in the injection rate, and changes in the design of the gating system. Note that a defect associated with gas entrapment is externally similar to a shrinkage cavity or a vacuole. Because the methods for eliminating these defects differ, it is initially necessary to determine the nature of these defects, which is not always an easy task. Thus, unlike the work by Knöpfle et al. [43], our study aimed to identify the causes of the defect to determine its type. Only after this can we develop reasonable solutions to prevent the occurrence of defects. The solution to this problem is to conduct a natural or computational experiment in which the absence or presence of the influence of technological casting parameters (except for the use of a vacuum) on the size of the defect indicates its nature.

Thus, an analysis of the literature provides the following. Despite the extensive knowledge base of typical design and technological solutions in molds and the accumulated worldwide experience in using MIM technology, the experimental selection of technological modes does not always allow manufacturers to launch the MIM method to produce new products of the proper quality. Simulation modeling is an effective tool for the technological preparation of new products. The use of the simulation modeling of feedstock casting to analyze the possibilities for defect-free molding of the part, which has proven problematic for MIM manufacturers, is presented below.

3. Objects and Methods

3.1. Objects of Research

The object of this study is a shaped product “guidepost” manufactured by the MIM method (Figure 1). The casting was made using feedstock MIM 4140, which was industrially produced by the company “TCK S.R.L.” (Santo Domingo, Dominican Republic). According to the manufacturer’s technical specifications, the feedstock filler material was a powder of chromium–molybdenum steel of grade 4140 (analog of steel 42CrMo4) with a spherical particle size ranging from 2 to 16 μm and an average D₅₀ of 7 ± 1 μm. The binder material was a polymer wax–polyolefin blend. The volume fraction of powder in the feedstock was 61.8%. The equilibrium density of feedstock at normal pressure P₀ = 1 atm at temperature T₀ = 23 ± 2 °C (corresponding to 296.15 ± 2 K) and humidity 50% was 5.1269 ± 0.0913 g/cm³. A more detailed description of the structure and properties of this feedstock is presented in previous publications [12,16,44].

The manufacturer could not produce this product without molding defects. The manufacturer used a two-cavity mold with a gating and feeding system, as shown in Figure 2. The feed system consisted of a riser and two runners. The mold had four cooling channels in the die plates, punch, and bed plate. This feeder automatically separates the casting from the feed system while the ejector system operates. A general view of the casting with the main overall dimensions is shown in Figure 2.

X-ray computed tomography of the composite semi-finished product, the green part, was conducted to assess the locations and sizes of defects that did not satisfy the requirements of the manufacturer. X-ray scanning was performed using an industrial microfocus computed tomography system Phoenix|x-ray V|tome|x S240, manufactured by Waygate Technologies Company (Wunstorf, Germany). The following parameters were used for tomography: an X-ray tube voltage of 180 kV and a detector resolution of 4 megapixels, which provided a result detail of approximately 1 µm. Figure 3 shows the reconstruction of 3D tomography data performed using Datos|x software in Myvgl 2.0.

Visual analysis of the tomography results showed that the defects formed during the molding of the semi-finished product had irregular shapes and dimensions, but their location in the massive part of the casting near the two horizontal molding marks was the same for all studied samples. Tomograms also illustrate the geometric configuration of a product: the presence of blind and through holes, cavities, and grooves located in different parts of the part, as well as significant changes in the thickness of the product walls. This complicates the mold design and requires additional tools to analyze and validate the design and technological solutions. For this purpose, further simulation modeling of the molding process was conducted.

The technological processes of the manufacturer were set as follows: the temperature of the material at the beginning, middle, and end of the screw, as well as in the nozzle, was the same and equal to 160 °C (corresponding to 433.15 K); the cooling time was 35 s; and the temperature of the coolant (water) was 40 °C (corresponding to 313.15 K). The volume of the material in the screw of the casting machine was 29 cm³. Of this volume, 11 cm³ was injected in the first stage and 7 cm³ in the second stage. Simultaneously, 11 cm³ remained in the screw for feeding. The corresponding injection velocities and pressures during the mold-filling process are listed in

Table A1. Initial parameters of injection molding of a green part used by the manufacturer under industrial conditions.

Process Stage	Injection		Feed
Parameter	1st step	2nd step	1 s, acceleration	1 s	0.1 s
Injection speed, cm3/s	40	25	15	15	15
Pressure, MPa	150	220	50	30	10

. The subjects of this study were the causes of casting defects in the product and the possibility of preventing the occurrence of these defects.

3.2. Research Methods

The methodological basis for the simulation modeling of feedstock injection molding processes has been covered in sufficient detail in many studies [18,45,46]. The methods used in this study are briefly described below.

Special attention should be paid to modeling data on thermophysical, rheological, mechanical, and PVT feedstock characteristics [14,17,46]. They determine the behavior and processability of feedstocks and directly affect the quality of green parts and sintered products. The properties of feedstocks significantly depend on the type of binder, volume filling of the mixture, and granulometric and morphological composition of the filler, which in turn determine the primary processing parameters: the melt and mold temperatures, compaction pressure, dwelling time, and mold filling rate required for softening and increasing fluidity.

The feedstock in the molding process is a highly filled suspension; therefore, simulation modeling of the flow of a single-phase fluid endowed with the efficient macro characteristics of the polymer powder is preferable. In addition to the base equations of hydrodynamics expressing the laws of the conservation of momentum, energy, and mass, the mathematical model of such a flow can include equations that make it possible to describe the redistribution of the powder filler in the binder volume during molding. Compared with two-phase or multi-phase modeling of powder slurry flow, these equations significantly reduce the computational resources required for calculations and the amount of empirical data on feedstock properties necessary.

The most widely used mathematical model describing the flow of feedstock and its yield strength is the Williams–Landell–Ferry model, supplemented by the Herschel–Bulkley model or Cross model [9,46,47]. A correct description of the complete rheological curve allows us to obtain the most reliable data when modeling the flow in the molding cavities of the mold, especially in places of low or, on the contrary, intensive shear influence; in thin sections of the part; and in places where a high melt pressure is necessary to ensure the complete filling of the mold.

Within the framework of this study, the previously developed methodology [46], which considers heat transfer in the mold and the presence of a cooling system, was used for simulation modeling. In the Autodesk Simulation Moldflow Insight (ASMI) 2019.0.2 software applied within the framework of this work for modeling molding processes, known equations describing the laws of the conservation of momentum, energy, and mass for liquids are used [48].

To perform the calculations, finite element models (FEMs) of the gating system, the casting system, the mold, and its cooling system were created using the mesh generator built into the ASMI. Tetrahedrons were used as finite-element mesh elements. The FEM of the casting system with the gating system consists of 2,276,277 finite elements and 418,526 nodes. In this case, the mold model was divided into two parts with different finite-element dimensions. In the model of the mold parts that are in direct contact with the casting and gating systems, the mesh was refined to improve accuracy. A coarser mesh was used for the remaining mold parts, in which the heat exchange was not very active. The total number of finite elements in the industrial mold model was 5,613,226, including 670,584 elements for the cooling system.

The configured FEMs were used in the ASMI for calculation in the {Cool + Fill + Pack} mode, which includes the analysis of the filling, packing, and cooling parameters. The iterative calculation process was completed when the difference between the results of the current and last iterations was not more than 0.5%. During the analysis, the initial melt temperature was 160 °C (corresponding to 433.15 K). The total process took 37.66 s. In this case, the mold opening time for product removal was set to 5 s. The injection rate during the simulation was 40 cm³/s, and switching to packing mode was carried out when the filling volume reached 99%. Based on the simulation results, a melt flow analysis was conducted in the mold to select the design and parameters of the gating and feeding system and their mutual location relative to the casting in the mold.

The thermophysical characteristics of the feedstock during modeling were set in tabular form based on data from a previously conducted study [12]. Figure 4 presents the graphical dependencies of the feedstock heat capacity and thermal conductivity on the temperature. The characteristic peaks on the graph of heat capacity change correspond to the latent heat of the phase transitions of the wax and polyolefin components of the polymer binder in the feedstock.

P-V-T dependence in the form of the Tait 2-domain model [15,45] describes the change in the specific volume of the feedstock in simulation modeling:

$$v(T,p)=v_0(T)\left[1-C·\mathit{ln}\left(1+{\displaystyle \frac{p}{B(T)}}\right)\right]+v_t(T,p)$$

(1)

where v(T, p) is specific volume as a function of temperature and pressure, m³/kg; v₀ is specific volume at zero overpressure, m³/kg; T is temperature, K; $p$ is pressure, Pa; C = const = 0.0894; and parameter B(T) is the parameter that considers the sensitivity of the material to pressure changes at different temperatures and is determined by the following formulas.

For the temperatures higher than the phase transition temperature, that is, T > T_t:

$$v_0\left(T\right)=b_{1m}+b_{2m}(T-b_5)$$

(2)

$$B\left(T\right)=b_{3m}{e}^{[-b_{4m}(T-b_s\left)\right]}$$

(3)

$$v_t(T,p)=0$$

(4)

where $b_{1m},b_{2m},b_{3m},b_{4m},b_5$ are empirical coefficients determined from measurements of PVT feedstock properties.

$$v_0\left(T\right)=b_{1s}+b_{2s}(T-b_5)$$

(5)

$$B\left(T\right)=b_{3s}{e}^{[-b_{4s}(T-b_5\left)\right]}$$

(6)

$$v_t(T,p)=b_7{e}^{\left[b_8\right(T-b_5)-b_{9}p]}$$

(7)

where $b_{1s},b_{2s},b_{3s},b_{4s},b_5,b_6,b_7,b_8,b_9$ are empirical coefficients determined from measurements of PVT feedstock properties.

In this case, the dependence of the phase transition temperature T_t(p) on pressure is expressed by the following equation:

$$T_t\left(p\right)=b_5+b_{6}p$$

(8)

where $b_5,b_6$ are also empirical coefficients determined from the results of analyzing experimental data.

The standard 2-domain Tait model embedded in the software used (as well as in the software of other developers) does not make it possible to make customized modifications, that is, to create customized mathematical models that make it possible to describe the PVT properties of feedstock with the multicomponent polymer binder. In this regard, it was not technically achievable to use PVT data previously measured for feedstock with a two-component wax–polyolefin binder in the modeling [16]. As a result, a model with empirical coefficients determined earlier by Ahn et al. [45] was used for simulation modeling. In this model, the nominal values of the specific volume of the material differ from the specific volume of the MIM 4140 feedstock; however, when calculating the feedstock shrinkage during cooling, the relative, but not the absolute, change in the specific volume was estimated. This can be attributed to the hypotheses and limitations of this study. For the model used, the relative specific volume change at T = const or P = const was higher than that established by the experiment that was considered when processing the simulation modeling results and was sufficient for formability analysis. Figure 5 presents the graphical dependence of the PVT characteristics of the feedstock used for the simulation modeling.

To describe the rheological behavior of the MIM 4140 feedstock, we used data previously measured by capillary rheometry over a wide range of shear rates, as described by the Cross-WLF model [49]. Figure 6 shows the dependence of viscosity on the shear rates used in this study.

Before the start of the simulation, the original mold of the manufacturer and the industrial injection molding machine (IMM) model Arburg 270 manufactured by Arburg GmbH + Co KG (Losburg, Germany) were used to check for the presence of defects in the casting when changing the process modes. In this case, three process parameters were varied: the pressing pressure, injection speed, and melt temperature.

Based on the simulation results, a new optimized mold with a modified melt supply point to the mold cavities was developed. The newly designed mold was tested under laboratory conditions using the IMM model RR/TSMP, manufactured by RAY-RAN Test Equipment LTD (Nuneaton, Germany).

During the field experiments on casting green parts, destructive testing was used, in which a cut in the green part was made in the plane of the probable defect. After that, visual control of the plane of the cut of the green part was carried out using a stereo microscope (brand “SZX12,” manufactured by the company “Sunny Optical Technology” (Ningbo, China)).

4. Results and Discussion

4.1. Analyzing the Causes of Defects by Varying Process Parameters

Two hypotheses were proposed as to the cause of a defect: (1) insufficient compensation of feedstock shrinkage in the process of its cooling in the massive part of the casting, as evidenced by the shape and defect location; and (2) air capture by the melt jet during the mold filling process because of a lack of ventilation channels in the mold (with the gaps between the mold surfaces and rod molding elements and between the pushers and their guide holes contributing). The possible cause of a defect could also include the non-uniform distribution of the powder filler in the slurry during the movement and subsequent cooling, so that cavitation voids could occur. However, this option was excluded because it would lead to warping and the appearance of shrinkage cavities in the defect zone that were not detected.

The manufacturer had already produced a mold for this product; therefore, the first and most rational step was to test different variants of the technological parameters of the injection molding process on the existing tooling. For the Arburg 270 IMM, the values of the pre-pressing pressure, injection speed, and melt temperature were varied. The main tested technologies are listed in

Table 2. Variation in parameters of the casting molding technological process.

Technological Parameter	Values
	Used by the Manufacturer	Experimental
Pre-pressing pressure, MPa	Not more than 120	120, 150, 180
Injection rate, cm3/s	40	10…40 with an interval of 5 cm3/s
Melt temperature, °C (K)	160 (333.15)	160, 170, 180 (333.15, 343.15, 353.15)

, and the full range of the experiments conducted is presented in

Table A2. Technological modes tested with the original mold to assess their impact on the presence of a defect when casting a green part.

Injection Speed, cm3/s	Injection Volume, cm3	Pressure Mode P–t, MPa–s	Temperature, °C (K)
Changing the prepress mode
40	16	120–1 50–1 30–1 10–0.1	160 (433.15)
40	16	120–1 120–2 50–1 10–0.1	160 (433.15)
40	16	150–1 120–2 50–1 10–0.1	160 (433.15)
40	16	150–1 150–2 50–1 10–0.1	160 (433.15)
40	16	180–1 150–2 50–1 10–0.1	160 (433.15)
40	16	180–1 180–2 50–1 10–0.1	160 (433.15)
Change in injection speed
10	16	150–1 150–1 50–1 10–0.1	170 (443.15)
15	16	150–1 150–1 50–1 10–0.1	170 (443.15)
20	16	150–1 150–1 50–1 10–0.1	170 (443.15)
25	16	150–1 150–1 50–1 10–0.1	170 (443.15)
30	16	150–1 150–1 50–1 10–0.1	170 (443.15)
Change in injection speed and temperature
30	16	150–1 150–1 50–1 10–0.1	180 (453.15)
Change in melt temperature
40	16	150–1 150–1 50–1 10–0.1	170 (443.15)
40	16	150–1 150–1 50–1 10–0.1	175 (448.15)

of the Appendix A. Increasing the pre-pressing pressure should reduce the volumetric shrinkage of the feedstock during cooling. The dwelling time of the nominal pre-pressing pressure in all cycles was at least 1 s, which was sufficient for a feedstock of a given shape and size for freezing. Changing the injection rate can change the flow pattern of the suspension in the mold: decreasing the rate results in a higher viscosity and a smoother melt flow front, whereas increasing the rate results in a lower viscosity and a more convex (jet-like) flow front. A temperature change will have a cumulative effect on the volume shrinkage and melt viscosity. Non-technological variants were excluded from the grid of experiments; for example, high melt temperature and low pre-pressing pressure, leading to even more significant shrinkage, and low temperature and low injection speed, leading to the rapid cooling of the slurry, the freezing of the feeder, and mold underfill―a “short shot.”—where excluded.

Varying the technological parameters during the modeling experiments did not eliminate defects; only a slight displacement of the defect from its initially known position was observed (Figure 7). There was no strict correlation between defect displacement and changes in technological parameters. Simultaneously, the absence of the influence of casting parameters on the defect size allowed us to conclude that the physical nature of the defect is trapped gas, not vacuoles (i.e., Hypothesis 1 was not confirmed).

4.2. Simulation Modeling and Analysis of Casting Processes in the Initial Mold

This stage of the study involved the simulation modeling of the injection molding process for the gating and feeding system of the mold used by the manufacturer. The results obtained by the “short shot” injection interruption method and the results of the mold-filling simulation modeling showed a high correlation between the physical and computational experiments (Figure 8).

The injection rate during the experiment was 40 cm³/s. Analysis of the obtained simulation results showed that during the filling process, the formation of a jet was observed at the transition of the melt flow front from one part of the mold to another, that is, a wider part of the mold (Figure 9). For the injection molding of polymers, this is the worst case in the production of products of variable thickness [50]. When jet flow occurs, the possibility of controlling the uniform filling of molding cavities is lost, which can lead to the capture of the existing air in the mold by the melt jet. In this case, the resulting jet was retained until it collided with the rod in the area of the defect fixed in the casting. Because of the absence of a branched system of ventilation channels, this version of defect occurrence in the product became a priority (i.e., Hypothesis 2 was confirmed).

Thus, to solve the problem of jet flow formation, it is necessary to ensure that the mold is filled from the more massive part of the casting to the thinner one. A smooth transition between the different casting parts was not considered because the product geometry was strictly regulated. The primary method to solve the identified problem is to change the feeder location such that it avoids melt flow from the thin section of the casting to the thick one. Thus, the optimization of gating and feeding systems is necessary.

4.3. Optimization of Mold Design

Based on the proposed assumption, different variants of the melt supply to the molding cavities were considered (Figure 10).

There are two blind holes in the massive part of the product that significantly influence the filling of the mold. Different variants of slot and point feeders with the ability to supply and stop the melt flow into the marks forming blind holes will not allow uniform flow around the molding elements and melt movement, leading to air escaping from the mold in possible locations in the ventilation ducts. Shifting the take-over point to the massive part of the casting and pushing the melt flow into the mold wall can ensure the uniform filling of the problematic part without trapping air and complete filling of the thinner parts of the casting.

A previously developed methodology by Semenov et al. [46] was used to analyze the casting process.

Table 3. Effect of feeder type and location of the feeder on injection molding quality.

No.	Type of Feeder	Description of Feeder	Result of Analysis
1	Right point feeder (Figure A1)	Truncated right cone, inlet sprue diameter 1.3 mm, located in the massive part of the casting	Unsatisfactory. Potential gas trapping between the rods.
2	Inclined point feeder (Figure A2)	Truncated inclined cone, inlet sprue diameter 1.3 mm, angle of inclination with horizontal plane 30°, melt flow is directed into the casting wall	Unsatisfactory. Potential gas trapping between the rods.
3	Slot feeder (Figure A3)	The cross-section is an equilateral trapezoid, the contact area with the casting is 6.7 mm2	Unsatisfactory. Potential gas trapping between the rods.
4	Tunnel feeder (Figure A4)	Truncated right cone, inlet sprue diameter 1.3 mm, melt flow is directed to the side wall	Unsatisfactory. Short molding. The character of the rod streamline is satisfactory.
5	Tunnel feeder with increased inlet sprue diameter (Figure A5)	Truncated right cone, inlet sprue diameter 2 mm, melt flow is directed to the side wall	The nature of the filling is satisfactory.
6	Circular point feeder, the first variant (Figure A6)	Four straight-point inlet sprues located at angles of 0 ° and 90° from the symmetry plane, inlet sprue diameter is 1.3 mm	Unsatisfactory. Potential gas trapping between the rods.
7	Circular point feeder, the second variant (Figure A7)	Four straight-point inlet sprues located at an angle of 45° from the symmetry plane, inlet sprue diameter is 1.3 mm	Unsatisfactory. Jet filling character.

lists the results of the numerical modeling for the considered types and variants of the feeder arrangement. The results of modeling the mold-filling process using different feeders are illustrated in Figure A1, Figure A2, Figure A3, Figure A4, Figure A5, Figure A6 and Figure A7.

The results of this study make it possible to design a new model laboratory mold with a modified take-over point of the melt to the molding cavities. The core technical solutions—in particular, the arrangement of rods and marks—were borrowed from the original design of the manufacturer’s mold, but the number of slots was reduced to one, and ventilation channels were added at the expected end of the mold filling. Testing was performed using a laboratory RR/TSMP IMM, manufactured by RAY-RAN Test Equipment LTD (Nuneaton, Germany).

For the melt supply, the design of the tunnel feeder was chosen (Figure 11) to be a truncated inclined cone with a sprue inlet opening diameter of 2 mm and a height of 6 mm. The chosen design best ensures the mold-filling process because the sprue inlet is located such that the melt flow at the feeder exit is directed into the wall of the molding cavity. The impact of the feedstock melt on the wall prevents the formation of a free-melt jet.

In the developed model laboratory tooling, the molding cavities in the die and punch retained the base geometric characteristics of the casting. However, to simplify the manufacturing, several changes were made: all the sharp corners of the casting were rounded, and a continuous smooth surface replaced the grooved surface of the casting. The arrangement of the rod elements did not differ from that of the molded rods in the manufacturer’s mold. However, to simplify the mold production, all sliders were replaced with fixed rods, and the geometry of the rod elements was simplified. Figure 12 shows a set of rod elements. The ejector system in the developed mold consisted of six pushers located in the cooling plate and die.

4.4. Analysis and Discussion of the Results Obtained with the Optimized Mold

Simulation modeling of the injection molding process was conducted for the proposed optimized mold design. Figure 13 shows the simulation modeling results characterizing the temperature distribution in the mold along the casting length and thickness at t = 1.24 s. This figure also shows the location of the analyzed nodes (nodes 589490, 518144, 589490, and 518144) in the mold sections.

Compared to the manufacturer’s process, the temperatures in the casting at the end of the cooling stage were lower, and the temperature in the massive part of the casting was 27 °C (corresponding to 300.15 K), whereas in the original process, the temperature in this area was 60 °C (corresponding to 333.15 K). It was possible to slightly reduce the temperature gradient along the length of the product, particularly at the end of the cooling stage. The maximum temperature difference along the length was 76 °C (corresponding to 349.15 K) at t = 3.4 s. This gradient was caused by a feature of the casting design, namely, a significant difference in the wall thickness.

Figure 14 shows the temperature distribution of the mold parts in contact with the casting over time. The temperature gradient along the length of the mold during the entire process varied slightly and did not exceed 2 °C (corresponding to 2 K). The maximum temperature gradient along the thickness was 8 °C (corresponding to 8 K) at t = 0.6 s. However, by the end of the pre-pressing stage, the thickness temperature gradient decreased to 3 °C (corresponding to 3 K). This evolution of the mold temperature gradient along the length and thickness of the casting was acceptable for injection molding. The nodes indicated in Figure 14, as well as the temperature distribution in the mold, are shown in Figure A8.

A full-scale experiment was performed to confirm the simulation modeling results for the possibility of defect-free molding using the new mold design. An optimized mold was manufactured using the geometric models developed for the physical experiment (Figure 15). In Figure 15, position 1 marks the rod and sign locations, and position 2 marks the added ventilation ducts at the intended end of the mold filling.

This study involved manufacturing ten experimental castings and investigating the massive areas of the manufactured samples for defect presence. The investigation proved that there were no voids in these areas of the green parts manufactured using the optimized mold (Figure 15, right).

Thus, the field experiment revealed that the castings obtained using the optimized tooling did not have the original defects that those produced by the manufacturer had, which confirms the validity of the conducted calculations, the results of the analysis, and the decisions made to change the mold design to prevent the occurrence of defects that prevented the manufacturer from setting up the launch of new products. In general, the results presented in this study can be useful for replenishing the knowledge base for the design of molds for MIM technology and ensuring the defect-free molding of green parts [51,52].

5. Conclusions

This paper describes a sequence of actions for eliminating possible casting defects and selecting design and technological solutions for MIM molds, based on an analysis of the results of simulation modeling of the formability of the raw material and the causes of possible defects. The main focus of this work is how to distinguish externally similar defects in the form of shrinkage cavities and voids due to air entrapment. Only after determining the causes of the defects does it become possible to choose reasonable strategies to prevent the occurrence of such defects.

To identify the causes of the defects in the part, advanced physical and numerical experiments were conducted. The present study corroborates the imperative of employing simulation modeling of the injection molding process of powder–polymer composites, even at the technological preparation stage of MIM production of new products with a complex and non-standard geometric structure for this technology. Notably, substantial qualitative agreement was achieved between the physical and computational experiments, despite the utilization of an incomplete set of input information regarding feedstock properties. The findings of this research enabled the verification of the hypothesis concerning the root cause of the defect, the design and fabrication of novel technological tools, and the elimination of defects identified by the manufacturer.

Based on simulation modeling, it was proven that the defect was associated with the formation of a free melt stream and the subsequent capture of air by this stream. Furthermore, based on modeling, various designs of the gating and feeding system were analyzed, which made it possible to select an option that ensures uniform filling of the gating cavities without the formation of a jet flow or a defect.

The constructed mold, in conjunction with its constituent elements, was subjected to testing in an injection molding laboratory machine, yielding flawless castings. Thus, the problem of casting defect formation can be solved by changing gating and feeding systems. This outcome substantiates the validity of the postulated hypothesis and the efficacy of the measures implemented to forestall the technological deficiencies. Consequently, the present study validates the efficacy of integrating numerical modeling and system analysis in technological processes.

The novelty of this work lies in how, by simulating the processes of raw material molding, manufacturers can optimize the mold design to prevent the occurrence of free melt flow and defects in the form of shrinkage cavities due to air entrapment in other MIM products. This approach allows not only the prediction of defects in MIM products, but also the improvement of product quality and reduction of production costs. In addition, the originality of this work is associated with how to distinguish between externally similar defects in the form of shrinkage cavities and voids caused by air entrapment. After determining the causes of the defect, it is possible to select reasonable methods for its elimination. At the same time, the limitations of the study include the fact that it does not consider the use of a vacuum, as this complicates and increases the cost of the technological process. A limitation of the presented results is that they are applicable specifically to the analysis of defects in the form of shrinkage cavities and voids with entrapped air. However, if the defects considered in this work were associated with shrinkage, this would require a change in the pressing pressure rather than an adjustment to the gating and feeding system to prevent jet flow. In this case, data on the PVT characteristics of the feedstock are of paramount importance. Moreover, the direction of further work in this case is toward the development of models for describing the PVT characteristics of feedstocks with a multicomponent binder, since now most studies are associated with the description of the PVT characteristics of feedstocks based on polyoxymethylene.