33. Increasing Cleanliness of Al-melts by Additon of Ceramic Fibers

This chapter deals with the increase in filtration efficiency by additional usage of ceramic fibers. The results shown here are generated in a water model system, which is already introduced earlier in subproject B01 and B04 [1‐5]. Based on the properties of water, mainly it’s dynamic viscosity, it can be suited as a model melt when additionally the poor wetting is considered. Since most of the impurities found in Al-melts are alumina based, they are also applied to the model melt. Besides experiments in water model system tensile test of castings are taken to determine the mechanical properties. Hence, the impact of ceramic fibers on a sand casted AlSi10Mg-alloy is described.

Usually, CFF’s are applied to remove particulate non-metallic inclusions as they lead to a decrease in mechanical properties of the casting [6]. A crude classification of these ceramic foam filters with porosities typically above 70% takes place with the number of pores counted per square inch (ppi-value). The higher the ppi-value the smaller the pores. The strut thickness decreases, but as the solid volume increases the total solids volume and hence the ceramic foam surface increases [5]. Thereby the contact probability between particles in the melt and the filter strut increases and additionally more surface is available for particle separation. It results from this correlation a higher filtration efficiency (Fig. 33.1).

A bar graph with error bars plots eta 1 versus 10 and 30 p p i. The given values are as follows. 10 p p i, 12.7. 30 p p i, 52.3.

The filtration efficiency \(\eta\) in water model system is given by the following Eq. 33.1:with the mass of particles separated in the CFF \({{\text{m}}}_{{\text{P}},{\text{CFF}}}\) and the mass of particles in the filtrate, respectively separated particles in the filter cloth \({{\text{m}}}_{{\text{P}},{\text{FC}}}\) subsequent to the CFF in the water model system. The filtration efficiency in the water model system is, as already stated dependent on the structure parameters of the CFF, as well as the approaching velocity of the melt, the particle system and the particle size distribution [5]. Investigations in Al-melts using LiMCA-system and / or micrographic analysis also depict these dependencies [7].

But, depending on the casting method the application of ceramic foam filters is limited to their ppi-value, as the pressure gradient is directly connected to the pore size. If the resistance of the CFF is too high for the melt to flow through the pores a freezing might occur as the melt cools down below liquidus temperature and leads to a casting stop (see Fig. 33.2).

A photograph of the melted aluminum solid placed on the casting.

Knowing that the ppi-value of the CFF needs to match the casting method in terms of process stability it follows from the above conclusions that small impurities in the single digit µm-range are difficult to be removed with CFF, as transport mechanism to the CFF strut are weak (Fig. 33.3). Thus in the case of sand casting the employed CFF commonly exhibit a ppi-value of around 10.

a. A scatter plot of the key value versus x. 4 plots for R e, N I, S t, and N S are distributed in an increasing manner between 1 E minus 04 and 1 E + 02. The plots for R e have the highest value. b. A dot plot with error bars of eta 1 versus P S D. The dots are marked at 2.1, 8.7, and 34.1.

The restricted application of CFF due to their ppi-value in respect of process stability during casting leads to a lack of impurity removal in the single digit µm-range which needs to be balanced. A promising approach to remove impurities in this order of magnitude is the application of ceramic fibers. The employed ceramic fibers CeraFib 99 from CeraFib GmbH are made of corundum with 99% of Al₂O₃ and 1% of oxidic additions [8]. These fibers feature a diameter of around 10–12 µm (Fig. 33.4) and exhibit a continuous operation resistance temperature of 1250 ℃ which is above common Al-melt casting temperatures (720–750 ℃). Due to these characteristics the application of these fibers for removing impurities in Al-melts is given.

a. A photograph of the CeraFib 99 fibers at 2 centimeters. b. A SEM image of the fiber bunch at 20 micrometers. The morphology of the sample has a stick like structure with a rough surface.

The implementation of small amounts of these ceramic fibers in the water model system with vertical flow through on top of the CFF`s already has a positive effect on the filtration efficiency (Fig. 33.5). Independent on the approaching velocity already minor amounts of 0.4 g of fibers added on top of a 30 ppi CFF with the dimensions of 50 × 50 × 20 mm³ lead to an increase of the integral filtration efficiency. Higher amounts of added fiber mass lead to a further increase in integral filtration efficiency. In this experimental setup the CFF acts as a carrier and prevents the fibers from entering in the filtrate, which is equal to a contamination of the casting. Therefore a fiber length of at least 15 to 20 mm ensures a detention when 10 and 30 ppi carrier CFF are used. The average pore diameter of these CFF is 4.92 mm respectively 2.30 mm. Computer tomography was used to gather this information. However, fiber mass can not be arbitrary raised, due to a lift in pressure drop as well which often goes along with a freezing of the melt.

A double bar graph plots eta 1 versus fiber mass. The values are plotted for 0.14 and 0.04 kilograms per second as follows. Fiber mass 0.1, 7.9, 8.7. Fiber mass 0.4, 19.8, 26.3. Fiber mass 1.0, 31.2, 37.8.

Two options of ceramic fiber addition with potential industrial application is introduced subsequently. The insertion of fibers into the impured model melt before the CFF derived from precoat filtration provides one means of fiber addition. Here, the fibers can start separating the impurities prior to the actual casting. During the casting process these fibers with separated impurities are washed up on the carrier CFF and can continue with the cleaning of the melt.

Placing the ceramic fibers in between two CFF´s is the second possibility. An optimal positioning of the fibers is ensured when the CFF´s exhibit a hollow space as it is shown in Fig. 33.6 and is consecutively stated as encapsulated arrangement. Furthermore, this adjustment guarantees the whole melt to pass through the fibers and raises the contact probability between impurities and fibers.

A photograph of 2 C F F at a size of 1 centimeter. The C F F has a porous structure with hollow space at its center.

Water model experiments with the encapsulated arrangement of 30 ppi CFF and varying fiber masses were carried out to determine the pressure drop \(\Delta P/\Delta L\). Two different mass fluxes of 0.15 and 0.5 kgs⁻¹, which corresponds to approaching velocities u of 12 and 40 cms⁻¹ respectively, were adjusted. The results indicate the relation between pressure drop and fiber mass but also approach velocity. However, in the case of the water model system no freezing of the model melt occurs and the evaluation of an upper limit for fiber mass needs to be evaluated for sand casting with Al-melt as well (Fig. 33.7).

A double bar graph plots delta P slash delta L versus fiber mass. The values are plotted for 0.15 and 0.5 kilograms per second as follows. Fiber mass 0, 3, 33.8. Fiber mass 0.3, 5.8, 45.8. Fiber mass 1, 9.3, 85.5.

Sand casting trials with an AlSi10Mg alloy were carried out at Formguss Dresden GmbH to determine the maximum amount of fiber mass that can be added in between two 10 ppi CFF (35 × 35 × 22 mm³). By placing 0.06 g of ceramic fibers in between the CFF´s the time for casting a mould (Fig. 33.8) of 4.5 kg increases from 14 to around 20 s, which means that the mass flux decrease from 0.32 kgs⁻¹ to 0.23 kgs⁻¹. This an indication for a raised pressure drop. The casting temperature was about 720 ℃. It can be concluded for sand casting with ceramic fibers and two 10 ppi CFF a maximum amount of 0.06 g of fibers can be added (Fig. 33.9).

a. A photograph of the mold stand consists of the lower mold half, sprue, C F F, mold cavity, molding plane, and closed riser. b. A photograph of the molten liquid is pouring into the mold.

A photograph of the fiber is sandwiched into the C F F.

Hence, further evaluation of filtration efficiency in water model system is done with 0.05 g of ceramic fibers. The addition of this amount of fibers seems to be few, but looking at the surface of the fibers with a diameter of 10 to 12 µm it becomes clear that they significantly increase the surface provided for possible particle separation (Table 33.1).

Moreover, the small diameter, compared to a 10 ppi CFF strut of around 1.6 mm, gathered with computer tomography (Zeiss Xradia 510 Versa) at the Institute of Mechanical Process Engineering and Mineral Processing, leads to a high contact probability between impurities and fibers. Whereas for the 10 ppi CFF with a depth of 20 mm the particles in the melt have a relatively low contact probability as only around 8 to 10 struts are passed. A 10 ppi CFF (50 × 50 × 22 mm³) further exhibits a medium pore size of around 4.9 mm. Having a material volume of 9.7 cm³ the surface provided for particle separation is 237 cm². In comparison to that, a 30 ppi CFF is characterised through a strut thickness of 0.9 mm and a pore size of 0.23 cm. The volume yields to 12.1 cm³ and results in a suface area of 520 cm².

The addition of 0.05 g of fibers with a diameter of 10–12 µm accompanies with additional surface of 47 cm², which is about 20% of surface for potential particle separation, but a volume of only 0.013 cm³. A characterization of fibers and CFF´s is summarized in Table 33.1.

Experiments in water model system with 30 ppi CFF in encapsulated arrangement (50 × 50x40mm³) and use of different fiber diameters from 10–12 µm and 300 µm illustrate the positive effect of fiber addition. There, 0.05 g of the fibers with a diameter of 10–12 µm is placed in between the CFF. In the case of the fibers with a diameter of 300 µm made of nylon thread and a density of 1.15 g/cm³ a mass of 0.13 g is applied. Hence, an almost tenfold higher volume of 0.0113 cm³ of fibers is added but the provided surface is only 15 cm² and thus less than one third. Nylon thread is employed to investigate the influence of fiber diameter on the filtration efficiency in water based model system. It can not be used for Al-melts. For this experiment, two different approaching velocities of 3.2 and 8 cm/s were choosen. The impurities are represented by Al₂O₃-particles with a particle size distribution of 2–20 µm. The results of the integral filtration efficiencies and the corresponding local separatetd masses can be seen in Fig. 33.10 respectively Fig. 33.11a and b.

A grouped bar graph plots eta 1 versus u. The values are plotted for no fibers, 300, and 10 micrometers as follows. u 3.2, 8.2, 12.2, 12.3. u 8, 4.5, 7.5, 10.3.

2 vertical stacked bar graphs of m P F versus u. The values are plotted for top C F F, fibers, and bottom C F F. a. The top C F F has the highest value for 300 micrometers at 0.10. b. The top C F F has the highest value for 10 micrometers at 0.08.

Approaching two 30 ppi CFF with 3.2 cm/s in the water based model system reaches an integral filtration efficiency of 8.2%. An increase in approach velocity decreases the integral filtration efficiency to 4.5%. The usage of fibers independent on the fiber diameter leads to an increase in integral filtration efficiency but at higher approaching velocities the influence of the fiber diameter becomes visible. Smaller fiber diameter have the advantage of providing more surface and raising the contact probability and thus increase the integral filtration efficiency. Thus, the lower the diameter the higher the integral filtration efficiency and fiber addition is more beneficial even when only small amounts are provided.

A more detailed view on the contribution of fibers is gained when analyzing the local separated particle masses (Fig. 33.11). For both approaching velocities, a lower fiber diameter indicates a stronger affection on the particle separation. However, in the case of a higher approaching velocity of around 8 cm/s an impact on the particle separation in the up- and downstream CFF is visible. Compared to the separated particle masses in the CFF alignment with no fiber use the particle separation in the up- and downstream CFF is increased. In the case of 10–12 µm fiber use the upstream CFF exhibits with 0.08 g even a doubled mass of separated particles.

As already mentioned, a fiber addition derived from precoat filtration is another possibility and can be directly compared with the encapsulated arrangement in water based model system. A application of precoat filtration in the semi-automated pilot plant which operates with a constant flow rate by adjusting the pressure in the vessel is not possible due to a time consuming cleaning subsequent to each filtration trial. Some fibers remain in the vessel without being washed up at the upper side of the approached CFF and gravimetric analysis is complicated due to complex dismantling of the storage vessel. Thus, these trials are carried out in a small-scale water based experimental setup with easy access to the storage vessel. But the experimental setup, as well as the vertical flow through of the CFF´s is similar to the semi-automated pilot plant. For these experiments the already well known Al₂O₃-fibers with a diameter of 10–12 µm are used by adding 0.05 g in the stirred storage vessel of the with 2 to 20 µm Al₂O₃-particles impured model melt. The approach velocity for these trials was set to 10 cm/s.

The results can be seen in Fig. 33.12. A comparison between no fiber addition, precoat fiber addition and encapsulated alignment is done.

A vertical stacked bar graph plots m P F versus no fibers precoat filtration, and encapsulated fibers. The given values are as follows. No fibers, 0.0084, 0.0113. Precoat filtration, 0.008, 0.0113, 0.0037. Encapsulated fibers, 0.0136, 0.0084, 0.0160.

Advantageously, the precoat fiber addition increases the residence time of the fibers in the water based model melt and hence should increase the contact probability too. Impurities already can adhere to the fibers in the storage vessel and are washed up at the CFF surface. However, as already mentioned some fibers may remain in the storage vessel. This behaviour is also observed when these fibers are added to an AlSi10Mg melt, where fiber addition becomes challenging due to the high surface tension of the Al-melt. In contrast to the precoat filtration, a placing of the fibers in between two CFF lowers residence time which is then directly related to the approach velocity.

However, the usage of 30 ppi CFF with a depth of 40 mm leads to a total separation of 0.02 g of particles. Thereby the upstream filter separates more particles than the downstream filter (0.0113 to 0.0084 g) which is already observed in B01 and literature as well [9]. Insertation of fibers in the storage vessel leads to an additional separation of 0.0037 g of impurities. A varying amount of particle loaded fibers still remains in the storage vessel and had to be taken out for evaluation of the mass of separated particles. With precoat filtration there is no affectation on the amount of separated particles in the up- and downstream CFF, where separation of 0.0113 and 0.0084 g respectively took place.

Encapsulated fiber addition leads to a total amount of 0.0375 g for this arrangement. Thereby the separated mass of particles is much higher and the advantageous placing of the fibers leads to more than a doubled amount of separated particles in the fibers (0.0084 g), compared to precoat filtration. One reason might be that this arrangement assures the entire water based model melt with all the impurities to pass through the encapsulated alignment.

The positive affectation on the particle separation of the surrounding CFF is obvious, as in the up- and downstream CFF´s the particle separation is raised to 50% compared to the precoat filtration. This finding in small scale water based experimental setup assists the results obtained with approaching velocities of 8 cm/s and illustrates the capability of encapsulated fiber addition, especially at higher approaching velocities on particle separation. This promising approach lead to a patent application for metal melt filtration [10]. SEM-analysis of loaded fibers with a diameter of 10–12 µm in water based model system proved evidence of particle separation in single-digit µm range (Fig. 33.13).

A microscopic image of the A l 2 O 3 fibers at 40 micrometers. The morphology of the fiber has deposition of A l based particles on its surface.

A transfer of precoat fiber addition to a sand casting with AlSi10Mg alloy needs to clear the addition of the fibers into the Al-melt. Therefore, several fiber addition strategies were testet in a hand ladle with ca. 2 kg of Al-melt. A simple adding on top of the melt with a subsequent mixing is not productive as the fibers form a bunch, surrounded by an oxide layer. A foregoing removal of the present oxide layer shows no positive effect, as the oxide layer forms immediately again. Placing the fibers on the bottom of the hand ladle and filling it with Al-melt also does not lead to a homogeneous dispersion of the fibers, as the fibers will remain at the interface of the melt. Wrapping the fibers in Al-foils before addition is also non-satisfying as the Al-foil melts immediately and the fibers are hold back at the interface and can not penetrate the melt. Further investigations were done with casting trials of cubes with the dimensions of 50 × 50 × 50 mm³. Besides the simple placing of the fibers on the bottom (Fig. 33.14a) more options were tested where half of the cube was filled with Al-melt. Then, the fibers were added on top of the still liquid melt and doused with further Al-melt. But the fibers were not wetted. Besides this liquid/liquid strategy the fibers were added on a semi-solid / mushy Al-melt which prevents the fibers from floating when further Al-melt is added (Fig. 33.14b). This method turned out to be the only possible investigated way of fiber addition. These cubes with a known amount of fibers then can be added to a crucible for precoat filtration. A subsequent casting of this fiber treated Al-melt into ingots was performed to estimate where these fibers are located and if impurities might adhere. Therefore, the ingots were cut into several specimen and the cross-sectional area was observed. But no fibers were found as they mostly remain in the hand ladle with other backlogs, such as oxide skins. A practical transfer of ceramic fiber precoat addition into an Al-melt casting process seems to be difficult in terms of fiber dispersion due to the high surface tension of the melt (Fig. 33.15).

a. A photograph of the sand mold cube with a size of 2 centimeters, and the fiber is placed at the top surface of the mold. b. A photograph of the sand mold cube with a size of 2 centimeters, and the molten fiber is poured into the mold.

a. A photograph of the rectangular ingot casting bar. b. A photograph of the casting bar, which is sliced into 5 unequal parts horizontally and vertically.

Besides precoat fiber addition the promising results of the water model experiments are transferred into a AlSi10Mg sand casting system with the experimental setup already shown in Fig. 33.8a. According to Formguss Dresden GmbH the alloy is refined with strontium. A mass of 0.03 and 0.06 g of the fibers were placed in between two 10 ppi CFF (35 × 35 × 22mm³).

AlSi10Mg alloy is a casting alloy, which is characterized by a dendritic solidification of the aluminium rich α-solid solution. As already stated, the impact of the ceramic fibers on the particle separation mainly in single digit µm-range should be observed. SEM analysis indicates the presence of gaseous inclusions with diameter of more than 100 µm (Fig. 33.16). Thereby, these pores superimpose the decreasing effect of the present particular inclusions below 10 µm on the mechanical properties and need to be removed with a hot isostatic pressing (HIP). The HIP was performed at bodycote GmbH and the application of a high temperature and a confining pressure closes the pores. Thus, the porosity is reduced to 0.1% and the impact of the fibers on the smallest particular inclusions in the casting can be evaluated.

a. A SEM image of the test specimen at a diameter of 16 millimeters with the distribution of gas inclusions. b. A microscope image of the specimen at 100 micrometers. The morphology of the sample has a cracked surface with several pores.

Subsequent to the HIP a T6 heat treatment, which consists of a solution annealing (525 ℃, 6 h), a quenching in water and an artificial ageing (167 ℃, 7 h), was performed. The heat treatment increases tensile strength as well as yield strength and decreases the elongation at fracture. Thus, the sensitivity in respect of present impurities increases when tensile testing is performed.

In order to clarify the experimental proceeding consisting of melt refining, HIP and T6 heat treatment it has to be pointed out that the mechanical properties of the test specimen are already shifted to elevated values as the largest defects and impurities are removed. Nevertheless, a positive influence by adding fibers is still visible (Fig. 33.17).

A double bar graph plots R P 0.2 and run versus specimen. The values are plotted for R p 0.2 and run as follows. No C F F, 220.5, 260. 10 p p i, 222.7, 261.3. 10 p i i + 0.03 grams fiber, 227.6, 266.4. 10 p p i + 0.06 grams fibers, 233.8, 272.2.

The mechanical properties of the as-cast specimen are 93.5 MPa respectively 139.0 MPa for yield strength and tensile strength. Elongation at fracture is about 2% and Brinell hardness number is 57.0 ± 1.9 HBW10. A HIP and T6 heat treatment increases the Brinell hardness to values of 99.0 ± 3.7 HBW10 due to fine dispersed precipitations.

It should be noted, that the application of a 10 ppi CFF is not apparently ineffective. But the effect of a HIP and T6 heat treatment is much higher, as they lead to a strong raise in the yield strength to 220.5 ± 8.2 MPa and tensile strength increases to 260.0 ± 8.7 MPa. Nevertheless, a fiber application of 0.03 g still increases the yield and tensile strength to values of 227.6 ± 3.4 MPa respectively 266.4 ± 2.4 MPa. Further increase of fiber application to 0.06 g again increases the yield and tensile strength to 233.8 ± 3.6 MPa respectively 272.2 ± 4.6 MPa and proves the effectiveness of ceramic fibers on the particle separation.

It also needs to be mentioned, that the application of 10 ppi CFF and fibers leads to a significant decrease in the iron content of up to 50% determined with spark spectrometry (Figs. 33.18 and 33.19). It can be assumed that the CFF as well as the fibers remove iron rich phases from the Al-melt, which has a positive effect on the mechanical properties. Structural analysis with SEM and light microscope shows the presence of intermetallic β-Al₇FeSi₂ phase which is known to be hard and brittle and thus unwanted. Although the presence of non-metallic inclusions, such as MgSi₂O, is also reduced by applying CFF and fibers (Fig. 33.19). Thereby, Mg-silicates are common non-metallic inclusions in AlSi-alloys [11]. The employed 10 ppi CFF reduces the proportion of MgSi₂O already to 32% and fiber use of 0.03 g leads to a proportion of 14%.

A scatter plots of F e content in percent versus specimen. The plots for no C F F, 10 p p i, 10 p p i + 0.03 grams fibers, and 10 p p i + 0.06 grams fibers are marked at 0.27, 0.16, 0.15, and 0.12.

A scatter plots of proportion versus specimen. M g S i 2 O plots for no C F F, 10 p p i, 10 p p i + 0.03 grams fibers, and 10 p p i + 0.06 grams fibers are marked at 0.17, 0.05, 0.02, and 0.02. Beta A l 7 F e S i 2 are marked at 3.5, 3.1, 1.2, 2.3.

The analysis of the ceramic fibers after a sand casting is performed. There, the fiber radius can be estimated with 5.4 ± 0.14 μm and is consistent with the manufacturer specification where a diameter in between 10 and 12 µm is stated (Fig. 33.20a). The fibers are molded in the AlSi10Mg-alloy. A detailed view of a microsection with ceramic fibers proves the separation of non-metallic inclusions with a diameter close to the single digit µm-range (Fig. 33.20b). The chemical composition of this Mg–silicate was determined with EDX-measurement to 58.5 at% oxygen, 21.5 at% magnesium and 14.1 at% silicon. This detection, as well as the verifiable decrease of the non-metallic inclusion MgSi₂O shown in Fig. 33.19 lead to the conclusion of a positive influence of the fibers on the melt filtration and hence the mechanical properties.

a. A microscopic image of the ceramic fiber at 20 micrometers. The morphology of the fiber has a fractured structure with the formation of inclusions with a radius of r 1 and r 2. b. A microscopic image of the fiber at 50 micrometers. The morphology has a cracked surface with inclusions.

A more detailed study of the particle separation on the ceramic fibers in water based model system was done with a cone beam computer tomograph Xradia 510 Versa from Zeiss with a resolution of 0.376 µm voxel size[12]. The approaching velocity was 3.2 cm/s. The impurities are represented by Al₂O₃ with a initial PSD of 2–20 µm. The segmentation was done in the image processing software ImageJ and the data evaluation was done with the software VG STUDIO MAX 3.3 (Fig. 33.21).

A 3 D simulation model plots the diameter of the ceramic fibers. The x, y, and z axes range from 100 to 250 micrometers. The fibers are generated in the graph, and the coherent particles are deposited on the surface of the fibers, which are represented by the different color codes.

For a better understanding of the data, it is important to note that the particle separation in Fig. 33.21 represents the end of the filtration experiment in the water based model system. With ongoing time of filtration, particle separation takes place and more and more particles are entrapped. Thus, it is not possible to conclude from this snapshot on the particle size of the impurities when they separate. But the evaluation of the number of particles according to their diameter proves evidence of the assumed particle separation in the single digit µm-range (Fig. 33.22). It becomes visible, that the majority of separated particles is in the single digit µm-range. 128 particles exhibit a radius of 5 µm or below. More 47 particles are in the range of 6 to 10 µm. Of course, the volume represented by these impurities is dwindling small compared to the largest separated coherent particles. But as the initial state of the particle diameter just before the separation can not be evaluated the data needs to be read as a snapshot at the time where the filtration experiment stopped.

A histogram plots the number N of particles versus radius. The y axis ranges from 0 to 140, and the x axis ranges from 0 to 95. The radius 0 to 5 has the highest value at 128, and the radius 36 to 40, 66 to 70, and 91 to 95 have the lowest value at 1.

An evaluation of this data set in VG STUDIO MAX 3.3 with the sphere method tool indicates that also the clustered coherent separated particles exhibit a diameter in the single digit µm-range with a modal value of 5.3 µm (Fig. 33.23). Thereby, detected particles range from 0.5 to 8.0 µm. During particle separation on the fibers it can be assumed that the particles start to separate on single fibers as it can be seen in Fig. 33.23 on the right side and with further separation the particles grow into the fiber interspace until they form bridges and coalesce to a large cluster.

A 3 D simulation model plots the diameter of the ceramic fibers. The fibers are generated in the graph, and the coherent particles are deposited on the surface of the fibers, which are represented by the different color codes, ranging from 0 to 13 micrometers.

The presented results obtained in water model system setup as well as in sand casting with an AlSi10Mg-alloy indicate a high potential of fiber addition on the particle separation for enhanced melt cleanliness and thus better mechanical properties. Two different possibilities of fiber addition were studied, the precoat fiber addition and the encapsulated fiber addition. The water based model melt results clearly indicate the advantageous fiber addition in encapsulated arrangement where they develop their full potential even with positive effect on the up- and downstream CFF with local enhanced particle separation. In contrast, the precoat fiber application may appear at first glance beneficial as the fibers are inserted into the storage vessel in the water model system respectively the crucible. But a satisfactory fiber addition into the Al-melt could not be achieved. Furthermore, the precoating of all fibers can not be fully ensured on the CFF as some fiber remain in the crucible.

A transfer of encapsulated fiber addition in sand casting of an AlSi10Mg-alloy confirms the findings in water based model system as the mechanical properties, in particular the yield and tensile strength are increased. The evaluation of loaded fibers in water based model system with high-resolution computer tomography proves the assumption of significant particle separation and provides information regarding the particle deposition to the fibers. Thereby, only small amounts of fibers in the range of 0.05 g are needed. The addition of the ceramic fibers can be simply made by placing them in between two CFF [10] and thus requires no extra investment of time or knowledge and could be immediately implemented in manufacturing process.

It should be noted that the increase of melt cleanliness can be achieved with a great many of actions through CFF application. In this context it is worth mentioning the investigations in B01 where, among others, the impact of graduated CFF structures with varying ppi-values or the CFF surface roughness was studied demonstrating the positive influence on the filtration efficiency [5, 13]. Also, alternative CFF arrangements were evaluated. The replacement of a CFF volume through such as a loose filling of diced CFF with equal volume is proved to increase the filtration efficiency in water model system [14]. A sand casting of this CFF alignment lead to an increase in tensile strength from 266.7 ± 5.3 to 269.8 ± 6.2 MPa. Hence, a patent application was generated (Fig. 33.24) [12].

A microscopic image of the C F F in the sand mold at 20 millimeters. The C F F has a cube like structure with a porous surface.

Concluding, the sum of all single actions will lead to an increase in filtration efficiency and thus melt cleanliness with elevated mechanical properties of the casting. Here, encapsulated ceramic fiber addition represents one promising measure that can be easily implemented in casting routine of foundry industry.