THE PENNSYLVANIA STATE UNIVERSITY SCHREYER HONORS COLLEGE DEPARTMENT OF MECHANICAL AND NUCLEAR ENGINEERING

Transcription

1 THE PENNSYLVANIA STATE UNIVERSITY SCHREYER HONORS COLLEGE DEPARTMENT OF MECHANICAL AND NUCLEAR ENGINEERING A FUNDAMENTAL STUDY OF SAND INVESTMENT SYSTEMS USING COMPUTATIONAL PARTICLE FLUID DYNAMICS JOSHUA D. ORLOWSKI Fall 2009 A thesis submitted in partial fulfillment of the requirements for a baccalaureate degree in Mechanical Engineering with honors in Mechanical Engineering Reviewed and approved* by the following: Robert C. Voigt Professor in Industrial Engineering Thesis Supervisor Mary Frecker Professor in Mechanical Engineering Honors Adviser * Signatures are on file in the Schreyer Honors College.

2 i ABSTRACT This thesis uses computational particle fluid dynamics software (CPFD) to study the sand investment system of a core blowing machine. Several aspects of the investment system are varied in order to determine the effects. A total of 8 simulations are conducted, testing 3 variables. Variables tested include the addition of vacuum to the corebox, as well as the addition of air inlets to the sand tube and sand head. The results of each of these design changes are discussed, and variables requiring further study are identified. It is shown in this thesis that the addition of vacuum to the vents in a corebox has little effect on the resulting core quality. Tests utilizing corebox vacuum do exhibit marginally better blow times, but only to a small degree. The addition of air inlets to the sand tube is shown to increase the fluidization of sand in the sand tube and sand head, but does not result in better corebox filling in these tests. Adding air inlets to the sand head has the largest effect on core filling, however in these simulations it has a negative impact on core quality and causes additional ratholing in the sand head. The addition of air inlets to both the sand tube and sand head are shown to have potential for improving the core blowing process. Future experiments are detailed that will focus on the proper application of these technologies.

3 ii TABLE OF CONTENTS LIST OF FIGURES... LIST OF TABLES... ACKNOWLEDGEMENTS... Chapter 1 Introduction Motivation Literature Review... Previous Work. Chapter 2 Testing Variables... Blow Pressure Location... Corebox Venting/Vacuum... Sand Head Air Inlets Chapter Summary Chapter 3 Model Generation.... Investment System.... Tooling Design.... Core Shape... Venting. Layout... Mesh Generation.. Constant Parameters..... Blow Pressure... Sand type.. Initial Sand Location.... Flowability Mesh Refinement..... Chapter 4 Simulation Results and Discussion.. Core Assessment Effects of Core Blowing Variables.. Corebox Vacuum.. Blow Pressure Locations.. Sand Head Air Inlets Discussion Chapter 5 Conclusion... Future work REFERENCES.. APPENDIX A Geometries.. APPENDIX B Detailed Run Parameters and Results.. iii v vi

4 iii LIST OF FIGURES Figure 1.1: Example of ratholing in sand head Figure 2.1: Comparison between standard sand tube and tube with additional air inlets Figure 2.3: Side view of sand head with additional air inlets.. Figure 2.4: Orthographic view of sand head with additional air inlets Figure 3.1: W Core model Figure 3.2: Original core venting and filling... Figure 3.3: Revised core venting and filling.... Figure 3.4: Corebox layout... Figure 3.5: View of meshed model Figure 3.6: Sand size distribution. Figure 3.7: Initial sand location for test 1 Figure 3.8: Flowability setting used for all simulations... Figure 4.1: Corebox for viewing results.. Figure 4.2: Test 1 filling by volume fraction Figure 4.3: Test 5 filling by volume fraction... Figure 4.4: Test 1 filling by density.... Figure 4.5: Test 5 filling by density.... Figure 4.6: Test 3 filling by volume fraction... Figure 4.7: Test 7 filling by volume fraction... Figure 4.8: Test 1 filling by volume fraction Figure 4.9: Test 3 filling by volume fraction

5 iv Figure 4.10: Test 5 filling by density Figure 4.11: Test 7 filling by density Figure 4.12: Test 2 filling by volume fraction..... Figure 4.13: Test 4 filling by volume fraction. Figure 4.14: Test 1 filling by volume fraction Figure 4.15: Test 2 filling by volume fraction Figure 4.16: Test 3 filling by density Figure 4.17: Test 4 filling by density Figure 4.18: Cross section of sand head with ratholing Figure 4.19: Sand head pressure at 0.1s in test Figure 4.20: Sand head pressure at 0.24s in test Figure 4.21: Sand head pressure at 0.1s in test Figure 4.22: Sand head pressure at 0.24s in test

6 v LIST OF TABLES Table 2.1: Simulation summary Table 3.1: Blow pressure data for runs 1 and 5 Table 3.2: Blow pressure data for runs 2 and 6 Table 3.3: Blow pressure data for runs 3 and 7 Table 3.4: Blow pressure data for runs 4 and 8 Table 3.5: Sand size distribution.. Table 3.6: Initial sand volume fraction.... Table 3.7: Mesh sizes used in resolution study.... Table 4.1: Blow time comparison for vacuum simulations..... Table 4.2: Result summary by blow time and filling order

7 vi ACKNOWLEDGEMENTS First and foremost, I would like to thank my family. They have supported me through all of my decisions, and constantly remind me that anything can be achieved through hard work. Next I would like to thank Dr. Robert Voigt for his work with me as this project has evolved. Dr. Voigt s insight has been invaluable in the completion of this thesis; this project would not have been possible without his guidance. Finally, I would like to thank Dr. John Cimbala for allowing me to join in his Arenaflow research. This project has been a great learning experience for me, one I would not have had the chance to partake in without Dr. Cimbala s help.

8 1 Chapter 1 Introduction This thesis will focus on the study of sand investment systems for core blowing machines. The primary objective of this thesis is to utilize Arena-flow computational particle fluid dynamics (CPFD) software to model particle flow in the sand tube, head, and corebox; in order to ultimately design more efficient and effective sand investment systems. Motivation There are many sources of motivation for this work. First, the sand investment systems used in core blowing machines have remained largely unchanged for the last half century. New CPFD software provides the ability to model sand flow in these systems in ways that were not previously available, opening up opportunities for improvement. There can be a great cost savings for core machine manufacturers with the ability to simulate new designs before the investment is made to actually build them. Beyond simply having new technology available for modeling core blowing, the metal casting industry as a whole stands to benefit from advances in core blowing technology. Modern castings are moving towards more complex and thinner walled structures. Making these types of difficult geometries requires cores to be of higher quality than in the past. Better core blowers will result in better dimensional tolerances, less scrap, and an increase in efficiency throughout the entire casting process. Finally, recent years have seen a strong push for foundries to become more ecofriendly. Forty compounds of the 189 hazardous air pollutants (HAPs) listed in the Clean Air Act of 1990 have been identified in foundry emissions, with a significant amount of emissions coming from high-production foundries (1). One of the major opportunities for

9 2 improvement is in the switch to low-emission binder systems. However, these binders require slightly different blowing techniques from conventional binders in order to be used effectively. One of the goals of this testing is to improve the ability to blow cores bound by low-emission binders. Literature Review Metalcasting has been around for thousands of years. Sources indicate that the first metal casting was most likely made over 5000 years ago (2). While the materials used and parts being formed have changed a great deal from the first castings made, the basic process of melting and pouring metal is the same. However, current technology provides the ability to model, monitor, and optimize the process in ways never before possible. In addition to great strides that have been made in the physical aspects of foundries, computational advances permit many changes to be simulated before they are actually constructed. One of the newest tools for modeling processes is computational fluid dynamics software. In recent years, computational fluid dynamics (CFD) software has been utilized to analyze/solve many different types of problems. These problems range from high-speed aerodynamics, to blast wave simulations, to HVAC modeling and more (3). CFD software works by dividing a large, continuous problem into many smaller volumes, sometimes called finite elements or cells. Then, the behavior of fluids within each of these elements can be calculated numerically. The greater the number of cells in a model, the higher the degree of accuracy that can be obtained from the results. However, this increase in accuracy comes at the expense of computational speed.

10 3 While typical CFD methods work well for single phase flows or flows containing small particles, their accuracy begins to break down when simulating larger particles, such as sand. For this reason, the software utilized in the simulations uses an Eulerian Lagrangian model for modeling two-phase flow in sand cores. This method, known as computational particle fluid dynamics (CPFD), consists of air modeled in a continuous manner in an Eulerian grid (as in typical CFD software), while the sand is modeled discretely in a Lagrangian frame of reference (4). This method is the most effective when some aspects of a problem are best modeled in a continuum, while others can be best represented by discrete entities (5). This type of modeling allows both the interactions between air and sand to be taken into account, as well as interactions between sand particles. This type of software has already been utilized to a large extent to model several different processes in the corebox, including multiple types of core blowing and core curing. However, less work has been done to study what occurs above the blowtubes in a core. That is, minimal study has gone into the happenings in the sand head and sand tube of a core blowing machine. More analysis of these aspects of the process is needed to address several problems commonly seen in core blowing. One of the largest core-machine related problems seen in the foundry industry, and one of the primary concern that this thesis would like to address is the phenomenon known as ratholing. Ratholing occurs when a granular material behaves as a cohesive mass, rather moving as a fluid (6). When this occurs in a sand head, the sand immediately above a blowtube will flow into the core, while the adjacent particles stay in place. This forms a tunnel, or rat hole. Figure 1.1 on the following page shows ratholing in a sand head and the effect it has on core filling.

11 4 Figure 1.1- "Ratholing" in the sand head. Note the amount of sand still available in the blow head. This phenomenon results in incomplete filling, as can be seen in the front core. Not limited to sand or core blowing, ratholing can be seen in many granular material handling operations. It is a primary concern for powdered or granular material stored in hoppers or silos. Many of these storage silos inject air through the material being stored in order to keep it in an aerated state, thereby increasing its ability to flow, and decreasing the tendency for rathole formation. While slightly different than the situation seen in core blowing machines, the lessons learned from rathole prevention in other granular materials can be of use. Previous Work Some equipment similar to one of the improved scenarios modeled in this thesis has been constructed by Harrison Machine Company and put into operation in foundries. While some anecdotal evidence has been seen that these sand investment systems are capable of making cores that are not possible with their standard investment system, no full study has

12 5 yet been completed. One of the goals of this thesis is to model, evaluate, and improve the Harrison design. In addition to some of the advancements that have already been made with Harrision machines, experiments utilizing vacuum drawn on the corebox are currently underway as well. Furness-Newburge, Inc. has seen promising results with vacuum up to 100 in H 2 O. The addition of vacuum to the corebox will be utilized as one of the variables in this study.

13 6 Chapter 2 Testing Variables There are many variables that affect the quality of a blown sand core. Core geometry, venting, blow pressure, tooling wear, and sand type are just a few aspects among the many that vary between foundries, core machines, and even individual cores in a corebox. The intention of this thesis was to select three variables associated with the sand investment system and tooling, test each variable on two levels, and study the effects of each on a sample core shape. The results obtained from these fundamental tests will serve to guide the way for future simulations. Blow Pressure Location The standard for core machines has been to apply pressure at the top of the sand tube, in order to force sand into the tooling. As an alternative to this approach, pressure will be applied through fifteen 12.5 mm diameter air inlets on the side of the blow tube, in addition to the pressure at the top of the blow tube. It has been seen that a critical step in reducing the occurrence of ratholes in a granular material is to increase the radial pressure gradient within that material (6). It is believed that by adding additional blow locations on the side of the sand tube, this radial pressure gradient in the sand can be increased, thereby reducing its tendency to form ratholes. Figure 2.1 on the following page show a comparison between the standard investment system and the system with additional sand tube blow locations.

14 7 Pressure across top plane of blowtube Pressure across top plane of blowtube B L O W T U B E Additional inlets on perimeter of blowtube B L O W T U B E S A N D H E A D S A N D H E A D Figure 2.1- Comparison between standard sand tube (left) and sand tube with additional air inlets (right). For full details and dimensions of tubes/inlets, as well as the pressures at each point, see Appendices A and B. In addition to decreasing ratholing, these extra air inlets will increase the amount of aeration in the sand. It has been shown that increased aeration has a large effect on the ability of powdered or granular materials to move as a fluid (7). It is thought that the more the sand in the sand tube and sand head can be made to behave as a fluid, the more readily it will move to fill the corebox. For full details on the sand tube air inlets and pressure configurations, see Appendices A and B. Corebox Venting/Vacuum While the number and location of vents on the corebox will be kept constant for all tests, the pressure at the vents will be varied during the testing. The standard for corebox venting is to have vents at atmospheric (0 gauge) pressure. This will be used as one testing scenario, as will having all vents at -10 kpa gauge pressure. This value is consistent with vacuum tests being conducted. Having a lower pressure at the vents will result in a higher

15 8 pressure differential between the sand tube and corebox, possibly leading to better filling of the cores. The vent sizing and locations are discussed in the next chapter. Sand Head Air Inlets The final variable for testing will be air inlets on the sides of the sand head. Conventional sand heads have no additional air inlets; all air comes from the sand tube above the sand head. Sixteen air inlets around the perimeter of the sand head will be simulated in these experiments. Each inlet is 10 mm wide and 20 mm high, for a total of 3200 mm 2 of additional blow area. Much like the additional blow locations in the sand tube, additional air inlets in the sand head will increase the radial pressure gradient in the head. This should help to cut down on the amount of ratholing in the sand head. In addition, air inlets in the sand head should result in greater aeration of the sand as it flows into the corebox. For detailed pressure information for the sand head air inlets, see Chapter 3. Figure 2.5 and Figure 2.6 show the layout of the air inlets in the sand head. Figure 2.3- Side view of sand head showing air inlet locations (yellow) in relation to blowtubes.

16 9 Figure 2.4- Location of additional air inlets in sand head. Chapter Summary Three variables related to the design of the corebox and sand investment system of a core blowing machine were chosen for testing. Each parameter will be tested on two levels, resulting in a total of 8 runs. Table 2.1 shows the value of each variable in each run. Run number 1 will be noted as the baseline case, using a standard sand tube, pressure at the top of the blow tube, no vacuum, and no sand head air inlets. This design is consistent with typical machines used in industry, and will provide a basis for comparison. Test Number Sand Tube Shape Table 2.1- Parameters for each simulation Vacuum Blow Pressure Location Sand Head Air Inlets 1 Standard None Top No 2 Standard None Top Yes 3 W/ Air Inlets None Top + Sand Tube No 4 W/ Air Inlets None Top + Sand Tube Yes 5 Standard 10 kpa Top No 6 Standard 10 kpa Top Yes 7 W/ Air Inlets 10 kpa Top + Sand Tube No 8 W/ Air Inlets 10 kpa Top + Sand Tube Yes

17 10 While there are many more levels that each variable could be tested on, the intent of this thesis is not to provide a comprehensive design analysis for every blowing possibility. Rather, the goal is to identify the parameters that have the largest effect on the success/failure of a core and designate these parameters for further study.

18 11 Chapter 3 Model Generation After variables for testing were chosen, it became necessary to create a model in which these parameters could be implemented. The model in this case consists of a sand tube, sand head, and corebox, modeled after a standard core blowing machine. In addition, constant parameters for each test include sand type, initial sand location, sand flowability, vent sizing and vent location. Investment System The combination of a sand hopper, sand tube, and sand head is known as the investment system. The hopper is unnecessary for the modeling to be done in these tests, as it is located outside of the area that is pressurized during core filling. As stated in chapter 2, the baseline case sand tube and sand head are modeled roughly after those on a Harrison Machine Co. Model 1016 core blowing machine. Appendix A has the dimensions of each geometry combination used in testing. Tooling Design While the primary concern of this thesis is the investment system of core blowing machines, the true test of each design is to see how well it can make a core. In order to compare differences in sand investment systems, a core was designed for use in filling simulations. The core used, which will be referred to as the W core, along with the one blowtube used for filling, can be seen in Figure 3.1.

19 12 Blowtube W core Figure 3.1- "W core" and blowtube used in simulation This W core was designed to be difficult to fill. As a complete fill in the baseline case would make it difficult to compare improved cases, this core was specified so to result in only near complete filling in the baseline configuration. More detail on the design of the core can be found in the next sections, while full dimensions are located in Appendix A. Core Shape The first consideration in designing a core for simulation was the shape. The W shape was chosen primarily the difficulty it presents in filling, especially for sands with lower degrees of flowability. The sharp corners in the core, combined with outer legs of equal height as the center, provide a challenge for blowing. For the purposes of this experiment, a single blowtube was located on the center leg of the core. The high outer legs and single blowtube require sand to enter the center of the core, make a 180 degree turn, and return to the same level at which it entered the core in order to create a complete fill. In

20 13 addition, the movement of sand through this core bears similarities to several cores used in previous Arena-Flow testing (4,8). Core Venting After choosing a shape for the core, the next step was to determine the appropriate venting. Much the same as designing cores for practical application, determining vent location and specifications was an iterative process. At first, venting was specified so as not to be the limiting factor in filling. Figure 3.2 shows the first venting configuration tried and the resulting fill. Figure 3.2- Original core venting configuration and the resulting fill. Note a total of 12 vents and minimal filling in the legs of the core. The original core venting configuration had a total of 12 vents. Each vent is 8 mm in diameter with 50% open area, resulting in a total open vent area of approximate 301 mm 2 for the twelve vents. This design proved to have far too much venting to adequately fill the core. For a two-phase flow, such as that seen within the corebox during filling, there is a minimum transport velocity that must be maintained or exceeded to keep particles suspended in the flow (7). The over-venting of the corebox led to a large amount of air escaping from vents 7 through 12 (Figure 3.2), rather than maintaining adequate velocity through the legs of the

21 14 core. The result is sand dropping out of the flow at the bottom of the legs, while still providing an adequate fill in the center of the core. This result was predicted by the overvented meandering core simulated by Schneider (8). In order to improve filling in the outer legs of the core, several vents on the center leg were closed. Figure 3.3 shows the core with limited venting and the resulting fill. The same size and style of vents were used, resulting in a total vent area of approximate 151 mm 2 in the revised configuration. This new venting configuration results in a much greater level of filling for the same blow parameters. This venting configuration was chosen for final simulations. Figure 3.3 Revised venting configuration and resulting fill. Note the number of vents has been reduced from 12 to 6 and the core fills to a much greater degree than previously. Core Layout After core shape and venting were determined, the final step in tooling design was to decide on a layout of the cores in the corebox. One option was to simulate only one core at a time, however this would eliminate the ability to study fluctuations in filling dependent on core location in the corebox. A design with five W cores along the length of the sand head was chosen. This is a number consistent with cores typically made on this size of machine,

22 15 and will also provide the ability to study variations between cores in a single blow. Figure 3.4 below shows the final design of the cores, blowtubes and sand head. Figure 3.4- Final layout of cores and sand head. Mesh Generation As with all CPFD software, the final geometries for testing had to be divided into a 3- dimensional mesh before any simulations could take place. The resolution of this mesh has an effect on both the accuracy of the solution, as well as the computational time required for each simulation. The tests included in this thesis averaged approximately 115,000 cells per simulation. Figure 3.5 shows the computational mesh for test 1. More information about the mesh creation/refinement process can be found later in this chapter, and details of the mesh for each test can be found in Appendix B.

23 16 Figure 3.5- Computational mesh used for simulations Constant Parameters While the main purpose of this thesis is to show the merits of each design aspect relative to one another, it was still important to choose realistic constant parameters. After choosing a geometry and vent layout to analyze, it is necessary to specify several other parameters before simulations can take place. Blow Pressure When choosing a blow pressure for a core, it is important to choose one that is high enough to move sand into the core effectively, but not unnecessarily high. For these tests, the highest blow pressure used was 300 kpa. This pressure was located at the the top plane of the sand tube for each test. This pressure represents a blow pressure that is considered average in

24 17 the foundry world. For tests with sand tube or sand head air inlets, lower blow pressures were specified at these locations. Full pressure values can be seen in Tables Table 3.1- Blow pressure data for tests 1 and 5 Blow Pressure (Runs 1,5) Location Number of inlets Pressure (kpa) Area Fraction Top 1 (Top Plane) Table 3.2- Blow pressure data for tests 2 and 6 Blow Pressure (Runs 2,6) Location Number of inlets Pressure (kpa) Area Fraction Top 1 (Top Plane) Sand Head Table 3.3- Blow pressure data for runs 3 and 7 Blow Pressure (Runs 3,7) Location Number of inlets Pressure (kpa) Area Fraction Top 1 (Top Plane) Sand Tube (Top) Sand Tube (Middle) Sand Tube (Bottom) Table 3.4- Blow pressure data for runs 4 and 8 Blow Pressure (Runs 4,8) Location Number of inlets Pressure (kpa) Area Fraction Top 1 (Top Plane) Sand Tube (Top) Sand Tube (Middle) Sand Tube (Bottom) Sand Head

25 18 Sand Type The software used for this analysis is capable of simulating the flow of many different types of sands. Beyond this, it can also simulate other particles, with a table consisting of particle percentage by weight, particle radius and bulk density as the main input for the program. The sand used in this testing is a slightly modified version of the 15LW type sand that is a default within Arena-flow. The characteristics of the sand used can be seen in Table 3.5, and Figure 3.6 shows a graph of the sand size distribution. Table 3.5- Sand size distribution used in simulations Sand Size Distribution % by wt Radius (m) Density (kg/m^3) E E E E E E E E E

26 Percentage by wt E E E E E E E E-04 Sand Radius (m) Figure 3.6- Sand distribution by weight. The data points on the graph represent user input values, while the lines represent the linear interpolation performed by Arena-flow. Initial Sand Location As these tests were primarily concerned with sand movement through the sand tube and sand head, it was important to choose initial sand locations that were representative of what would be seen in practice. The initial sand conditions for this testing were specified by volume fraction and location. At the bottom of the sand head is a volume of sand that is near its close packed volume fraction, simulating sand left in the head from a previous blow. On top of this close packed layer, extending almost to the top of the sand tube, is a volume of sand that is approximately 10% below the close packed volume fraction. This upper region of sand represents fresh sand dumped from the hopper before each run. Table 3.6 shows the volume fractions used for each test, while Figure 3.7 shows a sample of the initial sand

27 20 conditions for test 1. Note that the upper region of less packed sand does not extend to the edges of the sand head. This is due to the granular nature of sand, and its inability to flow to the edges of the sand head under gravity. Table 3.6- Sand volume fractions used in testing Initial Sand Volume Fraction Location Value Upper Lower Figure 3.7- Initial sand location and volume fraction for Test 1.

28 21 Flowability The final input for these simulations is a variable known as flowability. Flowability represents the effects of sand hardening during a blow (4). This variable is represented by merely a slider bar in the software used. Figure 3.8 shows this slider, as well as the flowability setting used in these tests. The setting used in all of these tests was average/high. The makers of Arena-flow caution against using values lower than this when full magazines are simulated, as it may lead to an over-prediction of ratholing. Figure 3.8- Flowability setting used in all simulations. Flowability represents the level of sand hardening effects during the blow. Mesh Refinement As stated earlier in this chapter, the mesh resolution used in simulation is an important factor in the accuracy of the results. While the software used touts its insensitivities to mesh sizing (4), it was still important to conduct a resolution study in order to get consistent and realistic results. The first step in determining a suitable mesh size involved creating a coarse mesh, just fine enough to resolve the geometric features of the sand tube, sand head, and corebox.

29 22 Table 3.7 below shows the number of cells used in several mesh refinement simulations. After running the coarse mesh simulation, the number of cells in the mesh was more than doubled (fine mesh in Table 3.7) and the simulation was run again. The results from these 2 runs had appreciable differences, indicating that the coarse mesh was not refined enough for use in the simulations. In order to arrive at a final mesh sizing suitable for use, the mesh was made finer still (finest mesh in Table 3.7) and the simulation was run again. Results from the simulation utilizing the finest mesh were nearly identical to those of the fine mesh, indicating that either mesh sizing would be appropriate for use. However, in the interest of reducing the computational time required for each run, it is most practical to use the coarsest mesh capable of giving accurate results. Several more tests of decreasing mesh resolution were conducted, before finally settling on the medium mesh size shown in Table 3.7 Table 3.7- Several mesh sizes used in resolution study. Mesh Mesh Sizes Number of cells Coarse Medium Fine Finest

30 23 Chapter 4 Simulation Results and Discussion After running all of the core blowing simulations, it was time to analyze the data obtained. There are several ways to analyze the success of blowing a core. First and most important is the quality of the produced core. Beyond this, the criteria for determining the success of a blow can vary depending on the priorities of the core maker. For many, time is an important factor. For some others, the ability to blow a core at a lower pressure may constitute a more successful blow. This chapter will lead with a discussion of the criteria used to judge the level of success for each simulation. Following this, the effects that each testing variable had on core formation will be discussed. Core Assessment The only true way for testing the quality of a core is to make a casting with it. However, in the simulated environment of these tests, using test cores to make a casting is not an option. Despite this, there are several other ways that these simulated cores can be judged for quality. The first and most obvious method of determining core quality is by visual inspection. The software used for this simulation provides several different ways to visually analyze cores. For the tests conducted in this thesis, filling can be viewed as plain sand (all sand grains are colored the same), by volume fraction (regions of the core are shaded by volume fraction) or by density (regions of the core are shaded by density). Viewing plain sand filling is suitable for identifying gross flaws in the core, however it is not very effective in distinguishing between sufficiently packed and less-packed regions of the core. For this

31 24 analysis, volume fraction and density are much better ways of judging core quality. In general, a core with a higher volume fraction or higher density will make a better casting. The majority of figures in this chapter will show regions of the core shaded by either volume fraction or density. In discussing the results of these simulations, it is important to use consistent nomenclature throughout the text. The majority of images in this chapter, as well as those in Appendix B will show a cropped view of the simulation. This view encompasses the entire corebox, as well as part of the blowtubes. The sand tube and sand head are removed for easier viewing of the corebox. Figure 4.1 shows this cropped view, as well as the numbering scheme that is used to refer to individual cores. For detailed results from each simulation see Appendix B. Core 5 Core 4 Core 3 Core 2 Core 1 Figure 4.1- Default view of the corebox to be used in reporting results. Note that cores are numbered from 1 to 5, starting at the front of the corebox and moving towards the back.

32 25 Effects of Core Blowing Variables The next sections of this thesis will discuss the effects that each core blowing variable had on the simulation. Some results will be displayed in the following sections, for full details and results of each simulation see Appendix B. Corebox Vacuum The first core blowing variable to be discussed is the addition of vacuum to the corebox. For the vacuum simulations, the pressure at all of the vents in the corebox was reduced from 0 kpa gauge in the baseline case to -10 kpa gauge. Figure 4.2 below shows the fill in the baseline case, while Figure 4.3 shows the resulting fill when vacuum is added to the vents. Figure 4.2- Final filling result, shown by volume fraction, for the baseline testing case.

33 26 Figure 4.3- Final filling result by volume fraction for test 5. The parameters for test 5 were identical to the baseline, except for the addition of vacuum to all vents in the corebox. When viewing simulation results by volume fraction, it can be seen that adding vacuum to the baseline case resulted in better filling in cores 3 and 4. However, the quality of fill in cores 1 and 2 was reduced. Core 5 filling was similar for both tests, with slightly better filling in both legs when vacuum is added. Figures 4.3 and 4.4 show the results for simulations 1 and 5 again, but with the corebox shaded by density rather than volume fraction.

34 27 Figure 4.4- Final filling result for test 1, shaded by density. Figure 4.5- Final filling result for test 5, shaded by density. The parameters for this blow were the same as test 1 (Figure 4.4), except for the addition of vacuum in test 5.

35 28 Viewing the results of simulations 1 and 5 by density leads to a similar conclusion as when the results are viewed by volume fraction. There is minimal change between the two simulations. The density in the center of core 1 in test 5 (Figure 4.5) is slightly lower than in test 1, however the density in the legs of cores 2, 3, 4 and 5 is slightly higher. A comparison of tests 3 and 7 shows similar results. Test 3, seen in Figure 4.6, is characterized by a sand tube with air inlets, while test 7 (Figure 4.7) has the same sand tube plus vacuum on the vents. Once again, there is little difference in the final filling in these two cases. Figure 4.6- Final filling for test 3, shaded by volume fraction. Test 3 utilized a sand tube with additional air inlets.

36 29 Figure 4.7- Final filling for test 7, shaded by volume fraction. Test 7 utilized the same sand tube as test 3, as well as 10 kpa of vacuum on the corebox vents. As Figures 4.2 through 4.7 show, there is minimal difference in the overall quality of the cores when vacuum is added to the corebox. The quality of some cores improves marginally, while that of other cores does not improve or decreases slightly. This effect is seen both when core filling is analyzed by volume fraction, as well as when it is viewed by density. These findings are further verified by comparisons between simulations 2 and 6, and 4 and 8. One aspect of core blowing that did improve across the board with the addition of vacuum was blow time. The total time required to blow the cores improved slightly in each case. Table 4.1 on the following page shows the time required to fill the center leg of each core in each simulation. Tests with less than 5 cores listed in Table 4.1 experienced ratholing that prevented the center leg of the unlisted cores from filling. Note the shorter blow times for the tests utilizing vacuum (right column of Table 4.1).

37 30 Table 4.1- Blow time comparison for tests with and without corebox vacuum. Vacuum Tests Standard Sand Tube, Top Pressure, No Sand Head Air Inlets Test 1 (no vacuum) Test 5 (10 kpa vacuum) Core Time Core Time 4.35 sec 4.34 sec 2.36 sec 2.35 sec 5.37 sec 3.35 sec 3.37 sec 5.35 sec 1.38 sec 1.36 sec Standard Sand Tube, Top Pressure, Sand Head Air Inlets Test 2 (no vacuum) Test 6 (10 kpa vacuum) Core Time Core Time 3.43 sec 5.40 sec 5.45 sec 3.42 sec 1.50 sec 1.50 sec Sand Tube with Air Inlets, No Sand Head Air Inlets Test 3 (no vacuum) Test 7 (10 kpa vacuum) Core Time Core Time 2.35 sec 3.34 sec 3.35 sec 2.35 sec 4.35 sec 4.35 sec 5.35 sec 5.35 sec 1.37 sec 1.36 sec Sand Tube with Air Inlets, Sand Head Air Inlets Test 4 (no vacuum) Test 8 (10 kpa vacuum) Core Time Core Time 3.41 sec 3.40 sec 5.43 sec 5.42 sec

38 31 Blow Pressure Locations The next variable to be analyzed is the location of blow pressure with respect to the sand tube. Typical core blowers, and the baseline case for these simulations, apply pressure only at the top of the sand tube. For the experimental case, 3 rows of 5 additional air inlets are added to the sand tube. Pressure values at each of these blow locations can be found in Tables 3.3 and 3.4, as well as in Appendix B. Figures 4.8 and 4.9 show the baseline filling case compared to one with additional blowtubes. Figure 4.8- Final filling result for test 1, shown by volume fraction.

39 32 Figure 4.9- Final filling by volume fraction for test 3. The design of test 3 has additional air inlets in the sand tube compared to test 1 (Figure 4.8). Note however, the minimal differences in core filling. Figures 4.8 and 4.9 show minimal differences between the case with sand tube air inlets and the case without. Some areas, such as the left leg of core 3, fill better in the case with additional air inlets. However, other locations such as the left leg of core 1 and the right leg of core 5 blow marginally worse when additional air inputs are added to the sand tube. Comparisons of other runs with and without sand tube air inlets are very similar to the comparison between test 1 and test 3. Figures 4.9 and 4.10 show the results of tests 5 and 7, respectively. There is no definitive advantage with regard to fill quality when using the sand tube with additional air inlets. A comparison of tests 2 and 4 (shown in Figures 4.11 and 4.12) and tests 6 and 8 yields similar, yet slightly different results.

40 33 Figure Final filling for test 5, shaded by density. This test utilized a standard sand tube, as well as 10 kpa of vacuum at the vents. Figure Final filling for test 7, shaded by density. Test 7 had the same parameters as test 5, with the exception of additional blow locations in the sand head. Notice that the sand density in the corebox is very similar in both tests.

41 34 Figure Final filling from test 2, shaded by volume fraction. Note the large voids in the center of cores 2 and 4. Figure Final filling from test 4, shaded by volume fraction. Note the large vacancies in the center of cores 2 and 4, similar to those seen in test 2. In addition, note the void in the center of core 1 not seen in test 2. While both test 2 and test 4 are unsuccessful blows due to the large vacancies in the center of cores 2 and 4 in each run (Figures 4.12 and 4.13), test 4 has an additional void in the center of core 1. A comparison of tests 6 and 8 shows a similar effect. Test 6 (no sand

42 35 tube air inlets) only has voids in the center of cores 2 and 4, while test 8 (with air inlets) shows these defects plus limited filling in core 1. Table 4.2 summarizes the results from all of the tests. The last column in Table 4.2 shows the unfilled cores in each test. Only tests with additional sand tube air inlets did not fill core 1. Table 4.2- Summary of runs showing core filling time and filling order. Total blow time is recorded at the time that the center leg of the core is completely filled. Tests with N/A for blow time did not fill the center leg of all 5 cores. Test Number Total Blow Time First Core To Fill Last Core To Fill Unfilled Cores sec 4 1 N/A 2 N/A 3 1 2, sec 2,3,4,5 1 N/A 4 N/A 3 5 1,2, sec 4 1 N/A 6 N/A 5 1 2, sec 3 1 N/A 8 N/A 3 5 1,2,4 With regard to blowing time, the addition of air inlets to the sand tube had a negligible effect. There is no significant difference in required blow time due to the addition of sand tube air inlets. Sand Head Air Inlets The last variable to be discussed, and the one with undoubtedly the largest effect on corebox filling in these simulations is the addition of air inlets to the sand head. However, contrary to the intention of adding these air inlets, every test utilizing sand head air inlets experienced a much larger degree of ratholing than the control case. Figure 4.14 and Figure 4.15 on show the resulting fills from run 1 (baseline case) and run 2. The only difference between these two runs is the addition of sand head air inlets in run 2.

43 36 Figures Baseline fill case, shown by volume fraction. Notice some voids exist in outer legs, but the center legs of all cores are completely filled. Figures Final filling for test 2, shaded by volume fraction Test 2 parameters were identical to test 1with the exception of sand head air inlets. Note the large voids in core 2 and core 4.

44 37 It is clear from Figures 4.14 and 4.15 that the addition of sand head air inlets had a negative effect on core filling. Similar results were seen between runs 3 and 4; 5 and 6; and 7 and 8. Each of these runs had identical parameters, with the exception of additional air inlets in the sand head for runs 4, 6, and 8. Each run showed significant ratholing when air inlets were added to the sand head, but none in the case without. Figure 4.16 and Figure 4.17 show a comparison between runs 3 and 4 by density, while Figure 4.18 shows a cross section of the sand head from test 4 in which ratholing is clearly visible. Figure Final filling for test 3, shaded by density. Test 3 utilized a sand tube with air inlets. Note the density in the center leg of each core compared to Figure 4.17.

45 38 Figure Filling from test 4, by density. Test 4 utilized the same sand tube as test 3, but also had air inlets in the sand head. Note the low density in the center of cores 2 and 4 corresponding to vacancies from ratholing. Figure Cross section of sand head in which ratholing can be seen. Note the amount of sand remaining in sand head.

46 39 The ratholing seen in Figure 4.18 prevented cores 1, 2, and 4 from filling completely. Possible reasons for rathole formation in cases with sand head air inlets will be discussed in the following section. Discussion The results of these 8 simulations turned out to be quite different than expected. First, the addition of vacuum provided marginal benefit to the blowing process. There are several reasons why this may have occurred. For example, a rather low value was used for the amount of vacuum on the corebox. A conservative value of 10 kpa of vacuum was used for these simulations, consistent with early vacuum testing. However, more recent experiments have been underway utilizing up to 100 in H 2 O of vacuum, roughly equivalent to 25 kpa. This value is over twice that which was used in these simulations. In addition to the amount of vacuum utilized in these tests being fairly small, a uniform value was used at all vents in the corebox. A more sophisticated vacuum system would have the ability to tailor the amount of vacuum at each vent in order to address specific filling problems. In essence, the addition of 10 kpa of vacuum at each vent merely serves to raise blow pressure by 10 kpa, or approximately 3 percent. Large changes in results would not be expected from this increase, although it does explain the small decrease in required blow time. The minimal effects due to the addition of vacuum, as well as the minimal effects from sand tube air inlets can also possibly be attributed to larger venting issues with the test core. The seemingly random vacancies at the ends of the core legs seen in almost every test are consistent with cores that are slightly undervented. In future tests, the venting of the core will be reevaluated and more venting may be added to the ends of the core legs.

47 40 We will now move on to the effects of adding air inlets to the sand tube. Making this change proved to have little effect on the final quality of the blown cores. However, in tests that experienced ratholing, the degree of ratholing was increased by the addition of sand tube air inlets. While underventing may have limited the ability to of the cores to fill, there are several other potential reasons why tests with sand tube air inlets provided relatively similar results to those without. First, the degree of aeration in the sand tube may have very little to do with the fluidity of the sand by the time it reaches the corebox. Although the mass of sand in the sand tube can move as a fluid, it loses this fluidity upon coming in contact with the stationary mass of sand in the sand head. In addition, the location of the air inlets in the sand tube may be at fault as well. There are countless ways in which the location, number, size and pressure of these air inlets could be varied; more experimentation is needed to determine if there is a combination that works better than the one used in this testing. Finally, the most interesting result from these experiments was the large increase in the amount of ratholing that was seen when air inlets were added to the sand head. It has been shown in other areas of study that increasing the level of aeration in a granular material is an effective way of reducing the formation of ratholes. Additional air inlets were added to the sand head in an attempt to do just this. However, the results of these tests show an opposite effect. The results of these simulations in which ratholing occurs seem counterintuitive, until one examines the pressure in the sand head during a blow with additional air inlets and one without. Figures 4.19 and 4.20 show the pressure in the sands head and corebox at 0.1 and 0.24 seconds into the blow for test 3. This case had additional air inlets in the sand tube, but none in the sand head. It can be seen that while a pressure gradient exists both between the

48 41 sand head and corebox, as well as within the corebox itself, the pressure is the sand head is relatively uniform. Figure Pressure within sand head and corebox at 0.1 seconds into test 3. The number in the upper right represents blow time in seconds. Figure Pressure within sand head and corebox at 0.24 seconds into test 3. Note that the pressure is still uniform throughout the sand head. However, the pressure in test 4 is a much different scenario. Figures 4.21 and 4.22 show the pressure within the corebox of test 4 at 0.1 and 0.24 seconds into the blow. At 0.1

49 42 seconds into the blow (Figure 4.21), the pressure is nearly uniform throughout the sand head, with one small area of high pressure near the center. However, by 0.24 seconds into the blow (Figure 4.22), a large area of high pressure, coinciding with the location of ratholing has developed in the center of the sand head. The air inlets around the perimeter of the sand head can clearly be seen as areas of low pressure. Figure Pressure within sand head and corebox at 0.1 seconds into test 4. Note that the pressure is still relatively uniform through the sand head. Figure Pressure within sand head and corebox at 0.24 seconds into test 4. Note the areas of high pressure that have developed in the center of the corebox. These areas coincide with the location of ratholing.

50 43 The existence of these low pressure areas around the edges of the sand head explain the amount of ratholing that was seen in these tests. The pressure at the sand head air inlets was specified to be lower than the blow pressure in order to eliminate the possibility of these inlets preventing the core from filling, and to maintain the pressure gradient between the sand tube and corebox. However, by specifying such a low value for these inlets, sand traveled to the edges of the corebox, rather than through the small blowtubes into the corebox. In future testing, the pressure at these sand head inlets will be increased. The areas of low pressure in the corebox during sand head air inlet simulations also help to explain why ratholing was more pronounced in tests with additional sand tube air inlets as to those without. Adding additional blow locations to the sand tube did work to increase the flowability of the sand. However, due to the locations of low pressure in the sand head, the sand flowed to these areas rather than to corebox. This effect shows that there is promise in adding additional inlets to the sand tube, however these tests were not the correct application.

51 44 Chapter 5 Conclusion The goal of this thesis was to model the sand investment system of a core blowing machine, while studying the effects of several variables on the blowing process. A total of 8 simulations were conducted, testing the effects of vacuum applied to the corebox, the addition of air inlets to the sand tube, and the addition of air inlets to the sand head. Several important conclusions can be drawn from the results of these experiments. First, adding 10 kpa of vacuum to every vent in the corebox proved to have little effect on the resulting fill quality. This addition did reduce the necessary filling time in some simulations, but effects were minor. From these results, vacuum will most likely be included in future coreboxes for the benefit it provides in the core curing process rather than the core blowing process. The results from simulations utilizing air inlets in the sand tube proved to be rather inconclusive. There is evidence that the addition of air inlets to the sand tube does increase the fluidization of the sand in the investment system, but these tests showed minimal changes in final core quality. Future testing will be conducted to further explore these effects. The variable with the largest effect on core quality was the addition of air inlets to the sand head. However, contrary to the intention of this thesis, this addition caused the amount of ratholing in the sand head to increase. The addition of low pressure blow locations in the sand head resulted in areas of low pressure near the perimeter of the sand head. Sand in the sand head flowed to these low pressures areas as opposed to the corebox.

52 45 Although the results of these simulations were not as positive as originally hoped, they did accomplish one major goal. At least two of the variables in this testing have been identified as worthy of further study. Design is an iterative process, it is unrealistic to think that the optimal values for each variable could be found in these 8 simulations. Further study will be conducted in order to focus on the proper implementation of the design aspects highlighted in this thesis. Finally, and perhaps most importantly, this thesis proved that it is possible and worthwhile to model sand flow in the investment system of a core blowing machine with CPFD software. Past usage of this software has focused primarily on tooling design, vent location, and other corebox related issues. However, the results in this thesis show that it is not only possible, but very beneficial to model what exists above the corebox. Computer simulation of sand investment system will be extremely important in designing the next generation of core blowing machines. Future Work While this thesis provides a starting point for future simulations, it barely scratched the surface of the work to be done with sand investment systems. There is still much research to be done with regard to core blowing machines, especially with the increasing push for low emission binder systems. There are several aspects of this thesis that will be addressed in future testing. First, core venting will be reevaluated to ensure it is not a limiting factor with regard to corebox filling. In addition, the geometry of the corebox may be changed in order to test cores more commonly seen in industry, allowing specific problems to be addressed.

53 46 With regard to the investment system, there are countless opportunities for continuing research. There are several results from this thesis that will be addressed in particular. First, the ratholing issues surrounding additional air inlets in the sand head will be explored further, as will the addition of air inlets to the sand tube. Both of these investment system changes have shown their potential to affect the core filling process. However, future experiments will work on the proper application of these aspects in order to improve sand investment systems. In particular, the pressure at the sand head inlets will be increased to prevent the formation of low pressure areas near the boundaries of the sand head. Beyond the results discussed in the last section, there was another issue raised in these tests that demands more attention. As can be seen in Table 4.2, either core 1 or core 5 were the last to fill in every simulation. These two cores are located at either end of the sand head. The delay in filling these cores is most likely due to effects stemming from a round sand tube feeding a rectangular hopper. Future simulations will include a rectangular sand tube in an attempt to minimize these effects. Finally, a large part of future research will be experimental verification of these simulations. The data obtained from computer simulations can be extremely valuable in design. However, these data are limited by assumptions that have to be made. Lab tests will be designed to verify the results of these simulations before moving ahead with full-scale design of core blowing machines.

54 47 REFERENCES (1) Wang, Y. J.; Cannon, F. S.; Salama M.; Goudzwaard, J.; Furness, J. C. Characterization of Hydrocarbon Emissions from Green Sand Foundry Core Binders by Analytical Pyrolysis. Environ. Sci. Technol. 2007, 41, (2) Geng, Hwaiyu. Manufacturing Engineering Handbook. New York: The McGraw-Hill Companies, 2004 (3) Rhodes, Norman., ed. Computational Fluid Dynamics in Practice. London: Professional Engineering Publishing Limited, (4) Arena-flow, LLC. Arena-flow User Guide. Albuquerque: Arena-flow, LLC (5) Löhner, Rainald. Applied CFD Techniques. Chichester: John Wiley & Sons Ltd., 2008 (6) Johanson, Kerry. Rathole Stability Analysis for Aerated Powder Materials. Powder Technology. 2004, 141, (7) Fan, Liang-Shih and Chao Zhu. Principles of Gas-Solid Flows. Cambridge: Cambridge University Press, (8) Schneider, M., et al. Experimental Investigation, Physical Modeling, and Simulation of Core Production Processes. AFS Transactions ,

55 48 APPENDIX A Geometry details W Core. Sandhead. Sandhead and corebox layout.. Standard sand tube Sand tube with air inlets

56 W Core Tests: All All dimensions in mm 49

57 Sandhead Tests: All All dimensions in mm 50

58 Sandhead and cores Tests: All All dimensions in mm 51

59 Standard sand tube Tests: 1,2,5,6 All dimensions in mm 52

60 Sand tube with air inlets Tests: 3,4,7,8 All dimensions in mm 53

61 54 APPENDIX B Detailed run data and results Run Summary.. Constant Parameters Test 1 Test 2.. Test 3 Test 4 Test 5 Test 6. Test 7. Test Note: The number in the upper right corner of all filling images denotes the blow time in seconds.

62 55 Run Summary Test Number Sand Tube Shape Vacuum Blow Pressure Location Sand Head Air Inlets 1 Standard None Top No 2 Standard None Top Yes 3 W/ Air Inlets None Top + Sand Tube No 4 W/ Air Inlets None Top + Sand Tube Yes 5 Standard 10 kpa Top No 6 Standard 10 kpa Top Yes 7 W/ Air Inlets 10 kpa Top + Sand Tube No 8 W/ Air Inlets 10 kpa Top + Sand Tube Yes

63 56 Constant Parameters Software Version Arena-flow 7.1 Computational Cells in Mesh 114,345 Total Vent Area (per core) (mm 2 ) Flowability Medium/High Sand Size Distribution Density % by wt Radius (m) (kg/m^3) E E E E E E E E E Initial Sand Volume Fraction Location Value Upper Lower 0.550

64 57 Test Number Sand Tube Vacuum Blow Pressure Location Sand Head Air 1 Standard None Top No Total Fill Time 0.38 sec First Core to Fill 4 Last Core to Fill 1 Unfilled Cores N/A Due to Ratholing Filling order 4.35 sec 2.36 sec 5.37 sec 3.37 sec 1.38 sec Defects Large voids L1, L3, L4, L5 Small voids R1, R2, R4, R5 Blow Pressure Data Location Number of inlets Area (each inlet) (mm2) Pressure (kpa) Area Fraction Top 1 (Top Plane) Total Blow Area

65 58

66 59

67 60 Test Number Sand Tube Vacuum Blow Pressure Location Sand Head Air 2 Standard None Top Yes Total Fill Time N/A First Core to Fill 3 Last Core to Fill 1 Unfilled Cores 2,4 Due to Ratholing Filling order 3.43 sec 5.45 sec 1.50 sec Defects Large voids L1, C2, L3, R3, C4, L5, R5 Small voids N/A Blow Pressure Location Number of inlets Area (each inlet) (mm 2 ) Pressure (kpa) Area Fraction Top 1 (Top Plane) Sand Head Total Blow Area

68 61

69 62

70 63

71 64 Test Number Sand Tube Vacuum Blow Pressure Location Sand Head Air 3 W/ Air Inlets None Top + Sand Tube No Total Fill Time 0.37 sec First Core to Fill 2,3,4,5 Last Core to Fill 1 Unfilled Cores N/A Due to Ratholing Filling order 2.35 sec 3.35 sec 4.35 sec 5.35 sec 1.37 sec Defects Large voids L1, L4 Small voids L2, L3, R4, L5, R5 Blow Pressure Location Number of inlets Area (each inlet) (mm 2 ) Pressure (kpa) Area Fraction Top 1 (Top Plane) Sand Tube (Top) Sand Tube (Middle) Sand Tube (Bottom) Total Blow Area

72 65

73 66

74 67 Test Number Sand Tube Vacuum Blow Pressure Location 4 W/ Air Inlets None Top + Sand Tube Sand Head Air Yes Total Fill Time N/A First Core to Fill 3 Last Core to Fill 5 Unfilled Cores 1,2,4 Due to Ratholing Filling order 3.41 sec 5.43 sec Defects Large voids 1L,1R,1C,2C,3L,4C,5L,5R Small voids Blow Pressure Location Number of inlets Area (each inlet) (mm 2 ) Pressure (kpa) Area Fraction Top 1 (Top Plane) Sand Tube (Top) Sand Tube (Middle) Sand Tube (Bottom) Sand Head Total Blow Area

75 68

76 69

77 70 Test Number Sand Tube Vacuum Blow Pressure Location Sand Head Air 5 Standard 10 kpa Top No Total Fill Time 0.36 sec First Core to Fill 4 Last Core to Fill 1 Unfilled Cores N/A Due to Ratholing Filling order 4.34 sec 2.35 sec 3.35 sec 5.35 sec 1.36 sec Defects Large voids 1L,5L Small voids 1R,2R,2L,5R Blow Pressure Data Location Number of inlets Area (each inlet) (mm 2 ) Pressure (kpa) Area Fraction Top 1 (Top Plane) Total Blow Area

78 71

79 72