As a vital branch of the foundry industry, pump valve castings are widely applied and demand high-quality standards, especially those with complex geometries. This thesis focuses on the integrated forming technology of pump valves with complex structures, combining rapid prototyping, computer simulation, and a novel submerged casting technique. The entire sand core is manufactured by additive manufacturing, enabling a one-piece core that replaces the traditional segmented core assembly. To explore the feasibility of applying submerged casting to complex pump valves, a series of water simulation experiments were carried out using an industrial robot to control the submerged pouring pipe. The flow field stability, slag removal efficiency, and dynamic response of the pipe were investigated. Based on the experimental results, an optimized motion curve was proposed to suppress turbulent fluctuations. A multi-way valve was selected as the research carrier. The casting scheme was designed and verified through numerical simulation using JSCAST. The sand mold was machined by CNC engraving, while the sand core was fabricated by PIRP rapid prototyping. After coating, drying, and assembly, the casting experiment was conducted with ductile iron. The as-cast part was inspected by three-dimensional scanning and sectioning. The results show that most dimensional deviations fall within -1.5 mm to -0.5 mm, and no severe internal defects were found. The research provides a reliable digital casting route for complex pump valve parts, and demonstrates the potential of submerged casting in industrial applications.
1 Introduction
1.1 Background and Significance
Casting is one of the oldest manufacturing processes, with a history of more than 6000 years. It plays a fundamental role in the machinery industry. In modern manufacturing, castings are used in aerospace, automotive, shipbuilding, energy, and many other sectors. The basic principle of casting is to melt metallic materials and pour the liquid metal into a mold cavity, where it solidifies to obtain the desired shape. Among all casting methods, sand casting remains the most widely used because of its low cost, flexibility, and suitability for a broad range of metal alloys and part sizes. However, conventional sand casting has several limitations, especially when dealing with complex internal cavities and high-precision requirements.
For complex pump valve castings, the traditional core-making processes usually require multiple core boxes, manual assembly, and adhesive bonding. These operations introduce unavoidable dimensional errors and reduce core strength. Moreover, the design of gating systems for complex parts is challenging. A poor gating system can cause turbulences, slag inclusions, gas porosity, shrinkage, and misruns. To overcome these difficulties, digitalized casting techniques have been developed in recent years. These techniques integrate computer-aided design, rapid prototyping, casting simulation, and three-dimensional scanning to improve the quality and efficiency of casting production. Among them, rapid prototyping (also called 3D printing) is a promising method for producing sand cores with complex geometries using a layer-by-layer approach. It eliminates the need for expensive core boxes and allows the direct fabrication of integral cores, thereby ensuring both geometric accuracy and structural integrity.
Submerged casting is a novel casting technology that combines computer-aided motion control and bottom filling. In this process, the pouring pipe is always kept below the liquid surface during mold filling. The liquid metal is filled smoothly from bottom to top without falling stream and without splashing. This is similar to a special metal additive manufacturing process in which the molten metal is deposited layer by layer. The temperature field in the mold exhibits a favorable bottom-up solidification pattern, which promotes directional solidification and effective feeding. Previous studies have shown that submerged casting can produce castings with reduce porosity and better mechanical properties. Nevertheless, this technology has not been widely applied to industrial castings such as pump valves. This thesis aims to investigate the feasibility of using submerged casting for a complex multi-way valve, and to integrate this new casting method with rapid prototyping to establish a complete digital casting route.
1.2 Current Research Status of Complex Pump Valve Castings
Pump valves are key components in hydraulic systems and fluid control devices. They are usually made of gray cast iron, ductile iron, or cast steel. Due to the high pressure and tight sealing requirements, pump valves must possess excellent internal soundness and dimensional accuracy. Many domestic and foreign researchers have contributed to pump valve casting technology. However, most studies focus on the optimization of gating systems, the development of new molding materials, or the application of special casting processes such as investment casting or lost foam casting. Few studies address the integral core fabrication for complex pump valves. Rapid prototyping offers a unique advantage for such components. By using PIRP (Profile Ineffective Rapid Prototyping) technology, a whole sand core with complex curved channels can be built from coated sand without any seam or bonding. This significantly improves the core strength and prevents core displacement during pouring.
The application of robots to casting is also an emerging trend. Industrial robots are capable of precise path control and real-time response. In this thesis, a robot manipulator was used to control the movement of the submerged pouring pipe. This robotic water simulation platform allowed visual observation of the flow field and enabled the verification of the velocity control curves. The combination of robot control and submerged casting creates a new direction for intelligent casting.
1.3 Main Research Contents
This thesis is organized as follows:
(1) A water simulation experimental platform was built using an industrial robot to simulate the submerged casting process. Different submerged pipes with different outlet diameters and configurations were tested. The flow velocity coefficient of each pipe was measured. The effect of intermediate ladle and slag collector on impurity removal was investigated. The influence of outlet orientation and cross-sectional area on flow stability was analyzed. The path and speed of the submerged pipe were optimized for actual pouring.
(2) The casting scheme for a multi-way valve was designed, including the gating system, the pouring temperature, and the processing allowance. The filling and solidification processes were simulated using the JSCAST software to verify the rationality of the design.
(3) The sand mold was machined by a CNC engraving machine, whereas the integral sand core was fabricated by PIRP rapid prototyping. The core was then post-processed by heating and stripping.
(4) The actual pouring experiment was carried out with ductile iron. The obtained casting was inspected by three-dimensional scanning and destructive sectioning. The results demonstrated the success of the proposed digital casting route.
2 Water Simulation Experiments for Submerged Casting
2.1 Theoretical Basis of Water Simulation
Water simulation is used to visualize the flow behavior of molten metal because the kinematic viscosity of water is close to that of some liquid metals. According to the similarity theory, the flow field in a water model can reflect the filling behavior in real casting under certain conditions. The basic principles of fluid mechanics, especially the Bernoulli equation, are applied to estimate the velocity of the submerged pipe.
For a submerged casting process, the molten metal flows out from the submerged pipe and fills the mold cavity. The outlet velocity can be expressed by the Bernoulli equation:
$$P_1+\rho g H_1+\frac12 \rho v_1^2 = P_2+\rho g H_2+\frac12 \rho v_2^2 + h_w$$
where \(P_1, H_1, v_1\) are the pressure, height, and velocity of the liquid surface in the pouring ladle; \(P_2, H_2, v_2\) are the corresponding values in the mold; and \(h_w\) is the head loss. For free outflow, \(H_2=0\), \(P_1=P_2=P_{atm}\), \(v_1 \approx 0\). The head loss can be approximated as
$$h_w = \zeta \frac{v_2^2}{2g}$$
Thus,
$$H_1 = (\alpha+\zeta)\frac{v_2^2}{2g}$$
Define the flow coefficient \(\mu_1 = 1/\sqrt{\alpha+\zeta}\), then the outlet velocity is
$$v_2 = \mu_1 \sqrt{2g H_1}$$
For submerged outflow, the liquid surface height in the mold is not zero. The equation becomes
$$H_1-H_2 = (\alpha+\zeta)\frac{v_2^2}{2g}$$
so
$$v_2 = \mu_1 \sqrt{2g(H_1-H_2)}$$
At any instant, the continuity equation must be satisfied:
$$Q = A_1 v_1 = A_2 v_2 = A_3 v_3$$
where \(A_1, A_2, A_3\) are the cross-sectional areas of the ladle, the submerged pipe outlet, and the mold cavity, respectively. \(v_1, v_2, v_3\) are the corresponding velocities. The rising speed of the liquid surface in the mold, \(v_3\), is equal to the required moving speed of the submerged pipe if the pipe is kept at a constant depth beneath the surface.
According to the volume conservation, the movement of the submerged pipe should be controlled so that the pipe outlet is always submerged. If the mold has a variable cross-section, the rising speed must be changed accordingly. For complex parts such as a multi-way valve, the cross-sectional area of the cavity varies with height because of the internal core. Therefore, the velocity curve is composed of several segments, each corresponding to a different cross-sectional area.
2.2 Water Simulation Equipment and Materials
The experimental platform consisted of an industrial robot, a transparent intermediate ladle, a clear acrylic mold, and a slag collector. The robot manipulator was equipped with a flange to hold the intermediate ladle. The submerged pipe was attached to the bottom of the intermediate ladle. The intermediate ladle had a partition plate to remove impurities from the water. The mold was made of transparent plexiglass so that the flow behavior could be recorded by a high-speed camera. The slag collector was placed at the bottom of the mold to trap the first stream of water, which contained numerous bubbles and impurities. The robot controller was programmed using a teach pendant. The program included the motion path and speed of the submerged pipe during the pouring process.
| Component | Specification | Function |
|---|---|---|
| Industrial robot | 6-axis manipulator | Control the motion of the submerged pipe |
| Intermediate ladle | Acrylic box with partition | Slag removal and flow damping |
| Slag collector | Acrylic small box | Collect the first stream and impurities |
| Submerged pipes | Different diameters and outlets | Vary the outlet flow patterns |
| Mold cavity | Transparent acrylic plates | Visual observation of flow field |
The general procedure of the water simulation experiment is as follows:
- Place the submerged pipe into the slag collector at the bottom of the mold.
- Open the electromagnetic valve by a remote control to start the water flow.
- At the same time, start the robot program that moves the pipe according to the pre-calculated velocity curve.
- Record the flow process using a camera and observe the interface behavior.
- Analyze the recorded images to evaluate the flow field stability and the position of the pipe below the free surface.
2.3 Measurement of Flow Velocity Coefficient
To achieve a correct velocity control, it was necessary to measure the actual flow velocity of different submerged pipes. The flow velocity coefficient depends on the pipe geometry, friction, and turbulence. We used a simple volume-time method. The submerged pipe was placed at the bottom of the slag collector. The time required to fill the slag collector from the moment of opening the valve to the moment when the water reaches a reference mark was measured. The volume of the slag collector was known. Then the average volumetric flow rate was calculated.
Table below shows the initial measured results and the corrected results after subtracting the time for water to travel along the pipe from the top to the outlet (about 0.8 s).
| Pipe type | Outlet diameter (mm) | Measured filling time (s) | Corrected filling time (s) | Outlet velocity (mm/s) | Liquid rise speed in collector (mm/s) | Liquid rise speed in mold (mm/s) |
|---|---|---|---|---|---|---|
| Straight | 20 | 2.8 | 1.5 | 1332.5 | 53.3 | 7.0 |
| Straight | 16 | 2.0 | 1.9 | 1644.5 | 42.1 | 5.5 |
| Straight | 12 | 2.8 | 2.3 | 2415.5 | 34.8 | 4.6 |
| Side | 8.5 | 3.9 | 3.0 | 1847.8 | 26.7 | 3.5 |
| Side | 11 | 2.6 | 2.4 | 1376.0 | 33.3 | 4.4 |
| Side | 14 | 2.3 | 2.0 | 1020.4 | 40.0 | 5.2 |
After the correction, the rising speed of the submerged pipe was adjusted accordingly. The verification experiments showed that the pipe could follow the liquid surface with a small constant depth below it. The difference between the calculated speed and the actual speed was within the acceptable range.
2.4 Effect of Intermediate Ladle and Slag Collector
During the initial stage of pouring, the first stream is turbulent and contains many air bubbles. In addition, impurities floating on the surface of the molten metal may be carried into the mold if the flow is too violent. To solve this problem, we added an intermediate ladle and a slag collector. The intermediate ladle had a vertical partition wall. Water entered the ladle on one side and left through a bottom opening on the other side. The partition wall prevented the floating impurities from entering the outlet if the water level was maintained above the bottom clearance.
Two modes were tested. In the first mode, the intermediate ladle was initially filled and then the outlet valve was opened without continuous water supply. It was observed that when the water level dropped below the partition wall, the impurities passed through the bottom clearance and entered the mold. Moreover, near the end of the discharge, air bubbles were drawn into the outlet. In the second mode, the water supply was continuous so that the intermediate ladle remained full. In this case, the impurities were trapped on the inlet side and no significant air entrainment was observed. Therefore, continuous pouring is recommended to achieve the best slag removal effect.
The slag collector at the bottom of the mold was designed to capture the first turbulent stream. When the submerged pipe was placed near the bottom of the slag collector, the outlet was initially below the water surface. As water flowed in, a lot of bubbles were generated. After the water level rose and the pipe outlet was submerged, the bubbles gradually disappeared. If the pipe was raised too early, the turbulent region could cause the impurities collected in the slag collector to be re-entrained into the main mold cavity. The experiment demonstrated that the slag collector combined with a continuous intermediate ladle effectively reduced the amount of bubbles and impurities entering the main cavity.
2.5 Influence of Outlet Configuration and Cross-Sectional Area
Six submerged pipes were tested: three straight pipes with internal diameters of 12, 16, and 20 mm, and three side-outlet pipes all with an internal diameter of 20 mm but side openings of 8.5, 11, and 14 mm. The flow field around the outlet was observed for each pipe.
For straight pipes, the water jet impinged vertically on the bottom region. The turbulent zone was directly below the pipe outlet. When the outlet diameter was large, the outlet velocity was low, and the turbulent region was wide but shallow. When the outlet diameter was small, the velocity was high, but the turbulent region was narrower and less noticeable. For side-outlet pipes, the water stream rushed out horizontally from the side openings. The turbulent zones appeared at both sides of the pipe, while the bottom was relatively calm. Similar to the straight pipes, a larger side opening produced a lower velocity and a shallower but wider turbulent zone.
Based on these observations, the selection of outlet configuration should consider the shape of the casting. If the casting has a long and narrow internal cavity, a straight pipe might be preferred because the turbulent region is confined to the bottom. If the casting has complex side structures, a side-outlet pipe can prevent direct impingement on the core surfaces. The cross-sectional area of the outlet determines the flow rate and should be matched with the required filling time.
Table: Flow field comparison of different outlet types
| Outlet type | Diameter | Velocity | Turbulent zone location | Turbulent zone size | Effect on overall flow |
|---|---|---|---|---|---|
| Straight | 20 mm | Low | Bottom center | Wide but shallow | Moderate |
| Straight | 16 mm | Medium | Bottom center | Narrow | Small |
| Straight | 12 mm | High | Bottom center | Very narrow | Minimal |
| Side | 14 mm | Low | Both sides | Wide near pipe | Moderate |
| Side | 11 mm | Medium | Both sides | Narrow | Small |
| Side | 8.5 mm | High | Both sides | Very narrow | Minimal |
2.6 Path Design and Optimization of the Submerged Pipe
In the initial experiments, the submerged pipe was placed at the bottom of the slag collector and then lifted at a constant speed. When the pipe exited the slag collector and entered the mold cavity, the flow through the narrow opening caused severe splashing. This was because the pipe velocity was still based on the cross-section of the collector, while the actual cross-section changed suddenly. As a result, the liquid level rose too quickly and the submerged depth became insufficient.
To solve this problem, the motion path was redesigned. The pipe was first held at the bottom of the collector until the water level had risen above the outlet. Then, the pipe was lifted at a velocity corresponding to the collector cross-section until the outlet reached the bottom level of the mold. After that, the pipe was stopped and waited for the water surface to rise above the outlet. Finally, the pipe resumed moving upward according to the mold cross-section. This optimized path effectively eliminated the splashing and maintained the liquid surface stable.
It was also found that the direction of the outlet influenced the stability during the transition stage. For a straight pipe, the turbulent jet could still affect the slag collector after the pipe had moved into the mold cavity, causing impurities to be stirred. For a side-outlet pipe, the turbulent zones remained around the pipe, so the impurities in the collector were not disturbed when the pipe moved upward. Thus, the side-outlet configuration is more favorable in the presence of a slag collector.
2.7 Water Simulation for Variable Cross-Section Castings
In practical castings, the cross-sectional area varies along the height. To verify the robot control algorithm, two artificial cores were inserted into the transparent mold. One core had a gradually varying cross-section, and the other had an abruptly varying cross-section. The goal was to check whether the submerged pipe could follow the liquid surface as the cross-section changed.
For abrupt changes, the control program could be set to a series of discrete velocities. The robot response time is negligible, so the transition between two velocities is almost instantaneous. The water simulation showed that the pipe stayed below the liquid surface, and the flow was stable after the liquid surface passed the cross-section transition.
For gradual changes, the cross-section was approximated by several small steps. The more segments used, the closer the approximation. The average cross-sectional area of each segment was used to calculate the average velocity. The experiment proved that the robot could execute the multi-segment velocity program smoothly. The pipe remained at a desired depth below the liquid surface, and no obvious disturbance was observed.
2.8 Water Simulation for the Multi-Way Valve Casting
To evaluate the feasibility of submerged casting for a real complex pump valve, the actual multi-way valve sand core was placed inside the transparent mold. The core had twelve discrete sections along the vertical direction. The cross-sectional area of each section was extracted from the CAD model using the Magics software. A plane with a thickness of 0.3 mm was used to perform a Boolean operation with the core to obtain the cross-sectional area. It was found that a 0.1 mm plane and a 0.3 mm plane gave similar results, while a 1.0 mm plane overestimated the area. Therefore, the 0.3 mm thickness was used.
The table below shows the section divisions and the corresponding cross-sectional areas and velocities.
| Section | Height (mm) | Cross-sectional area (mm²) | Required rising speed (mm/s) |
|---|---|---|---|
| 1-1 | 10 | 28452 | 8.9 |
| 1-2 | 15 | 28797 | 8.3 |
| 1-3 | 20 | 30682 | 7.2 |
| 2 | 25 | 32600 | 6.6 |
| 3 | 41 | 39800 | 5.4 |
| 4 | 51 | 37700 | 5.7 |
| 5 | 63 | 40500 | 5.3 |
| 6 | 68 | 40500 | 5.3 |
| 7 | 83 | 35800 | 6.0 |
| 8 | 87 | 41300 | 5.2 |
| 9 | 92 | 44800 | 4.8 |
| 10 | 97 | 42100 | 5.1 |
| 11 | 115.5 | 44800 | 4.8 |
| 12 | 130.5 | 40500 | 5.3 |
The velocity curve was entered into the robot controller. The water simulation results showed that the pipe was always kept under the free surface and rose smoothly with the liquid level. The flow remained stable even when the liquid passed through the complex core sections. No violent splashing or air entrainment was observed.
3 Casting Scheme Design and Numerical Simulation
3.1 Mold and Core Information
The research object is a multi-way valve used in hydraulic systems. The CAD model consists of the valve body and the complex internal core. The core defines several intersecting channels that are difficult to produce by conventional core-making. In this study, the core was made as a single piece by PIRP rapid prototyping. The overall dimensions of the casting are about 180 mm × 120 mm × 130 mm, and the wall thickness varies from 6 mm to 15 mm.
3.2 Casting Process Design
The casting material was ductile iron. The pouring position was chosen so that the large core plane was vertical and the main channels were oriented upward, allowing easy core support and gas venting. The processing allowance was determined according to GB/T6414-1999. The linear shrinkage coefficient was set to 0.9%.
The gating system was designed as a bottom-gated, open system to ensure smooth filling. The cross-sectional area ratio of the sprue, runner, and ingate was chosen as \(A_{sprue}:A_{runner}:A_{ingate}=1.5:2:1\). The ingate was the choke area. The total casting weight including the gating system was about 20 kg. The filling time was calculated by
$$t = S\sqrt{m}$$
where \(S=1.7\), \(m=20\) kg. Thus,
$$t = 1.7 \times \sqrt{20} \approx 7.6\ \text{s}$$
The average static head was \(H_p=20\ \text{cm}\). The choke area is given by
$$A_{ingate} = \frac{m}{\mu \rho t \sqrt{2g H_p}}$$
Using \(\mu=0.6\), \(\rho=7.3\ \text{g/cm}^3\), \(t=7.6\ \text{s}\), \(H_p=20\ \text{cm}\),
$$A_{ingate} \approx 8.6\ \text{cm}^2$$
and accordingly \(A_{sprue}=12.9\ \text{cm}^2\), \(A_{runner}=17.2\ \text{cm}^2\). A step-wise slag trap was added at the bottom of the runner to prevent oxide inclusions. No riser was equipped because ductile iron can self-feed through graphite expansion if the pouring temperature is adequately controlled.
3.3 Numerical Simulation with JSCAST
The CAD model of the casting and gating system was converted to STL format and imported into JSCAST. The mold material was defined as sand, and the casting material was selected as ductile iron. The meshing and calculation parameters were set automatically. The filling and solidification processes were simulated to evaluate the flow pattern, the temperature distribution, and the last solidification regions.
The simulation results of the filling process showed that the liquid metal rose smoothly from the bottom of the mold to the top. No apparent misrun or cold shut was observed. The flow front advanced uniformly, and the core remained stable. The solidification simulation indicated that the last solidification region was located near the center of the valve body. Because ductile iron has a strong tendency to form graphite expansion, the risk of shrinkage porosity was considered acceptable. The maximum center temperature after filling was about 1280 °C, and the complete solidification time was approximately 320 s.
| Simulation parameter | Value |
|---|---|
| Pouring temperature | 1400 °C |
| Mold material | Furan sand |
| Core material | Coated sand |
| Heat transfer coefficient | 500 W/(m²·K) |
| Initial mold temperature | 20 °C |
| Filling time | 7.6 s |
| Total solidification time | 320 s |
The simulation confirmed the reliability of the designed casting process. Therefore, the actual pouring experiment was carried out without further changes.
4 Rapid Manufacturing of Sand Mold and Core
4.1 Parting Design of the Sand Mold
The sand mold was designed as a rectangular block with a minimum molding thickness of 40 mm. To facilitate CNC machining, the mold was divided into three parts: the upper part, the middle part, and the lower part. The upper part contained the pouring cup and the upper cavity surface. The middle part contained most of the cavity and the gating system. The lower part was designed to machine the bottom of the cavity and the slag trap. This parting design allowed the engraving cutter to access all internal surfaces without requiring manual undercutting.
4.2 CNC Engraving of the Sand Mold
The three parts of the sand mold were machined from pre-cured sand blocks using a CNC engraving machine. The upper and lower parts were relatively simple. The middle part required two-sided machining. The machining parameters were selected to achieve a surface roughness suitable for coating. After engraving, the mold parts were cleaned and inspected. The dimensions were verified by measuring with calipers and by trial assembly.
4.3 PIRP Rapid Prototyping Principle
PIRP (Profile Ineffective Rapid Prototyping) is a novel layer-by-layer manufacturing method developed for sand parts. It uses a coated sand material that cures at low temperature and becomes ineffective at high temperature. The process is as follows: A layer of coated sand is spread on a platform. A heated plate is pressed onto the layer to cure the whole layer. Then a laser beam scans the outer contour of the part, heating the sand to a high temperature and causing the resin to lose its binding ability. By repeating this process, each layer has a weak outer contour. After all layers are stacked, the loose sand outside the contour can be removed, leaving the desired part. The main advantage of PIRP is its high speed and low cost for large parts, and it does not require expensive laser sintering of the entire area.
4.4 Data Processing for the Core
The CAD model of the multi-way valve core was first added with auxiliary split blocks. These blocks are intended to facilitate the removal of waste material during post-processing. The combined model was converted to STL format and then to CLI format, which is recognized by the PIRP machine. The slicing thickness was set to 0.3 mm. The process data included the contour file for the laser scanning and the heating cure file for the whole layer.
4.5 Core Fabrication and Post-Processing
The PIRP machine used in this study was a PIRP-1500 built by the research group. The coated sand was of a commercial grade used for sand casting. After the forming process, the core was still weak because the resin had only been partially cured. The surrounding loose sand was carefully removed as much as possible. Then the core was placed in a heating furnace at about 180 °C for 30 minutes to fully cure the resin. To prevent deformation during heating, the core was embedded in silica sand. After curing, the auxiliary blocks were separated, and the residual support material was manually removed. The outer surfaces of the core were lightly sanded to remove the step effect. The finished core showed good integrity and accurate dimensions.
| Process step | Equipment | Parameters | Remarks |
|---|---|---|---|
| Spreading sand | PIRP-1500 | Layer thickness 0.3 mm | Coated sand |
| Laser scanning | CO2 laser | Power 60 W | Outer contour only |
| Post-curing | Furnace | 180 °C, 30 min | Full resin curing |
| Stripping | Manual | — | Remove waste blocks |
5 Casting Experiment and Inspection
5.1 Coating and Preparation
Before pouring, both the sand mold and the core were coated with a refractory coating. The coating formula was an alcohol-based zirconia coating, which is typically used for steel castings. The high-quality coating was chosen to provide excellent surface finish and to prevent sand-metal reaction. The upper and lower mold parts were brushed with a coating of Baume degree 55–60. The middle mold part was coated by dipping or flow coating with a Baume degree of 45–50. The complex core was coated by flow coating to ensure all internal channels received a uniform thin layer. After coating, the parts were ignited immediately to evaporate the alcohol, leaving a solid refractory layer.
Vent holes were drilled on the upper mold part to allow easy escape of gases generated by the decomposition of resin during pouring. The mold parts and core were then dried in an oven at 80 °C for 12 hours to remove residual moisture. Finally, the core was positioned in the mold, and the three mold parts were assembled. To prevent runout, the mold was clamped and surrounded by molding sand. A weight was placed on the top to avoid floating of the mold due to metallostatic pressure.
5.2 Melting and Pouring
The melting experiment was carried out in a steel plant foundry. The green strength of the mold was sufficient to withstand the metallostatic pressure. The pouring temperature was controlled between 1360 °C and 1420 °C. The pouring process was performed manually with a ladle. The actual pouring time was roughly 8 s, which agreed well with the design value. The total weight of the poured metal was about 25 kg. After pouring, the casting was allowed to cool completely in the mold.
| Parameter | Value |
|---|---|
| Alloy | Ductile iron |
| Melting furnace | Medium frequency induction furnace |
| Pouring temperature | 1380 °C |
| Mold temperature | 20 °C |
| Filling time | 7.8 s |
| Casting weight | 16.5 kg |
| Total weight with gating | 24 kg |
The casting after knockout was free of visible surface defects such as cold shuts, misruns, or excessive flash. The surface was smooth due to the coating, and no burn-on was observed.
5.3 Post-Processing and Dimensional Inspection
The gating system was cut off using a band saw. The remaining riser stubs were ground. The cast part was then cleaned by sand blasting. Three-dimensional scanning was performed using a structured-light scanner. The scanning data were reconstructed into a triangulated mesh and compared with the original CAD model. The deviation analysis showed that the majority (more than 85%) of the surface points had deviations between -1.5 mm and -0.5 mm. The standard deviation was about 0.7 mm, indicating a concentrated distribution of errors. Considering the sand casting process and the complex internal core, this accuracy is satisfactory.
To inspect the internal quality, the casting was cut along a plane that revealed the main oil channels. The cross-section showed that all internal channels were fully formed and free of core displacement or blockage. The surfaces inside the channels were smooth, and no sand inclusion was visible. An additional dye penetrant test was carried out on the section plane. The inspection results did not reveal any major cracks or shrinkage cavities. Minor micro-porosity might exist, but it is within the acceptable level for sand castings.
Conclusions
In this thesis, an integrated digital manufacturing route for complex pump valve castings was established. The main conclusions can be summarized as follows:
- Industrial robots can be successfully used to control the submerged casting process. The water simulation platform proved that the robot executed the programmed motion accurately and smoothly. The submerged pipe remained under the free surface at all times, demonstrating the feasibility of robotic submerged casting.
- Water simulation experiments provided useful insights into the flow behavior during submerged filling. The intermediate ladle with a partition wall and a continuous pouring mode effectively removed impurities from the liquid metal. The slag collector prevented the first turbulent stream and entrained bubbles from entering the main cavity. The side-outlet configuration was preferred to avoid the re-entrainment of impurities from the slag collector.
- Variable cross-section mold cavities can be handled by dividing the filling process into multiple segments. The robot’s fast response allows abrupt velocity changes. Gradual changes are approximated by a series of segments, and the accuracy can be improved by using more segments. The water simulation with the actual multi-way valve core confirmed that the proposed velocity control curve was correct.
- The integral sand core for the multi-way valve was successfully manufactured by PIRP rapid prototyping. The core geometry was accurate and the core strength after curing was adequate for the pouring process.
- The designed bottom-gated open gating system, combined with numerical simulation, resulted in a smooth filling pattern and sound casting. The actual pouring experiment produced a defect-free multi-way valve. Three-dimensional scanning showed that dimensional deviations were within -1.5 mm to -0.5 mm for most surface points, and sectioning revealed complete internal channels without obvious defects.
This research confirms that the combination of rapid prototyping, submerged casting, and robot automation provides a promising approach for manufacturing complex sand casting parts with high quality and reliability.