The conventional sand 3D printers used in the casting industry have long suffered from low printing efficiency, which has become a bottleneck for the industrial-scale application of additive manufacturing in sand casting. In our work, we developed a novel multi-working box sand 3D printer based on the 3DP (3D printing) principle, aiming to significantly improve the production throughput while maintaining high precision and reliability. This article presents the key mechanical design aspects, the working principle, and the performance evaluation of the system. We will discuss the structural design of the main frame, the multi‑working‑box configuration, the material selection and deformation control, and the overall workflow. Through systematic analysis and experimental validation, the proposed design achieves a maximum printing efficiency of 400 L/h, which is a substantial improvement over the conventional single‑box systems.
Sand casting remains the most widely used casting method, accounting for 80%–90% of total casting production. The traditional sand mold manufacturing process involves multiple manual steps, high energy consumption, and significant environmental pollution. The emergence of sand 3D printing, based on the 3DP process (binder jetting onto a sand bed), has revolutionized the foundry industry by eliminating pattern costs, reducing lead times, and enabling complex geometries. However, the printing speed of early sand 3D printers was limited to around 108 L/h, which was insufficient for mass production. To address this challenge, we integrated multiple independent working boxes into a single machine, allowing simultaneous or sequential printing on different boxes. This parallel approach dramatically increases the overall throughput while keeping the machine footprint relatively compact.
The working principle of a sand 3D printer follows a layer‑by‑layer deposition process. First, a 3D model is sliced into thin layers (typically 0.1–1.0 mm). The build platform (working box bottom plate) is raised to one layer thickness below the top of the box. A recoater then spreads a uniform layer of sand across the entire working area. Subsequently, the print head moves over the sand bed and selectively deposits a binder (usually furan resin or phenolic resin) according to the slice data. After completing one layer, the platform descends by one layer thickness, and the process repeats until the entire part is built. After printing, the unbound sand is removed, leaving the finished sand mold or core. Our multi‑working box system extends this principle by having several independent boxes that can be operated either in parallel or in a staggered manner.
The overall mechanical architecture of the multi‑working box sand 3D printer is shown in the following structural diagram (the image is placed at the end of the discussion on the upper frame). The machine comprises a sand mixer, an upper frame, a print head system, a cleaning unit, a recoating system, multiple working boxes, a box conveying system, a box lifting system, a lower frame, and a liquid supply system. The upper and lower frames are constructed from welded steel columns to create several independent working zones. Each zone is equipped with a dedicated lifting mechanism driven by a servomotor with a brake, using a belt drive and ball screw to achieve precise Z‑axis motion with an accuracy of ±0.1 mm. The working boxes have movable bottom plates that are closely coupled with the lifting mechanism. The box conveying system, driven by low‑speed AC motors through chains and rollers, transports boxes between the printing zone and the buffer zone along the Y‑direction.
A key innovation lies in the recoaters: each working box is served by an independent recoater mounted on a linear module along the Y‑axis. The recoater contains a rotating motor and a vibration motor to ensure uniform sand distribution. The print head is suspended from the upper frame and can move in both the X‑direction (via a linear motor) and the Y‑direction (via a servo motor). This dual‑axis movement allows the print head to cover all working boxes. The sand mixer is mounted on top of the upper frame and supplies sand to all recoaters through a distribution system. The cleaning station and liquid supply unit are fixed at one end of the print head’s Y‑travel.
The print head must maintain a constant gap (typically 0.2–0.5 mm) above the sand bed during operation. Any vertical deflection of the upper frame would cause uneven binder penetration and reduce print quality. Therefore, the deformation of the upper frame under moving loads must be strictly controlled. We compared two candidate materials for the crossbeam: Q345B welded steel beams and HT300 cast iron beams. The key mechanical properties are summarized in Table 1.
| Property | Q345B welded steel | HT300 cast iron |
|---|---|---|
| Density (kg/m³) | 7850 | 7200 |
| Young’s modulus (GPa) | 200 | 120 |
| Yield strength (MPa) | 345 | 300 |
| Damping ratio | 0.01–0.02 | 0.05–0.08 |
| Deflection under self‑weight (mm) | 0.15 | 0.22 |
| Cost index | 1.0 | 1.5 |
Although cast iron offers better vibration damping, its lower Young’s modulus results in larger deflection, and its cost is higher. The Q345B steel also has superior anti‑deflection performance. However, finite element analysis (FEA) of the initial design with a straight crossbeam showed a total deflection of 0.208 mm under the combined load of the linear motor, the print head carriage, and the moving mass (see Figure in the original paper). This exceeds the required tolerance of ±0.05 mm. To compensate, we adopted a pre‑curved machining strategy: the mounting surface of the upper crossbeam was machined with a convex upward curve as illustrated in Figure 5 of the original paper. The curve equation was derived from the deflection shape under the concentrated load of the print head assembly. The key design parameters are given in Table 2.
| Parameter | Value |
|---|---|
| Span length (m) | 3.5 |
| Maximum deflection under load (mm) | 0.208 |
| Required straightness after assembly (mm) | ±0.05 |
| Machined convex rise at mid‑span (mm) | 0.25 |
| Curve profile | Parabolic: y(x) = 4h x (L‑x)/L² |
The parabolic pre‑curve ensures that when the moving mass (print head and carriage) is placed, the beam flattens to within the tolerance band. The FEA results after pre‑curving showed a maximum deviation of only 0.038 mm under concentrated load (excluding self‑weight), meeting the design requirement.
The lifting system for each working box must maintain a consistent layer thickness throughout the build. The Z‑axis motion is achieved using a ball screw driven by a servomotor with a brake. The screw pitch is selected to provide a resolution of 0.01 mm per pulse. The lifting force required depends on the weight of the sand, the box, and the platform. The maximum load for a single box (dimensions 1.2 m × 0.6 m × 0.8 m) is approximately:
$$
F_{\text{max}} = m_{\text{sand}} \cdot g + m_{\text{platform}} \cdot g
$$
Where the sand mass \( m_{\text{sand}} = \rho_{\text{sand}} \cdot V_{\text{box}} \). Taking sand density \( \rho_{\text{sand}} = 1.6 \, \text{g/cm}^3 = 1600 \, \text{kg/m}^3 \), and box volume \( V_{\text{box}} = 1.2 \times 0.6 \times 0.8 = 0.576 \, \text{m}^3 \), we get \( m_{\text{sand}} = 921.6 \, \text{kg} \). Including the platform mass of 50 kg, the total mass is about 972 kg, giving a force of approximately 9530 N. The ball screw efficiency is about 0.9, and the servomotor torque required is:
$$
T = \frac{F \cdot P}{2\pi \eta} = \frac{9530 \times 0.01}{2\pi \times 0.9} \approx 16.9 \, \text{N·m}
$$
We selected a 2 kW servomotor with a brake to provide ample torque and safety margin.
The recoaters are critical for achieving uniform sand layers. Each recoater is driven by a dedicated linear module along the Y‑axis. The recoater linear speed is set at 0.3 m/s to ensure a consistent layer without disturbing previously deposited binder. The sand is supplied from the mixer via a flexible hose. Inside the recoater, a rotating auger and a vibrator ensure that the sand falls evenly through a slot. The slot width is adjustable from 1.0 mm to 3.0 mm to control the layer thickness. The relationship between the recoater speed, slot width, and layer thickness is given by:
$$
h = \frac{Q}{v \cdot w}
$$
Where \( h \) is the layer thickness (mm), \( Q \) is the sand flow rate (mm³/s), \( v \) is the recoater speed (mm/s), and \( w \) is the slot width (mm). For a typical operation, \( h = 0.3 \, \text{mm} \), \( v = 300 \, \text{mm/s} \), \( w = 2.0 \, \text{mm} \), we require \( Q = h \cdot v \cdot w = 0.3 \times 300 \times 2 = 180 \, \text{mm}^3/\text{s} \). The mixer is designed to supply up to 500 mm³/s to multiple recoaters simultaneously.
The print head assembly uses a linear motor in the X‑direction and a servo‑driven Y‑module. The X‑axis linear motor can achieve acceleration up to 20 m/s² and a maximum speed of 2 m/s. The Y‑axis uses a ball screw with a servo motor to achieve positioning accuracy of ±0.02 mm. The total mass of the print head and carriage is about 80 kg. The inertia forces during acceleration are:
$$
F = m a = 80 \times 20 = 1600 \, \text{N}
$$
These forces must be absorbed by the upper frame without causing excessive vibration. The pre‑curved beam design, combined with the high damping of the welded Q345B structure, successfully limits the vibration amplitude to below 0.05 mm during acceleration.
The overall workflow of the multi‑working box sand 3D printer is as follows: Before printing, a working box is conveyed into position by the roller system. The lifting system raises the box platform to the top limit. The corresponding recoater checks its sand level via a level sensor. If sand is insufficient, the mixer prepares and delivers fresh sand mixed with catalyst. Then the recoater moves across the box to deposit a uniform sand layer. After recoating, the print head moves to that box and performs one full pass (forward and backward) to deposit binder. The print head then moves to the next box or waits. The lifting system lowers the platform by one layer thickness (0.1–1.0 mm, typically 0.2–0.5 mm). The cycle repeats until the box is full. Multiple boxes can be operated concurrently: while one box is being printed, another can be recoated or conveyed out. This parallelism dramatically increases efficiency. The calculated printing efficiency for a single box is about 100 L/h. With four boxes operating in a pipelined fashion, the overall efficiency can reach 400 L/h, as shown in Table 3.
| Configuration | Number of boxes | Efficiency per box (L/h) | Total throughput (L/h) |
|---|---|---|---|
| Conventional single‑box | 1 | 108 | 108 |
| Multi‑box (our design) | 4 | 100 | 400 |
The liquid supply system delivers binder, catalyst, and cleaning fluids to the print head. A key parameter is the binder jetting frequency and droplet volume. Each droplet is approximately 80 pL. The print head has 512 nozzles operating at 50 kHz, giving a maximum flow rate of:
$$
Q_{\text{binder}} = 512 \times 80 \times 10^{-12} \times 50 \times 10^3 = 2.048 \times 10^{-3} \, \text{L/s} = 7.37 \, \text{L/h}
$$
This is sufficient for the target application. The cleaning station uses a combination of ultrasonic cleaning and solvent flushing to prevent nozzle clogging.
To further increase productivity, the machine can also be operated in a mixed‑mode: some boxes can be used for printing while others are being cooled or removed. The conveying system uses a chain drive with a linear speed of 0.2 m/s, allowing a full box to be moved out within 10 s.
The mechanical design of the multi‑working box sand 3D printer has been successfully implemented in industrial applications. Field tests demonstrate that the machine can produce sand molds and cores with surface finish comparable to traditional methods, while reducing production time by over 60%. The elimination of pattern tooling also reduces costs and lead times significantly. Furthermore, the machine operates with minimal noise and dust pollution, aligning with green foundry initiatives.
In conclusion, we have developed a multi‑working box sand 3D printer that effectively addresses the low‑efficiency bottleneck of conventional sand 3D printers. By incorporating multiple independent working boxes, a pre‑curved upper frame, and a synchronized recoater‑print head system, the machine achieves a throughput of 400 L/h, which is nearly four times that of a single‑box machine. The design details, including material selection, deformation compensation, and mechanical parameter calculations, have been validated through FEA and practical testing. This machine paves the way for the widespread industrial adoption of 3D printing in sand casting, enabling rapid prototyping, low‑volume production, and complex geometries with reduced environmental impact. The ongoing work includes optimizing the recoater design for even higher speeds and integrating intelligent monitoring systems for predictive maintenance.