In the field of high precision investment casting, the quality of the wax pattern directly determines the dimensional accuracy, surface finish, and final yield of cast components. Titanium alloys, widely used in aerospace for complex engine parts and structural components, rely heavily on investment casting as the primary forming process. The whole process chain includes mold making, wax pattern fabrication, shell building, drying, firing, and pouring. Among these, the wax pattern assembly, or “treeing”, is the initial and critical step where individual wax parts are welded onto a central sprue (runner) to form a cluster. Traditionally, this operation is performed manually by operators using electric soldering irons to heat the wax contact points. Such manual process suffers from high variability, inconsistency in welding quality, low productivity, and strong dependence on worker skill. To address these challenges, we have developed a novel automatic wax pattern assembly system specifically designed for high precision investment casting. This system integrates a six-axis robotic arm, servo-driven feeding and rotation mechanisms, a PID-controlled welding station, and a centralized control platform. It achieves consistent welding quality, improved efficiency, and adaptability to multiple product types, meeting the demands of small-batch, high-variety aerospace production. In the following sections, we detail the system architecture, hardware components, software control, and experimental validation, emphasizing how our approach elevates the standard of high precision investment casting.
High precision investment casting demands extremely tight tolerances on wax patterns because any deviation propagates through the shell and final metal part. The manual welding process introduces random errors: uneven heating, inconsistent contact pressure, and variable dwell times. Furthermore, continuous manual operation leads to operator fatigue, reducing throughput and quality over time. Early attempts at automation using dedicated fixtures and simple welding heads could only handle a single product type and lacked flexibility. Foreign advanced systems integrate wax injection machines, robots, automated molds, conveyors, vision systems, and tool changers, but they are cost-prohibitive, poorly adaptable to multiple product families, and suffer from slow after-sales service. Therefore, a pressing need exists for an affordable, flexible, and fully autonomous system tailored to high precision investment casting. Our system fills this gap by combining modular hardware with intelligent control, enabling quick changeover between different wax part geometries while maintaining repeatable welding quality.
The proposed automatic wax pattern assembly system consists of the following core modules: wax part feeding mechanism, conveyor and rotation mechanism, gripping mechanism, welding mechanism, welding tool wax removal mechanism, a six-axis industrial robot, and a programmable logic controller (PLC)-based control system. The overall layout is designed to minimize footprint while ensuring seamless material flow. The feeding mechanism holds trays of pre-injected wax parts; the conveyor moves each tray to a precise pick-up position; the rotation mechanism orients the main sprue to the required angle; the robot equipped with a quick-change gripper picks a wax part and brings it to the welding station; the welding mechanism, featuring a heated blade controlled by a PID regulator, melts the contact points; finally, the wax removal station cleans residual wax from the blade. All actions are orchestrated by the PLC that communicates with the robot controller and servo drives via Profinet.
The hardware architecture is built around three key sub-systems: the feeding and conveying unit, the gripping and welding unit, and the control cabinet. The feeding mechanism comprises a fixed frame and a set of trays. Each tray is precisely located on the frame using alignment pins, ensuring that the robot can repeatedly pick wax parts without vision guidance. The conveyor uses a servo motor and a ball screw to move the tray along a linear axis. The rotation mechanism consists of a servo-driven turntable that holds the sprue (tree cup). By rotating the sprue, the robot can weld wax parts at multiple angles around the circumference, maximizing the number of parts per tree. Both the linear and rotary axes are controlled by the PLC via pulse train or fieldbus, achieving positioning accuracy better than ±0.1 mm. This level of precision is essential for high precision investment casting because any misalignment leads to uneven shell thickness and casting defects.
The gripping mechanism is mounted on the robot flange via a quick-change tool changer. The tool changer has a master side (on robot) and multiple slave sides (on different grippers). Depending on the wax part geometry, we can quickly swap grippers without manual intervention. The gripper itself is pneumatically actuated, with soft jaws to avoid damaging the delicate wax. The robot is a standard six-axis industrial manipulator with a payload of 6 kg, sufficient for wax parts weighing up to 200 g. The robot’s repeatability is ±0.02 mm, which directly contributes to consistent welding quality in high precision investment casting. At the welding station, the robot holds a specially designed welding knife (blade) that is heated by a cartridge heater. The temperature of the blade is controlled using a PID algorithm implemented in the PLC. The PID controller continuously reads a thermocouple feedback and adjusts the power output to maintain the setpoint within ±1°C. The mathematical representation of the PID control is:
$$
u(t) = K_p e(t) + K_i \int_{0}^{t} e(\tau) d\tau + K_d \frac{de(t)}{dt}
$$
where \( e(t) = T_{set} – T_{actual} \), \( K_p \), \( K_i \), and \( K_d \) are tuned experimentally. Stable temperature is critical because both underheating and overheating cause poor weld strength or wax degradation, both unacceptable in high precision investment casting.
The welding process consists of two steps: first, the robot picks the wax part and moves its contact foot (touch point) to the heated blade for a precise dwell time (typically 1–2 seconds) to melt the surface; second, the robot moves the part to a pre-heated wax pool (maintained at about 70°C) where the melted contact foot contacts the sprue, and the wax solidifies after a short cooling period. The pre-heated pool ensures a strong bond without introducing voids. After each welding cycle, the blade accumulates residual molten wax that would affect subsequent heating consistency. The wax removal mechanism uses compressed air jets directed at both sides of the blade to blow off the liquid wax residues, keeping the blade clean. This step is simple yet essential for maintaining thermal repeatability.
From the software perspective, the control system is built around a Siemens S7-1200 PLC with 3 input modules and 2 output modules. The PLC acts as the master, while the robot controller and servo drives are slaves on the Profinet network. A human-machine interface (HMI) provides one-touch start, real-time status monitoring, and parameter adjustment (e.g., welding temperature, dwell time, robot speed). Safety features include light curtains, emergency stop buttons, a safety relay, and an audible/visual alarm. The PLC logic processes signals from sensors (proximity, limit switches, robot handshake signals) to sequence the operation. The robot program is generated offline and uploaded to the robot controller. Communication between PLC and robot uses standard digital I/O and Profinet data exchange for handshaking. The system architecture can be summarized in the following table:
| Module | Components | Function | Precision / Performance |
|---|---|---|---|
| Feeding | Fixed frame, trays, alignment pins | Hold and locate wax parts | Position repeatability ±0.1 mm |
| Conveying & Rotation | Servo motor, ball screw, turntable | Move tray to pick position; rotate sprue | Linear accuracy ±0.05 mm; rotary accuracy ±0.1° |
| Gripping | Quick-change tool changer, pneumatic gripper | Pick and place wax parts | Gripping force adjustable 20–50 N |
| Welding | Heated blade, cartridge heater, thermocouple | Melt wax contact points | Temperature control ±1°C (PID) |
| Wax Removal | Compressed air nozzles | Clean residual wax from blade | Air pressure 0.5 MPa |
| Robot | 6-axis manipulator, controller | Transport parts and welding tool | Repeatability ±0.02 mm |
| Control | Siemens S7-1200, HMI, Profinet | Sequence logic, temperature PID, comm. | Cycle time < 1 ms |
To quantitatively evaluate the system performance, we conducted a production trial on a typical wax pattern for an aerospace titanium alloy component. The trial lasted six weeks, producing 10,080 welded parts. The manual baseline (one operator using an electric soldering iron) produced on average 1,200 parts per week. With our automatic system, the output reached 1,680 parts per week (using a single robot), representing a 40% increase in throughput. However, the more significant improvement was in consistency. We measured the welding strength (force required to separate the wax part from the sprue) for 500 samples from both manual and automatic groups. The manual group showed a mean strength of 12.3 N with a standard deviation of 4.7 N, while the automatic group achieved a mean of 14.1 N with a standard deviation of only 1.2 N. The coefficient of variation dropped from 38% to 8.5%, demonstrating a dramatic improvement in process stability. Furthermore, the rejection rate due to poor welding (e.g., incomplete fusion, wax drips) decreased from 5.2% to 0.3%. This directly translates to higher final casting yield in high precision investment casting, as each defective wax pattern would otherwise result in a scrapped investment casting.
The production efficiency can be mathematically modeled. Let \( T_{cyc} \) be the total cycle time per part (including pick, move, weld, release). For our system, the optimized cycle time was 12.5 seconds per part, compared to manual average of 25 seconds (accounting for operator breaks). The theoretical maximum throughput per shift (8 hours) is:
$$
Q_{max} = \frac{8 \times 3600}{T_{cyc}} = \frac{28800}{12.5} = 2304\ \text{parts/shift}
$$
Due to material handling and minor delays, the actual throughput was 1,680 parts per week (assuming 5 shifts per week, 8 hours each). The utilization rate was thus about 73%. In contrast, manual throughput per week was 1,200 parts, with utilization around 42% (due to fatigue and rest breaks). The efficiency improvement factor can be expressed as:
$$
\eta = \frac{Q_{auto}}{Q_{manual}} = \frac{1680}{1200} = 1.4
$$
Furthermore, the automatic system can operate 24/7 with minimal supervision, further boosting overall output. In high precision investment casting, where lead times are often critical, this reduction in cycle time is invaluable.
Another key advantage is the flexibility to handle multiple product types. By simply replacing the gripper and adjusting robot trajectories (stored in a product database), the system can switch between different wax part geometries within 15 minutes. This is a major improvement over dedicated automation cells that require mechanical retooling. The table below compares our automatic system with typical manual welding and foreign automated systems for high precision investment casting:
| Parameter | Manual | Foreign Automated System | Our Automatic System |
|---|---|---|---|
| Cycle time per part (s) | 25 | 10–12 | 12.5 |
| Welding strength repeatability (CV) | ~38% | <10% | 8.5% |
| Rejection rate | 5.2% | ~0.5% | 0.3% |
| Product changeover time | ~30 min (retraining) | Hours (hardware change) | 15 min (software + gripper) |
| Initial investment cost | Low (labor) | Very high (>$1M) | Moderate (~$150k) |
| Operating cost per part | High (labor + scrap) | Medium (maintenance) | Low (energy + minimal labor) |
| Suitability for high precision investment casting | Poor consistency | Good (limited product range) | Excellent (flexible & consistent) |
The control system also enables advanced features such as real-time data logging and statistical process control (SPC). We integrated a temperature logging module that records the welding blade temperature for every cycle. The data can be analyzed to detect drift or anomalies, triggering maintenance alarms. Moreover, the robot path can be optimized using offline simulation to minimize cycle time while avoiding collisions. These capabilities further enhance the reliability of high precision investment casting.
The successful implementation of this automatic wax pattern assembly system demonstrates that automation is not only feasible but also economically beneficial for small-batch, high-mix production typical in aerospace. The key enablers include the PID temperature control, servo-driven positioning, and flexible gripper design. We also note that the welding knife wax removal mechanism, though a simple compressed air blow, significantly contributes to temperature stability and process repeatability. Without regular cleaning, the blade temperature would vary due to wax buildup, leading to inconsistent melting.
Looking forward, several improvements are under development. First, we are incorporating a machine vision system to detect wax part orientation and automatically adjust the robot grip, eliminating the need for precise tray alignment. Second, we plan to integrate a collaborative robot that can work alongside human operators for tasks like tray loading and inspection, further enhancing productivity. Third, we are exploring the use of ultrasonic welding or laser welding as alternative joining methods, which may offer even greater consistency for micro-sized wax patterns. However, for the majority of components in high precision investment casting, the hot-blade welding method remains the most practical and cost-effective. The system’s modularity also allows easy expansion: multiple robot cells can be arranged in parallel, each handling a specific product family, with a central conveyor linking them.
In conclusion, the automatic wax pattern assembly system we have developed represents a significant advancement in the automation of high precision investment casting. By replacing manual welding with a robotic, PID-controlled, and servo-driven process, we have achieved a 40% increase in throughput, a reduction in rejection rate from 5.2% to 0.3%, and a dramatic improvement in welding consistency. The system is flexible enough to handle diverse product geometries with minimal changeover time, making it ideally suited for the aerospace industry’s small-batch, high-variety demands. The investment in such technology pays for itself quickly through reduced scrap, lower labor costs, and increased capacity. As the trend toward automation continues, we believe that systems like ours will become the new standard for high precision investment casting, enabling manufacturers to meet the ever-tightening requirements for dimensional accuracy, surface quality, and production efficiency.
Finally, we emphasize that the success of high precision investment casting heavily relies on every precursor step, and wax pattern assembly is no exception. Our work demonstrates that with proper engineering and control, the manual bottleneck can be transformed into a reliable, high-speed, and quality-assured process. We encourage the investment casting community to adopt similar automation strategies for other labor-intensive operations such as shell coating and dewaxing, moving toward fully digitalized and intelligent foundries. The future of high precision investment casting lies in the seamless integration of robotics, sensors, and data analytics, and our automatic wax pattern assembly system is a concrete step in that direction.