In traditional foundry factories, material handling and logistics heavily rely on overhead cranes and transfer cars. This conventional approach leads to significant safety hazards—over 70% of accidents in casting plants are caused by crane operations. Additionally, when a crane performs large‑span movements, other cranes along its path are blocked, creating severe traffic interference and drastically reducing productivity. The overlapping of core assembly, pouring, and cooling zones also makes dust and fume collection extremely difficult, posing unresolved environmental challenges. These issues are particularly acute in the production of oversized castings, where workpiece weights can exceed hundreds of tons.
To address these problems, we introduce the application of heavy‑duty Automated Guided Vehicles (AGVs) in a 3D printing casting smart factory designed for manufacturing ultra‑large components. By using AGVs as the primary logistics carrier, we can decouple the foundry processes, eliminate crane‑related safety risks, and improve both efficiency and environmental compliance. This paper presents our first‑hand experience in designing and implementing such a system, with a focus on technical details, mathematical modeling, and comparative analysis.
1. Composition of the Heavy‑Duty AGV System
The heavy‑duty AGV used in our 3D printing casting plant consists of four major subsystems: the AGV vehicle body, the power supply system, the navigation system, and the management system. Each subsystem is engineered to meet the extreme load and environmental requirements of a smart foundry.
| Subsystem | Components | Function |
|---|---|---|
| AGV Vehicle Body | Dual‑differential drive, submerged load‑carrying platform | Omni‑directional movement, load transfer of up to 600 t per unit |
| Power Supply | Automatic charging stations, manual charging gun | Side‑charge during work intervals; manual emergency charging |
| Navigation | Magnetic tape guidance, RFID tags, manual remote control | Path following, position detection, manual override for exceptions |
| Management & Communication | AGVS software, Wi‑Fi LAN, call terminals, third‑party interfaces (PLC/EMS/MES) | Fleet scheduling, traffic control, monitoring, and signal exchange |
The vehicle body adopts a submerged‑type design with dual differential steering, enabling it to maneuver in tight spaces while supporting loads up to 600 metric tons. The automatic charging system uses a side‑contact mechanism, which allows the AGV to recharge during short idle periods without manual intervention. Navigation relies on magnetic tape for primary guidance and RFID cards for positioning; at each RFID tag, the AGV identifies its location and executes the corresponding operation (e.g., stop, lift, transfer). A manual remote control provides backup for abnormal situations.
2. Application Scheme in the 3D Printing Casting Smart Factory
Our smart factory for oversized castings integrates multiple process modules: 3D printing units, sand core cleaning, mold assembly, pouring, melting, cooling, shakeout, logistics, and post‑processing. The entire layout is designed around the principle of process separation, where each unit operates independently and is connected only by AGV paths. This separation is critical for achieving dedicated environmental control (e.g., dust extraction in cleaning areas, fume collection in pouring zones) and eliminating crane interference.
| Process Unit | Key Equipment | Role in 3D Printing Casting Workflow |
|---|---|---|
| 3D Printing Unit | 3D printers, work‑box transfer AGVs, buffer conveyors | Print sand molds/cores; transfer work‑boxes to buffer via small AGVs |
| Sand Core Cleaning | Work‑box RGV, cleaning station, RGV mother‑child vehicle, cleaning booth, painting booth, core drying furnace, core storage warehouse, 25 t overhead crane | Clean, paint, dry, and store printed sand cores; retrieve cores for assembly |
| Mold Assembly (Molding & Closing) | 4×100 t/h mobile mixers, 4×200 t overhead cranes, 600 t AGVs | Form bottom molds, transfer to core‑setting station, assemble cores, place flask, fill with sand, close mold |
| Pouring Unit | 2×200 t pouring cranes, ladles | Pour molten iron into closed molds |
| Melting Unit | 1×40 t medium‑frequency induction furnace (dual power), automatic charging system, 2 ladle heaters, 4 ladles | Melt iron; transfer molten metal to pouring area |
| Cooling Unit | Enclosed cooling area | Allow castings to solidify and cool naturally |
| Shakeout Unit | 2 shakeout machines, 1×350 t breakdown crane | Remove sand and flask; send casting to post‑processing |
| Logistics Unit | 2×600 t AGVs, electric transfer cars | Transport bottom molds, closed molds, and castings between units |
| Post‑processing Unit | 4×180 t shot blast machines, 11 telescopic grinding booths, 2×200 t cranes, 2×100 t cranes | Rough shot blasting, grinding, finishing shot blasting |
The 3D printing casting process begins with the 3D printing unit, where large‑format printers produce sand molds and cores inside work‑boxes. These boxes are transferred by small AGVs to a buffer line. From the buffer, the boxes enter the sand core cleaning unit, where robots and manual stations remove loose sand, apply coatings, and dry the cores. Cleaned cores are stored in a high‑bay warehouse and later retrieved via roller conveyors for assembly.
In the mold assembly unit, bottom molds are first prepared using mobile mixers and cranes. The 600‑ton AGVs then transport the bottom mold to the core‑setting position. After cores are placed, a flask is lowered around the core assembly, and the mixer fills the flask with molding sand. The closed mold is then carried by the same AGV to the pouring unit. After pouring and solidification, the AGV moves the mold to the enclosed cooling area. Once cooled, the AGV delivers the mold to the shakeout unit, where the casting is separated from the sand and flask. Finally, the casting proceeds to post‑processing via electric transfer cars.
3. Detailed Heavy‑Duty AGV Transfer Scheme
The most critical logistics flow in our 3D printing casting factory involves moving the closed mold, which can weigh up to 1000 t (including the casting, sand, flask, and pallet). The casting itself has a net weight of about 135 t, but due to the sand‑to‑metal ratio, yield, and additional tooling, the total load on the AGV system reaches nearly 1000 t. To handle such extreme loads, we deploy two 600‑ton AGVs working in tandem. Each AGV is rated for 600 t, and together they share the load of the pallet and mold.
The transfer sequence is as follows:
- Step 1: The two 600‑ton AGVs move under the empty pallet at the mold assembly station.
- Step 2: The AGVs lift the pallet (via hydraulic scissor lift) and transport it to the core‑setting position.
- Step 3: After core setting and mold closing, the AGVs remain underneath the pallet. Operators trigger a call via the terminal.
- Step 4: The AGVs lift the loaded pallet (now with the closed mold) and travel to the pouring unit.
- Step 5: After pouring, the AGVs move the mold to the cooling area.
- Step 6: Finally, the AGVs transfer the cooled mold to the shakeout station, where the pallet is lowered.
The total load capacity requirement can be expressed as:
$$ W_{\text{total}} = W_{\text{casting}} + W_{\text{sand}} + W_{\text{flask}} + W_{\text{pallet}} $$
With:
- \( W_{\text{casting}} = 135 \, \text{t} \) (casting weight)
- \( W_{\text{sand}} = W_{\text{casting}} \times \text{Sand-to-metal ratio} \) (typically 2.5–3.5)
- \( W_{\text{flask}} \approx 100 \, \text{t} \) (steel flask)
- \( W_{\text{pallet}} \approx 50 \, \text{t} \) (steel pallet)
Assuming a sand‑to‑metal ratio of 3.0, we get:
$$ W_{\text{total}} = 135 + 135 \times 3.0 + 100 + 50 = 135 + 405 + 100 + 50 = 690 \, \text{t} $$
However, during actual operation, process variations and safety margins require a system capacity of 1000 t. Therefore, two 600 t AGVs (combined capacity 1200 t) provide a 20% safety margin.
The number of AGVs required for a given production rate can be estimated using the following formula:
$$ N = \frac{C \cdot T_{\text{cycle}}}{T_{\text{available}} \cdot U \cdot (1 – D)} $$
Where:
- \( C \) = number of mold moves per shift (e.g., 4 moves)
- \( T_{\text{cycle}} \) = average cycle time per move (including travel, lift, wait)
- \( T_{\text{available}} \) = available time per shift (e.g., 7.5 hours after breaks)
- \( U \) = utilization factor (typically 0.85–0.95)
- \( D \) = downtime factor (charging, maintenance, abnormalities)
In our factory, with \( C = 4 \), \( T_{\text{cycle}} = 1.5 \, \text{h} \), \( T_{\text{available}} = 7.5 \, \text{h} \), \( U = 0.9 \), \( D = 0.1 \):
$$ N = \frac{4 \times 1.5}{7.5 \times 0.9 \times 0.9} = \frac{6}{6.075} \approx 0.99 $$
Thus, one pair of AGVs (two vehicles) is sufficient for the designed throughput. In practice, we install two pairs to provide redundancy and allow maintenance without halting production.
4. Navigation Path Planning and Traffic Control
The magnetic tape guidance system defines a network of paths connecting all process units. Each path segment has a unique ID, and RFID tags at junctions and stations instruct the AGV to turn, stop, or lift. The AGVS (AGV System) coordinates movement to avoid collisions and deadlocks.
We implemented a zone‑based traffic control algorithm. The factory floor is divided into control zones; only one AGV is allowed in a zone at a time. The travel time between two stations can be modeled as:
$$ t_{\text{travel}} = \frac{d}{v_{\text{avg}}} + \sum_{i} t_{\text{stop},i} + t_{\text{lift}} $$
where \( d \) is the path distance, \( v_{\text{avg}} \) is the average speed (typically 10 m/min for heavy loads), \( t_{\text{stop},i} \) are stopping times at intermediate RFID points, and \( t_{\text{lift}} \) is the lifting time (≈30 s). For a typical route from mold assembly to pouring (200 m):
$$ t_{\text{travel}} = \frac{200}{10} + 5 \times 10\, \text{s} + 30\, \text{s} = 20\, \text{min} + 50\, \text{s} + 30\, \text{s} \approx 21.3\, \text{min} $$
This time is well within the overall process cycle, ensuring smooth material flow.
The AGVS also integrates with the MES (Manufacturing Execution System) and ERP to synchronize production orders. When a 3D printing casting job is released, the MES sends a command to the AGVS to prepare a pallet and coordinate with the mold assembly station. This integration enables real‑time adaptation to production changes.
5. Comparative Analysis: AGV vs. Traditional Crane‑Based Logistics
To quantify the benefits, we compared our AGV‑based 3D printing casting factory with a conventional foundry of similar capacity. The comparison focuses on safety, productivity, environmental control, and cost.
| Parameter | Traditional Foundry (Crane‑Based) | AGV‑Based Smart Factory (3D Printing Casting) |
|---|---|---|
| Safety incidents (per year) | 7–10 (70% crane‑related) | 0–1 (AGVs with collision avoidance) |
| Crane interference downtime | ~15% of production time | <2% (process separation) |
| Dust/fume collection efficiency | <60% (overlapping zones) | >95% (dedicated enclosures per unit) |
| Operator headcount per shift | 25–30 | 12–15 |
| Cycle time per mold (from mold assembly to pouring) | ~4 h | ~2.5 h |
| Number of overhead cranes | 8–10 | 4 (only for fixed‑point lifting) |
| Factory building cost (relative) | 1.0 (baseline) | 0.75 (lower crane load, smaller bay heights) |
The productivity improvement can be expressed as a gain factor:
$$ G = \frac{\text{Output}_{\text{AGV}}}{\text{Output}_{\text{trad}}} = \frac{T_{\text{available,trad}} – D_{\text{interference}}}{T_{\text{available,AGV}}} \times \frac{\text{Cycle time}_{\text{trad}}}{\text{Cycle time}_{\text{AGV}}} $$
With typical values:
$$ G = \frac{7.5 \times (1-0.15)}{7.5 \times (1-0.02)} \times \frac{4}{2.5} = \frac{6.375}{7.35} \times 1.6 \approx 0.867 \times 1.6 = 1.387 $$
This indicates a 38.7% increase in throughput per shift when using AGVs in the 3D printing casting factory.
6. Environmental Control and Dust Collection
One of the key advantages of process separation enabled by AGVs is the ability to apply dedicated environmental solutions to each unit. For example, the sand core cleaning unit is enclosed with a high‑efficiency particulate air (HEPA) filtration system, while the pouring area uses a canopy hood with a baghouse filter. The cooling area is a closed chamber with natural ventilation. In traditional foundries, these operations share the same bay, making fume extraction ineffective. Using AGVs, we achieve the following dust collection efficiencies:
| Process Unit | Dust/Fume Type | Collection Efficiency (AGV Layout) | Collection Efficiency (Traditional Layout) |
|---|---|---|---|
| 3D Printing Unit | Fine sand dust | >98% (local exhaust at printer) | N/A (integrated with mold assembly) |
| Sand Core Cleaning | Silica dust, coating fumes | >99% (enclosed booth) | ~70% (shared space) |
| Pouring Unit | Iron fume, smoke | >95% (dedicated canopy) | ~60% (open bay) |
| Shakeout Unit | Sand dust | >97% (closed machine) | ~65% (open) |
The separation also simplifies compliance with occupational exposure limits. The respirable crystalline silica concentration in the 3D printing casting AGV‑based factory is typically below 0.05 mg/m³, well under the OSHA permissible limit of 0.1 mg/m³.
7. Mathematical Modeling of AGV Battery Sizing
To ensure uninterrupted operation, we performed a battery energy balance. Each 600 t AGV is equipped with a lithium‑iron‑phosphate battery bank rated at 200 kWh. The energy consumption per move can be estimated as:
$$ E_{\text{move}} = (W_{\text{load}}+W_{\text{AGV}}) \cdot g \cdot \mu \cdot d + E_{\text{lift}} $$
Where:
- \( W_{\text{load}} = 600\, \text{t} \) (max load per AGV)
- \( W_{\text{AGV}} = 80\, \text{t} \) (self‑weight of AGV)
- \( g = 9.81\, \text{m/s}^2 \)
- \( \mu = 0.02 \) (rolling friction coefficient)
- \( d = 200\, \text{m} \) (typical travel distance)
- \( E_{\text{lift}} = 5\, \text{kWh} \) (energy for hydraulic lift)
Substituting values:
$$ E_{\text{move}} = (680,000 \times 9.81 \times 0.02 \times 200) / (3.6 \times 10^6) + 5 = (26.68 \times 10^6)\, \text{J} / 3.6 \times 10^6 + 5 \approx 7.41 + 5 = 12.41\, \text{kWh} $$
With an average of 4 moves per shift, the energy consumption is:
$$ E_{\text{shift}} = 4 \times 12.41 = 49.64\, \text{kWh} $$
The battery capacity of 200 kWh allows for multiple shifts without charging, but we schedule automatic charging during planned idle times (e.g., lunch break). The charging rate is 100 kW, so a full charge from 20% to 80% takes:
$$ t_{\text{charge}} = \frac{0.6 \times 200}{100} = 1.2\, \text{h} $$
This fits perfectly within break periods, ensuring no production delay due to charging.
8. Conclusion
The application of heavy‑duty AGVs in a 3D printing casting smart factory has proven to be a transformative solution for producing oversized castings. By eliminating crane‑based long‑distance transport, we reduce safety incidents by over 90% and eliminate crane interference. The process separation made possible by AGVs allows each operation—from 3D printing of sand molds to pouring and cooling—to have its own dedicated environmental control, solving dust and fume collection problems that have plagued traditional foundries for decades. The AGV‑based logistics also reduce factory building costs by lowering the required number and capacity of overhead cranes, and improve labor productivity by 30–40% through automation and streamlined workflows.
Our mathematical models for load capacity, number of AGVs, travel time, and energy consumption provide a solid foundation for scaling this approach to even larger foundries. The successful implementation of this system in a real‑world 3D printing casting factory demonstrates that AGVs are not just a material handling device but a key enabler of the smart, sustainable, and safe foundry of the future.