1. Overview

Manifolds distribute and collect coolant between the CDU and the cold plates in a liquid-cooling system. Depending on the design, a part may carry several ports, connection holes, threaded holes, mounting positions and sealing faces, which places demands on part location, operation sequencing and finished-part protection. This project covers the CNC machining stage: pallet-based blank feeding, robot loading and unloading, a multi-machine line and finished-part recovery bring handling, clamping and collection into a single flow.

2. Production background and retrofit need

• Blanks must be presented in a consistent orientation so the robot can grip them reliably.
• Different ports and machined faces may need different clamping references, which have to match the actual process.
• Parts must move between several machines, so machining-complete signals and robot service order have to be coordinated.
• Machined sealing faces, threads and fittings need dedicated grip positions and finished-part placement.
• During long runs the cell has to manage blank shortage, machine alarms, clamping faults and finished-store status.

The source material does not describe how the part was produced before automation. The proposal notes that manual piece-by-piece handling becomes a heavier burden as the number of machines grows, but it does not confirm that the customer previously worked manually; that point still requires a site record.

3. Workpiece and operation characteristics

What the source material records about the workpiece:

• The manifold connects the CDU of a liquid-cooling cabinet to the cold plates and distributes the coolant.
• Depending on the product structure it may involve face machining, hole machining, thread machining, port machining and multi-operation finishing.
• A part may carry several ports, connection holes, threaded holes, mounting positions and sealing faces.

4. Automation challenges

• Keeping one datum between pallet location, robot grip and machine fixture.
• Balancing cycle times across several machines instead of judging by machine count.
• Protecting sealing faces and threads during handling.
• Mixed specifications: pallet zoning, gripper compatibility, program matching and poka-yoke.
• Blank shortage, full store and recovery after a fault during continuous operation.

5. System design

• Pallet feeding — blanks are placed in pallets by specification and grip orientation and located by the pallet structure.
• Robot pick — parts are picked from fixed positions and re-oriented for machine clamping.
• Automatic loading — the part is placed in the fixture and confirmed by part-present, clamp-closed and machine-ready signals.
• Sequence handling — after each operation the cell unloads and transfers the part to the next machine according to the real routing.
• Finished-part recovery — completed parts go into a finished pallet or store for inspection and transfer.
• Exception handling — failed picks, machine alarms, clamp faults, empty pallets and full stores are detected and handled safely.

6. Loading and unloading sequence

1. Pallet loaded into the cell → part located and picked
2. Robot loads the target machine → clamp confirmed
3. CNC machining → machining-complete signal
4. Robot unloads → transfer by routing or to the finished store
5. Cell checks machine and store status → next cycle

7. Machine, robot and storage coordination

What the proposal requires of the cell — design requirements, not measured behaviour:

• The robot must interlock with the machine door, clamp-closed signal, machining-complete signal and the safety interlock, so it never enters the machining area while the machine is not in a safe state.
• On a multi-machine line, machine status, robot motion and part routing have to be coordinated.
• For mixed specifications, a defined pallet zone and program-matching relationship is needed to prevent mixing.
• Optional functions named in the proposal are blank-level monitoring, full-store warning, alarm reporting, cycle counting and status feedback.

8. Key selection basis

The proposal refers only to a “robot” and states explicitly that it has not been decided whether a gantry robot or a 6-axis industrial robot would be used. It also notes that whether one robot can serve several machines depends on layout, service range, machining time and load/unload time, and cannot be judged from machine count alone.

9. Machine make and model fit

10. Part storage and automatic feeding

What the proposal records about storage:

• A pallet-based blank store combined with a finished-part store.
• Pallet structure is designed around the manifold's size, weight, geometry and locating datum so parts can be gripped reliably.
• Planning condition: 300 blanks, with a storage configuration targeted at 8 hours of continuous running (a planning condition, not an acceptance result).

Note: the proposal text quotes 300 blanks, while the confirmed store has 256 positions. The two figures are not on the same basis; how they reconcile is still to be confirmed.

11. Multi-machine line (where applicable)

The proposal positions the project as a multi-machine CNC line but does not record how many machines are connected, and states that “if several operations are involved, the part routing must follow the actual process route; no unconfirmed number of operations is assumed”.

12. Control, safety and exception handling

Control and safety requirements stated in the proposal:

• The machine door, clamp-closed signal, machining-complete signal and safety interlock must be interlocked with robot motion.
• Failed picks, machine alarms, clamp faults, empty pallets and a full store must be detected and handled safely.

13. Commissioning and verification

Verification items listed in the proposal — proposed at concept stage; no acceptance record exists:

1. Pallet location and blank pick success rate
2. Part placement and clamp-state confirmation
3. Coordination and waiting time between machines
4. Unload and finished-part placement stability
5. Blank shortage, full store and fault recovery during continuous running
6. Changeover and anti-mix logic between specifications
7. Measured cycle time, uninterrupted runtime and the fault log

The proposal states plainly that “verification data are still to be supplied, and no success rate, productivity gain or failure rate may be entered without acceptance records”. No measured value is given in this section.

14. Results and data scope

The proposal gives 300 blanks over a target of 8 continuous hours, i.e. an average blank consumption rate of 37.5 parts per hour, or about 96 s per part. It states explicitly that this is a planning calculation, that it does not represent the machining cycle time, and that an 8-hour unattended run has not been validated.

Single-part machining cycle time is 68 s (business confirmation). The line output interval is not provided. The ~96 s figure below is a planning calculation from 300 blanks over 8 hours and is not the machining cycle time.

15. Scope and limitations

• The concept gives manifold batch production a repeatable flow — pallet feeding, robot tending, multi-machine transfer and finished-part recovery — reducing repeated manual handling and creating the conditions for longer unattended running.
• Delivered, but without a written record: pick reliability, machine-to-machine coordination and waiting time, unload stability, uninterrupted runtime, changeover and anti-mix performance, and measured cycle time.
• Not provided in the source material: customer name, acceptance records, productivity gains, operator reductions, first-pass yield.

16. FAQ

Q: Why is multi-machine tending considered for manifolds?
A: When a part needs several operations on different machines, an automated cell can transfer it between machines and remove repeated manual handling. Whether a line pays off still depends on layout and cycle balance.

Q: How long do 300 blanks last?
A: Under the planning figures, 300 blanks correspond to 8 hours. Note that the confirmed store holds 256 positions, so the planning figure and the physical capacity are not on the same basis. Changing specification, routing or machine set requires recalculating the consumption rate.

Q: Can one robot serve several machines?
A: It depends on layout, robot reach, machining times and load/unload times. Machine count alone is not a basis for the decision.

Q: Does automatic tending guarantee sealing-face quality?
A: Automation makes handling, clamping and placement consistent and reduces the risk of damage, but sealing-face quality still depends on the process, tooling, fixtures and inspection.

Q: What is the machining cycle time?
A: 68 s per part for a single machining cycle, confirmed by the business. The line output interval depends on the number of machines and the operation split, which are still to be confirmed.

Reference images

The two images below show comparable automation lines and equipment forms (including our own assembly workshop). They are not photographs of this project’s customer site; the actual configuration is subject to project assessment.

Source and verification status

Facts on this page are graded in the table below. No figure, customer name or acceptance conclusion appears without a stated origin.

Business-confirmed values come from a verbal confirmation on 21 October 2026. Machine models, machine count, material grades and dimensions are still to be confirmed from written records, and none of them has been estimated here.

Related pages

To discuss automation for a similar workpiece, send us the blank drawing or photos, machine make and model, and target output.

This page follows the same 19-block structure as the Chinese version. All facts and data-scope statements from the original proposal and case manuscript are retained. Items with no record remain marked “to be confirmed”; business-confirmed values are shown separately and were not mixed into the document-derived text.