End-Return and Wrap-Around Welding on Ship Stiffeners: Define the Corner Before Automating It

The short detail that can govern automation coverage

A long, open stiffener fillet is usually easy to identify on a drawing. The difficult scope discussion often begins at its end. A required return around a stiffener termination can change the torch orientation, shorten the available approach and bring the nozzle close to an edge or adjacent bracket. Its length may be small compared with the longitudinal seams, while its effect on automatic completion is substantial.

In Chinese project discussions, 包脚 and 包角 can refer to wrapping or returning the weld around a local end or corner. Translate the required geometry with a marked drawing, not with the term alone. Depending on the detail, the intended work may be an end return, a corner continuation or another specified termination. This article uses end return as a practical discussion label, not as a substitute for a design symbol or a universal welding requirement.

TitanWeld's marine brief covers a highly non-standard double-sided, double-arc project, with two arms in the basic unit and a current ceiling of four units or eight arms. It does not establish automatic coverage of every end detail. End-return coverage, alternative completion methods and acceptance evidence must be agreed for the specific assembly and tool arrangement.

Resolve the design instruction before choosing a path

Mark the exact extent of the required weld on the current detail drawing. Identify the two longitudinal sides, the termination face, any intentionally unwelded region and the relationship to neighboring joints. Confirm the weld size, whether the weld must continue around the end and whether a particular start or stop location is specified. Refer an unclear symbol to the responsible designer rather than filling the uncertainty with a robot-programming convention.

Give the return its own identifier or an explicit segment within a persistent weld identifier. This allows an inspection record to say what was examined and an output report to say what remains unfinished. A single completed flag for the whole stiffener can otherwise conceal a missing return, especially when both long sides were welded successfully.

The WPS defines the welding variables while qualification evidence supports its applicable scope, as explained by TWI. [1] For the project review, ask how the selected instructions address the end detail and any change in position or technique. A procedure used for an unrestricted straight fillet should not be treated as evidence for every corner simply because the same wire and power source are used.

Check the complete tool envelope around the end

Model or inspect the nozzle, torch neck, sensor, wrist, hose routing and adjacent fixtures together. The wire tip can sometimes reach the nominal corner while the gas nozzle cannot follow the required attitude. Similarly, the robot may have a feasible static pose but lack a clear approach or departure. Evaluate the complete motion, including service and recovery movements, with the actual tool package proposed for the project.

Review the return from both sides of the assembly. The end of a stiffener can be open on one side and crowded on the other by a bracket, temporary attachment or clamp. A symmetric part model may fail to include temporary production hardware. The trial drawing should therefore show the fixture state and identify which attachments are present during welding.

TWI identifies geometry sensors as tools for weld searching and tracking and notes that welding conditions can interfere with sensing. [2] At an end, verify what the selected sensor can observe before and during the transition. A sensing demonstration on a long straight joint does not establish that a short edge, abrupt geometry change or partly hidden corner will be interpreted correctly.

Divide the motion into reviewable regions

Discuss the end detail as an approach region, a corner transition, a return segment and a departure region. This division does not prescribe separate weld starts; it provides a way to inspect tool orientation, commanded speed and process changes at each part of the path. Keep any decision to stop and restart within the approved process development plan.

At the transition, review whether the robot can preserve the required tool relationship without excessive reorientation or an unplanned dwell. A path preview should show the tool body as well as the centerline. If a smoothing function changes the nominal corner trajectory, establish how the actual deposited weld will be assessed. A mathematically smooth curve is useful only if it produces the required joint geometry under the agreed procedure.

Make the seam-end reference observable. If the system uses a nominal model endpoint, identify how placement and manufacturing variation are handled. If it uses a measured feature, record the feature definition, the accepted detection result and the action for an ambiguous reading. The correct response to uncertainty may be a controlled hold for review rather than an automatic extrapolation beyond the part.

Coordinate the two arms at a shared termination

Two arms working on opposite longitudinal fillets can approach the same termination at similar times. The space around the end may not allow both complete tool envelopes to occupy their desired positions simultaneously. Examine this region as a separate shared-access problem, including the wrists and hoses. Do not infer a safe sequence from a drawing that shows only two wire tips.

Possible project strategies include assigning the return to one arm, staggering the local operation or completing it in a separate agreed stage. These are options for evaluation, not claims of installed TitanWeld functionality. The chosen method must account for the required weld continuity, local access, procedure instructions and the unfinished work left for the other arm.

Write the handover explicitly: which longitudinal segment is complete, which arm owns the termination operation, where the other arm waits and what releases the next movement. Preserve weld identity through the handover. Otherwise a controller might consider both programs complete while the required connecting detail has no recorded owner. Test this condition using the actual scope map rather than relying only on the program sequence names.

Use representative end details in development trials

Choose trial pieces that reproduce the relevant end geometry, nearby obstructions, support condition and surface condition. A short isolated coupon can help develop a local technique, but it may remove the fixture or neighboring stiffener that causes the real access problem. Use the development stages deliberately: first learn about the local weld, then confirm that the process fits the representative assembly.

Observe the approach and transition, then examine the finished detail using the specified inspection route. Select any destructive or additional examination with the responsible welding and inspection personnel. Do not invent a mandatory sectioning location or test quantity from an editorial example. The point is to make the evidence capable of answering the project's particular concern about the corner and termination.

TWI describes visual inspection as a method used alongside other examination techniques and assessed against applicable criteria. [3] A photograph of a smooth return is therefore useful context, but it should not be presented as complete proof of internal soundness or structural suitability. Keep the photo, the measured result and the formal acceptance record distinct and traceable to the same trial detail.

Measure the operational cost of the return

Separate longitudinal arc time from the total time needed to complete the end. The additional work may include sensing, reorientation, waiting for the opposite arm, another start, a special inspection position or an agreed manual operation. These activities are part of the assembly cycle even when the deposited return is short. Ignoring them can make a favorable weld-length coverage ratio misleading.

Count return details by type and frequency in the expected production mix. One rare obstructed corner may be an acceptable exception; a similar detail repeated at every stiffener can dominate intervention frequency. Report both the number of automatically completed returns and the number requiring another operation. Keep acceptance status visible rather than equating attempted coverage with accepted coverage.

Use the same boundary in comparisons. If a baseline includes the fitter's preparation and the return weld while a proposed robotic time covers only the straight sides, the two figures do not describe the same work. An honest comparison may still support automation strongly, but the improvement should emerge from a complete and agreed scope rather than from a convenient stopping point in the stopwatch record.

Prepare for interruption at the corner

An interrupted return needs a location-specific recovery instruction. Identify whether the stop occurred before the transition, during the return or after the required termination. The operator should be able to locate the incomplete region on the assembly drawing. Retain the original operation record and add the recovery operation so that subsequent examination can follow both.

The responsible welding engineer should define any inspection, cleaning or preparation required before restarting. The automation sequence should make that intervention possible with the necessary access and controlled machine state. Returning automatically to a stored coordinate is not enough if the joint surface has changed during preparation or if the assembly was moved to gain access.

During acceptance, include a recoverable interruption at a representative end detail when it is relevant to the agreed scope. Evaluate whether the operator can understand the event, access the work, choose the approved recovery and complete the record. This test examines the usability of the entire process. It can reveal a practical weakness that remains invisible during an uninterrupted demonstration of a single attractive weld.

State coverage in drawing terms

A useful end-return acceptance statement names the drawing revision, detail family, incoming condition, tool arrangement, procedure reference and examination result. It also identifies any excluded corner and the approved way to finish it. This provides a buyer with a basis for judging a proposed production mix and gives the integrator a stable target for further development.

Avoid a blanket statement that the system can wrap every stiffener foot or corner. The same phrase can conceal substantially different geometry across ship assemblies. Equally, do not assume that one failed detail excludes all end-return automation. Classify the details and preserve the reason for each result so that changes in tooling or fixture design can be evaluated against specific constraints.

The goal is complete, reviewable joint coverage. When a return is required by the design, it deserves a named operation, an access assessment and an acceptance record even if it contributes little to total weld length. That discipline keeps the most difficult few centimeters of a joint from becoming an unplanned manual task at the end of every production cycle.

Project review checklist

• Locate each required end return on the controlled drawing.
• Record the stiffener-end geometry and the required joint detail.
• Check torch attitude, nozzle clearance and fixture obstruction throughout the turn.
• Define start, stop and transition treatment in the applicable procedure.
• Record demonstrated coverage, excluded corners and the accepted finishing method.