Every shop has a tool that is only just coping. It might be a forming die that has been reworked twice, a fixture that takes an hour to align, or a program that only one person can adjust. It works, until it does not, and it tends to fail when the schedule is tightest.
A schedule that assumes all tooling is healthy is not a schedule. It is a wish. The test of a plan is not how it performs on a good day, but what happens on the day the die cracks at the start of a shift with three jobs depending on it. The range of fabricated parts such a plan must cover is listed under other parts.
This is where mixed-work shops differ most from dedicated lines. A dedicated line can hold a spare die, because it only needs one. A mixed shop has hundreds of tools and cannot duplicate them all, so it has to plan for failure instead of preventing it entirely.
Recovery speed is decided long before the failure. It depends on whether the tool was documented, whether a substitute process exists, and whether anyone has the authority to switch to it without a meeting.
A shop that keeps a tooling register with revision history can identify a replacement or a repair path quickly. A shop that relies on the memory of one experienced setter is fast until that setter is on holiday. The difference is not skill, it is records.
Authority matters as much as records. If every deviation needs sign-off from someone in a meeting, recovery takes days regardless of how good the records are. Shops that recover well have pre-agreed rules: which substitutions are allowed, who may approve them, and what has to be re-measured afterwards.
Slack placed at the end of a schedule is useless, because by then the commitment has already been missed. Slack has to sit immediately after the operations that are most likely to fail, so a failure is absorbed before it propagates.
In fabrication, that usually means buffer after forming and after welding, because both depend on tooling and on skill. Machining is more predictable once the program is proven, and inspection is predictable if the bench is staffed. Putting buffer where variability is low just hides capacity.
The consequence is that a resilient schedule looks inefficient. Machines have gaps. A planner defending those gaps will be asked why, and the answer has to be a specific failure mode, not a general preference for safety.
The fastest recovery is usually a different process, not a repaired tool. A part that was to be formed in a die can sometimes be machined from solid, or fabricated in sections and welded. Each alternative has different cost and different inspection needs, but it exists.
These alternatives are only usable if they were considered at quotation. A shop that quoted one route has no fallback. A shop that quoted a primary route and noted a secondary one can switch in hours, provided the buyer accepts the change.
That acceptance is the sticking point. Buyers often refuse substitutions on principle, which removes the factory's only fast option. A better position is to pre-approve substitutions that stay within the drawing, and to require notification rather than permission.
Customers see a date move. They rarely see why, and the silence during the gap is worse than the news. A shop that calls on the day of the failure with a revised window keeps the relationship; a shop that calls at the end of the week with an apology does not.
The content of the call matters too. "We have a tooling problem" is not information. "The forming die failed, we are machining the first twenty from solid to protect your date, and the balance will follow on the original route" is information, and it gives the buyer something to plan against.
This requires the shop to know, within a shift, which jobs are affected and what the alternatives cost. That knowledge comes from having the register, the authority rules, and a planner who can re-sequence quickly.
Recovery capability can be assessed in a single visit or call, and it does not require seeing a failure.
The answers separate shops that have a recovery process from shops that have a recovery story.
A fully loaded schedule is fragile in a way that is easy to miss, because it looks efficient right up to the moment it fails. One broken tool consumes the buffer of every job behind it, and the recovery cost is paid by customers who had nothing to do with the failure.
Shops that run this way tend to have volatile delivery performance and good unit costs. Shops that hold slack have stable delivery and slightly higher unit costs. Neither is wrong; they serve different buyers. The error is a buyer choosing on price and then expecting the delivery behaviour of the other model.
Mixed work makes this sharper, because the number of tools in play is high and the probability that one of them is marginal is close to one. In that environment, slack is not caution. It is the only way the plan survives contact with the floor.
A purchase order can encode recovery expectations without dictating process. It can require notification within a defined period of any event that threatens the agreed window. It can pre-approve substitutions that remain within the drawing, subject to inspection. It can require the inspection record for any part made by an alternative route.
For buyers sourcing fabricated parts where tooling is involved, the practical starting point is a shop that can describe its tooling records and its substitution rules before anything goes wrong. The background is on the company page, the process for new work on the customisation page, and the product ranges under stainless exhaust systems and titanium components, with volume terms on the dealership page. The questions above can be asked against any of it.
