Why Aerospace Schedules Slip
Standard job shop scheduling assumes the machine is the bottleneck. In aerospace, the constraint is often elsewhere: a single CMM with a backlog of FAIs, an inspector approving first piece before the lot can run, a plater who batches parts weekly, or a customer source inspector who visits Tuesdays. Add expedites from primes, rework loops after nonconformance, and engineering changes that restart a job.
Each hold point is small but they stack. A job with 12 operations including three outside processes can spend 70 percent or more of its calendar time waiting rather than being worked. Plan lead times from actual history, not machine hours alone.
What to Put on the Router
Model the constraints explicitly so the scheduler sees them.
- Outside operations with vendor-specific turnaround days and pickup or shipping lags
- First piece inspection and FAI as operations with inspector or CMM resource and a hold before the remaining lot proceeds
- Source inspection or customer witness points as dated milestones
- Material receipt and cert review before the first operation
- Final inspection, certificate of conformance and packaging as scheduled operations
- Queue and move time between operations based on measured averages
Finite Capacity, Load by Work Center and Dispatch Lists
Infinite-loaded MRP will tell you everything can ship on time. Finite scheduling checks load on each work center and CMM and shows the real due date. Shops on SyteLine, Epicor Kinetic, Global Shop Solutions, JobBOSS or E2 can use built-in scheduling; others add a scheduler.
Keep it simple at first: load the five critical resources, publish a daily dispatch list per work center, and review a late-job report each morning. The common failure is garbage in. If setup and run times in the ERP are quoted numbers never updated from actuals, the schedule will be wrong however good the algorithm.
Feed actuals back from shop floor reporting before investing in optimization.
Try it here, free
Outside Processing Purchase Order Template for Job Shops
A purchase order template for sending parts out for heat treat, plating, anodize, grinding or paint, with process spec, certs and return instructions.
Where AI Helps and Where It Does Not
AI is useful for three things: predicting actual lead time for a job from similar past jobs including outside process delays, flagging jobs likely to miss due dates early, and answering planner questions about the shop against ERP data in plain language. It is not a replacement for a planner who knows that the customer's inspector is on vacation or a press is down. Optimization engines can also improve sequencing to reduce changeovers, but only after data quality is fixed.
A private model running over your ERP keeps job and customer data inside your network, which matters for ITAR-restricted programs.
Quote Lead Times You Can Keep
The same data should feed quoting. If the quote promises 4 weeks and your history for similar parts with anodize and FPI is 7 weeks, you will be late by design. Compute quoted lead time from historical elapsed time by operation mix, then add the current backlog on constrained resources.
Giving a buyer an honest date and holding to it is a competitive edge with primes who score delivery.
Frequently asked questions
How do you schedule outside processing in a job shop?
Treat it as an operation with a vendor-specific lead time and transit lag, and update that lead time from actual receipt dates. Group shipments where it does not hurt delivery, and give the scheduler visibility of the vendor's batch days so you do not miss a run.
What is the biggest bottleneck in aerospace machining?
It depends on the shop, but inspection capacity (CMM and FAI), outside special processes and material availability frequently constrain output before machine time does. Measure queue time per work center for a month to find yours.
Do I need finite-capacity scheduling software?
If your delivery performance is poor and your shop has more than a few constrained resources, yes, it can help. Under about 15 people with a visible backlog, a disciplined dispatch list and a capacity board are often enough.
Can AI predict late jobs?
It can flag jobs resembling past late ones based on routing, vendors, material and queue, giving planners early warning. Accuracy depends on how complete your history is. Treat it as a risk signal for the planner, not an automatic reschedule.
Ask Netray to analyze your ERP job history and show actual lead time by operation mix, including outside processing, so your quotes and schedule match reality.