How should a make-or-buy decision be made for machined parts when the in-house shop is overloaded?

How do you know the make-or-buy question has become urgent?

The question is urgent when the same machines are the constraint week after week, overtime and expediting have become routine, and on-time delivery is falling even though the shop is busy. At that point every new job you keep inside delays some other job you already promised.

Look for these signs before anyone calls it a capacity crisis:

• A stable bottleneck. Work in process piles up in front of the same machine group, often the five-axis machining centers, a twin-spindle lathe cell or the one machine that holds the tightest bores. A shop whose queue moves around from week to week has a scheduling problem; one whose queue stays put has a capacity problem.
• Queue time growing faster than load. Queueing theory explains why a nearly full machine feels much worse than a moderately loaded one: Kingman’s heavy-traffic result for a single-server queue shows expected waiting time growing roughly in proportion to 1/(1 − utilization) as the load approaches capacity, so the last increments of load add far more waiting than the first (Kingman, 1961).
• Premium cost as a habit. Weekend shifts, rush freight on raw material and repeated schedule breaks for “hot” orders.
• Engineering work squeezed out. Prototypes, fixtures and process improvements wait behind production, which is usually the first sign that the shop is spending its scarcest hours on the wrong parts.

If those signs are present, the decision is not whether to outsource, but which part families to move so the constraint is freed for the work that earns the most from it.

How do you compare the real cost of making a part with buying it?

Compare the costs that change: on the make side, material, direct labor, variable overhead, any fixed cost you would really shed, and the value of the constrained capacity the part uses; on the buy side, the quoted price plus every cost of getting conforming parts to your line.

Build the comparison for one part number at its expected annual volume, then repeat it for the family.

• Make, relevant cost. Material at replacement price, direct labor at the rate you would actually pay for those hours, tooling and inserts consumed, and variable overhead such as power and coolant. Add a fixed cost only if it disappears when the part leaves, such as a dedicated fixture lease or a machine you would sell.
• Make, capacity charge. If the part runs on the bottleneck, add the contribution margin of the work that is displaced, or the overtime premium you pay to avoid displacing it. In the United States, the Fair Labor Standards Act requires covered, non-exempt employees to be paid at least one and one-half times their regular rate for hours worked over 40 in a workweek (US Department of Labor), so capacity bought with overtime costs more than the standard labor rate.
• Buy, landed price. Unit price, tooling or setup charges spread over volume, packaging, freight, duty and fees. The landed cost guide walks through the import lines.
• Buy, one-time transfer cost. Supplier search and audit, drawing cleanup, first article inspection, and the safety stock built while the new source is proven.
• Buy, recurring management cost. Receiving inspection, supplier communication, change control and the cost of poor quality if lots are rejected.

The capacity charge is simple arithmetic once the routing is known: bottleneck hours per piece multiplied by the contribution margin per bottleneck hour of the work it displaces. A bracket that spends a large share of an hour on the five-axis cell can carry a capacity charge larger than the whole gap between its make cost and a supplier’s landed price, even though it looks cheap at the standard shop rate. Run the same check in reverse for a part that uses only idle machines: its capacity charge is zero, and buying it frees nothing on the constraint.

Then ask the qualitative questions the same accounting text lists: whether the supplier’s parts will meet your quality requirements, whether the supplier will keep the quoted price, whether it can keep producing the quantity you need, whether the move displaces your workers or hurts morale, and whether it affects your reputation. A part whose numbers are close should be decided on those questions, not on the last decimal of the spreadsheet.

What should the drawing and RFQ say when a part moves to a supplier?

Everything your own shop knew without being told: the revision, every critical feature and how it is measured, material and certification requirements, the finish, the packaging, and the volumes you expect. A supplier can only build what the documents say.

Use this checklist before the RFQ goes out:

• Released drawing and model. Current revision, with the 3D model and the drawing stating which governs if they disagree.
• Critical characteristics marked. The features your assembly depends on, with the datum scheme and the inspection method you will use at receiving. The first article inspection guide covers the report you should ask for.
• Unwritten shop knowledge written down. Deburr expectations, edge breaks, cosmetic surfaces, thread gauging, and anything the internal team “just knows.”
• Material and records. Alloy and temper or condition, material certificates, and any restriction on substitutes.
• Commercial basis. Annual volume, release quantities, any tooling or fixtures you own and will transfer, and the date you need first articles.
• Exit and fallback. Keep the program, fixtures and setup notes for the parts you move, so the work can come back inside if the peak ends or the supplier fails.

Put the transfer through formal change control as well: the engineering change notice guide explains how to keep a supplier on the same revision as your own floor.