Process & DFM
Which Manufacturing Process at What Volume: Running the Crossover
The short answer
Choosing between 3D printing, CNC machining, sheet metal and injection molding is a crossover calculation. Each process charges you a fixed cost to start and a variable cost per part. The right process at your quantity is the one where fixed cost divided by the per-part saving is smaller than the number of parts you plan to buy. Rules of thumb about which process suits which volume are shorthand for that arithmetic, and they break whenever your part is unusually large, unusually simple or made from an expensive material.
The only question that matters: where do the lines cross?
Process comparison charts tell you that 3D printing suits 1 to 100 parts, machining suits 1 to 500, and injection molding starts making sense in the thousands. Those numbers are averages of other people's parts, and your part is not average.
Underneath the chart is one piece of arithmetic. Every process charges you twice: a fixed cost to get started, and a variable cost for each part that follows. 3D printing has almost no fixed cost and a per-part cost that barely falls with quantity. Machining has a moderate fixed cost in programming, fixturing and first-article setup, then a per-part cost that drops as the setup spreads out. Injection molding has a large fixed cost in tooling and a per-part cost so low it is often a rounding error next to the tool.
The crossover quantity between two processes is the extra fixed cost divided by the per-part saving:
Crossover = (Fixed cost B - Fixed cost A) / (Unit cost A - Unit cost B)
If you expect to buy more parts than the crossover, the higher-fixed-cost process wins. If you expect fewer, it loses. Everything else in this article is about the cases where that arithmetic gives you the wrong answer.
What each process is really charging you for
The fixed and variable split is not arbitrary. It follows from what the process physically has to do before it can make your second part.
| Process | Fixed cost is | Per-part cost is driven by | Falls with quantity? | Punishes |
|---|---|---|---|---|
| 3D printing | Almost nothing. File preparation and a build setup. | Material volume and machine time on the build plate. | Barely. Nesting more parts per build is the only real lever. | Bulk. Solid, chunky parts are expensive because you pay for volume. |
| CNC machining | Programming, fixturing, first-article inspection. | Machine time, which is roughly how much material is removed and how many setups it takes. | Yes, steadily. Setup spreads across the run, and the program is written once. | Complexity and tight tolerance. Every extra setup is a step change, not a gradual increase. |
| Sheet metal fabrication | Programming, and tooling for any non-standard bend or form. | Cut length, number of bends, hardware insertion, finishing. | Yes, sharply at first. Nesting improves and bend setups amortize. | Bends more than cuts. A flat part with 40 holes is cheap; the same part with 12 bends is not. |
| Injection molding | The tool. This dominates everything else early on. | Cycle time and material, both very low per part. | Dramatically. The tool cost divided across parts is the whole curve. | Design change. Every change after the tool is cut is paid for in steel. |
The pattern: as you move down the table, you pay more before the first part and less for every part after it.
A worked crossover
Take a small aluminum enclosure lid, 120 by 80 mm, 3 mm thick, with a bent flange, six holes and a slot. It can be machined from plate or cut and bent from sheet. Suppose your quotes come back like this:
Machined from 6061 plate: no tooling, $38 per part at 25 off, $29 per part at 250 off. Laser cut and press braked from 5052 sheet: $220 in programming and setup, $11 per part at 25 off, $7 per part at 250 off.
At 25 parts, machining costs $950 and sheet metal costs $495. Sheet metal has already won, because the $220 setup is smaller than the per-part gap across 25 parts.
Run the formula at the 250-off prices instead: $220 divided by ($29 - $7) is 10 parts. Above ten parts, sheet metal is cheaper on this design, and by 250 parts it is not close: $1,970 against $7,250.
Now change one thing. Make the lid 20 mm thick with a machined pocket on the underside, and sheet metal is no longer a candidate at any quantity. The crossover only exists between processes that can both make the part.
Four things that move the crossover, in the order they matter
Design stability. This is the one that decides most projects, and it is not on any volume chart. Tooling turns your design into steel. If the design is still moving, the crossover is irrelevant because you will pay for the tool twice. Cut tooling after the design is frozen, which in practice means at DVT rather than EVT.
Material cost share. When material is a large part of the unit price, the whole curve flattens. A part machined from titanium or carbon fiber plate is dominated by the blank, so buying 500 instead of 50 saves far less than it would in aluminum, and the crossover into tooling arrives later than you would expect.
Part size. Large parts break the usual ordering. A one-meter panel is cheap to route and expensive to print, because printing charges you for volume while routing charges you for the perimeter you cut. This is why the standard chart misleads on large flat parts, and why routing is priced on path length rather than volume.
Secondary operations. Anodizing, powder coating, tapping, inserts and inspection attach to the part, not to the process, and they often have their own minimum charges. A $7 sheet metal part with a $9 finishing minimum is a $16 part. Compare finished parts, not raw ones.
When volume is the wrong axis entirely
Three cases where the crossover calculation should be ignored:
The geometry only one process can make. Internal channels that no cutter can reach, undercuts that no mold can release, a lattice that only powder bed fusion can produce. Decide feasibility first, economics second.
The tolerance sets the floor. If a bore needs to hold a few hundredths of a millimeter, printing and molding are out regardless of quantity, or they become a first operation followed by machining anyway. Tolerance narrows the candidate list before volume touches it.
Time is the binding constraint. A tool that pays for itself at 400 parts is worthless if it arrives after your launch. Bridge production exists for exactly this: machine or print the first few hundred units while the tool is being cut, then switch. You pay more per part for a while in exchange for not moving the date.
Getting a real answer in about a week
The crossover is only as good as the numbers you put in it, and the numbers come from quotes, not from charts. A practical way to get them:
Quote the same part two ways. Send the model to be quoted as machined and as fabricated, or as machined and as molded, and say plainly that you are comparing processes. A shop that knows why you are asking will give you a more useful answer.
Ask for quantity breaks, not one price. Ask for your realistic quantity, half of it, and four times it. Three points show you the shape of the curve, which is what you are buying the information for.
Ask what the tooling buys you. For molding, ask how many shots the tool is rated for, whether it is aluminum or steel, and what a change to one feature would cost after it is cut. That last number is the real risk in the decision.
Compare finished and delivered. Include finishing, inspection and freight. A process that wins on the part can lose on the shipment, which is the landed cost argument.