Lights Out Machining: What It Takes and Which Parts Suit It
A machine that runs from 6 pm to 6 am without an operator produces twice the parts of one that does not, at roughly the same depreciation. That is why lights out machining comes up in every capacity review, and why so many first attempts fail. The constraint is rarely the machine; it is the twenty small things that must be true before you can walk away.
This is for the buyer told a supplier runs unattended, who wants to know whether the claim is real and what it should do to the price.
What unattended running actually buys
Take a 3-axis vertical machining centre with a 12-station pallet pool. Attended, in a two-shift operation, it might produce 14 productive hours a day. Unattended, with six hours of lights-out running at 70% of attended efficiency, the same machine produces roughly 18 productive hours. That is a 29% capacity increase from an asset you already own.
The cost side matters more. A fully loaded operator costs $26-34 an hour, and one operator can tend three machines. An unattended hour costs machine time, power and tooling only — typically $14-20 an hour for a mid-size VMC, against $45-60 attended once labor is spread across the parts.
| Capability | Indicative cost | What it unlocks | Realistic payback |
|---|---|---|---|
| Pallet pool, 8-12 stations | $38,000-70,000 | Work available without an operator present | 18-30 months at two-shift loading |
| Tool life monitoring with breakage detection | $3,500-6,000 | Ability to walk away without a broken tool ruining a batch | 4-9 months |
| Spindle probing with automatic offset | $8,000-13,000 | Automatic work offset and first-part verification | 6-12 months on family parts |
| High-pressure coolant and chip conveyor | $9,000-18,000 | Chip control in deep pockets and aluminium | 8-14 months |
| Bar feeder (turning) | $18,000-32,000 | Continuous unattended stock feed | 12-24 months |
| Fire detection and suppression | $2,500-7,000 | Insurance compliance for unattended operation | Not a payback item — a precondition |
| Automated part and chip removal | $6,000-20,000 | No operator intervention per part | 10-20 months |
Note the fire suppression row. Insurers increasingly ask what happens if a coolant line fails at 2 am.
The four systems that have to be in place
Tool life management. Every critical tool needs a defined life in metres or minutes, a counter in the control, and a strategy for what happens at expiry: a sister tool takes over, or the machine stops. Without this, a 6 mm end mill that should last 40 parts will break at part 34, and you will find 34 scrapped or suspect parts on the pallet in the morning.
Chip control. Chips are the main cause of unattended failure, and the failure mode is rarely a crash — it is a recut, a bad surface or a blocked conveyor that floods the enclosure. Deep pockets in aluminium need high-pressure coolant, and pocket depth should be checked at quoting. A job that machines well with an operator clearing chips every twelve minutes can be unsuitable overnight.
Probing and in-process verification. A probe that measures the first part on each pallet, compares to nominal, and adjusts the work offset converts thermal drift from a scrap event into a correction. On aluminium parts with long cycles, the machine can grow 20-30 µm over six hours as the spindle and ballscrews warm. If your tolerance is ±0.01 mm, that drift is the whole tolerance band.
Workholding and part handling. Hydraulic or pneumatic fixtures with pressure confirmation, pallets that locate repeatably to within 5 µm, and a pallet changer that does not jam. The unglamorous failure is a part not seated correctly that then gets machined 0.3 mm off.
See also: Battery Housing Machining: Gasket Grooves, Flatness and Leak Test Prep and Sending CAD Data to a Machine Shop: What to Control.
Which part types actually suit it
Lights out machining pays on parts that are long-cycled, geometrically forgiving, and low in tool count.
Good candidates:
- Aluminium enclosures and brackets with 4-10 minute cycles and 8-14 tools.
- Turned pins, spacers and bushings running from bar stock with 60-180 second cycles.
- Parts where the tolerance is ±0.02 mm or looser and the finish is defined by the tool, not by hand work.
Poor candidates:
- Thin walls under 1 mm, where vibration and chip load vary unpredictably.
- Deep pockets with poor chip evacuation that need an operator with an air gun.
- High-temperature alloys where tool life is short and unpredictable, so the sister-tool logic runs out.
- Cosmetic parts requiring in-process inspection or hand blending.
- Very short cycles under 60 seconds, where pallet changes and probing dominate the time.
A rough screen: if the part needs more than 15 tools, more than one operation with a manual intervention, or a tolerance tighter than ±0.01 mm on a long dimension, assume it will not run unattended until proven otherwise.
Why it changes the quote on volume parts
Unattended hours are cheaper hours, and a shop that has genuinely solved them can quote 8-15% below a competitor on the right part family without cutting margin. The saving is real and it comes from three places: labour spread over more spindle hours, fewer setups because pallets carry multiple operations, and better machine utilisation on capital already spent.
The saving does not appear on parts that do not suit the method. This is where buyers get burned. A supplier quoting aggressively on a thin-walled housing because "we run lights out" will either run it attended anyway and lose money, or run it unattended and ship you a batch with chatter on three parts in twenty. Ask which part families actually run overnight, and ask how many hours a week, not whether the capability exists.
There is a second commercial effect. Unattended capacity is inflexible capacity: the machine runs what is loaded at 6 pm. A supplier optimising for lights out wants predictable releases and fewer changeovers, which suits a blanket order with monthly call-offs. It does not suit a purchase order arriving Tuesday for 40 pieces on Friday, which is always an attended job at an attended price.
The failure modes nobody budgets for
Broken tools are the obvious one. The expensive ones are quieter.
Thermal drift on long cycles produces a slow taper that passes a sampling inspection and fails at assembly. Coolant concentration drifting over a week changes chip evacuation and finish. Chip accumulation stops conveyor flow at 3 am and floods the enclosure. Fixture seal failure loses clamping pressure and the part moves.
The mitigation is boring: a morning check sheet recording spindle load, coolant concentration, probe results and first-part measurement, plus a rule that any program change gets a dry run on the first pallet. Shops that run unattended well have a night-shift procedure that fits on one page.
For how we sequence production and hand-offs on volume work, see how we work. Our standard tolerance is ±0.01 mm across milling, turning and finishing, which is what makes the probing step in an unattended cell worth the capital — the drift matters at that level.
Frequently asked questions
How many hours a week does a shop need to run lights out before it makes sense?
Below about 30 unattended hours a week, the capital rarely pays back inside two years unless the parts are high-margin. The sweet spot for most job shops is 40-50 hours a week.
Does unattended running reduce part quality?
Not inherently, but it removes the operator who would have caught a problem. The outcome depends on whether probing and tool monitoring are in place and whether the morning check is performed. A pallet pool without in-process verification is a risk that may not be priced.
Can lights out machining hold ±0.01 mm?
It can, with probing on the first part of each pallet and a stable thermal environment. Without those, day-shift accuracy is usually better than night-shift accuracy because someone is watching. If the drawing is tight on a long dimension, ask specifically how drift is compensated, not whether the shop can hold the tolerance.
Why do some suppliers charge more for unattended parts?
Because they are amortising engineering: fixture design, prove-out and tool life trials. That cost is one-off per part number, so the premium should shrink with volume and show up as a setup charge.
Related reading
Send a drawing for a quote and DFM review.
STEP, IGES, X_T, PDF or DWG up to 50 MB. Tell us quantity and finish, and the quote comes back with tolerance and process notes, not just a price.