The bench has no machine to pay for

A shop that machines parts and then bolts them together has two rates to work out, and most shops only ever work out one. The spindle gets a rate calculated to the penny. The bench gets whatever number was handy — usually the machine rate, because it is the only rate anybody ever wrote down.

Work out your own shop rate → Free, no account, five minutes.

That is wrong in both directions, and which direction depends on the shop. Some apply the machining center rate to a guy with a torque wrench and a bench block, quote assembly at ninety-seven dollars an hour, and lose the job to somebody who priced it honestly. Others decide assembly is not really machine work, leave it off the quote entirely, and hand over twelve or thirteen hours of burdened labor for free on every lot. The first shop loses work it should have won. The second wins work it should have charged for. Neither one can tell you which it is doing, because neither one has a bench rate.

The bench rate is not hard to build. It is the same short arithmetic you use for a machine, with one line deleted and two lines shrunk. The line you delete is the big one.

Take a $60,000 turning center with ten years left on it. Every hour it runs it is paying for its own replacement: at 1,200 spindle hours a year it has 12,000 hours left in it, so $5.00 of every hour goes to depreciation. It draws around ten kilowatts, so at twelve cents a kilowatt-hour it burns $1.20 an hour in electricity before you count the compressor. It eats inserts, coolant, way oil and filters — call it $5.00 an hour across a mixed job. And it occupies eight hundred square feet of your floor once you count the chip cart, the tool cabinet and the material staged behind it.

Now walk twenty feet to the assembly bench. There is no machine. There is a steel bench, a vise, a set of torque wrenches, an arbor press, a couple of air tools, some go/no-go gauges and a rack of sockets. All of it together might be nine thousand dollars, and it does not wear out on a schedule the way a spindle does — you replace a torque wrench when somebody drops it, not when it hits a service-hour count. The power draw is the compressor cycling and the light over the bench. The consumables are thread locker, anti-seize, shims, wipes, gloves and the occasional fastener that was not on the BOM.

The one thing that does not change is the person. If the same machinist does both jobs he costs $30.00 an hour, plus about 40% for payroll tax, workers comp, holiday and the rest, so $42.00 an hour burdened — at the bench and at the lathe alike. That is why the two rates end up closer than people expect once you work them out: burdened labor is the largest line in both. It is also why they are not the same. Everything wrapped around that $42.00 is different.

Start with the shop, because two of the lines come out of shop-wide numbers rather than anything sitting at the bench. Take an eight-person shop with three machines and one assembly area. Fixed overhead — rent, insurance, phone and internet, software, heat and light, the half-time person in the front office — runs $9,000 a month, so $108,000 a year. That has to be carried by hours you sell, and in a small shop the honest way to split it is by floor space, because floor space is what the rent actually buys.

Three machine cells at eight hundred square feet each is 2,400 square feet. The assembly area — bench, subassembly racking, the cart the finished units sit on — is 600. That is 3,000 square feet of chargeable floor, so the machines carry 80% of overhead and the bench carries 20%. Aisles, the break room and the office go into the same pot and get shared out with everything else. They are not free, and pretending they are is how shops end up with a rate that does not cover the rent.

So the three machines carry $86,400 a year between them, $28,800 each. At 1,200 spindle hours a machine that is $24.00 an hour. The bench carries $21,600, and if the bench is genuinely occupied twenty hours a week for fifty weeks — a thousand hours a year — that is $21.60 an hour. Notice that the bench's overhead line is nearly as heavy as the machine's, and for a good reason: it takes up real square footage and it is busy fewer hours than the machines are.

The tooling line needs one decision. Nine thousand dollars of bench kit over a six-year life is $1,500 a year, and across a thousand bench hours that is $1.50 an hour. If you have built dedicated assembly fixtures for one customer's product, keep those out of the rate — they belong on that customer's job as tooling, the same way a set of soft jaws does.

$66.90 against $77.20 — a difference of $10.30 an hour, all of it capital and current. Run the same exercise against a $250,000 machining center, where depreciation alone is $20.83 an hour and the power is closer to $2.16, and the machine break-even comes to just under $97.00. Against that, the bench is about thirty dollars an hour cheaper. The bigger the iron, the wider the gap, and the more it costs you to quote bench work at the machine number.

Take a small gearbox subassembly. You turn the shafts and bore the housings in house, then somebody at the bench presses in the bearings, sets the shim stack, drops in the gears, torques the cover and turns it over by hand to check for drag. Once he is up to speed that is about half an hour a unit. The customer wants 25 of them, so 12.5 hours of assembly labor sits on the lot. Here is what those 12.5 hours cost you depending on which rate you reach for.

Quote the assembly at your machining center rate and you are $376.25 over the shop that used a bench rate. On a job worth a couple of thousand dollars that is plenty to lose it, and nobody will ever tell you that is why. Quote it at the lathe rate and you are $128.75 over: survivable, but you gave away a margin advantage you actually had, for nothing. Leave it off the quote and you handed the customer 12.5 hours of a burdened machinist — and if that part repeats four times a year, fifty hours.

Every shop that has tried to cost assembly hits the same wall. You stopwatch the first unit, it takes fifty-five minutes, and now you have a number you do not believe and nothing to replace it with.

You are right not to believe it. That fifty-five minutes contained things that will never happen again on this job: reading the print for the first time, working out which bearing goes in which end, hunting for the right circlip pliers, walking to the storeroom twice, and one press operation done backwards that had to come apart. None of that is assembly time. Some of it is bench setup and some of it is pure first-article learning, and neither repeats twenty-five times.

So split the bench the way you already split the machine. Bench setup is kitting the parts, laying them out in build order, staging the fasteners and knocking together whatever simple fixture holds the housing square. Call it 45 minutes, and it happens once per lot: across 25 units that is 1.8 minutes a unit, on a lot of two it is 22.5 minutes a unit. That is the same setup-versus-run split that decides your machining price, and it decides your assembly price the same way.

Then there is the learning, which is a different animal, because it does not vanish after the first one. It fades across the whole lot.

Hand assembly follows a learning curve reliably enough to plan around it. The shorthand is that unit time drops to about 85% of what it was every time cumulative quantity doubles: the second one takes 85% of the first, the fourth takes 85% of the second, the eighth takes 85% of the fourth. That is an 85% curve, and for bench work with hand tools it is a sound working assumption.

Run our fifty-five minute first unit through it and the shape is steep, then flat — which is exactly how it feels on the floor. By the tenth one the guy has stopped thinking about the order of operations, and everything after that is small refinement.

Add up all twenty-five unit times and the lot takes 814 minutes — 13.6 hours, an average of 32.6 minutes a unit. That is the number that belonged in the quote. Now put it next to the two numbers that usually go in instead.

The average of the first three is the seductive one, because it looks like you did the responsible thing and took a sample. You did not. You sampled the three most expensive units in the lot and averaged them, which lands you 47% over the true lot average and $428 over on the price. On a competitive job that is the entire difference between winning and not.

The error at the other end is real too. Quote the steady-state time you eventually reach and you are $187 short, because the first eight or ten units genuinely did take longer and somebody paid for that. Neither the first unit nor the last unit is the right number. The lot average is, and on short lots the lot average sits much closer to the first unit than anybody's gut says it does.

The fix is boring and it works: log actual minutes per unit on the next three assembly jobs you run, not just the lot total. The lot total tells you what happened. The per-unit series tells you where your curve flattens and hands you a first-unit multiplier you can quote the next job with — and after three jobs it stops being theory and becomes your own number. Keeping that history is what Spindlerate is for, but a clipboard taped to the bench does the same job if somebody actually fills it in.

In an eight-person shop the guy at the bench came off the lathe an hour ago and goes back on it after lunch. That raises two questions people get tangled in.

The first is whether he should be costed differently at the bench. No. He costs $42.00 burdened wherever he stands. It is the machine, the power and the floor that change, not the man. Do not invent a lower assembly wage for the same employee — you will build a rate that looks great on paper and does not match your payroll.

The second question is the one that matters. What does that bench hour really cost when the lathe is booked solid? If the lathe sells at $110 and breaks even at $77.20, every spindle hour carries $32.80 of contribution. Move the operator to the bench for an hour while there is a queue behind that lathe and you have paid $66.90 for the assembly hour and given up $32.80 you would otherwise have earned. The true cost of that hour is $99.70 — higher than the machine rate you were trying not to charge.

That is not an argument for quoting assembly at $99.70. It is an argument about scheduling. If a machine is your bottleneck, assembly belongs in the hours that machine is not the constraint — after the spindle stops, during a long unattended run, or on the morning you are waiting on material. The shops that seem to get more out of the same three machines are usually the ones doing this without calling it anything.

One more thing to watch on mixed days: do not bill the same hour twice. If your job tracking has the operator clocked to a machining operation while he is standing at the bench, you are charging a machine rate for bench work and your machine utilization figure is lying to you as well — which means the spindle-hour divisor in your machine rate is wrong too, and the error spreads. One hour, one work center, one rate.

None of this makes assembly profitable on its own. What it does is stop assembly from being the invisible line that quietly eats the margin you earned at the spindle — and stop you from losing jobs you had priced correctly everywhere except the last half hour.