Power Cost

Managing Power Cost Per Run: An Energy Accountability Playbook

A measured kW figure and one multiplication give you power cost per run and cents per part. How to build, use, and defend those numbers.

Power cost per run is the difference between guessing your margins and knowing them. Electricity runs 2 to 10 percent of cost of goods sold in most discrete manufacturing, and up to 30 percent in heat treating, melting, and forming operations. A plant paying 40,000 dollars a month for electricity with no allocation below the meter is quoting every job with a smeared average, undercharging the energy hogs and overcharging the light work. Getting to a defensible dollars per run and cents per part figure takes one measurement and one multiplication, and it changes quoting, scheduling, and capital decisions from that day forward.

The math is kilowatts times hours times rate. A CNC cell drawing a measured average of 75 kW through a 6 hour run at 0.12 dollars per kWh costs 75 times 6 times 0.12, or 54 dollars per run. If the run yields 900 good parts, energy is 6 cents per part. The Power Cost calculator does this per run, per shift, or per part once you feed it real numbers. The trap is the input: use a measured average draw from a clamp-on meter or the machine's own energy monitor, logged across a representative run, because nameplate ratings and spot readings will both lie to you.

Nameplate versus actual is the first correction to make. Most machines draw 30 to 60 percent of nameplate during real production because nameplate reflects worst case simultaneous loads. Meanwhile idle draw is bigger than anyone expects: a machining center can pull 20 to 40 percent of its loaded power while sitting in cycle-stop with hydraulics, chillers, and controls live. A machine idling 10 hours a week at 15 kW burns 780 kWh a month, about 94 dollars at 0.12 per kWh, times however many machines you own. Log one week of data per major asset and you will find the idle waste within days.

Understand your tariff, because the rate is not one number. Demand charges of 10 to 20 dollars per kW per month are set by your single highest 15 or 30 minute peak, and they commonly run 30 to 50 percent of an industrial bill. A plant that hits a 500 kW peak at 15 dollars per kW pays 7,500 dollars that month for the peak alone. Staggering the startup of large loads by 15 minutes, or shifting one energy intensive process off the peak window, can cut demand 10 to 20 percent without touching production volume. Time-of-use rates add another lever, with off-peak power often 2 to 3 times cheaper than on-peak.

The operating levers rank by payback. Idle shutdown discipline and programmed sleep states are free and typically recover 5 to 10 percent of machine energy. Scheduling energy intensive runs off-peak captures the tariff spread with zero capital. Compressed air leak repair, covered by its own playbook, usually returns its cost in under six months. Variable speed drives on pumps and fans return 20 to 50 percent of that motor's energy where flow varies. Power factor correction below 0.95 stops utility penalties that can add 1 to 3 percent to the whole bill. Work the free items first; most plants find 8 to 12 percent before spending anything.

The failure modes are accounting failures. Allocating electricity by square footage or headcount tells you nothing and hides the three machines using 40 percent of the power. Costing energy at the average blended rate ignores demand charges, so off-peak shifting looks worthless on paper when it is not. Using nameplate data overstates machine energy 2 to 3 times and makes every efficiency project look better than it is. And measuring once, building a spreadsheet, and never re-logging means your per-part costs quietly rot as tooling, programs, and rates change. Treat energy data like quality data: it expires.

Deploy on a cadence a small plant can actually sustain. Daily, nothing; do not build a daily energy ritual, it will die. Weekly, review kWh per unit produced for the top five loads, submetered or logged, and flag any 10 percent drift. Monthly, reconcile the utility bill: verify the peak demand event, identify what caused it, and check the power factor line. Quarterly, re-log one major machine per month on rotation so every big asset gets fresh data annually, and refresh the cents per part figures in your quote model. Annually, review the tariff itself; plants change and the rate schedule that fit five years ago often no longer does.

World class energy management in discrete manufacturing looks like this: the top ten loads submetered, energy at cents per part known and quoted for every product family, kWh per unit trending down 3 to 5 percent per year, and demand peaks managed deliberately rather than discovered on the bill. Plants running this discipline typically carry 10 to 20 percent lower energy cost per unit than peers within three years, with most of the gap coming from idle elimination and peak management rather than equipment purchases. The number that proves maturity is simple: when someone asks what a run costs in power, the answer takes five seconds.

Published 2026-07-02.