Calculations

How to Calculate Valve Manufacturing Metrics: Casting Yield, Test Time, and Actuator Torque

The core formulas a valve shop actually runs, from casting yield to actuator torque, worked through with real units and numbers.

A valve line runs on five recurring calculations: casting yield, seat leakage test time, pressure test throughput, actuator torque sizing, and the pour quantity needed to ship a complete order. Each one has a clean formula, but the inputs come from specific places on the floor, and mixing up a unit (minutes versus hours, poured versus machined count) is where numbers go wrong. This guide works each formula with concrete inputs. It teaches only the math; pricing and target ranges are covered in the sibling cost and benchmark guides. Keep a calculator like Valve Body Casting Yield open to check your arithmetic as you go.

Start with casting yield, because it sets how much of everything downstream you actually pay for. Yield = accepted castings / total poured x 100. Pull the accepted count from final inspection acceptance (visual, dimensional, and NDE all passed), not the count that merely reached machining, and pull total poured from the mold count including immediate scrap. With 87 accepted from 100 poured, yield is 87.0 percent. The trap: if you count only the 96 bodies that survived to the machine cell as your denominator, you compute 87/96 = 90.6 percent and hide nine early-scrap bodies that still burned melt and mold.

Yield drives the pour quantity, which is a separate step people forget. To ship a complete order you must pour order quantity / yield, not the order quantity itself. At 87 percent yield an order for 100 good bodies needs 100 / 0.87 = 114.9, so you release 115 to the pattern. Budget at 100 and you are short 13 bodies and chasing a late re-pour that blows the ship date. Round up, never down, and re-run the division whenever yield shifts more than 2 or 3 points, since a drop from 87 to 82 percent moves the release from 115 to 122.

Seat leakage test time converts a queue into bench hours. Base hours = valves x cycle time per valve, converted to hours, then Total = base x (1 + allowance/100). The unit conversion is the whole game: cycle time is in minutes per valve, so divide by 60. For 40 valves at 15 minutes each on a single serial bench, that is 40 x 15 = 600 minutes = 10.0 base hours, and with a 20 percent setup, documentation, and retest allowance, 10.0 x 1.20 = 12.0 hours. Use measured bench timestamps for the cycle time, covering fixturing, fill, the API 598 or ISO 5208 dwell, leakage measurement, and bleed, not the nameplate dwell alone. Seat Leakage Test Time runs this directly.

Pressure test throughput answers the opposite question: how many valves clear a bench per shift. Gross = valves per cycle x cycles per shift, then Effective = gross x availability x first-pass yield, both as fractions. Take a two-station bench at 24 cycles: gross is 2 x 24 = 48 valves. Apply 85 percent availability (logged uptime after fill, bleed, fixturing, and maintenance) and 95 percent first-pass yield: 48 x 0.85 x 0.95 = 38.76 effective valves. The two losses are worth splitting out, 48 x (1 - 0.85) = 7.2 valves lost to downtime and 48 x 0.85 x (1 - 0.95) = 2.04 to first-test failures. Pressure Test Throughput reports all three.

Actuator torque sizing is the one dimensional calculation. Required actuator output = valve operating torque x safety factor, and you size to the worst-case point on the stroke. For a quarter-turn valve the governing values are usually breakaway (break-to-open) and seating torque; take the larger. If measured breakaway is 340 newton-meters and you apply a 1.5 safety factor, required output is 340 x 1.5 = 510 newton-meters, and the chosen actuator must deliver at least that at the minimum available supply pressure, say 4.5 bar, not its catalog rating at 6 bar. Derating the actuator to real air pressure is the step that prevents a unit that cannot seat. Actuator Sizing Workload scopes the engineering hours for a batch of these.

Two habits keep all five clean. First, lock your unit before you multiply: minutes to hours needs a divide by 60, percentages become fractions (85 becomes 0.85) before they chain, and torque stays in one system, newton-meters or pound-feet, never both. Second, never average across incomparable items in a single number; a 24-inch metal-seated gate valve dwells far longer than a 2-inch soft-seated ball valve, so a blended 15-minute cycle time understates the batch whenever large bore is present. When sizes are mixed, split into classes and run each separately, then sum. Do that and the arithmetic above holds against your actual floor logs.

Published 2026-07-01.