Reliability, Verification & Maintainability

Design from failure backward

A machine is reliable when its likely failures have been found and designed out before it’s built. The tool for that is an FMEA (failure mode and effects analysis): list how each part could fail, what happens, and score three things from 1 to 10:

• Severity (S): how bad the effect is.
• Occurrence (O): how likely it is.
• Detection (D): how hard it is to notice before it causes harm (10 means hidden).

Their product, the risk priority number (RPN = S × O × D), ranks what to fix first. For the treadle lathe:

The scores are judgment, which is fine: the value is in being forced to think about each part. Notice what the earlier lessons already did: lesson 2 found that bearing fatigue isn’t the risk and dust is ¹; lesson 3 found that fatigue from knots and grain slope, not static strength, governs the treadle ²; lesson 4 found the seasonal movement that loosens joints. The FMEA collects those into one list.

Verify with numbers

Each requirement from the first lesson gets a test you can pass or fail:

• Foot stroke: measure the foot pad’s travel. Target about 9 inches.
• Flywheel: bring it to 100 rpm, stop treadling, and time the coast-down. Then make a heavy cut and count rpm before and during it. Target: no stall at the dead points, speed dip under about 10 percent.
• Spindle speed: count spindle revolutions in 15 seconds with a chalk mark, on each step of the pulley. Targets about 730, 550, and 440 rpm.
• Runout: a dial indicator on the faceplate face and on the spindle nose while turning by hand. Target: under a few hundredths of a millimeter at the nose.
• Center alignment: bring the centers together and check height with a straightedge, or turn a test cylinder and measure both ends. Target 0.5 mm. Repeat in winter and in summer, because lesson 4’s seasonal term only shows up then.
• Bearings: turn the spindle by hand. It should spin smoothly with no roughness or play.

MIT’s machine design course treats this as the second half of design: characterize the device you built against the requirements you set ³.

Availability and repair

How much of the time the lathe is ready to use depends on how often it fails and how long a fix takes. With mean time between failures (MTBF) and mean time to repair (MTTR), availability A = MTBF ÷ (MTBF + MTTR). A machine that fails every 500 hours of use and takes 5 hours to fix is available 500 ÷ 505 ≈ 99.0 percent of the time; cut the repair to 1 hour and it’s 99.8 percent, without changing how often it fails.

For the treadle lathe, that means: keep a spare pitman pin and bushing, a spare belt, and a pair of bearings on the shelf; make the pitman pin removable without taking the treadle off; and keep the shim stack for the tailstock labeled. A worn part that takes ten minutes to swap barely counts as a failure.

Make maintenance a design input

Write the maintenance down and build it into the machine:

• Each session: oil the pitman pin; listen for knocks; check belt tension.
• Monthly: spin the spindle by hand for roughness; check the crank pin and its lock nut.
• Each fall, when the heat goes on: tap the wedges and drawpins tight; recheck center alignment.

Then design so those jobs are easy: an oil cup on the pitman bushing, a wrench-sized flat on every nut, wedges you can reach with a mallet. Maintenance that’s hard to do doesn’t get done.

What to take from this

Find the failures before they happen: an FMEA scores severity, occurrence, and detection, and the top RPNs here (bearing dust, loosening wedges, a loose crank pin) each get a design answer. Verify every requirement with a number, and check center alignment in both seasons. Availability is MTBF ÷ (MTBF + MTTR), so spares and quick-change parts count as much as durability. Build the maintenance checklist into the design.