I spent a fair chunk of last week performing emergency surgery on a laser welder whose nozzle accidentally hit a clamp. A 15 amp, 240 volt stepper motor can apply one hell of a lot of force with incredible speed; the nozzle didn't stand a chance. The easily replaceable nozzle, though, wasn't the thing that gave way; other mechanical components higher up in the assembly took the brunt of the impact.
This got me thinking about designed-in weak links. Where expensive equipment is likely to be in jeopardy, it's good practice to consider the likely chain of failure points and design the strength of each one accordingly. Some examples of this philosophy include:
- Snowblowers have a shear bolt on the auger, so that if it snags on something, the $2 bolt snaps off before the shafts or gears can be damaged.
- Outboard and sterndrive engines have a rubber-splined shock absorbing hub that, in most cases, will slip or tear at lower loads than would be required to shear off a driveshaft or gear tooth. If the hub can't absorb the entire load, the propeller blades are designed to break off, rather than transmitting damaging forces into the more expensive internal parts.
- Cars are designed to fail progressively in a collision. The bumper is sacrificed first; if there's still impact energy to dissipate after it has failed, then the bumper mounts, fenders, front subframe and engine mounting braces absorb the impact in turn, thus preventing failure of the passenger compartment cage.
There is often a temptation to include a designed point of failure wherever it's possible to do so. It is not, however, necessarily appropriate to include a designed-in weak link in every structure. Some examples might include:
- A sailboat's rig is ordinarily designed so that all vital components are of similar strength (relative to their expected loads). Strengthening some parts more than others would add weight without increasing the overall strength of the assembly.
- There is no good reason to make the struts holding a Space Shuttle Orbiter to its fuel tank any stronger or weaker than their attachment brackets on the tank; the failure of any component of this system would result in the loss of the vehicle.
The pattern should, by now, be clear.
If we can reasonably expect that a system will be damaged in ordinary use, and we can make that damage survivable by controlling the order in which things fail, then we should definitely do so. That laser nozzle should have sheared off on impact, just as an outboard engine's propeller blades should break before its driveshaft, or how a milling machine's cutter is supposed to snap off in a collision before its bearings or drives are damaged. A sailboat's rudder should always be designed to fail in a particular sequence when it hits something: first the edge and tip get shredded up, then a chunk of the blade breaks away, then more of the blade gets banged up, and then – only after the blade has been completely destroyed – is it OK for the shaft to bend or crack.
If, on the other hand, the failure of any component of the system will necessarily result in the failure of the entire system, there is little point in designing a particular part to fail first. This is the case with the stays, shrouds and turnbuckles of a Bermudan-rigged sailboat (although I would argue that the chainplates should be substantially stronger than the standing rig, as they should survive a dismasting so that the boat can be repaired and fitted with a new mast after salvage.)