What you're seeing: R(t) = exp(−(t/η)^β). β under 1 means infant mortality, ≈1 is random, above 1 is wear-out. The cyan dot marks B10, the time by which 10% are predicted to fail. The white dot marks mission reliability. This two-parameter model does not display confidence bounds or fit uncertainty.

What you're seeing: a MIL-STD-810-style acceleration PSD. Grms is the square root of the area under the acceleration PSD curve; Miles' equation estimates how hard a resonant part rings: Grms,resp = √(½π · fₙ · Q · ASD(fₙ)). The animation synthesizes a repeatable random-phase signal across 20–2,000 Hz and applies a linear base-excited single-mode response with damping ζ = 1/(2Q). Table and article share one displacement magnification; playback is slowed by 125×. The flexible mounts represent the component’s effective compliance, not a fixture design. This is an illustrative profile, not a prescribed qualification level. The response estimate assumes a linear, lightly damped, single-mode system and locally broadband excitation; it does not predict pass/fail.

What you're seeing: the short-form torque equation T = K · F · d. The diagram shows the clamp force F squeezing two plates together; the bars show how the nut factor K (friction) changes the torque needed for that same clamp force. These nominal nut factors illustrate sensitivity to friction; they are not calibrated installation values. Joint stiffness, external loading, scatter, and stripping limits are not modeled. Note: 304 (A2-70) and 316 (A4-70) stainless share the same strength (≈450 MPa proof). 316 offers better corrosion resistance. A threadlocker like Loctite 242 adds a prevailing (locking) torque on removal that this short form doesn't capture.

What you're seeing: Little's Law: WIP = Throughput × Lead time. Below the critical WIP the line is starved, so adding WIP buys more output; above it the bottleneck is saturated, so extra WIP only stretches lead time without adding a single unit of output. That gap is the quantitative case for running lean. (Ideal "best-case" line; real lines sit a bit below these curves.)