In a very generic sense, inefficiency(ies) could mean anything from inappropriate choice(s) of material(s), to improper locations of joints making up assemblies, or nature of joints themselves in the design. An everyday example is a chair that has somewhat poor stiffening under the seat that causes it to distort perceptibly, or a rather inefficient design of the legs group for torsional loads, that often causes the chair to twist annoyingly. Millions of such examples abound, from undesirable degrees of freedom of hand grip loops in metro trains and buses to inefficient locations or inadequacies of bearing supports of a turbo- shaft. Monte- Carlo simulations could help estimate the long term impact, from an appropriate model here.

Performance issues arise after usage has been on for a while. Such issues are the most common in the world of design, often arising from an inadequate understanding of the environment the design lives in. Millions of performance issues arise everyday, such as a door squeal that wasn't there yesterday, or an air conditioner's noise that's now taking front stage in the office. Or on a much bigger note, a (scarier) unusually high structural response to routine load in a girder supporting a flyover or a metro train track. Performance issues need multiple diagnoses in the field, supported by signal- processing algorithms and of course, intelligent reasoning!

"If something can go wrong, it will! "- Murphy's Law.

Failure(s) are almost inevitably the result of either an exceedingly major flaw in the design, negligence of performance issue(s), exposing the design to an environment well outside what it was meant to live in, or combinations of these. History is full of failures across nations, advanced and developing. We've read and still read about terrifying chopper and fixed- wing aircraft crashes, bridge crashes and such. It's hardest to investigate design failures because often any (key) distress signals might never be on record or could possibly get destroyed by the highly non- linear destructive path to failure. Nevertheless, given enough time to investigate, adequate simulations and a dose of good luck, one could likely crack the root cause(s) reliably. That could help in building features to prevent similar failure(s).