Worried about acceleration?

Ambraseys first law of earthquake
engineering, Imperial College, 1965.
Solid line good, dashed line, bad

In the old days engineers who were old timers (back then, I was a new kid) used to talk about horizontal ground acceleration as the principal source of   disturbance, damage, and failure in structures. Then in the late twentieth century we had some medium sized earthquakes like San Fernando and Northridge down in LA, with these big sharp accelerations, one g or more, twice as much as what experts like George Housner and Nathan Newmark, whom we all respected, had said would be likely in earthquakes; these leaders had led us to believe (we said to ourselves) that we didn't have to worry about more than 0.5 or 0.6g, that's about all the geophysics of the earth could produce and even my prof, Nick Ambraseys, (I've got it her somewhere in my 1965 lecture notes at Imperial College, something about stress drop and velocity and limits.) So everyone, seeing these high accelerations began to waffle around: There must be something wrong about this record. Or: just hold on here, my structure, it doesn't care about high frequency vibration, it worries about longer period motion. 


And then the vision of those old WW2 army barracks being pulled down comes back, that long noisy gradual fall.

Cal Tech engineer Tom Heaton, whose style is much in the solid practical tradition of George Housner or Nick Ambraseys, suggests that designing for peak ground acceleration or for targeted response frequencies, could even be dangerously misleading:

While I understand that many structural fragilities are described in terms of PGA, its use only leads to dangerous mischaracterization of earthquake risk....5% damped response spectral acceleration (sa) is also a common parameterization of shaking intensity. While SA is certainly more useful than PGA, there are serious concerns about using it to predict structural demand. In particular, sa is based on a linear analysis of a structure about its undeformed state. However, there will always be significant ductile yielding prior to catastrophic failure of a structure, and it is more meaningful to use parameters that better characterize a structure that is in its highly deformed state. For example, structures that yield plastically have much lower effective stiffness and much higher effective damping than is typically assumed in current practice.