NeW CONVERSION METHOD FROM RESPONSE SPECTRUM TO.
PSD FUNCTIONS
By Young J. Park," Associate Member, ASCE
‘AostmAct: A new approach to directly convert a target response spectrum to an equivalent power spectral
{density (PSD) function is presented that requires no numerical iterations. Following the same approximation
Scheme used in Monte Carlo simulations, the unknown PSD function is discretized and expressed as the sum
(of Dirae delta functions. The discretized power components can be obtained by solving a standard least-
squares problem. The method was applied to response spectra with steep slopes and sharp turning points
The Monte Carlo simulation results slightly overestimated the zero-period acceleration values. Otherwise, 2
lose match between the target and simulated response spectra was observed. ‘The proposed conversion
algorithm isso simple that the method can easily be implemented
4 Small subprogram.
INTRODUCTION
In the application of near random vibration theories to
seismic deign and analyses, ts offen necessary to convert
2 response spectrum (RS) to an equivalent power spectral
desnity (PSD) faction. Most ofthe conversion metho pub-
lished fa [eg fatfinger(1983)] require a rather complex
fmerica tration scheme, and often result nan unaiti-
factory match wth Monte Carlo simulation results
‘Anew approach (o ditectly convert target RS oan equiv-
alent PSD function is presented that requires no numerical
iterations. The proposed conversion algorithm i so simple
thatthe method can esily be implemented in existing stu.
tural analysis programs a8 a small subprogram.
CONVERSION ALGORITHM
‘The objective here isto obtain an equivalent PSD function
S(w) given a target acceleration spectrum Ri(w, fh) and a
‘modulating envelope function 1() that can be represented by
the effective duration 7.
By following the same ‘approximation scheme used in Monte
Carlo simulations, the unknown PSD function is discretized
at radial frequencies w, and expressed as the sum of a series
of discretized power components p, as
3,
S stopav,s(e) = 3 p,A¢e) o
where 8(a) = the Dirac delta function; and Aw, = the in-