convergence study of the truncated karhunen–loeve expansion.pdfVIP

INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING Int. J. Numer. Meth. Engng 2001; 52:1029–1043 (DOI: 10.1002/nme.255) Convergenc estudy of th etruncated Karhunen–Loev expansion for simulation of stochastic processes S. P. Huang, S. T. Quek and K. K. Phoon∗;† Department of Civil Engineering; National University of Singapore; Block E1 A#07-03; 1 Engineering Drive 2; Singapore 117576 SUMMARY A random process can b erepresented as a series expansion involving a complet eset of deterministic functions with corresponding random coecients. Karhunen–Loev e(K–L) series expansion is based on th igen-decomposition of th ecovarianc efunction. Its applicability as a simulation tool for both stationary and non-stationary Gaussian random processes is examined numerically in this paper. Th study is based on v ecommon covarianc emodels. Th econvergenc eand accuracy of th eK–L expansion ar einvestigated by comparing th esecond-order statistics of th esimulated random process with that of th etarget process. It is shown that th efactors aecting convergenc eare: (a) ratio of th elength of th process over correlation parameter, (b) form of th ecovarianc efunction, and (c) method of solving for th igen-solutions of th ecovarianc efunction (namely, analytical or numerical). Comparison with th established and commonly used spectral representation method is made. K–L expansion has an edg over th espectral method for highly correlated processes. For long stationary processes, th espectral method is generally mor cie