“Correlation
of Various Response Spectrum Methods with Time History Analysis for a Tuned Subsystem”
Developing appropriate
methods for designing structures subject to seismic stress concerns all engineers, but none more so than nuclear engineers. For the nuclear industry, the NRC has set strict guidelines on design criteria for
primary systems (‘supporting” structures) and secondary systems (“supported” structures or equipment).
Often, estimates of the seismic
response of coupled primary and secondary structures are actually calculated separately (decoupled). There are two reasons for this: (1) the sheer number of possible interactions exceeds the limits of accuracy
of standard computational methods, and (2) appropriate seismic response spectra at the connection between primary and secondary
systems are not available.
However, when designing primary
and secondary structures in nuclear power plants, it is necessary to analyze their response as coupled systems, often using
complex analytical models. In their paper “Correlation of Various Response
Spectrum Methods with Time History Analysis for a Tuned Subsystem,” Bechtel engineers M. Dong and H. M. Lee explore
refined methodologies for analyzing the seismic response of these systems. They
address the physical conditions appearing when the natural frequency of an appendage subsystem is close to one of the natural
frequencies of the primary structure. This is called a “tuned” branch
system.
While analyzing minor changes
to a subsystem design, Dong and Lee applied the closely spaced modal combination rules of NRC Regulatory Guide 1.92. They found that Regulatory Guide 1.92 leads to gross overestimation of subsystem behavior
when responses from two very closely spaced modes are added absolutely, but in reality they are out of phase and have a canceling
effect.
Dong and Lee found that when
using modal response spectra techniques, unrealistic modal coupling effects will most likely occur when the subsystem is either
tuned or nearly tuned to the primary system. The current revision of Regulatory
Guide1.92’s rules yields results that are grossly conservative when compared with results of a time history analysis
of the tuned subsystem. More refined modal coupling methods in use in the
industry provide good correlation with solutions from time history analysis with synthetic time history.
Dong and Lee’s study indicates
that the subsystem decoupled according to NRC’s SRP 3.7.2 (NUREG-0800) rules yields reasonable results as compared with
time history solutions. They recommend the use of independent support motion
analysis (ISM) with the absolute grouping method together with seismic anchor motion (SAM) analysis to evaluate systems and
subsystems. This decoupled approach is simple and has significant safety margins
even when the subsystem is tuned to the primary system. An NRC committee is evaluating
their findings.
BECHTEL ENGINEERING BULLETIN
Page
20, Volume 5, Number 1, Year 1989