According to most seismic design and assessment standards, it seems that regardless of any plan or vertical stiffness irregularities, a linear elastic MRSA is allowed to be used as the primary analysis approach for all structures.
The codes are not clear on the appropriate analysis method for cases that require a nonlinear analysis such as when a soil-structure interaction analysis or tension/compression only elements are required.
ASCE41 does not allow for an elastic analysis including MRSA when the performance of the structure at a certain earthquake level involves considerable inelasticity.
However, the question is if there is any other limitation with the use of MRSA?
One example is when we have a structure with completely different lateral resisting systems along its height such as a flexible structure sitting on top of a stiff structure. This is quite common for structures with a podium or when more storeys are added to an existing building. In such cases the modal behaviour of the two parts of the building can be quite different and in some cases even out of phase.
Even if we assume such different modal behaviour can be captured by a MRSA, a large number of modes have to be introduced to achieve an acceptable total mass participation.
But how much SRSS or CQC the results of a large number of modes and calculate the forces and displacements based on that are reliable? And what they mean from a physical point of view?
Another shortfall of a MRSA is that its base shear needs to be matched with the Equivalent Static Analysis (ESA) one. In cases where the periods of the bottom and top are considerably different, choosing the appropriate period for the ESA base shear calculation can be quite challenging.
This is an issue for a MRSA, while for example a Time History Analysis (THA) can capture the correct base shear. Would a large scaling factor indicate anything as to whether a MRSA is appropriate or not?
Another challenge is when the top and bottom structures have different ductility demand at a certain earthquake level. How can we capture the seismic performance of the structure using a MRSA in such cases?
There are other concerns with the accuracy and reliability of MRSA that may suggests the need for further investigations into adopting the most appropriate analysis method for different structures.
Since a pushover analysis also won’t be able to capture the correct behaviour in most of these cases, the only other option is a nonlinear time history analysis. But is it really where we need to go in such cases? Are there more simple ways to deal with these issues?