Water tower analysis

For an existing 110' high elevated water tank located in seismic zone 4 would you recommend a static analysis per 97' UBC standards or a full blown site specific dynamic analysis. Analysis is being performed for a retrofit consideration. Per 97' UBC if the structures period is less than 0.7 sec then per code you are allowed to perform a static analysis, above 0.7 sec code recommends dynamic. We have heard from a few source that a dynamic analysis should definitely be performed on a inverted pendulum type structure like this. Please advise on your theories on how to accurately model this type of structure.

Thanks,
KSP Engineers

 

[twjag]

KSP Engineers,
There is a third option: perform a 97 UBC Scaled Response Spectra (SRA) analysis based upon Ca & Cv from tables. Dynamic solution allows for a reduction to UBC static shear.

Based upon your "inverted pendulum" description, I interpret geometry as a water tank "egg" supported on a single cylindrical pedestal. UBC requires 90% mass participation, which is difficult (if not impossible) to obtain if you model with plate elements: this is due to the fact that the large number of plate elements (& theri local participations) requires an unreasonably high number of modes (& extended process time) to obtain participation.

We have found that taking the following two step procedure yields quick & accurate results:
1. Model the pedestal using cylindrical beam elements with vertical load applied at top (elevation = tank + water combined c.g.). Run Modal Freguency Analyses (MFA) to obtain freguencies & mass participaton. Subdivide vertical column as needed to obtain 90% mass participation using no more than 10 modes per direction. Next run Response Spectra Analyses (RSA). Scale RSA shears to UBC & run DL+Seismic Load Combinations to obtain shear & moment reactions at foundation.
2. Model the tank & pedestal using plate elements with surface loads applied to tank to simulate weight of contents (water). Run MFA to obtain freguencies & mass participaton. Use sufficient modes to obtain >60% mass participation (30 or so modes will usual suffice). Run RSA using significant modes. Scale RSA shears to UBC & use for Seismic Load Combinations. Compare base reactions to those for Model 1. If in reasonable agreement, OK. In final report Model 2 is submitted as the basis for design, & Model 1 is an attached addenda to support using Model 2 at mass participations < 90%. This should be thoroughly addressed in the forward of the report.

UBC rules are based upon regular building structures. The rules tend to penalize nonbuilding structures such as vessels & stacks. The UBC 90% mass requirement is to insure adequate mass participation in all parts of standard building structures. Mass participation is not a problem in single cylindrical cantilevered structures such as yours. This is proved using the two step procedure given above. Valuable insights on mass participaton & seismic design of nonbuilding structures can be found in the SEAOC Blue Book & Seismic Design Manual Vol 1. They are available on a single &quot;Seismic CD&quot; from UBC at a very reasonable price. The CD contains a number of other good references, including Vol 2 of the UBC, all in PDF format.