UCLA

Earthquake engineering practice has traditionally focused on the horizontal components of ground motion, while the vertical component and its effects on building response have received less attention. This dissertation addresses this gap by developing a data-driven framework to characterize vertical structural response in instrumented buildings and to improve understanding of vertical dynamic properties, floor-level acceleration demands, and code-based vertical seismic load effects. Using sensor recordings from past earthquakes, complemented by numerical simulation and code-oriented evaluation, the research provides recommendations to inform future provisions of the United States building codes. The study uses dense triaxial acceleration recordings from the Community Seismic Network at the NASA Jet Propulsion Laboratory campus. Six mid-rise JPL buildings, ranging from four to nine stories, are analyzed using earthquake records that were consistently recorded across the selected buildings. A Python-based data preparation and processing pipeline is developed to retrieve, verify, convert, filter, and align the recorded building response. The processed records are then used to identify vertical and horizontal structural response frequencies through Fourier amplitude spectra, power spectral density, transfer functions, and coherence functions. The results show that vertical response frequencies are generally higher than horizontal frequencies. Across the six buildings, dominant vertical frequency is typically concentrated in the approximately 5–10 Hz range, while dominant horizontal frequency occurs at the fundamental horizontal mode, approximately 0.7–1.8 Hz. To interpret the measured vertical frequency ranges captured using sensor recordings, a refined three-dimensional numerical model of JPL 183 is analyzed. The simulation results show that vertical modal mass is distributed across many modes, and that substantially more modes are required to capture vertical response than are required for horizontal modal response. Stiffness-perturbation studies are then used to distinguish global vertical response from slab-dominated vibration. A key finding is that the dominant vertical response is often not the lowest identified vertical frequency but instead occurs at higher frequency ranges associated with global vertical response, slab vibration, or coupled global-local behavior. This dissertation also evaluates vertical and horizontal floor response spectra using 5%-damped pseudo-spectral acceleration computed from measured floor accelerations. The vertical floor response spectra show strong period dependence, with the largest vertical amplification generally concentrated in the short-period range of approximately 0.08–0.25 seconds. Vertical acceleration demand generally increases with height. Horizontal floor response spectra show a different pattern: short-period horizontal amplification is generally more modest, while stronger amplification occurs at intermediate-to-longer periods associated with horizontal modal response. Comparisons with ASCE/SEI 7-22 Chapter 13 show that current horizontal height-amplification assumptions capture the general concept of increasing acceleration demand with height but do not fully represent the measured period dependence. For vertical response, the results indicate that the current code framework does not provide an explicit vertical floor-response-spectrum or height-amplification model. Therefore, a separate vertical floor-amplification factor is strongly recommended for future consideration in nonstructural component design. Finally, the dissertation evaluates two approaches in the current building codes (ASCE, 2022) for calculating vertical seismic load effect: the period-independent expression based on 0.2S_DS and the period-dependent expression based on 0.3S_av. The two approaches are compared using NGA-West2 ground-motion models for more than 3,000 seismic scenarios and ASCE 7 Hazard Tool data for 19,316 California sites. The results show that the relationship between the two expressions depends strongly on the vertical period and seismic design category. The 0.2S_DS expression can underestimate vertical seismic demand for moderate vertical periods, especially around 0.05–0.20 seconds in higher seismic design categories, while it can be overly conservative for very short and long vertical periods compared to results derived from 0.3S_av. Based on these findings, the dissertation provides period- and seismic-design-category-dependent recommendations for calculating vertical seismic load effects and proposes refinements to improve the long-period behavior of the vertical design spectrum. Overall, this dissertation demonstrates that vertical structural response is an important structural dynamic phenomenon that should be characterized using measured data, physical interpretation, and period-dependent design considerations. The findings provide empirical evidence and building code-oriented recommendations for improving the treatment of vertical response in future building code provisions.