Conclusions and Recommendations
8.3 Future Research Needs
Future research needs in soil-structure interaction have been organized into two general thematic areas. The first theme involves relatively short-term
recommendations expanding on current studies to: (1) provide tangible insights into the benefits of SSI analysis for owners and practicing engineers; and (2) further explore the benefits and limitations of SSI response history analysis procedures, possibly leading to improved procedures. The second theme involves relatively long-term recommendations intended to address fundamental limitations in the state of SSI knowledge, which limit the accuracy and reliability of SSI models available for use in engineering practice.
8.3.1 Theme 1: Expansion of Current Studies
Example applications presented in Chapter 7 presented case studies on buildings with available earthquake recordings. This effectively limited range of possible structural configurations that could be investigated to the types of structural systems for which data were available. In particular, case-study buildings had modest SSI effects and lacked strongly nonlinear responses. Additional studies should be performed on a common building configuration that is specifically selected to produce large SSI effects.
One such example might include a low-rise dual system building (i.e., shear wall plus moment frame system) with a basement. The building could be designed using a fixed base model (i.e., Model 2 from Figure 6-2) using a typical design spectrum and then redesigned using a bathtub model (i.e., Model 4 from Figure 6-2) for the same design spectrum.
Both large-amplitude and modest-amplitude ground motions could be applied to the model to evaluate the effects of nonlinearity in the structural response on the impact of SSI effects. Alternative SSI element configurations that better account for uplift and nonlinear soil behavior, as described in Section 2.4, could also be applied to assess their relative impact on elastic-plastic procedures. Analysis using a direct approach (as opposed to a substructure approach) could be performed, especially in the case of strongly nonlinear response.
Expanded studies on such a building could be used to provide quantitative
comparisons of the effects of different foundation modeling approaches on structural member sizes and, ultimately, construction costs.
8.3.2 Theme 2: Research to Address Knowledge Gaps
Research needs for expanding the state of knowledge for soil-structure interaction are as follows:
The foundation damping model of Veletsos et al. (various) produces different results than similar models by others. A critical examination of the derivation of that model is needed, followed by the development of equations for foundation damping that properly consider hysteretic damping from soil response, radiation damping from rotational and translational vibration modes, and the sensitivity of radiation damping to different soil stiffness profiles. This problem is discussed in Section 2.1.
The rotational stiffness of shallow foundation systems with non-rigid structural foundation elements is poorly understood. In particular, the effects of coupled versus uncoupled rotations at the base of lateral-load bearing elements on radiation damping and overall system impedance need to be investigated. This problem is discussed in Section 2.2.3.
The impedance of pile-supported foundations is poorly understood for realistic pile and soil conditions, especially for pile groups. Elasto-dynamic solutions for piles in idealized soil profiles exist in the literature, but are not used in practice, in part because they only apply at low displacement levels. The discrete element models that are used (e.g., LPILE, APILE) are poorly constrained for stiffness, and are intended for non-seismic problems. Next-generation element models for dynamic loading of piles are needed that accurately capture the stiffness from
elasto-dynamic analyses and the capacity from discrete element models. This problem is discussed in Section 2.3.
The kinematic interaction problem for pile-supported mat foundations subjected to incoherent wave fields has not been explored to a sufficient degree. Data from Japan suggest that existing models (for base-slab averaging and kinematic
response of pile foundations) are unable to capture observed foundation/free-field transfer functions, which decay rapidly with frequency. This problem is
discussed in Section 3.3.
The conversion of transfer functions, derived from kinematic interaction analyses, to ratios of response spectra currently relies on guidelines provided in FEMA 440, Improvement of Nonlinear Static Seismic Analysis Procedures (FEMA, 2005). A relatively robust statistical model is needed for the relationships between these ordinates as a function of frequency and ground motion characteristics. This problem is discussed in Section 3.4.
Values of the R-factor used in force-based methods for seismic design are based on engineering judgment, considering observations of building performance in past earthquakes and anticipated performance of similarly designed buildings in future earthquakes. Nonlinear response (without collapse) of good performing systems is used to justify high R-values. However, for some buildings, good performance may result, in part, from soil-structure interaction effects that also serve to reduce seismic demands. SSI effects may be partially reflected in current values of the R-factor, resulting in a potential for double-counting. There is a need to revisit the definition of R factors with respect to SSI effects, to make sure that the specified values represent the effects of structural ductility alone.
This issue should be considered in the implementation of NIST GCR 12-917-20, Tentative Framework for Development of Advanced Seismic Design Criteria for New Buildings (NIST, 2012). This problem is discussed in Section 4.1.
Available experimental data described in Chapter 5 should be distilled into a common format of impedance ordinates, and compared to a consistent set of predictions utilizing the procedures and parameter selection protocols given in this report.
Case studies of buildings with recordings from seismic instrumentation are extremely valuable. Unfortunately, current protocols for structural
instrumentation seldom provide the information needed for SSI studies, so additional data are needed. At a minimum, sensors are needed to record
structural translations at the foundation and roof, at least two vertical sensors on the foundation to record rocking, and a ground instrument near the building.