Performance-based Design of Soil-Foundation Interface in Buildings - 1412005a

Project Title—ID Number	Performance-based Design of Soil-Foundation Interface in Buildings - 1412005a
Start/End Dates	10/1/05 – 9/30/06
Funding Source	PEER-NSF
Project Leader (boldface) and Other Team Members	Tara Hutchinson (UCI/F), Prishati Raychowdhury (UCI/GS), Barbara Chang (UCI/GS)

Project goals and objectives

Seismic loading on a building with shear walls can potentially cause the foundations of the shear walls to rock. Current building codes discourage designs that allow rocking. For existing buildings, however, it is often very expensive to retrofit foundations to prevent rocking. Furthermore, it has been suggested that building performance might actually be enhanced if rocking is allowed because rocking could reduce seismic demands on a building and dissipate energy in hysteresis at the foundation soil interface. Although previous work has focused on shear walls, results should be applicable to shallow foundations for building and bridge columns as well.

The ultimate goal of this project is to develop the necessary tools to predict rotations and translations at the soil – shallow foundation interface and to allow engineers to assess, through quantitative analysis, the trade off between the benefits (energy dissipation and isolation) and the detriments (e.g., permanent and cyclic settlement and/or tilt) associated with foundation nonlinearity.

The objectives of this work in Year 9 are to:

1. Refine and finish a BNWF foundation model implementation in OpenSees, assuring it reasonably captures foundation nonlinearity; moment-rotation, shear-sliding, and axial- settlement behavior. This includes documenting the model for use by others and testing the model convergence and sensitivity to input parameters. In addition, it includes clearly identifying and quantifying all empirical model parameters, presenting these guidelines for their quantification, evaluating the sensitivity of the results to engineering assumptions inherent in these parameters, and delineating boundaries beyond which application of these models is not appropriate.
2. Complete data analysis from the frame-wall-foundation model centrifuge tests and archive data report with UC Davis.
3. Develop and document a process for using the models to extract a basic engineering foundation interface model for wall or column footings

Role of this project in supporting PEER's mission (vision)

This project supports the PEER strategic plan by providing performance data, validation tests, and nonlinear models to advance the simulation capabilities of OpenSees. In addition, understanding the behavior of shallow foundations is critical to development of performance based design procedures for buildings.

Methodology employed

Our approach has been to use a Beam-on-Nonlinear-Winkler (BNWF) framework (e.g. using spring, dashpot, and gap elements) for modeling the nonlinear soil response. This allows us to capture the salient features of the rocking foundation, including distributed soil nonlinearity and uplifting of the foundation. The intent of subgrade type modeling has always been to strike a balance between theoretically more rigorous solutions and practicality and ease of use in routine geotechnical engineering practice. Therefore, the approach of using the BNWF model is directly applicable to many practical applications. Complementary numerical modeling is being conducted at UC Davis using a plasticity based 'macro-element' representation.

Centrifuge testing with UC Davis has been conducted on two different wall-frame-foundation models to develop a database for evaluating the models. In addition, UC Davis has an extensive database of isolated footing-wall models, considering different factors of safety, embedment moment/shear ratios, and soil conditions for model evaluation.

Brief Description of previous year's achievements, with emphasis on accomplishments during last year (Year 8)

BNWF Model Implementation, Testing, Sensitivity Evaluation and Documentation The UC Irvine implementation involves use of a number of independent nonlinear springs at the base of the shear wall or column. This model has been tested on a number of isolated wall-footing model experiments conducted by UC Davis and has worked well for large amplitude rotations on fairly strong soils (see Figure, SSG03#2c test, with FSv = 11.5). However, for softer foundation soils and for small amplitude rotations, the model underestimates settlements observed in the centrifuge tests. We are evaluating existing axial loadsettlement curves from centrifuge experiments and plan to use these to refine the backbone curves for shallow footings on softer/looser soils to correct this shortcoming. Datasets from SSG02 and 03 series tests include datasets on medium sands, which are ideal for this purpose. After regressing through the test datasets backbone curves, the existing qzMaterial model may be modified to account for the softer initial axial loadsettlement backbone curve observed in these centrifuge tests.

To test the sensitivity of the model, we participated in a case study building evaluation with UC Davis, led by UCLA. The shearwall-footing portion of an example 4 story building was modeled and is being used to study the sensitivity of response to derived input and model parameters. We find from this study that Rayleigh damping and mesh discretization have the most pronounced affect on the models response. We plan to further evaluate the model sensitivity for other case buildings, and other ground motions. For example, one additional case study is a model of the wall-frame-foundation structure tested on the centrifuge. Here we considered five different base conditions in the modeling: (i) fixed base, (ii) elastic base with no sliding, (iii) elastic base with sliding allowed, (iv) nonlinear base with no sliding, and (v) nonlinear base with sliding allowed. We considered 10 ground motions applied to the model and a range of FSv (3, 5, 7, 10). Results show that FSv has a significant affect on the response of the building (on floor acceleration, drift, moment and shear). The capacity curves as expected also show the dramatic difference in stiffness, period, yield point, etc. for the different foundation cases.

Model Frame-Wall-Foundation Centrifuge Test Data Analysis

We have conducted a series of tests with UC Davis where one and two-bay frame structures were attached to shearwalls and supported by square and strip footings. These models were subjected to slow cyclic lateral loading and dynamic base excitation (sinusoidal and earthquake motion). A unique feature of the models was the use of designed plastic hinges at the beam-column joints. This was intended to model the energy dissipation within the superstructure. The models were heavily instrumented, with nearly 70 sensors on the two-bay model and therefore the data processing is a significant task. Preliminary findings show that the desirable yielding and nonlinear behavior of the beam-column joints is occurring during the experiments. Energy is also dissipated by rocking, sliding, and settlement of the footings. However, we observe highly asymmetric hysteretic loops for both structures, due to the asymmetry of the lateral force resisting system. The figure below shows the percentage of energy dissipated by the footings EF divided by the total energy dissipated (by footings and beam-column fuses) S E (F-B) in % versus the drift ratio (%). This preliminary data shows that even at low drift ratios, the footings can provide a good portion of the total dissipated energy (as much as 35% of the total). For higher drift ratios, one would expect the portion of energy dissipated due to the yielding of the joints to be more substantial, which is observed for the static push test data.

Other similar work being conducted within and outside PEER and how this project differs

Within PEER, this work is closely coordinated with work by USC (Martin) and UCD (Kutter) and UCLA (Stewart). Work conducted at UCD includes providing experimental data and guidance for use of the data in analytical modeling. In addition, UCD is developing a complementary macro-element model. Work conducted at USC involves the oversight and integration of work performed at UCD and UCI. Work performed at USC also includes interfacing with practicing engineers in the US and Europe involved in implementation of nonlinear SSI into seismic design guidelines or codes. There also useful related work on this topic being conducted in France, Italy, and England. A new project at UCLA (Stewart) is attempting to use the findings and developments at Davis and Irvine and implementing them in the context of the PEER PBEE framework.

Our numerical studies on this topic will be unique in that most studies either include nonlinearity of soil elements, or nonlinearity of structural components of the system. Both the testing and numerical simulations incorporate these two contributions and study system response.

Describe any instances where you are aware that your results have been used in industry

We have held a workshop in which we have discussed our results with several structural and geotechnical practicing engineers. Thus the results have been used to clarify the issues and mechanisms, which require attention.

Expected milestones & deliverables

1. Data report on frame-wall-foundation tests in collaboration with UC Davis – Summer 2006.
2. BNWF mesh implementation (including code examples, documentation, and sensitivity studies) in C++ - Fall 2006.
3. PhD dissertations: Chang and Raychowdhury, expected 2007

Member company benefits

We have engaged significantly with Mark Moore of Rutherford and Chekene on this work.