The design and analysis of buried, reinforced concrete box structures subjected to dynamic destructive loads

Thomas Randolph Slawson, Purdue University

Abstract

An engineer facing the design of a blast loaded buried structure has two choices for predicting the response of a trial structural design: (a) SDOF models as recommended by current protective construction design manuals or (b) general purpose finite element methods. General purpose finite element methods are cost prohibitive due to the labor and computer costs relative to typical design budgets. Most engineers follow the accepted practice of using simplified SDOF approximations. The result is an overly conservative structural design which results in higher construction costs. To bridge the gap between overly simplified SDOF approaches and cost prohibitive general purpose finite element methods, simple models that accurately model the primary responses of the structure are required. The suite of models developed in this research range from a three-degree-of-freedom (3DOF) model to a 3-D model using shell elements. The 3DOF model includes the deformation of the blast loaded wall, rigid body motion of the structure, and deformation of the back wall of the structure. These degrees-of-freedom were identified from a review of the experimental data base on blast loaded structures. The 2-D plane frame model is an improvement over the 3DOF model since the primary response modes of the structure do not have to be assumed. The 3-D model includes 3-D effects of loading and response that can not easily be considered in the simpler models. All the models use a decoupled approach where free-field loads, the loads in the soil surrounding the structure, are predicted from fits to ground shock data, and structural loads are determined through an approximate soil-structure interaction model. This approach enables a reduction in computational resources required since the soil does not have to be modeled explicitly in the structural model. The models reproduce the important behaviors observed in the data and should be considered as improvements to current practice on an engineering level.

Degree

Ph.D.

Advisors

Ting, Purdue University.

Subject Area

Civil engineering

COinS