You are on page 1of 5

Designing an Underground Structure with STAAD

Dr. Mukti L. Das Principal Engineer Bechtel Power Corporation Frederick, MD 1. Introduction The purpose of this paper is to share the authors experience of STAAD that was used to solve an important underground structure in the Boston Central Artery, popularly known as Big Dig Project. As an independent consultant, the author was involved in investigating the design adequacy of a section of the underground transit system as shown in Figure 1. 2. Development of STAAD Model Starting from the top, there is a varying depth of soil cover followed by the top concrete slab of the tunnel. There are two viaduct columns directly supported on the concrete slab. The second column is on the left of the one that is shown on the upper left corner of Figure 1 and is not shown. This sketch also shows 10 wide segmented T-shape slurry wall with 10 wide flange facing outside at regular interval. The bottoms of these concrete walls extend into the bed rock. At the top, these segmented slurry wall strips support T-shape slurry walls with wider flange facing outside and budding to each other along the length of the tunnel forming its wall. These wider T-shape walls with web protruding inside the tunnel resist the earth pressure. There are concrete beams connecting two tunnel walls. These connecting beams are shown as small rectangles just under the bottom slab of the tunnel. The train rails are supported on this bottom slab.

Figure 1: Longitudinal view of a section of the tunnel

A finite element model of the tunnel section left of and including the viaduct column shown in Figure 1 was developed with STAAD. The T-shape slurry walls were modeled as BEAM members as they were not a monolithic wall. The top and bottom slabs were modeled as finite element mesh as shown in the Figure 2. The last left portion of the tunnel model is a monolithic rectangular concrete box consisting two walls, top and bottom slabs. This box was modeled as a finite element mesh. The soil was modeled as Winkler spring calculated on the basis of the modulus of subgrade reaction of soil. The beams members representing the slurry walls were divided into several parts. Each joint of beam members was attached to a soil spring calculated on the basis of the influence area of that joint. Similarly, each joint of the plate/shell elements was attached also to a soil spring representing the influence area of that plate element. All the appropriate loads including hydrostatic pressure on the slurry walls and the bottom slab were input in STAAD.

Figure 2: Finite element mesh with slurry walls idealized as BEAM members

Figure 3: Finite element meshes of top slab, part of bottom slab and boxed shape left end are
shown in red

3. Analysis Output The text output and the graphical presentation of the resulting deflections and stresses on beam and plate/shell elements were thoroughly investigated before acceptance. This investigation was based on hand calculation on several simplified models.

Figure 4: Dead load deflections under two columns supported by top slab

Figure 5: Moment, Mx, contour due to dead load under viaduct Columns

4. Conclusion Like any other software output, caution should be exercised before accepting the output. During the process of converting the real world physical problem to a mathematical model to be analyzed by software, several assumptions are made. These assumptions are vital to preserve the principal characteristics of the structure for reliable result. Following

suggestions are made especially for creating finite element models such as the one that was used in this project. Aspect ratio: The ratio of the shortest side to the longest side of an element should not be larger than 1:4. The ratio, 1:1 will give best result. Try to use rectangular elements as much as possible. Element size: Use closer mesh size near the region where either the geometry or the load changes abruptly. Neighborhood: Do not place a very small element in the neighborhood of a very large element. The transition from very large elements to very small elements should be gradual. Element number: The total number of finite elements to represent a continuum should be appropriate. Then, the question may arise, what is that number? With some experience, you will know that number. Certainly, one element to represent a floor of a building is not appropriate. Probably, one hundred (100) is the number. Again, this will depend on the shape and any abruptness in geometry or applied load. Convergence: This is very difficult to test. Ideally, the mesh size should be made finer and finer until there is no change in result. For a big model, it is not practical. If the structure is discretized appropriately, the convergence will not be a problem. Shell element: Most problems in structural engineering require shell elements. If possible avoid use of Solid (Brick) elements. Local Axis: STAAD uses a special local axis system for Shell elements. Try to understand it thoroughly before you interpret the output. Units: The resulting force/moment is known as Force Resultant and is expressed as force/moment per length-unit of the shell cross section. All commercial shell analyses programs use this concept.

You might also like