Sunteți pe pagina 1din 17

Draft 1.

2012
Draft 1.3

August 10, 2012

Construction Company
2259 Ward Avenue, Suite 200, Simi Valley, CA 93065

Alexandre Basso

[SHORWALL HALF MODEL ANALYSIS]


[Type the abstract of the document here. The abstract is typically a short summary of the contents of the document. Type the abstract of the document here. The abstract is typically a short summary of the contents of the document.]
Page 1 of 17 Document1

Draft 1.3

August 10, 2012

[TABLE OF CONTENTS]
1 2 3 3.1 3.2 3.3 3.4 3.5 4 4.1 4.2 4.3 5 6 Executive Summary ....................................................................................................................... 4 Introduction .................................................................................................................................. 4 ShorWall Model ............................................................................................................................ 5 Model Description..................................................................................................................... 5 Model Setup .............................................................................................................................. 7 Material Properties ................................................................................................................... 8 Loading Conditions.................................................................................................................. 12 Symmetrical Boundary Conditions.......................................................................................... 12 Analysis Results ........................................................................................................................... 12 Half vs. Full Model ................................................................................................................... 12 Tendon Sensibility Analysis Results ........................................................................................ 13 Concrete Strength (fc) Sensitivity Analysis Results ................................................................. 14 Conclusion ................................................................................................................................... 16 References .................................................................................................................................. 17

Page 2 of 17 Document1

Draft 1.3

August 10, 2012

[TABLE OF FIGURES]
Figure 1. ShorWall Model Geometrical Imperfection ........................................................................... 6 Figure 2. ShorWall Half Model with Soil Springs ................................................................................... 7 Figure 3. ShorWall Model Elements ...................................................................................................... 8 Figure 4. Free Length for Reinforcements through the Joint Interface .............................................. 10 Figure 5. Interface Element Normal Function ..................................................................................... 11 Figure 6. Interface Element Shear Function ........................................................................................ 11 Figure 7. Symmetrical Boundary Conditions ....................................................................................... 12 Figure 8. Comparison of Full and Half Model Results ......................................................................... 13 Figure 9. Tendon Sensitivity Study Results .......................................................................................... 14 Figure 10. Concrete Strength Sensitivity Study Results....................................................................... 15

Page 3 of 17 Document1

Draft 1.3

August 10, 2012

Executive Summary

The unconventional shape and construction method of the ShorWall makes it difficult to determine the structural behavior and material stresses using conventional calculations. Therefore a full finite element model was created to analyze the wall but it was taking up to one hour to solve the model. As a result, a study was conducted with a half model and it was determined that it was feasible to use this model and still get the same results as the full model. The major benefit is that half model could be solved in less than twenty minutes thus expediting the answers to different scenarios. The following recommendations are based on sensitivity studies and should be taken into account when designing the ShorWall. Pre-stressed or post-tensioned tendons should not be used since they do not increase the buckling load, cause large local stresses, and may make the wall joints stiffer than they are in reality. Plus, the load-displacement curves were the same for the models with and without the pre-stressed tendons. The ShorWall should be designed as a rigid wall deflecting less than one inch under service loads in order to eliminate destabilization of adjacent foundations. The wall thickness and/or the concrete strength should be increased to raise the loaddeflection curve thereby raising the buckling failure load.

Introduction

Common commercial building design and construction practice of relatively shallow underground structures generally incorporates a two-step process: installation of temporary shoring to support earth loads during excavation, followed by construction of the permanent structure. These structures are typically rectilinear in shape. This approach is both time and cost intensive due primarily to the necessity to provide temporary shoring and the time required to complete its construction. The ShorWall utilizing a circular design which is capable of self shoring during construction alleviates the inefficiencies found in the current practice. This structural system consists of a stacked series of circular rings. Each ring is made up of curved precast concrete segments which are joined together utilizing shear fittings and post tension structural elements. The process of constructing this type of structure will usually begin with some form of perimeter soil improvement followed by structural excavation of the soil up to 6 feet (1.8 m) deep and installation of damp-proofing material against the soil face. Precast segments are then installed end-to-end forming a complete circle or ring. Following ring completion, grout is applied under prescribed pressure to fill the space (annulus) between the ring and the damp-proofing/soil behind the ring, thereby engaging the ring in resisting lateral soil pressure. During and following pressure grouting, the lateral soil pressure bearing on the ring applies compressive forces that are carried by hoop stress throughout the ring. The friction between the segments and the soil (resulting from the soil pressure), together with construction means and methods, resists the gravitational weight of the segments, enabling the next phase of excavation below each

Page 4 of 17 Document1

Draft 1.3

August 10, 2012

completed ring (underpinning). Any number of additional rings can be constructed below a completed ring by repeating these steps one ring at a time, until the design depth is achieved. This construction sequencing is defined as a top down construction method. Designing the ShorWall can prove to be a difficult task due to its structural behavior that is not similar to conventional construction. Thus finite element models were created to analyze the behavior of the structure and to determine the buckling load of the wall. This report will discuss the finite element model simplification and the sensitivity studies that have been conducted to determine the most efficient design for the ShorWall.

ShorWall Model

In order to reduce the computational time of the ShorWall analysis, the full model was reduced in half by identifying a symmetrical plane and removing the elements on one side of said plane. Next, symmetrical boundary conditions were added to the wall on the symmetrical plane in order to represent the influence of the removed half of the wall. To determine that the half wall was modeled properly, a study was conducted to compare the deflection results of the half model to that of the full model, see Analysis Results. The ShorWall half model is based on the full model (PFM_2a3_v4_c45) created by Kasidit Chansawat (KC). This model was in turn edited by TNO DIANA to mitigate the parallelism problems that KC ran into as explained in the Nonlinear FE Modeling and Analysis of the Full and Detailed ShorWall dated October 18, 2011.

3.1 Model Description


The following are the geometrical properties that were used to model both the ShorWall full and half models in DIANA 9.4.4. Inside Radius (nominal) Inside Diameter (nominal) Total Height Precast Segment Height Precast Segment Thickness Precast Segment Length Geometrical Imperfection Ellipse Major Radius Ellipse Minor Radius Mesh Width Mesh Depth Mesh Thickness = = = = = = = = = = = = 115 ft or 1,380 in (35 m or 3,505 cm) 230 ft or 2,760 in (70 m or 7,010 cm) 40 ft or 480 in (12 m or 1,219 cm) 5 ft or 60 in (1.5 m or 152 cm) 12 in (30.5 cm) 10 or about 20.16 ft (6.14 m) 0.5 in (12.7 mm) 1,380.5 in (3,506 cm) 1,379.5 in (3,504 cm) 1 or about 24.3 in (61.7 cm) 20 in (50.8 cm) 12 in (30.5 cm)

Page 5 of 17 Document1

Draft 1.3

August 10, 2012

Number of Rings Number of Wall Segments

= =

8 rings per 40 ft depth 36 segments per ring, 9 per quadrant

The loading onto the ShorWall is converted into compressive hoop stresses which induce the wall to behave similarly to a column under compression. When a column is loaded under compression, it will follow the primary path which is the original load-displacement curve until the critical load is reached and then it will track the Perfect Circle secondary path known as the post-buckling curve. The point in which the two paths intersect is called the bifurcation point. The primary path becomes unstable after the bifurcation point even though it is still mathematically possible to continue on that path, nonetheless, a real structure will follow the secondary path. In order to force the finite element model to jump from the primary to the secondary path, a geometrical imperfection needs to be introduced to nudge the model into buckling. Figure 1. ShorWall Model Geometrical Imperfection Halcrow conducted a study on the ShorWall buckling and determined that a geometrical imperfection of 0.5 inches was appropriate to induce buckling without extensively de-stabilizing the structure. This results in a slightly elliptical shaped shaft rather than a perfectly round shaft as illustrated in Figure 1. Special consideration should be taken when designing the ShorWall since it is a thin-walled shell structure that is sensitive to imperfections, meaning that the buckling load is affected by small changes in geometry, distribution or orientation of loads, and manner of support.

Page 6 of 17 Document1

Draft 1.3

August 10, 2012

3.2 Model Setup

Concrete Panels (CHX60)

Soil Springs (SP2TR) Figure 2. ShorWall Half Model with Soil Springs

The finite element half model, Figure 2, is made up of a variety of elements that represent the different components of the wall. The pre-cast concrete panels were modeled with a twenty node solid brick element (CHX60) which is based on quadratic and Gauss interpolation in order to provide accurate results without greatly reducing the mesh size. The brick elements are meshed together to build a full pre-cast segment, the panels are identified by different colors in the image above. In between each concrete segment there are vertical and horizontal interface elements that represent the joint material, in this case a rubber packer. The interface is modeled with a plane quadrilateral element (CQ48I) that uses two planes, one for each face, with eight nodes each. This interface was also used to model the alignment pins which are positioned within the post-tension ducts. Each ring of pre-cast concrete segments has one horizontal tendon and each segment has four vertical tendons that have been modeled as pre-stressed embedded reinforcement bars that are bonded to the concrete (mother element). The tendons were not modeled as post-tensioned reinforcement bars because they do not interact with the model unless they are bonded to the mother element. The soil springs were modeled as a two node translational spring (SP2TR) and were connected to the corners of each solid brick element. Figure 3 depicts all of the elements that were used to model the ShorWall in DIANA.

Page 7 of 17 Document1

Draft 1.3

August 10, 2012

Soil Springs (SP2TR)

Horizontal Tendons

Alignment Pins (CQ48I) Vertical and Horizontal Interface Elements (CQ48I)

Vertical Tendons

Figure 3. ShorWall Model Elements

The ShorWall model was created to be slightly elliptical in order for the wall to buckle, but this made the model more complicated to create. For example, a single segment cannot be created in DIANA and then copied radially about the center to create a ring of segments because it will either form a circular shape or the nodes will not lineup properly. Also, after creating the first ring of segments, the ring cannot be duplicated below the first ring and rotated to produce the running bond because the nodes will not lineup in between rings due to the elliptical shape. Thus the first and second rings were created separately with the running bond and than they were duplicated below to form the full height wall.

3.3 Material Properties


Concrete Various concrete properties were used to model the pre-cast concrete segments in order to determine which variables affected the buckling capacity of the ShorWall. Due to the limitations in Diana, the model had to be converted to SI units before the concrete properties for the ACI 209R-92 model could be entered. The properties are listed below for the different half model versions that it applies to. Half Model v1 to v4 Compressive Strength (f'c) Youngs Modulus (E) Poison Ratio () Half Model v5 & v8 The ACI 209R-92 model code regulation was used to derive the non-linear material properties for concrete. Below are the input parameters for DIANA 9.4.4. Compressive Strength at 28 days (FCM28) Poisson Ratio () = = 55 MPa (8,000 psi) 0.2 = = = 8,000 psi (55 MPa) 57,000 x sqrt(8,000 psi) = 5.0E+09 psi (3.45E+07 MPa) 0.2

Page 8 of 17 Document1

Draft 1.3

August 10, 2012

Density () Cement Type Curing Method Analysis Type

= = = =

2.4 kN/g III Steam Cracking Function (CRACKN) Plasticity Function (PLASTN) Concrete Tensile Strength Limit (FTMODN)

Half Model v6 & v9 The compressive strength at 28 days was changed to 27.5 MPa (4.0 ksi). Half Model v7 & v10 The compressive strength at 28 days was changed to 82.5 MPa (12.0 ksi). Tendon (Steel) In DIANA, the tendons do not interact with the mother elements (in this case the pre-cast segments) unless they are bonded to them. In the post-tension element model, the tendons have to be manually bonded to the mother element after they are tensioned. Therefore, the tendons were modeled as pre-stressed elements since they are automatically bonded. Also, the tendons do not need any boundary conditions at the symmetric plane because they are embedded into the mother element before the pre-stressing load is applied. The Youngs Modulus used for the steel tendons is lower than an actual post-tension cable in order to stay consistent with KCs work. However, it still gives a good determination of the systems behavior. Diameter Area Youngs Modulus (E) Poison Ratio () Yield Strength (fy) Free Length = = = = = = 0.5 in (12.7 mm) 0.1963 in (127 mm) 29.0E+09 psi (2.0E+08 MPa) 0.3 1.0E+05 psi (690 MPa) 1.0

The reinforcement free length is used to determine the stiffness in the normal and shear direction of the tendon in the interface element. The value for the free length can only be entered in the DIANA Mesh Editor and the stiffness per unit area in the normal direction (kn) and in the shear directions (ks & kt) are determined as follows

The stiffness in the shear direction is caused by the dowel effect of the reinforcement bars through the interface and is assumed to be half the stiffness of the normal direction, see Figure

Page 9 of 17 Document1

Draft 1.3

August 10, 2012

4. The dowel effect means that the reinforcement has the ability to be a major contributor to shear resistance at the joint much like a dowel or alignment pin.

Figure 4. Free Length for Reinforcements through the Joint Interface

If the free length is not specified, DIANA uses the thickness of the interface element. But if the thickness is zero, then DIANA uses a virtual thickness of 10-5 times the distance from the first to the second node of the interface element. This could potentially make the steel reinforcement in the joint very stiff therefore the free length was manually specified to be one. For more information see section 14.5.1 in the Element Library manual. Soil Spring Soil springs were used to represent the interaction between the ShorWall and the surrounding soil. The springs have been modeled to resist a maximum compressive load of 562,302 pounds per inch and to not resist any tensile loads. Below are the parameters used to determine the spring stiffness. Subgrade Reaction Modulus Mesh Size Spring Stiffness Interface Elements The vertical and horizontal interface elements have the same properties as the alignment pin, see Figure 5 and 6. TNO DIANA made this change because KC modeled the vertical and horizontal interface elements as Coulomb friction which made the model unstable. The instability originated from the fact that the shear resisting force at the joints is calculated using the normal force which is derived from the hoop stress in the horizontal direction and the self weight of the segments in the vertical direction. As a result the normal force in the vertical direction was not large enough to generate the appropriate shear resisting force to keep the panels in place. = = = 1,157 psi/in (314,000 kPa/m) 24.3 in x 20 in (61.7 cm x 50.8 cm) 1,157 x 24.3 x 20 = 562,302 lb/in (98,474 kN/m)

Page 10 of 17 Document1

Draft 1.3

August 10, 2012

Normal Function Displacement (in) Traction (psi) -1.0E+09 -7,600 -0.00113 -7,500 0 0 0.08 23 0.20 31 1.0E+09 41
Figure 5. Interface Element Normal Function

Shear Function Displacement (in) Traction (psi) -1.0E+09 -87 -0.43 -77 0 0 0.43 77 1.0E+09 87
Figure 6. Interface Element Shear Function

Page 11 of 17 Document1

Draft 1.3

August 10, 2012

3.4 Loading Conditions


The loads were applied in construction phases with the first phase adding the weight of the structure to the model in one load step. In the second construction phase, the steel tendons were prestressed to 80 ksi (80% of the yield strength of the steel) in one load step. Next, a uniformly distributed lateral pressure of 60 psi was gradually applied to the exterior face of the ShorWall to represent the soil loading. The load was applied in small incremental load steps in order to capture the behavior of the wall as the load increased.

3.5 Symmetrical Boundary Conditions


Once the model was cut in half, symmetric boundary conditions had to be applied to mimic the interaction between the two halves. In order to keep the model symmetrical, the precast segments on the odd rings (1, 3, 5, and 7) were left at full size with the vertical interface element on the outside. Boundary conditions in the direction normal to the segment face were applied to the joint interface to model the panel, on the other side, resisting the hoop stresses. The even ring (2, 4, 6, and 8) precast segments had to be cut in half and both normal and moment reactions were attributed to the nodes to model the internal forces.

Reaction Normal to the Segment Face (Red)

Internal Moment Reaction (Blue)

Dummy shear boundary conditions, radially perpendicular Figure 7. Symmetrical Boundary Conditions to the joint face, were not needed to make the model stable since the soil springs stabilized the model. The dummy shear reaction would be used on one of the symmetrical planes in order to anchor the model and keep it from freely moving in the 3D environment during the finite element analysis of the structure.

Analysis Results

4.1 Half vs. Full Model


In order to determine that the ShorWall half model was working properly, the results of the half model were compared to that of the full model analyses. Figure 8 is a comparison between three models. The first model, PFM_2a_Base1, is a full model of the ShorWall without any joints or posttension cables; please refer to KCs work for additional information on the model. The PFM_2a3_v4_c45 is the analysis result for KCs full model which was edited by TNO DIANA. Half Model v1 (w/PT) is the half model results after all the appropriate changes were made from the full ShorWall model. The results of the half model are the same as the full model which proves that it was modeled correctly and that it can be used to analyze the structural behavior of the ShorWall.

Page 12 of 17 Document1

Draft 1.3

August 10, 2012

30

25

20 PFM_2a_Base1 Load (psi) 15 PFM_2a3_v4_c45 Half Model v1 (w/PT) 10

0 0 -1 -2 -3 -4 -5 -6 -7 Displacement (in) -8 -9 -10 -11 -12

Figure 8. Comparison of Full and Half Model Results

4.2 Tendon Sensibility Analysis Results


The ShorWall half model was analyzed to determine the systems sensibility to the pre-stressed cables in an attempt to determine how much the walls buckling resistance is influenced by the tendons. As it can be seen from Figure 9, the results from models v1 through v4 yielded the same curve for inward deflections of up to three inches. When looking at the data past the three inch deflection, it can be noticed that the third model with only vertical pre-stressed tendons and fifth model without prestressed cables had similar responses and failed sooner than the fourth model with only horizontal prestressed tendons. Thus the inclusion of only horizontal tendons is more beneficial in delaying failure of the wall rather than stiffening it. Therefore, inclusion of the tendons does not help to increase the buckling failure load or, in other words, raise the load-deflection curve. This means that for the same deflection, the stiffer model would be able to carry a higher load than the less stiff model.

Page 13 of 17 Document1

Draft 1.3

August 10, 2012

30

25

20

PFM_2a_Base1 Load (psi) 15 Half Model v1 (w/PT) Half Model v2 (vert PT only) 10 Half Model v3 (horiz PT only) Half Model v4 (no PT) 5

0 0 -1 -2 -3 -4 -5 -6 -7 Displacement (in) -8 -9 -10 -11 -12

Figure 9. Tendon Sensitivity Study Results

In the half model without the tendons, the pre-cast panels began to shift out of their original configuration under high inward deflection due to the lack of tendons. This could mean that the joint stiffness is higher when the pre-stressed tendons are used. The bonded tendon feature in DIANA which only allows the part of the cable in the joint interface to deform might be contributing to the higher joint stiffness. It should be noted that this is not a realistic representation since the post-tension cables will not be grouted to the panels thus allowing the whole cable to elongate and not just the section going through the joint. Thus it is recommended that the ShorWall be modeled without the pre-stressed cables in order to make the design more conservative. It is also recommended that the ShorWall design limits the deflection of the wall to less than one inch when under service load conditions to mitigate the mobilization of the surrounding soil which can cause adjacent foundations to fail. Consequently, the ShorWall needs to be designed as a rigid structure rather than flexible one.

4.3 Concrete Strength (fc) Sensitivity Analysis Results


The concrete strength sensitivity study was conducted to determine how much the strength of the concrete (fc) affects the buckling resistance strength of the ShorWall. The half model was tested with three different concrete strengths (4.0, 8.0, and 12.0 ksi) and with and without pre-stressed cables. These models used non-linear concrete properties with cracking. At the lower fc, the model

Page 14 of 17 Document1

Draft 1.3

August 10, 2012

experienced failure in buckling but as fc increased, the model was more prone to suddenly failing due to crack formation when steel tendons were included in the model. From Figure 10 it can be seen that increasing or decreasing the concrete strength by 4.0 ksi causes the load-displacement curve to significantly move up or down, respectively. At one and two inch deflections, model v6 with a concrete strength of 4.0 ksi had a 29% lower load capacity than model v5 (fc = 8.0 ksi) while model v7 (12.0 ksi) had a 23% higher load capacity. Thus, it can be safely stated that the buckling resistance of the ShorWall is highly dependent on the strength of the concrete.
30 Wall Failed After Crack Formation

25

20

Load (psi)

15

PFM_2a_Base1 Half Model v5 (f'c = 8.0ksi) Half Model v6 (f'c = 4.0ksi)

10

Half Model v7 (f'c = 12.0ksi) Half Model v8 (no PT, f'c = 8.0 ksi)

Half Model v9 (no PT, f'c = 4.0 ksi) Half Model v10 (no PT, f'c = 12.0 ksi)

0 0 -1 -2 -3 -4 -5 -6 -7 Displacement (in) -8 -9 -10 -11 -12

Figure 10. Concrete Strength Sensitivity Study Results

The models with and without the tendons generated the same load-deflection curve. However, interestingly enough, the models with the pre-stressed cables failed sooner than the models without the tendons when non-linear concrete properties were used. This phenomenon should be further investigated if the steel tendons are to be used in the design of the ShorWall. Also, models v5 and v8 had a load deflection curve slightly higher than the base model (PFM_2a_Base1) which can be attributed to rounding down the Youngs modulus from 5.098235E+09 psi to 5.0E+09 psi in the concrete properties of the base model.

Page 15 of 17 Document1

Draft 1.3

August 10, 2012

Conclusion

The DIANA half model provides the same solution as the full model and is adequate to be used in the design of the ShorWall. This will reduce the analysis time, providing quick answers to different scenarios. The following recommendations are based on the studies outlined above and should be taken into account when designing the ShorWall. Pre-stressed or post-tensioned tendons should not be used since they do not increase the buckling load, introduce high local stresses at the joints, and may make the wall joints stiffer. Plus, the load-displacement curves were the same for the models with and without the pre-stressed tendons. The ShorWall should be designed as a rigid wall deflecting less than one inch under service loads in order to eliminate destabilization of adjacent foundations. The wall thickness and/or the concrete strength should be increased to raise the loaddeflection curve thereby raising the buckling failure load.

Page 16 of 17 Document1

Draft 1.3

August 10, 2012

References

Cook, R. D., Malkus, D. S., Plesha, M. E., & Witt, R. J. (2007). Concepts and Applications of Finite Element Analysis (4th ed.). Singapore: John Wiley & Sons.

Page 17 of 17 Document1

S-ar putea să vă placă și