'FEA' vs. Seepage Tank Model

Seepage Analysis

FEA vs. Seepage Tank Model

Presented by Loai Naamani, Diala Turk, & Hasan Osman

June 17^th, 2002

Figure1: Drainage and Seepage

Figure2: Pyrex piezometer tubes

Figure3: Cutoff Dam model

Figure4: Foundation Plate

Figure5: Cutoff Wall model

Figure6: Pile model

Figure7: Dam showing patty sealing

Figure8: Experiment 1 – Fabricated

Figure9: Experiment 2 – Cutoff Wall

Figure10: Experiment 3 – Cutoff Pile/ Foundation Plate

Figure11: Piezometer readings

Figure12: Flow line visualization

Figure13: Measuring Hydraulic Conductivity

Figure14: Sources of Error

(Top)

Prelude

This project involves the experimental versus finite element modeling of a series of precisely defined groundwater seepage problem setup variations. Each model was first constructed in an experimental setup (using the ‘Seepage Tank’ available in the hydraulics lab), and then imported into ActiveSEEP (an upgrade of the AutoCAD port for 2-D groundwater seepage modeling software application originally developed for the ‘CVEV118/698 Computer Methods for Civil Engineering course’.)

The modeled setups were of a steady-state confined flow nature, where different combinations of impervious structures were used: a pile, a cut-off wall, and a sinking dam. Piezometer readings were taken at random points throughout the soil body. On the other hand, FEM grids of varying complexities were constructed on an adaptation of the experimental setups in ActiveSEEP (ActiveSEEP page). Grid lines were chosen to intersect at nodes corresponding to those at which piezometers were paced in the seepage tank, and head results were accordingly extracted and compared with the experimental ones. The data is documented, analyzed, graphed, and compared in multiple Excel workbooks that are innovatively linked to contain all necessary material: AutoCAD drawings, ActiveSEEP grid levels, LISA’s potential head contour plots, raw and filtered head data, and graphs used to visualize the deviation between the two seepage analysis methods.

(Top)

Introduction

The following set of experiments consists of typical setups that demonstrate real life situations that arise when dealing with water as it moves through a permeable medium. The trials described involve three main “engineered” elements that consist of a cutoff wall model, a dam model and a pile model. The main equipment used in all three runs is the Drainage and Seepage tank (see figure below). A stand supports the Drainage and Seepage tank, the sump tank, the centrifugal pump and the motor. A pump switch, which controls the pump, and hence the supply of water through the inlet hose to the tank from the sump tank, is connected to the stand. Water is returned to the sump tank by means of the adjustable overflows.

Theoretically, water is to be supplied through the inlet hose, but due to the high turbulence in the sump tank, water supply from an external hose was preferred in this experiment.

All three experiments consist of a sand bed of a fixed height, on which the permeability test was conducted and the hydraulic conductivity obtained, and for all three experiments, the boundaries of the problems are well defined, since they depend on the simple geometry of the models.

A detailed step-by-step description of each operation is given in the set-up of the first experiment under the “procedure” section. In the following experiments, recurring basic operations are not repeated.

The first set-up models a wooden dam embedded in sand, with piezometers penetrating the dam, to measure the heads at points under it due to the difference in water levels between the upstream and downstream sides of the tank.

The second set-up models a cutoff wall penetrating through the sand bed with a head difference resulting from a higher water level to the left side of the cutoff wall.

The third consists of a pile model placed next to a foundation pressure plate, with a head difference between the left and right sides of the system resources.

(Top)

Apparatus

The following main equipment was used in the experiment runs:

(Top)

Experimental Procedure

q Setup 2: Seepage around a dam

A frequently encountered type of discharge in hydraulic engineering is seepage around a dam. A segment of wood dam of rectangular cross section, with the base 22.5 cm above the bottom of the tank was designed and fabricated to serve in the experiment. The height of the dam is 30 cm upstream and 20 cm downstream and its base is a 40 x10 cm rectangle.

After the dam segment was fabricated in the shop labs to meet the pre specified model we designed, its base was perforated to allow for placement of piezometers inside the dam.

All piezometers used in the experiment had their ends glued with a piece of a pad tissue serving as a semi permeable membrane to prevent the upward movement of sand particles in the tubes. Three piezometers were placed inside the perforated holes, at random heights, and sealed with silicone at the contact area with the base. The dam was then painted with a water resistant varnish to provide certain impermeability.

Once the dam dried, it was positioned in the tank and fixed in place using patty seals to ensure that the contact areas between the dam and the tank were firmly sealed. Placing patty along the contact area between the glass and underneath the dam to make sure that no leakage will be allowed was a relatively difficult task since practically it was hard to manipulate the area under the dam with the dam placed inside the tank.

(Top)

Next, water was slowly poured in the tank through the inlet hose until the water level in the upstream region reached the overflow hose level previously adjusted to meet the desired upstream water level. The tank was then filled with sand to a depth of 26cm. Piezometers were then randomly placed along the length of the tank. It was important to adjust the sand levels after placement of the piezometers to the previously specified height.

Time was given for sand particles to consolidate and settle. It was specifically important to ensure that the area underneath the dam and on its sides was consolidated, so that it wouldn’t allow erosion due to turbulence near the edges of the dam. It took on average an hour and a half for total consolidation and settlement to take place.

q Setup 1 : Seepage underneath a cutoff wall

Seepage underneath a cutoff wall or a sheet pile in real practice is one of the seepage problems that are most common. Sheet pile walls are used to reduce seepage under all types of dams, sea walls, dividing walls and other similar structures.

An already existing vertical lateral pressure plate was used to model the sheet pile structure. The vertical plate is sealed with rubber on the sides, and is perforated along its length where piezometers are sealed with semi-permeable membranes. These piezometers measure the heads along the depth of the pressure plate.

The tank was filled with sand to a level of 25 cm above the bottom of the tank. The upstream overflow level was adjusted to a higher level than the downstream overflow (refer to AutoCAD drawings for numerical data), taking the datum to be the bottom of the tank. The vertical pressure plate was adjusted at approximately the middle of the tank. A clearance of about 8 cm was left between the vertical plate and the bottom of the tank. Contact between the cutoff and the tank walls was sealed with the aid of the rubber glued to the sides of the cutoff. Piezometers were randomly placed in the tank. Water was then slowly poured into the tank.

(Top)

q Experiment 3: Seepage underneath a cutoff pile and the foundation pressure plate

The object of the experiment is to demonstrate direct uplift pressures using the horizontal bottom of a submerged structure as the basis for the experiment.

The setup is basically the same as in the experiment with the dam, but consisting of a vertical rectangular wooden structure, 30cm in height and 10cm in width, with an additional element consisting of a simulated foundation of a structure. The foundation is modeled by means of a PVC sheet about 6 mm thick, 460 mm long and 150 mm wide, with both long edges covered with the elastic rubber packing. The horizontal plate has 5 standpipes fitted along the centerline.

The procedure followed is exactly the same as with the dam experiment.

(Top)

Complementary Tests

q Dye injection/tracer test

In order to visualize the flow lines in the experiment, we chose a simple experiment on which to conduct the tracing of flow lines (refer to figure under apparatus). Thus the cutoff pile, for its simple structure, was chosen to serve the purpose of the experiment.

After almost steady flow conditions were achieved and sand was allowed to settle and consolidate, pen ink, which served as the dye, was introduced by the means of a vertical syringe, at two different points along the surface of the sand. After some time, the flow line pattern is traced and it can be identified, as can be seen in the pictures.

(Top)

q Hydraulic Conductivity test

The soil chosen for the experiment consists of sieved sand available in the soil mechanics laboratory, it is a narrow range grained sand on which the permeability test was conducted and a hydraulic conductivity of 2.6 E –03 cm/sec obtained.

(Top)

Sources of Error

The tank system itself constitutes a major source of leakage, through the sump tank drain and the overflows. In fact, before any experiment was conducted, the tank was filled with water and kept for a period of 24 hours with all possible outlets closed, and it was found that the level of water in the tank decreased by 17.5 cm, which means that the system in itself is not.

Although care was taken in sealing the dam and pile model as efficiently as possible with the sides of the seepage tank by means of patty and rubber, leakage through the system could not be avoided, and it was apparent throughout the experiment that water was leaking through the sides of the dam and pile and accumulating between the glass membrane of the tank and the modeled structure.

The seepage rate equations usually apply for steady conditions. However, although this was assumed in the experiment, the fact that an external hose was used to supply water doesn’t guarantee that the rate of supply of water was equal to that of removal by the downstream overflow hose.

(Top)

ActiveSEEP

ActiveSEEP (ActiveSEEP page) is an AutoCAD 2002 port for 2-D groundwater seepage modeling powered by ‘LISA's finite element analysis engine (‘LISA’ page). It is meant to provide a highly flexible CAD environment for the user to define his seepage problem and the corresponding finite element grid/mesh. The accompanying ActiveSEEP Manual (available here), provides the necessary documentation along with a sample problem walkthrough.

Version 2.01.Beta of ActiveSEEP has been specially conceived for this project; it contains an additional module developed to manage (in run-time) the drawing and meshing (at different complexity levels) of the 3 problem setups described earlier and modeled using the seepage tank. The problems are automatically drawn with grid lines that intersect at the nodes corresponding to the piezometer tips in the seepage tank model, at which head readings were taken. The other grid complexity alternatives (3 for each problem setup) are meant to provide finer FEA solutions.

This version of ActiveSEEP also provides links to the corresponding Excel files (and comprehensive report) in which the documentation, analysis, and comparison of FEA vs. seepage tank model results are presented. (For the links to function properly, you must abide by the installation instructions provided in ‘Readme - ActiveSEEP.txt’ that guarantee the integrity of the relevant folders & files structure remains functional.)

Besides the additional module, ActiveSEEP Version 2.01.Beta includes many bug-fixes and refinements over earlier version, namely in the "Auto Mesh" tool(s).

(Top)

FEA Results from ActiveSEEP

As mentioned earlier, all 3 problem setups are already embedded in ActiveSEEP and can be regenerated in run-time with the click of a button (ActiveSEEP ---> FEM vs. Seepage Tank ---> Cut-Off Wall ---> Draw Setup). This draws the problem, and then draws the essential problem and impervious structure boundary grid lines. That is, the minimum amount of lines needed to correctly define the problem. Those ‘Critical Grid Lines’ embody the coarsest grid possible to correctly define a given problem.

After drawing the problems, a grid/mesh is required that places nodes or junctions at the points corresponding to those at which piezometer readings were taken in the seepage tank model. This is what the ‘Grid Complexity: Minimal’ menu item does: the minimum amount of grid lines to define all piezometer reading points, for it’s the FEA results at those points that will enter the analysis and consequent comparison. Menu items ‘Grid Complexity: Complex’ and ‘Grid Complexity: Maximum’ represent 2 other levels of grid complexity, around double and triple those needed for the minimum respectively. Those finer grids are meant to provide added accuracy, if desired.

Once the problem has been correctly meshed, ActiveSEEP is ready to have LISA run the analysis. However, the analysis for all problems at all complexity levels has been already done and is instantly accessible from the ‘View FEM vs. Tank Model results in Excel’ submenu item.

The documentation, analysis, and comparison of the FEA vs. seepage tank results is presented in a shrink-wrapped manner through 3 problem-specific Excel workbooks, each containing a diverse set of embedded object and drawings. Just as all the documents presented contain internal links (mostly hyperlinks) between the different Microsoft Office and AutoCAD files to make the process of navigating through this 'multifaceted' endeavor easy and efficient, the 3 Excel files contain everything one might need to keep track of the different aspects of the analysis; be that from embedded AutoCAD drawings to LISA run results and contour plots. Hence, the reader is asked to revert to the accompanying compact disk, and enjoy the detailed analysis therein. The following is an overview of the spreadsheet contents.

(Top)

Results Analysis & Comparison via Excel

The three spreadsheets are of an identical structure and format. They can be accessed at:

m Setup 1 - Cut-Off Wall.xls

m Setup 2 - Sinking Dam.xls (upon request)

m Setup 3 - Pile + Dam Extension.xls (upon request)

Each spreadsheet contains 5 worksheets:

m Seepage Tank Model (CAD) – An AutoCAD sketch of the seepage tank model of this problem setup.

m Readings vs. FEM Results – Piezometer head readings, FEA results at the corresponding points for all complexity levels, coordinates and node numbers of corresponding points, comparison of Tank vs. FEA (finest), Minimal vs. Complex, and Complex vs. Finest results with complete statistical summaries, 2 multi-series charts showing % error between results.

m ActiveSEEP (Complexity Min.) – LISA run of ActiveSEEP version of setups; contains complete nodal results (node number, head, Vx, and Vy), and an ActiveSEEP-produced CAD drawing of problem with minimum complexity grid.

m ActiveSEEP (Complexity Med.) – LISA run of ActiveSEEP version of setups; contains complete nodal results (node number, head, Vx, and Vy), and an ActiveSEEP-produced CAD drawing of problem with more complex grid, also contains the LISA-generated contour plots of the potential heads.

m ActiveSEEP (Complexity Max.) – LISA run of ActiveSEEP version of setups; contains complete nodal results (node number, head, Vx, and Vy), and an ActiveSEEP-produced CAD drawing of problem finest grid.

(Top)

Epilogue

It showed to be a fact that more stringent control over experiment conditions is a must. A better seal and fitting of the impervious structures between the tank walls would have definitely better prevented cross-leakage and dampened the variation in head measurements/calculations, namely in the vicinity of the structure where such variation was most pronounced.

Furthermore, after extensive testing of different meshes with varying complexity levels and critical regions, a better appreciation for the ‘art of meshing’ has been developed. A more grounded understanding of how economical we can get when designing a grid (perspective of computer resources), which locations in the modeled body are more critical than others , what is results ‘convergence’, etc…

Finally, the experience of designing, building, and testing a given scenario experimentally and then using an in-house software tool to model the same system and reach comparable results is very gratifying. One gets to touch on the overall and interrelated conception of governing theory, different solution methodology, experimental means and skills, and the potential of information technology to correctly depict real-life scenarios.

(Top)

References

§ Armfield Limited^® Drainage and Seepage Tank - Model S1 Instruction manual (1993)

§ U.S. Army Corps of Engineers (September 1986), “Seepage Analysis and Control for Dams”, Washington

§ Gibb, John W. and Kramer, Bill (1999). “AutoCAD VBA Programming Tools and Techniques”; Miller Freeman Books.

§ Clark, Jeffrey E. (2002). “VBA FOR AutoCAD 2002: Writing AutoCAD Macros”; Prentice Hall

§ MSDN Library Visual Studio 6.0a

§ LISA Help Documentation

§ Das, Braja M. (1998). “Principles of Geotechnical Engineering” 4^th Edition; PWS Publishing Company.

§ Al-Khafaji, Amir Wadi and Tooley John R. (1986). “Numerical Methods in Engineering Practice”; HOLT, RINEHART AND WINSTON.

§ “Introduction to Groundwater Modeling” (from FEA Library)

§ “Seepage, Drainage and Flow Nets” (from FEA Library)

(Top)