Prediction of seepage from earthen dams using data mining methods

Number of pages: 136 File Format: word File Code: 31454
Year: 2014 University Degree: Master's degree Category: Civil Engineering
  • Part of the Content
  • Contents & Resources
  • Summary of Prediction of seepage from earthen dams using data mining methods

    Dissertation for Master's Degree in Civil Engineering

    Soil and Foundation Mechanics

    Abstract

    Dams are always considered as infrastructure structures and have vital value. In the past, the creation of dams was mainly for the purposes of providing drinking water and irrigation of agricultural fields, but today it has been developed more due to the need for hydroelectric energy and other purposes. The estimate of 20 billion cubic meters of fresh water in the world is a proof of the importance of dam construction in today's world.  Therefore, checking and preventing damage to dams is of particular importance. Although in the past, the overtopping phenomenon was the first reason for the destruction of dams, but today, with the increase of the flood design period, the main problem that has attracted the attention of engineers is the problem of seepage. Existence of seepage in earthen dams is unavoidable, but if there are suitable conditions for soil erosion, it causes washing away of susceptible points, and if necessary measures are not taken at the beginning of erosion, it will lead to dam destruction. Basically, the occurrence of seepage in earthen dams is inevitable. However, seepage must be controlled so that it cannot harm the stability and safety of the dam during the 50-100 years of operation of the dam. Despite all the advances made in the science of geotechnical engineering, the problem of seepage is still the main problem that occurs in dams.

    In this research, an attempt was made to predict seepage from the body of the "Star Khan" earthen dam by using artificial neural network as one of the most powerful and famous data mining methods. In order to achieve this goal, a data set including 1684 piezometric data was used. The data set was divided into two sections, training and validation, with a ratio of 80 to 20. The application of appropriate and practical statistical parameters showed that the presented network is well trained and has a high capability in predicting seepage phenomenon.

    Comprehensive examination of dam failure statistics, various causes of earth dam failure and recognition of seepage phenomenon as the most important causes of earth dam failure are other important parts of this research.

    Key words: dam construction statistics, earth dams, seepage, causes of dam failure, data mining, artificial neural network

     

     

     

    Chapter 1: Overview

     

     

     

    1-1-  

    The country of Iran is located on the dry belt of the planet. The average rainfall in Iran is about one-third of the world's rainfall and less than one-half of the average rainfall in Asia. Therefore, the importance of planning and managing the use of existing water resources is considered vital. Therefore, the climatic conditions of the country and its need to build water storage structures have put the construction of dams on the agenda of the planners, which provide the possibility of more use of river water as surface water containment structures and flood control. Curbing floods and running water with the help of building a dam is one of the infrastructure issues in the growth and development of any country, including Iran.

    In the past, building a dam was mainly for the purposes of providing drinking water and irrigation of agricultural fields, but today it has been developed more due to the need for hydroelectric energy and other purposes. The estimate of 20 billion cubic meters of fresh water in the world is a proof of the importance of dam construction in today's world. Another important goal of dam construction is the improvement and development of the irrigation and agricultural network of the downstream lands. In countries like Iran, where the temporal and spatial distribution of rainfall is inappropriate and rainfall occurs in seasons when there is less need for water or most of the rainfall is concentrated in some areas, the only solution to deal with this issue is to build a dam, and this is inevitable especially in countries that rely on agriculture. Dam construction brings business and national income. At present, irrigation networks and agricultural water supply in Iran have led to the development, improvement and economic prosperity of the regions. Another major purpose of dam construction is the use of electricity. The use of this source, which is the cheapest type of energy in most countries of the world, has different purposes depending on the needs and construction characteristics. Nowadays, the construction of a dam with the aim of producing hydroelectric power is a common thing, and developed and even developing countries take full advantage of this existing potential.

    If the construction conditions are not paid attention to and the studies are insufficient, the risk of failure threatens the dam.. The statistical study of damage in damaged dams, due to the existence of similarities in the conditions, provides the possibility of providing suitable solutions in the design of dams by engineers. Damage in dams can be seen in different ways, the most common cause of dam failure, especially in recent years, is erosion due to seepage or seepage. Basically, the occurrence of seepage in earthen dams is inevitable. However, seepage must be controlled so that it cannot harm the stability and safety of the dam during the 50-100 years of operation of the dam. Despite all the advances made in the science of geotechnical engineering, the problem of seepage is still the main problem that occurs in dams.

    1-2- Statement of the problem

    The climatic conditions of the country and its need to build water storage structures have put the construction of earthen dams on the agenda of planners, which provide the possibility of more use of river water as surface water containment and flood control structures. According to the studies that are usually done before the construction of the dam, it is not always possible to accurately predict the hydraulic behavior of the dam body or its adjacent geological formations. Therefore, the possibility of seepage after the construction of the dam seems almost certain. The intensity of seepage is acceptable in many cases as long as the safety of the dam is not compromised. From the point of view of safety, it is important to investigate the risks caused by seepage and penetration due to the inherent complexity of their characteristics. Many reservoirs of dams built in the world have seepage. This seepage may occur from the geological formations of the construction site or the foundation of the dam or from the body of the dam. Among the consequences of poor seepage, we can mention economic issues, high hydraulic gradient that leads to phenomena such as seepage or boiling, and an increase in pore pressure that leads to a reduction in effective stress. Therefore, one of the most important points in the study stages, during the implementation operations and after the construction of earthen dams, is the issue of seepage from the foundation and body of the dam, which has been a problem facing the designers of the dams. Therefore, it is necessary to accurately calculate the amount of seepage from the body and foundation of the dam and examine the methods of controlling or reducing it, technically and economically, in order to prevent life and financial risks.

    The accurate estimation of seepage from the body of earthen dams is an important challenge in the design of these huge structures. The use of instrumentation may provide an accurate estimate of this phenomenon to some extent, but the upcoming problems such as the failure of the instruments due to time, expenditure of money and manpower for reading, etc., have made this method difficult. It is a common thing to use the analytical solution methods proposed by different researchers to evaluate the amount of seepage from the body of earthen dams located on impermeable bed due to their ease of use. However, these analytical methods use assumptions to simplify the construction of equations that may lead to large errors. Therefore, according to the contents stated in the previous sections, the aim of this study is to provide an artificial neural network model [1] to more accurately predict the amount of seepage from the body of earthen dams and eliminate the above problems. Therefore, it is tried to predict the phenomenon of seepage in earthen dams based on the data of precise instruments of a specific dam and the use of data mining methods.

    1-3- Importance and necessity of research

    The country of Iran is located on the dry belt of the earth. The average rainfall in Iran is about one-third of the world's rainfall and less than one-half of the average rainfall in Asia. Therefore, the importance of planning and managing the use of existing water resources is considered vital. Therefore, the climatic conditions of the country and its need to build water storage structures have put the construction of dams on the agenda of the planners, which provide the possibility of more use of river water as surface water containment structures and flood control. Curbing floods and running water with the help of building a dam is one of the infrastructure issues in the growth and development of any country, including Iran.

    In the past, building a dam was mainly for the purposes of providing drinking water and irrigation of agricultural fields, but today it has been developed more due to the need for hydroelectric energy and other purposes. The estimate of 20 billion cubic meters of fresh water in the world is a proof of the importance of dam construction in today's world. Another important goal of dam construction is the improvement and development of the irrigation and agricultural network of the downstream lands.

  • Contents & References of Prediction of seepage from earthen dams using data mining methods

    List:

    Abstract

    Chapter One: General 1

    1-1- Introduction. 2

    1-2- statement of the problem. 3

    1-3- The importance and necessity of research. 5

    1-4- research variables. 8

    1-5- research variables. 8

    1-5-1- The main (general) goal of the research. 8

    1-5-2- Sub-goals (specific) 8

    1-6- Research questions. 9

    1-6-1- The main research question: 9

    1-6-2- Sub-questions (special) 9

    1-7- Research hypotheses. 9

    1-8- Definition of technical and specialized words and terms (conceptually and operationally) 10

    1-8-1- Conceptual definitions. 10

    1-8-2- operational definitions. 11

    1-9- research limitations. 11

    Chapter Two: Theoretical foundations and research background 12

    2-1- The theory of seepage phenomenon. 13

    2-1-1- Introduction. 13

    2-1-2- Flow in porous media. 13

    2-1-3- Inhomogeneous isotropic steady state seepage. 17

    2-1-4- Steady state, non-isotropic and heterogeneous seepage. 18

    2-1-5- one-dimensional flow. 19

    2-1-6- Darcy's law in unsaturated soils. 21

    2-1-7- Permeability coefficient of unsaturated soils. 23

    2-1-8- Boundary conditions in seepage analysis problems. 26

    permeable boundary. 27

    2-1-8-1- Inputs and outputs 27

    2-1-8-2- Seepage level. 28

    2-1-8-3- seepage line. 28

    2-2- Dam construction statistics in different countries. 28

    2-2-1- Dam failure 31

    2-2-2- Statistics of dam failure 35

    2-2-3- Statistics of various causes of dam failure 41

    2-2-4- Causes of increased seepage. 46

    2-2-5- Permitted and acceptable volume of seepage. 48

    2-2-6- Consequences of improper leakage. 51

    2-3- Recent disputes in the field of seepage. 54

    2-3-1- The study of Ersain (2006) 54

    2-3-2- The study of Miao et al. (2012) 56

    2-3-3- The study of Noorani et al. (2012) 56

    2-3-4- The study of Pourkrimi et al. (2013) 57

    2-3-5- Kamanbehdast and Delvari's study 58

    3-1- Artificial neural networks

    3-1-2- Neural network model 64

    3-1-2- Neron. 64

    3-1-2- Multi-neural layers 67

    3-1-3- Multi-layer networks 67

    3-1-3- Stimulus functions (transformation function) 69

    3-1-4- Network training and parameter setting 72

    3-2- Adaptive neural-fuzzy inference system (ANFIS) 3-2-1- The history of fuzzy logic. 74- 3-2- Types of fuzzy systems. 76- 3-2- The structure of fuzzy systems. 82

    3-3-Introduction

    3-3-Project location 83

    3-3-Materials used in the dam body. 85

    3-3-4-1- Materials used in the seal core. 85

    3-3-4-2- Materials used in filter layers. 85

    3-3-4-3- Materials used in drainage layers. 85

    3-3-4-4- Materials used in the shells of pebbles. 86

    3-3-4-5- The materials used in the protective layer of the mirage slopes and abutments of the dam. 86

    3-3-5- Geological and geotechnical features of Sattar Khan Dam construction. 87

    3-3-5-1- Geology. 87

    3-3-5-2- Geotechnics of dam construction. 88

    3-3-5-3- Support stones and alluvium. 89

    3-3-5-4- Pi. 89

    3-4- Second stage geotechnical studies. 90

    3-5- Sealing of the dam by concrete sealing curtains. 92

    3-6- Tooling. 93

    3-6-1- Pipe opener piezometers. 95

    3-6-2- Vibrating wire piezometers. 95

    3-7- Check the data of precision instruments in the body of Sattar Khan Dam. 98

    Chapter 4: Research results 99

    4-1- Introduction. 100

    4-2- Data set 100

    4-3- Structure of the proposed neural network model. 102

    4-4- Evaluation and comparison of the performance of the proposed model. 106

    4-5- Summary and conclusion. 118

    Chapter 5: discussion, conclusions and suggestions 119

    5-1- Introduction. 120

    5-2- Results. 120

    5-3- Suggestions 121

     

    Source:

     

    [1]Fredlund, D. G. Rahardjo, H. (1993) "Soil Mechanics for Unsaturated Soils," Newyork, John Wiley and Sons, Chap. 7, P.P.

    [2]Das, B. M. (2001) "Principal of Geotechnical Engineering," Thomson-Engineering, Fifth Edition.

    [3]Fredlund, D. G. Rahardjo, H. (1993) "Soil Mechanics for Unsaturated Soils," Newyork, John Wiley and Sons, Chap. 5, P.P. 110-117. [4] US Army Corps of Engineers. (1986) "Engineering and Design Seepage Analysis and Control for Dams," chapter 4, seepage principles, EM-1110-2-190,.

    [5]International Commission on Large Dams, (http://www.icold-cigb.org/)

    [6]Middlebrooks, J. L, (1983). Report of the dam failure. Engineering Geology, 24(1), 239-256.

    [7] Gruner, C. S, (1994). Seepage in earthfill dams. Journal of Geotechnical Engineering, 109(7), 946-960.

    [8] Takase, J. M, (1996). The failure of Teton dam. Engineering Geology, 12(3), 173-205.

    [9] Babb, C. L., (1987). Embankment dam instrumentation manual. Published by Pearson Education, ISBN 81-7808-300-0.

    [10]Kawasaki, H. et al., (2010) "Case study of a Behavior Monitoring in a Fill Dam for Accurate and Long Term Measurement," Commission Internationale Des Grands Barrages, Hanoi,, May.

    [11] Ersayin, D, (2006). Studying seepage in a body of earth-fill dam by (Artificial Neural Networks, Doctoral dissertation, Master Thesis, ?zmir Institute of Technology.

    [12]Miao, X. Y., Chu, J. K., Qiao, J., and Zhang, L. H, (2012). Predicting seepage of earth dams using neural network and genetic algorithm. Advanced Materials Research, 403, 3081-3085.

    [13] Nourani, V., and Aminfar, M. H., (2012). Integrated ANN model for earthfill dams seepage analysis, 1(2), 22 pages. [14] Poorkarimi, S., MolaAbasi, H., Kordnaeij, A dam using Group Method of Data Handling (GMDH) type neural network. Scientific Research and Essays Journal, 8(3) 120-127. [15] Kamanbedast, A., (2013).

    [16] Patterson, D. W, (1996). Artificial neural networks: theory and applications. Prentice Hall, 477 pages.

    ]17[Minhaj.M.B., "Fundamentals of Neural Networks", Amirkabir University Publications, 1377.

     

    [18]Zadeh, L.A, "Fuzzy Sets," Information and Control, vol. 8, pp. 338-353, 1965.

    [19]Nauck, D., et al., Foundations of Neuro-Fuzzy Systems: J. Wiley & Sons, 1997.

    [20]Abraham, A, "Neuro Fuzzy Systems: State-of-the-art Modeling Techniques," presented at the Proceedings of the 6th International Work Conference on Artificial and Natural Neural Networks Lecture Notes in Computer Science, 2001.

    [21]Rutkowski, L. and Cpalka, K, "Flexible Neuro-Fuzzy Systems, Vol. 14, No. 3, May." IEEE Transactions on Neural Networks, vol. 14, pp. 554-574, 2003.

    [22] Bartholomew, C. L., B. C. Murray and D. L. Goins, “Embankment Dam Instrumentation Manual”, United States Department of the Interior, Bureau of Reclamation, Washington D. C., (1987)

    [23] Fell, R., “Measurement of positive pore pressure” , Geotechnical Field Instrumentation, Extension Course, Australian Geomechanics Society, Melbourne, (1987)

    [24] Hanna. T. H., "Field Instrumentation in Geotechnical Engineering", Tran Tech. Publication. , (1985)

    [25] Kohonen, T, "An Introduction to Neural Computing," presented at the Proc. IEEE First Int. Conf. on  Neural Networks, 1988.

    [26] Paulin, F. and Santhakumaran, A, "Classification of breast cancer by comparing back propagation training algorithms," International Journal on Computer Science and Engineering, vol. 3, pp. 327-332, 2011. [27] Wang, C.M. and Huang, Y.

Prediction of seepage from earthen dams using data mining methods