COST FUNCTION BASED ALTERNATIVE SITE EVALUATION TOOL
FOR RIVER DIVERSION STRUCTURES IN AMHARA REGION, ETHIOPIA
Bahir Dar Institute of Technology
Faculty of Civil and Water Resource Engineering
School of Research and Post Graduate Studies
Hydraulic Engineering Post Graduate Program
December, 2015 COST FUNCTION BASED ALTERNATIVE SITE EVALUATION TOOL FOR RIVER DIVERSION STRUCTURES IN AMHARA REGION, ETHIOPIA
By
EWUNETU TARKEGN
A Thesis work in Partial Fulfillment of the Requirements for the Degree of Master of Science in Hydraulic Engineering Submitted to the Faculty of Civil and Water Resources Engineering, Bahir Dar Institute of Technology, Bahir Dar University
I, the undersigned, declare that the thesis comprises my own work. In compliance with internationally accepted practices, I have dually acknowledged and refereed all materials used in this work. I understand that non-adherence to the principles of academic honesty and integrity, misrepresentation/ fabrication of any idea/data/fact/source will constitute sufficient ground for disciplinary action by the university and can also evoke penal action from the sources which have not been properly cited or acknowledged.
ii EWUNETU TARKEGN
APPROVAL SHEET
Approved by Board of Examiners
iii DEDICATION
I am glad to dedicate this thesis to my precious Grand Father, Chekol Mersha, who brought me up to attend School during my early stage of boyhood and for all his support, advice and encouragement.
iv ABSTRACT
The usual method of river diversion irrigation project development involves construction of headwork and main canals. Given the fact that the large portion of total cost of diversion works is shared by the two components and the topographic layout usually allows for a choice of several alternative sites from which attention should be paid to the optimum layout design of the diversion works. With regard to evaluating alternative sites, the cost estimation of river diversion structures plays an important role. The total cost can be empirically correlated with sensitive (statistically significant) design variables of headwork and main canal. The objective of this study was to develop quantitative planning tool for river diversion structures. To achieve this objective about 35 diversion structures were collected and from which the main design variables and their quantity of works have been extracted. Beside to this, the unit rate data were extracted from more than 70 number of relevant construction agreement documents in the entire study area.
With the help of multiple regression analysis technique, an empirical cost function was developed for Amhara Regional state based upon the cost estimates and the main design variables which are considered as decision variables for predicting the cost. The cost estimates were obtained using combination of the unit rate trend function and the respective quantities of engineering works. The unit rate trend function was developed from the unit rate data for the commonly adopted construction engineering work items. Moreover, the cost function parameters were defined for the seven clustered study areas which have rocky and alluvial foundation conditions.
The overall validation results show that there is an agreement between the predicted and observed costs with an R2 ranging 0.813 to 0.908 and 0.97 to 0.99 minimum and maximum values for rocky and alluvial foundations respectively. The new developed cost function can be applied to predict the cost for planning and evaluation in river diversion site selection procedure. Further, it can improve the existing qualitative based site selection approaches. Therefore, having reconnaissance or desk work survey data and some hydrologic data, we can apply the cost function for the desired spatial zone to identify the least cost feasible river diversion site among several alternatives.
v ACKNOWLEDGEMENT
First of all, I would like to thank ‘Almighty God’ who blessed me to accomplish this research work successfully. I am glad to express my sincere gratitude to my research advisor Mr.Fasikaw Atanaw for all his willingness and devotion to advise this research. And then, I would like to acknowledge my co-Advisor Mr.Mengiste Abate who also guided my work to complete successful.
In addition, I thank the Federal Democratic Republic of Ethiopia Roads Authority for granting me this MSc study opportunity. Next to this my heartfelt thanks goes to the data providers ANRS Water Resource Development Bureau, Amhara Design and Supervision Works Enterprise, Amhara Water Works Construction Enterprise and Organization for Rehabilitation and Development in Amhara (ORDA).
vi TABLE OF CONTENTS DECLARATION ................................................................................................................ii DEDICATION ..................................................................................................................iv ABSTRACT ..................................................................................................................... v ACKNOWLEDGEMENT ..................................................................................................vi TABLE OF CONTENTS ................................................................................................. vii LIST OF TABLES ............................................................................................................ix LIST OF FIGURES .......................................................................................................... x NOMENCLATURE ........................................................................................................ xiv 1 INTRODUCTION ...................................................................................................... 1 Background ........................................................................................................ 1
Problem Statement ............................................................................................ 2
Research Questions........................................................................................... 3
Research Objectives .......................................................................................... 3
Research Significance ....................................................................................... 3
Scope of Research............................................................................................. 4
Research Outline ............................................................................................... 4
2 LITERATURE REVIEW ............................................................................................ 6
General .............................................................................................................. 6
Headwork and Main Canal ................................................................................. 7
Existing Approaches on Diversion Site Selection ............................................. 10
Related Past Studies ........................................................................................ 11
Requirement of Statistical Techniques ............................................................. 12
3 MATERIALS AND METHODS ................................................................................ 16
Study Area Description .................................................................................... 16
Table 3-2 Summary table of common item of engineering works .................................. 20
Table 5-1 CT-MLR analysis result with sort of design variables for rocky class diversions in Awi &W/Gojjam area of year 2015............................................................................. 47
Table 5-2 CT- MLR analysis result with sort of design variables for alluvial class diversions in Awi &W/Gojjam area of year 2015............................................................................. 48
Table 5-3 CT- MLR analysis result with combined design variables for rocky class diversions in Awi &W/Gojjam area of year 2015 ........................................................... 50
Table 5-4 CT- MLR analysis result with combined design variables for alluvial class diversions in Awi &W/Gojjam area of year 2015 ........................................................... 51
Table 5-5 Summary of cost function parameters for calibrated equations..................... 54
Table 8-1 Summary of table for sample diversion weirs used in the research .............. 64
ix LIST OF FIGURES
Figure 2-1 River diversion structure components (Arora, 2002) ...................................... 7
Figure 3-1 Study Area Map ........................................................................................... 17
Figure 4-1Schematic layout of river diversion structure................................................. 24
Figure 4-2 Schematic river cross section with left side command peak field level (Baban, 1995) ............................................................................................................................. 25
Figure 5-1 Earth work unit rate trend function for Awi&W/Gojjam area ........................ 35
Figure 5-2 Concrete and masonry work unit rate trend function for Awi&W/Gojjam area ...................................................................................................................................... 36
Figure 5-3 Finishing and bar work unit rate trend function for Awi&W/Gojjam area ...... 37
Figure 5-4 Matrix plot showing trend of canal section area against its design variables 38
Figure 5-5 Curve fitting of canal section area (Ax) against canal conveyance (K) ....... 39
Figure 5-6 Summary of sampled irrigation design Duty................................................. 40
Figure 5-7 Line plot showing Irrigation area against design discharge .......................... 40
Figure 5-8 Summary of diversion weirs heights ............................................................ 41
Figure 5-9 summary of diversion weirs length ............................................................... 42
Figure 5-10 Summary of diversion weir overturning factor of safety ............................. 43
Figure 5-13 Summary of stable weir body (H/B) ratio.................................................... 45
Figure 5-14 Summary of stable weir body (T/B) ratio .................................................... 45
x Figure 5-15 Matrix plot showing trend of Weir body stable section area against relevant design variables ............................................................................................................ 46
Figure 5-16 Curve fitting of weir body stable section area (A) against height (H) .......... 46
Figure 5-17 Residual frequency histogram of CT-MLR analysis result for rocky class diversions in Awi &W/Gojjam area of year 2015 ........................................................... 48
Figure 5-18 Residual frequency histogram of CT- MLR analysis result for alluvial class diversions in Awi &W/Gojjam area of year 2015 ........................................................... 49
Figure 5-19 Residual frequency histogram of CT- MLR analysis result for rocky class diversions in Awi &W/Gojjam area of year 2015 ........................................................... 51
Figure 5-20 Residual frequency histogram of CT- MLR analysis result for alluvial class diversions in Awi &W/Gojjam area of year 2015 ........................................................... 52
Figure 5-21 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in Awi & W/Gojjam spatial zone ............................................................................................... 57
Figure 5-22 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in N/Gondar spatial zone................................................................................................... 57
Figure 5-23 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in S/Wollo &Oromiya spatial zone ..................................................................................... 57
Figure 8-1 Earth work unit rate trend function for N/Gondar spatial Zone .................... 65
Figure 8-2 Concrete and masonry works unit rate trend function for N/Gondar spatial Zone ...................................................................................................................................... 65
Figure 8-3 Bar and finishing works unit rate trend function for N/Gondar spatial Zone . 66
Figure 8-4 Earth work unit rate trend function for S/Gondar spatial Zone ..................... 66
Figure 8-5 Concrete and masonry works unit rate trend function for S/Gondar spatial Zone ...................................................................................................................................... 67
Figure 8-6 Bar and finishing works unit rate trend function for S/Gondar spatial Zone . 67
Figure 8-7 Earth work unit rate trend function for E/Gojjam spatial Zone ...................... 68
Figure 8-8 Concrete and masonry works unit rate trend function for E/Gojjam spatial Zone ...................................................................................................................................... 68
xi Figure 8-9 Bar and finishing works unit rate trend function for E/Gojjam spatial Zone . 69
Figure 8-10 Earth work unit rate trend function for N/Wollo&waghimira spatial Zone ... 69
Figure 8-11 Concrete and masonry works unit rate trend function for N/Wollo&waghimira spatial Zone................................................................................................................... 70
Figure 8-12 Bar and finishing works unit rate trend function for N/Wollo&waghimira spatial Zone .............................................................................................................................. 70
Figure 8-13 Earth work unit rate trend function for S/Wollo&Oromiya spatial Zone ...... 71
Figure 8-14 Concrete and masonry works unit rate trend function for S/Wollo&Oromiya spatial Zone................................................................................................................... 71
Figure 8-15 Bar and finishing works unit rate trend function for S/Wollo&Oromiya spatial Zone .............................................................................................................................. 72
Figure 8-16 Earth work unit rate trend function for N/Shewa spatial Zone .................... 72
Figure 8-17 Concrete and masonry works unit rate trend function for N/Shewa spatial Zone .............................................................................................................................. 73
Figure 8-18 Bar and finishing works unit rate trend function for N/Shewa spatial Zone 73
Figure 8-19 summary of main canal (b/d) ratio ............................................................. 74
Figure 8-20 Matrix plot showing trend of wing wall stable section area against relevant design variables ............................................................................................................ 74
Figure 8-21 summary of masonry wing wall (b/d) ratio .................................................. 75
Figure 8-22 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in S/Gondar spatial zone ................................................................................................... 75
Figure 8-23 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in E/Gojjam spatial zone ................................................................................................... 76
Figure 8-24 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in N/Wollo& Waghimira spatial zone ................................................................................. 76
Figure 8-25 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in N/shewa spatial zone .................................................................................................... 77
xii Figure 8-26 graphical summery of ratio of apron thickness (t a) to weir height (Hb) for alluvial foundation diversions ........................................................................................ 77
Figure 8-27 graphical summery of ratio of d/s cutoff depth (d 2) to weir height (Hb) for alluvial foundation diversions ........................................................................................ 78
Figure 8-28 graphical summery of ratio of cutoff depths (d2) to (d1) for alluvial foundation diversions ...................................................................................................................... 78
Figure 8-29 graphical summery of ratio of weir height (Hb) to apron length (La) for alluvial foundation diversion ...................................................................................................... 79
xii i NOMENCLATURE
ADB - African Development Bank,
ADF - African Development Fund,
AMSL- Above Mean Sea Level,
ANRS - Amhara National Regional State,
CT- The combined cost estimates of headwork and main canal expressed per 1000 birr,
DSS - Decision Support System,
ESRDF- Ethiopian Social Rehabilitation and Development Fund,
ETB- Ethiopian Birr; Currency,
FAO - Food and Agriculture Organization,
IFAD - International Fund for Agricultural Development,
L/s/ha - liter per second per hectare unit,
IWMI - International Water Management Institute,
MDG - Millennium Development Goal,
MLR - Multiple Linear Regressions,
MoWE - Ministry of Water and Energy,
MoWR - Ministry of Water Resources,
ORDA - Organization for Rehabilitation and Development in Amhara,
Irrigation had been practiced in Ethiopia since ancient times by producing subsistence food crops. Eighty years back, in 1930’s the first modern irrigation development was started along the Awash River by local and international entrepreneurs. In that time, the study and design of irrigation schemes was practiced with help of foreign experts and limited local expert. But later on, (2002) gradual local capacity building in field of water resource development, planning, the development of guidelines and procedures was provided with help of international agencies like FAO (Continental Consultants, 2002b). Besides to this, other international financing agency like FAO, UNDP, IFAD, WB and ADB activated the water sector by enrolling and assigning local and international consultants. However, materials prepared by these bodies were found less reflective on the local conditions (Continental Consultants, 2002b).
Since then, the government of Ethiopia has given much emphasis for the upgrading of traditional irrigation systems and implementation of small and medium scale irrigation projects. This is to secure food self-sufficiency for alarmingly increasing population and to strengthen the agricultural based national economy. In this time, the Government has been striving to enhance the development of medium and mega scale projects.
Because of the availability of Perennial River flow, topographic suitability and its low investment cost , river diversion water resource projects is widely favored (Awulachew et al., 2007). Before the addition of 493,000 hectares in a year 2010, the per capita irrigated area only reached 45 m2 per head by the year 2015. This figure was small compared to 450m2 global value. The country total irrigation development was approximated 7.34% out of the overall development potential of 3.7 million ha (MoWE, 2013). Despite potentials and demands, the irrigation development is still not satisfactory. An attention was noticed to enhance water resources for irrigation development and the country has been increasingly investing in the sector (Awulachew et al., 2007). So, we have ample of assignment on development of irrigation potential using river diversion structures.
Site selection for river diversion is one of the main components that influence the overall project feasibility. Hence, it should be well quantified and optimized for the potential alternatives.
1 Existing approaches of river diversion structure site selection procedure is usually conducted on the base of qualitative ideal site check list and limited cost analysis for predefined feasible site among candidates (Arora, 2002; Baban, 1995).
However, for planning and management of a water resources system, decision support mathematical models can be employed to provide quantitative information to the decision makers and able to make a better selection from alternative measures (Loucks, Van Beek, Stedinger, Dijkman, & Villars, 2005). These Mathematical Models are often the simplified representation of the real system to capture the key characteristics in an approximate mathematical framework (Saliha, 2012).
To come up with filling the above gap of knowledge, this research was conducted to develop new quantitative planning model for evaluating alternatives and selecting the most feasible diversion site in Amhara Region (Ethiopia) so that the present day river diversion design and site selection procedure can be improved.
Problem Statement
Engineers usually plan and design a water resources system by making compromises and negotiation among competing alternatives (Loucks et al., 2005). River diversion structure site selection and layout design procedure is usually conducted on the basis of qualitative ideal site check list and preliminary cost estimates (Arora, 2002).
Unlike for other hydraulic structures, the topography can often allow several alternative sites suitable for diversion layouts. However, these alternative sites should be quantified for a better evaluation of feasibility and for making objective decisions. In spite of this, Engineers commonly use non quantitative approaches to select diversion sites among several alternatives. This possibly limits investigations on potential least cost topographic layout options.
This research was conducted to develop a cost function or a quantitative planning tool that can be used to select financially feasible diversion site among different alternative potential sites.
2 Research Questions
The research has been desired to investigate for the most feasible diversion site selection and evaluating candidate sites. After serious of statistical technique investigation on the existing diversion project sample data, it was possible to develop a quantitative planning tool.
The following research questions are used to give an insight into the research objective setup and methods formulation.
1. How can we formulate the cost function that can be expressed in terms of sensitive design variables of diversion headwork and main canal? 2. Can we develop unit rate trend function for common item of engineering works? 3. Can we apply the cost function to predict the cost in order to identify the least cost river diversion site from other alternative sites?
Research Objectives
The main objective of this study was to develop a quantitative decision support planning tool, cost function, capable of predicting cost estimates. This can be applied for river diversion site selection and planning purposes in Amhara Region. It can be derived by means of empirical investigation using sampled data collected from the study area. In achieving this, the complementary research specific objectives are;
To identify sensitive design variables in order to formulate cost function from measured quantities of engineering works, To develop unit rate trend function for common items of construction engineering works.
Research Significance
As introduced before, diversions are important hydraulic structures used for irrigation schemes and other water resource development. The demand is increasing every year, despite the limited improvement in planning method of diversion. Therefore, it has to be planned and technically quantified for better efficiency and improvement. That is river diversion site selection procedures are likely to have a problem of investigating alternative sites using quantitative planning models. In Ethiopia river diversion construction is increasing every year due to the high irrigation demand, however the technical planning and layout design improvement is still remaining. Hence, the
3 new planning model for river diversion site selection is a helpful tool to provide quantitative information and evaluate several candidate sites. This thesis is conducted to develop a quantitative planning model for solving complex variable problems in river diversion planning and lay outing. Furthermore, it helps us to identify the feasible site and understand the relationship of all relevant variables in the system of several alternative sites evaluation.
Scope of Research
The scope of this research was limited to the development of river diversion structure planning tool in the study area, Amhara region. The tool can be important for feasible diversion site selection procedure. The study approach was to formulate an engineering cost estimating empirical function for widely adopted river diversion weirs. Moreover, only the main canal and the headwork structures were considered as the principal component of the diversion structure or weir. Here, the total cost was simplified to account the combined cost of these two components. Beside to this, rocky and alluvial type general foundation geologic conditions were addressed as a significant cost factor for the diversion headwork. About 35 river diversion weir projects were purposively sampled in the entire study area.
Eventually, the cost function was formulated from empirically correlation between total cost estimates and set of sensitive (statistically significant) design variables of main canal and headwork. As result, the cost function can be further applied as a quantitative planning tool in identifying the least cost feasible river diversion site. Details of basic assumption and limitations are well verified in section (4.3) under model development.
Research Outline
This research is structured in 6 main chapters. It is opened with chapter-1 introducing general background, problem statement, research questions, research objectives, research significance, scope of research and research outlines. Chapter-2 summarizes the state of the-art review of literatures beginning with highlighting about the diversion structure head work and main canal, requirements of statistical techniques and the existing approaches on diversion site selection procedure were clearly explained. Related past studies were also reviewed and the gaps of knowledge’s have been identified. Similarly, chapter-3 provides the study area general description, materials, methods and data preparation. Chapter 4 discusses detail of modeling
4 approach for a new quantitative planning model development. Beside to this, basic assumptions and limitations are elaborated here. In chapter 5, the data analysis graphical and tabular summary results are presented and discussed in detail with application of planning model. After all, Chapter 6 summarizes the entire research work and concludes to the main findings.
5 2 LITERATURE REVIEW
General
River water resource development is widely interfered by the construction of river diversion projects. Diversions are constructed across a perennial rivers and can be used for the purposes of irrigation, power generation, navigation, flood control, water supply(Arora, 2002; Baban, 1995; Novák, Moffat, Nalluri, & Narayanan, 2007). This is primarily desired for raising water head so that it regulates and divert the river water into the off-take canal. River diversion weirs are low cost and widely adopted gravity type hydraulic structures (Baban, 1995). The diversion site selection is usually conducted on basis of topographic survey and other technical feasibility studies.
The general design study of head work design includes hydrological, geological, topographical, agronomical and environmental studies(Arora, 2002; Baban, 1995; Garg, 1987). The hydrological study is primarily to determine the maximum flood level, the minimum, mean and maximum discharges of the river (Baban, 1995). Geological study is often needed to characterize the site geologic formation and soil type. The geologic and geotechnical considerations are also important for analysis and detail design of the cutoff piles, impervious aprons and to specify suited construction materials (Arora, 2002; Garg, 1987). Likewise, the topographic survey usually desired to determine the peak field level of the command area, the cross-sectional and longitudinal profile of the head work site, extension of river training works, canal alignment and location of crossing structures (Baban, 1995; Yazdandoost & Attari, 2004). In addition, the agronomic study aims to estimate crop water requirement and other water demands and to delineate irrigation suitability land (Garg, 1987). The environmental and socio- economic studies are also important to assure sustainability of project by creating coherence and participatory approach with local society (Baban, 1995).
In the view of the design principle, diversion structures are desired to be safe against any hydraulic, structural and geotechnical failure (Arora, 2002; Baban, 1995; Garg, 1987; Novák et al., 2007).
6 Headwork and Main Canal
River diversion structure layout generally includes main canal and head works as shown in figure 2.1. The headwork components usually includes weir body, wing walls, divide walls, apron, cutoffs piles, operation slabs, gates, off takes, under sluices and others (Baban, 1995).
Figure 2-1 River diversion structure components (Arora, 2002)
Each components are to be designed according detail of hydrologic, hydraulic, and structural and geotechnical analysis and requirements (Arora, 2002; Baban, 1995; Garg, 1987).
Hydrologic analysis is necessary to determine maximum design flood discharge and the design mean river flow discharge. The mean flow discharge is desired to satisfy total crop water requirement considering downstream water release. It is primarily important to design intake and main canal. For most irrigation projects the design base flow is usually determined as 80% dependability of river flow duration curve (Baban, 1995).
Similarly, the design flood discharges is required to provide adequate dimension of head work and its components. This is to convey the incoming peak flood discharge safely with no
7 overtopping to side and accommodating hydraulic jump. The design flood discharge magnitude is usually determined with help of flood frequency analysis considering the headwork site as watershed outlet (Baban, 1995).
The Hydraulic requirements of diversion structures include checking section adequacy for the hydrodynamic and hydrostatic loadings (Arora, 2002). Diversion structures primarily require topographic and river geometry survey. These data will be used to select weir type, to estimate weir height, to compute different hydraulic analysis like; control section flow, hydraulic jump, and seepage analysis; and to design section of relevant components (Arora, 2002; Baban, 1995; Garg, 1987; Novák et al., 2007). As sample data indicates, the most common diversion type is vertical drop weirs with broad crested and sometimes ogee shape. The ogee type has higher hydraulic efficiency but its applicability is limited because of construction difficulties. On the other hand, the broad crested type is usually preferred for ease of practicability (Baban, 1995). Similarly, the wing wall of diversion structure mostly designed with trapezoidal section geometry.
The headwork site foundation generally required to be water tight and scouring resistant hard geological formation. The vertical cutoff piles are often designed to guard against scouring effect and to reduce the exit gradient within safe limits (Arora, 2002; Baban, 1995; Garg, 1987; Novák et al., 2007). The subsurface flow analysis theories; Bligh’s and Khosla’s creep theories are commonly used for determination of the uplift pressures exerted by seeping water and the safety of the structure against uplift, downstream scour or piping failures (Arora, 2002; Baban, 1995; Garg, 1987).
The structural requirement is to make the structure safe and stable against overturning, sliding and tension failures under critical loading condition (Baban, 1995). In feasibility study and layout design of diversion structure the background about these basic design principles and technical considerations are important.
Furthermore, most of the studied irrigation schemes in Ethiopia use water from perennial rivers through diversion weirs (Girma & Seleshi, 2007). This shows necessity of river diversion weir structures on the development of irrigation potential in Ethiopia as well as in the study area, Amhara Region.
As shown in figure (2.2), design variables of headwork weir include the weir height, base width and top width. Based on construction materials, diversion structures can be made of concrete or cyclopean concrete type, rock-fill type and masonry type (Continental Consultants, 2002a).
Figure 2-3 Ogee type river diversion weir (Continental Consultants, 2002a).
Main canal is water conveyance component running from diversion head work to the command peak field level. The route selection of main canal usually depends on actual site topography and geology(Arora, 2002; Baban, 1995; Garg, 1987).
Where; a1- is canal width, h1-is full supply depth, A-is canal cross sectional area,
The canal section dimension is usually designed using Manning’s uniform flow equation (Te Chow, 1959). An irrigation canal can be made of concrete, masonry, earth materials (Arora, 2002).
Existing Approaches on Diversion Site Selection
The diversion sites mostly preferred are closer to the command area (Arora, 2002). However, the topographic layout usually allows several number of candidate sites to be considered. Site feasibility study is usually conducted to identify the least cost diversion layout in the presence of competing alternatives sites. Hence, feasible diversion site selection procedure can better be improved by means of quantitative measurement and preliminary cost estimate (Baban, 1995).
According to (Arora, 2002) and (Novák et al., 2007) project appraisal and planning processes of diversion structures usually have three stage of site feasibility investigation. The first stage is reconnaissance which mainly to assess the field based on the existing topographic, geologic map and to suggest possible number of sites. The second stage is preliminary study or Identification primarily to determine the most suitable site among candidate sites. It includes hydrological analysis, water demand assessment, preliminary survey and tentative layout and general foundation investigation. The third stage is final study mainly focused on the detail technical study of pre identified suitable sites for final design purpose. The investigation includes topographic survey and contour map generation, layout of structural components, detail of hydrological analysis and subsurface investigation. An ideal site for river diversion is characterized by the following points(Arora, 2002; Baban, 1995; Garg, 1987);
10 1. Availability of a stable abutment with narrower and well defined reach of river, 2. River bank configuration with least effect of weir construction on upstream submergence and downstream scouring so as to reduce river training works, 3. Feasible water intake site during the dry season flow and least sediment inflow, 4. Suitable canal layout that allow maximum command size and minimum idle canal length; 5. Suitable site configuration for diversion even during construction period; 6. Structural Stability requirement with economical construction costs, convenience for operation and maintenance, 7. Perpendicular alignment of headwork to the direction of river flow; 8. Suitable location for under sluice and head regulator position; 9. Less submergence effect on upstream areas; 10. Sound rock foundation condition; 11. Nearby availability of construction materials; 12. Accessible location of site; 13. Minimum overall cost of project.
So far, at least in Amhara Region, practicing engineers usually exercise these purely qualitative diversion site selection and layout design approaches. This possibly limits the quantitative measurement of works in making informed decision in site selection processes.
Related Past Studies
It is hardly possible to obtain past studies in the proposed type of research. Some thing abroad, (Sedighizadeh, Mansoori, Pirestani, & Sedighizadeh, 2011) developed a model for economic optimization of Water diversion system during dam construction. For this model, the developer formulated empirical cost function for tunnel and cofferdam works. The cost function was then minimized with the aid of PSO Algorithm (particle swarm optimization). The study was found that the optimum tunnel diameter and the total diversion cost are directly related to the river flood discharge (Q). In addition to this, river length and tunnel length are mainly a function of the ratios of the unit prices. The results of optimization used in some of the case study showed a significant change in the cost. This research is merely important for supposed quantitative planning model development for diversion structures. However, the study was limited to planning and designing
11 of river diversion tunnels and coffer dam for dam construction only. So it cannot be applicable for the usual river diversion structure planning and layout using diversion weir and main canal under consideration.
A research by (TURAN, 2004) Development of a Computer Program for Optimum Design of
Diversion Weirs, but still now it need detail final survey data for selected diversion site. More of it focuses on pure engineering design procedures upon detail fielded data gathered. It does not allow us to compare candidate diversion sites in project study and site appraisal process.
Decision support systems for large dam planning and operation in Africa (McCartney, 2007) emphasizes the role of decision support systems (DSS) for dam planners and operators in situations of conflicting objectives and problematic decision makings. Even if, the paper focus on DSS based large dam planning and operation, it also has overviewed the range of applications of DSS in water resource planning including river diversion structures.
Requirement of Statistical Techniques
In 1930s, Fisher had first introduced the statistical principles in experimental investigation of agricultural science in England (Reale, 1993). Later on, their application was extended to industrial settings and planning by the development of response surface methodology (RSM) by Box and Wilson (1951). Along with the need of western industry quality improvement in 1970s, Taguchi and Wu (1980), Kackar (1985), and Taguchi (1987, 1991) had significantly expanded the wide range application of statistical techniques in the field of applied science and engineering (Reale, 1993). Now days, statistical or numerical techniques are becoming advanced and widely applicable in engineering and applied science research areas (Chapra & Canale, 2012; Singh & Masuku, 2012).
Engineers and scientists have noticed that empirical studies upon observation and experiments are vital to embody the cumulative wisdom of past experience in generalized mathematical models (Chapra & Canale, 2012). Moreover, engineering problems can usually be expressed in simplified empirical function(Chapra & Canale, 2012). Such principle of empirical generalization is found important to synthesize observations and experimental results into a descriptive mathematical framework or model. Thus, derived mathematical models can be further used to solve practical engineering problems (Chapra & Canale, 2012). Regression
12 models are the basis and frequently applied for such empirical trend function development (Reale, 1993).
In most cases, an appropriate functional relationship can be deduced from the behavior of a physical problem (Reale, 1993). The trend function is usually formulated using least squares regression method of curves fitting that minimizes the discrepancy between the data points and the trend curve. It is important to determine the regression coefficients in formulating the trend function (Chapra & Canale, 2012).
In this study, regression analysis was desired to develop an empirical model relating important design variables to the total cost estimate. Multiple linear regression statistical analysis method can be applied in the presence of multiple explanatory variables (Helsel & Hirsch, 1992). This is important to explain possible variation of response variable with regard to multiple explanatory variables. The Multiple linear regressions model mathematically expressed as follow (Helsel & Hirsch, 1992; Reale, 1993):
Y i x x 0 1 i1 2 i2 ... x .................(2.1) k ik i
This can be rewritten as:
x ..........................................(2.2) k
Y i 0 j 1 j ij i ;
j=1, 2, 3, …, k, Where; Yi- is the response variable, Xj - is the explanatory variables of (jth) order,
j - is the coefficient for (jth) order explanatory variables, t - is unexplained noise of data,
The least squares function can be expressed as below (Chapra & Canale, 2012);
(Y i 0 kj 1 j X ij ) ............(2.3) n n
L 2 2 i 1 i i 1 i 1
Then least squares function will be minimized with respect to the least square estimator, regression coefficients so that the sum of the squares of the errors can be minimized. That is;
d Li 0..............................................................................(2.4) d j
13 Moreover, about (K+1) number of least square normal equations can be expressed in matrix notation (Reale, 1993).
The solution of matrix will give us the least squares estimators of the regression coefficients. The statistical significance of regression predictor can be tested with test statistics p-value for 90% and 95% of conventional confidence levels (Helsel & Hirsch, 1992).
Confidence interval is a range of values, derived from sample statistics, that is likely to contain the value of an unknown population parameter (Singh & Masuku, 2012). Because of their random nature, it is unlikely that different samples from a given population will yield identical confidence interval. But if you repeated your sample many times, a certain percentage of the resulting confidence interval would include for the unknown population parameter (Reale, 1993). The percentage of confidence interval that contains true population parameters is called the confidence level. A given percentage of confidence interval indicates that much of the stated percentage of sample will produce confidence interval of that contain the true population parameter (Helsel & Hirsch, 1992).
P-value is the test statistics parameter that shows the probability to observe effects as big as those seen in the sample study (Singh & Masuku, 2012). It is important to determine the significance of explanatory variables of regression analysis in light of probability. If the p-value of a test statistics is found less than the stated significance level, alpha, the explanatory variables can be regarded as sensitive or statistically significant for the response variable (Reale, 1993).
14 So empirical model can be derived using statistical techniques. Moreover, they are important to express the results of an experiment quantitatively so that it will able to facilitate understanding, interpretation, and implementation process (Chapra & Canale, 2012; Loucks et al., 2005). In the planning and design of water resource systems comprising a number of alternatives, Constrained Optimization technique are usually applied to maximize the desired total benefits and minimize the cost (Loucks et al., 2005; McCartney, 2007; Vedula & Mujumdar, 2005). Optimization is the numerical procedure of searching for a point on the functions that either minimize or maximize the predefined objective function. The objective function can be any mathematical model derived to simulate an engineering system (Chapra & Canale, 2012).
Figure 2-5 Engineering problem solving approach (Chapra & Canale, 2012)
As shown in figure (2.5), practical problems first be defined and solved with combination of theory and data inputs into a mathematical model designed for the system. For this study the cost function has been formulated as a mathematical model to solve feasible diversion site selection problems.
15 3 MATERIALS AND METHODS
Study Area Description
This study was conducted within 153,000 Km2 geographical area of Amhara regional state that is part of Ethiopia. Because of two important reasons, the region was selected as study area for an entitled research. The first reason was that the region is part of country where potential of river diversion water resource projects has been carried out. The second reason was to use the advantage of the regional extent of diversion projects data available at the Regional Water resource Bureau which is owner of projects.
The study area is located within approximate UTM coordinates of North (965607_1523723) m and East (92735_656503) m. It has also rugged topography with altitude ranging from 509m to 4494m AMSL. The area shares the Abbay river basin geology character dominantly with thick tertiary and quaternary volcanic formations, Precambrian metamorphic rocks, Mesozoic sedimentary formations (MoWR, 1998). Beside to this, the annual rainfall of the study area varies between 800mm to 2200mm with a mean value about 1420mm (MoWR, 1998). The area is also predominantly characterized by rain fed subsistence agriculture practice.
The Amhara regional state has an estimated total population about19,120,000, of which 88.5 % are estimated to be rural inhabitants, and 11.5 % are urban according to the Central Statistical Agency of Ethiopia (CSA 2005) censes (Awulachew et al., 2007).
Irrigation development has been largely incited in the country as well as in the study region including both small and medium scale projects. The region has enormous potential both in land and water resources. It shares the major portion of Abbay river basin which has 54.8 Bm3 surface water potential and 815,581ha irrigation potentials (MoWR, 1998). By the year 2007, there were about 310 modern irrigation schemes developed in Amhara Region that covered 8,469.26 ha Irrigation area with 17,443 beneficiary people. Out of this total irrigated area, small- scale and medium-scale irrigation schemes accounts about 5,718.68ha and 2,750.58ha portion respectively (Awulachew et al., 2007). The population number also gets increasing with time. These pressurize demand for irrigation potential development over available river water and land resources.
16 The Regional Water Resource Bureau, an authorized regional water institution, has been implementing significant number of small and medium scale irrigation projects mostly using river diversion weir structures. This implies that irrigation development has been significantly activated within study area and recent diversion project’s data is available at the Bureau.
Figure 3-1 Study Area Map
The study area map figure (3.1) shows the spatial extent of the 35 river diversion projects that were sampled for an entitled research. It also depicts main rivers and administrative zones found in the region.
Materials
Different materials have been intensively used for this research. Some of these are software’s; MINITAB-16, Microsoft Excel, Arc GIS desktop, and documents like diversion weir design,
17 drawings (BoWRD, 2009_2015), construction agreements documents of irrigation and related water projects.
After the comprehensive data checklist had been prepared, Microsoft excel was employed not only to build data set but also to do important statistical analysis. Another statistical investigation software package, MINITAB-16, has been used for further analysis. Majority of data were extracted from the design document and drawings. Beside to this, construction documents were primarily used to extract unit rate data that are required to develop the unit rate trend function for the common item of civil works within the study area. Arc GIS Desktop (10.1) was also used for mapping the study area.
Methods Cost estimation of civil works is an important element of decision making procedure in project feasibility study and site selection process (Sedighizadeh et al., 2011). In design and construction of an engineering system, engineers usually confront quantitative measurement and optimization techniques to minimize the cost and maximize the desired benefit. For ease of such decision making process, practical problems can be simplified and expressed in mathematical function of certain decision variables (Sedighizadeh et al., 2011). Moreover, statistical techniques can be applied to formulate data driven mathematical models of engineering problems. These are important to empirically synthesize and to define the key parameters of system equation (Chapra & Canale, 2012; Chen, de Kee Fong, & Kaloni, 2003; Elshorbagy, Corzo, Srinivasulu, & Solomatine, 2009).
As the study was desired to develop a new mathematical planning model for site selection of river diversion structures based upon main canal and headwork prioritized river diversion structure components. This can be provided by developing the total cost function. Cost function is a mathematical equation used by economists and scholars to depict the cost behavior under the variation of cost driver factors or decision variables (C.Mitchell, 2015). In this study sensitive design variables were considered as cost driver factors. Furthermore, the least-square regression analysis techniques can be employed to empirically synthesize engineering problems (Chapra & Canale, 2012; Reale, 1993).
18 So that, with the help of least-square regression analysis, it was possibly to develop the cost function as well as the unit rate trend equations using relevant data that has been extracted from design and construction agreement documents, and drawings.
The study was conducted with 35 river diversion projects sampled in the study area. Purposive sampling technique was used to fulfill the requirement of the comprehensive data check list. In addition, the rocky and alluvial type general geology scenarios were considered for headwork structure. This was found important to account the variation of civil works and design considerations. Among the total 35 sample size, the 22 are rocky type and the rest 13 are alluvial types. Beside to this, more than 70 number of construction agreement documents mainly of irrigation projects were used over 10 years (2006 to 2015) and then clustered under the seven major spatial zones. The clustering was done according to the unit rate data uniformity observed within the administrative zones of Amhara region or the study area. This was required to extract unit rate data for the main item of civil works; earth work, concrete & bar works, masonry and finishing works.
19 This is because the unit rate is an important factor that makes the cost of diversion structure variable both in space and time dimensions.
The cost of civil works usually varies both in spatial and temporal dimensions. To minimize such variations, the work quantities and unit rates were considered independently. Moreover, the temporal variation can be better controlled with development of unit rate trend function over clustered spatial zones of study area while the quantity of engineering works and respective design variables can be considered as spatially lumped and time invariant. The unit rate trend function was also desired to predict the up to date unit rates of construction work items.
In more general sense, the total cost estimate can be rewritten as follow; (Sedighizadeh et al., 2011). i n
Cost total (quantity) * (unit.rate) ...............................................(3.1) j j i 1
As expressed in the above equation (3.1), the cost of civil works is usually estimated as sum of the product of work quantity with its corresponding unit rate.
Table 3-2 Summary table of common item of engineering works
Engineering works Usual items of engineering works quantities
Site clearing, Sc(m2), 1. Earth work Normal soil excavation, Nsx (m3), Loose bed & soft formation excavation, Lbx(m3) Hard Rock excavation, HRx(m3), Back fill & compaction, Bfc(m3), Cyclopean concrete , Cpn(m3), 2. Concrete, masonry & bar works Lean concrete, C-10(m3), Capping concrete, C-20(m3), Concrete, C-25 (m3), Masonry work, Mas(m3), Reinforcement bar (kg) Plastering work, Pst(m2) 3. Finishing works Pointing work, Pst(m2)
20 The data of these work quantities and their corresponding design variables shown in table (3.1) and (3.2) were extracted from the design documents. The unit rate values can be traced from the developed unit rate trend function.
Some of analysis techniques employed in this study are summarized below:
Matrix type scatter plot was used to identify empirically relevant design variables, Least-square regression method of curve fitting was employed to develop of unit rate trend function for common work items over a period 10 years, Multiple linear regression (MLR) statistical analysis technique was also used to formulate the cost function of diversion structure based on main canal and headwork. The matrix scatter plot with smoother can be used to give us an insight about the data pattern and the relationship between dependent and independent variables(Helsel & Hirsch, 1992).
Furthermore, measured quantities of engineering work items are usually considered proportional to the dimensions or design variables (Sedighizadeh et al., 2011). Hence, the design variables can be correlated with the cost which was obtained from the combination of measured work quantities and unit rate trend function. Consequently, with the application of least-square method of multiple linear regression analysis, it was possible to develop the cost function expressed in terms of important design variables shown in table (3.1). The detail mathematical description of unit rate trend function development and multiple regressions based cost function formulation is presented in chapter (4).
Data Preparation
To develop the cost function data such as; unit rate, design variables and measured quantities of civil works were prepared for analysis. At first, the main canal and the headwork were prioritized as the principal components of river diversion structure. This is because these two components can account for the major volume of civil works and can possibly implies the diversion site variability. And then, a comprehensive data checklist has been prepared that comprises; site name and UTM location, general geologic conditions, hydrologic design data (canal design discharge, base flow, design flood discharge), agronomic design data(irrigation duty, command area), geometric design variables like dimension of weir, apron, canal, wing wall and divide wall of the headwork components and main canal. These data were extracted from
21 35 diversion weir design documents and drawings. With review of relevant design and construction documents, the usual engineering work items were identified. And then, the design variables with respective quantity of engineering works were extracted from collected sample projects.
This study was limited about the 35 sample river diversion projects. This was because of limited data sources availability, data collection time and budget requirements. Moreover, the data nature of both design variables and quantity of engineering works is discrete type. Sample diversion projects were designed at different years, and they are spatially distributed within the study area.
With merit of adequate data requirements of statistical analysis techniques and above mentioned constraints, design variables and quantity of engineering works were considered spatially lumped and time invariant. On other words, these data were replicated to the seven spatial zones. This is important to determine the more precise factors with adequate data set. In our case, design variables were considered as factors that bring change to the cost or the response variable.
In this study unit rate trend function development was desired for the usual work items over a period of 10 year (2006_2015). Based on the uniformity observed in the unit rate sample data, the rate data were clustered into 7 major administrative zones; North Gondar, South Gondar, Awi &W/Gojjam, E/Gojjam, N/wollo & WagHimira, S/Wollo & Oromiya and N/Shewa. This unit rate trend function was developed on base of median unit rate value per year. This was to avoid the effect of some outlier value probably exhibited in the unit rate data.
Eventually, cost can be synthesized with combination of spatially lumped quantities of works and unit rate trend function. Furthermore, the total cost can be formulated as an empirical function of most characterizing design variables of main canal and headwork.
22 4 MODEL DEVELOPMENT
General
As it was stated in the introduction chapter, decision support mathematical models are usually desired to provide quantitative information to the decision makers in order to make a better selection from alternative measures in planning a water resources system (Loucks et al., 2005). These mathematical models are often the simplified representation of the real system to capture the key characteristics in approximate mathematical framework (Saliha, 2012)
The existing procedure on river diversion layout and design was much limited to the purely qualitative approaches. So, we need to develop a quantitative mathematical planning model for evaluating alternatives and selecting the feasible diversion site. The modeling approach was based on the total cost function formulation using readily determined quantities of civil works and respective unit rate trend function of the respective work items. Design variables and measured engineering work quantities were determined form major components of river diversion structure like head work and main canal. Furthermore, the cost function can be used as an objective function of constrained optimization.
Above all, the quantitative planning model described below predicts engineering cost estimates in terms of sensitive design variables. This will help diversion layout designers to investigate competing alternatives sites on quantitative based engineering cost estimates.
Schematic Layout
The usual method of river diversion study involves feasibility study of headwork site, river flow adequacy, relative distance of command area from the headwork and other technical considerations (Baban, 1995). During site visit and with the help of topographic map, designers can decide the likely headwork sites and canal layouts. As discussed in chapter 2, the ideal sites requirements for the headwork site selection is mainly includes narrower, straight and stable reach of river configured to the shorter length of main canal alignment. It is clear that river diversion topographic layout allows us several alternative sites. At feasibility stage, the designer was supposed to suggest several number of alternative sites (Arora, 2002).
23 In general principle the morphology of natural streams gets deeper and wider as we go to the downstream. In other words, the stream section becomes narrower and shallower to the upstream direction (Leopold & Maddock Jr, 1953).
One may choose the downstream reach site with larger dimension of headwork and shorter length of main canal to the command area. Such layout brings the higher cost of head work and lower cost for main canal. On the contrary, the upstream reach site with smaller dimension of headwork and longer main canal with some idle length gives us lower cost of head work and higher cost of main canal. Between these two extremes, there lies several possible site layout or configurations with different cost and benefit from the incremental command size. Still the designer is in charge of choosing the most feasible layout that minimizes the combined cost of headwork and main canal.
Figure 4-1Schematic layout of river diversion structure
Where; HW-is headwork, MC-is main canal
24 The diversion height required to head up the river water flow is equal to command area peak elevation plus sum of all hydraulic head losses in conveyance as well as regulating system, canal full supply water depth and allowable free board. As shown in the figure below, the required height of diversion structure can be expressed as (Baban, 1995):
H Z H L Dw F...............................(4.1)
Where: Z (m)-is static head difference between peak field level and river bed level, H L (m)–is total hydraulic head loss in the conveyance canal, Dw (m)-is canal full supply water depth, F(m) – is free board allowance.
Figure 4-2 Schematic river cross section with left side command peak field level (Baban, 1995)
The main technical factors such as river geologic formation, base flow, irrigation duty, and command area, length of main canal and river section geometry of river are important to identify alternative sites. Besides to technical considerations, the designer should be able to optimize for the feasible site based on possible economic merits.
The set of sensitive design variables of main canal and headwork are varying with alternative sites considered. The design variables of main canal includes canal design discharge Qd (m3/s), canal length Lc(m) and conveyance Kc. Similarly, the headwork main design variables are; weir height ,Hb (m); weir length ,Lb(m);height of wing wall, Hw(m); length of wing wall, Lw(m); downstream apron length, La(m) and apron thickness, ta(m).
The total cost of diversion layout depends on components such as irrigation field infrastructure cost, main canal cost and the headwork cost. Given the fact that the total cost is sum of individual components, the most feasible diversion site among alternative sites is the one that gives us the minimum value for total cost per command area.
25 CT C FI (CMC C HW ) ...........................(4.2) A A A
Where: CT-is total cost, A-is command area, CFI-is cost of field infrastructure, CMC-is cost of main canal, CHW-is cost of Headwork including components.
It can be assumed that the field infrastructure density is uniform for a given irrigation project. In this study the cost of field infrastructure per command area is assumed constant. But main canal and headwork can be alternated with different degree of feasibility. That is, once the peak field level is selected upon the river lean flow and the maximum command size, alternative site layouts can be subjected to quantitative evaluation and optimization using the combined cost of main canal and head work per command area. Furthermore, the optimization requires an objective function formulation that empirically relates the total cost with sensitive decision variables.
Basic Assumptions and Limitations
Some basic assumptions were made in this study to attain the necessary guides in formulation of the cost function. River diversion site selection includes considerations like socio-economic and environmental viability beside to the technical engineering requirements such as topographic, hydrologic, structural and geological feasibility. In this study the importance of socio-economic and environmental factors were assumed to weight equal for all alternative sites to be evaluated. Their relative difference should rely on quantitatively measurable design variables of hydraulic, structural and geotechnical requirements.
More importantly, the total cost was simplified to the combined cost of main canal and headwork in doing quantitative evaluation of alternative diversion sites. In addition, future design principles and construction method of diversion irrigation projects were supposed to be similar to already existing sampled projects. Diversion design variables and the corresponding measured quantities of engineering works were spatially clustered and made time invariant under the study area. This was employed to reduce variations of costs in space and time dimensions. The unit rate trend function was developed for usual engineering work items to account and capture the time variation of cost and spatially clustered to spatial zones.
26 Two general geologic foundation conditions of rocky and alluvial formations were considered to account cost variations of the diversion headwork substructures. But for the main canal, the sample indicates, the normal soil type geologic condition was entirely used.
Another limitation of this study was that the main canal possible crossing structure like flume, drops, culverts were not included to the main canal. Owing to this, the designer is supposed to give his own (%) allowance to the main canal so as to account for cost of possible cross drainage structures.
Modeling Approach
The modeling approach of the new quantitative DSS planning tool for river diversion site selection is shown in the chart below.
Main Canal & Headwork Components
Unit Rate Trend Function for Design Variables & Quantity of
Engineering Work Items Engineering Works
Cost Function formulation
Planning Tool Application
Figure 4-3 Flow chart of developing mathematical planning tool
27 4.4.1 Cost Function Formulation
As partly discussed in section (3.3) of materials and methods, cost estimation of civil works is an important procedure used in project feasibility study particularly to evaluate competing alternative sites. The cost of civil works is usually estimated by the combination of measured work quantity with its corresponding unit rate. Moreover, the cost function can be formulated to solve cost estimation engineering problems (Sedighizadeh et al., 2011).
The cost function is an empirical equation synthesized from the data that expresses the relationship between the total cost as a response variable and the relevant design variables which are considered as cost driver factors. The sort of design variables indicated in table (3.1) were considered as multiple explanatory variables influencing the cost of the diversion structure. As a result, we can drive the cost function using multiple linear regression (MLR) statistical analysis technique. The least square method of MLR analysis was used to fit set of sample data to the true functional relationship between cost and sort of design variables which are considered as decision variables. Eventually, the total cost was tested for its empirical correlation with above relevant design variables using MLR analysis technique.
This can be expressed in the following general mathematical form.
C T f {Q , K c , Lc , H , Lb , Lw , H w , La}...............(4.3) c b
Or as an alternative test with meaning full combination of design variables,
C T f {Lc K c , H b Lb , La Lb , Lw H w}.........................(4.4)
Where; CT - is total cost per thousands of Ethiopian birr, Qc-is canal discharge, Kc-is canal conveyance, Lc-is canal length, Hb- is weir height, Lb-is weir length, La- is apron length, Hw-is height of wing wall, Lw – is the length of a wing wall, (Lc*Kc)-is product of canal length and conveyance, (HbLb)-is product of weir height and length, (La*Lb)-is product of apron length and weir length, (Lw*Hw) – is product of wing wall length to its height.
Beside to the design principle, the above listed design variables were tested for their empirical significance particularly with the corresponding measured quantities of work. The detail is presented in the next section (4.4.2).
The combination of design variables was made in such a way that it can proportionate for the respective section quantities of works.
28 Accordingly, likely regarded design variables were combined as; Lc*Kc, Hb*Lb, La*Lb, Hw*Lw for main canal, weir body, apron and wing walls respectively. Thus combined variables can be considered as explanatory variables and tested for cost function formulation as shown in equation (4.4). Further justification for other design variables of headwork is presented section (5.2.2).
Finally using MLR analysis, the general cost function can be mathematically expressed as;
C L k H L L L . . L H T 0 1 c c 2 b b 3 a b 4 w w ..................(4.5)
Where: Where; CT - is total cost estimated per thousands of Ethiopian birr, parameters; a 1, a2, a3, a4- are coefficients of Lc*Kc, Hb*Lb, La*Lb, Hw*Lw respectively, (LC*Kc)-is product design variable of canal length and conveyance, (Hb*Lb)-is product design variable of weir height and length, (La*Lb)-is product of apron length and weir length, (Lw*Hw) – is product of wing wall length to its height.
The least square regression analysis was used to determine both the parameters of cost function and the sensitivity (statistical significance) of each explanatory variable. This was possibly done with help of test statistics p-value for the 90% confidence interval considered. That is, the explanatory variables as well as the constant of regression equation were regarded as sensitive for their P-value less than (0.1) and insignificant for their P-value greater than (0.1). Accordingly, the tentative MLR cost function equation was rewritten in terms of sensitive design variables of headwork and main canal as shown in equation (4.5).
The predictive accuracy of the various mathematical models can be evaluated using the root mean squared error (RMSE), the coefficient of determination (R2) and the Nash–Sutcliffe efficiency (NSE) (Elshorbagy et al., 2009) ,(Nash and Sutcliffe ,1970). These can be given as follows;
(Y oi Y pi) .......................................................(4.6) n 2
i 1 RMSE n
29 2 in1 (Y pi Y pi )(Y oi Y oi ) R 2 ............( 4.7) 2 2 i 1 (Y pi Y pi) * i 1 (Y oi Y oi) n n
(Y pi Y oi) .....................................................(4.8) n 2
i 1 NSE 1 2
(Y oi Y ) n i 1 oi
Where; Yoi- is observed value, Y`oi – is observed mean value, Ypi- is predicted value, Y`pi -is predicted mean value.
Some regression model builders prefer to use an adjusted (R2) statistics, which can account for the sample size and the number of independent variables. The adjusted coefficient of determination (R2 adj) can be written as(Reale, 1993);
2 R adj n 1 1 n 1 k 2 1 R ....................................................(4.9)
Where; n- is number of points, K-is no of predictors.
The adjusted R2 statistics will not always increase as variables are added to the model. In fact, if unnecessary terms are added, the value of Radj will often decrease. In addition, the standardized residuals (di) can be used to examine the fitted model to ensure that it provides an adequate approximation to the true system (Reale, 1993).
ei d i MSE ....................................................(4.10)
Where, ei-is error, MSE-is mean square error.
The standardized residuals have mean zero and approximately unit variance; consequently, they are useful in looking for outliers. Most of the standardized residuals should lie in the interval (- 2 to 2), and any observation with a standardized residual outside of this interval is potentially unusual with respect to its observed response (Reale, 1993).
The above all statistical analysis of total cost function formulation was manipulated with help of MINITAB-16, statistical analysis software package.
30 The cost function can be further used as an objective function of optimization problem in selecting feasible diversion site among potential alternatives. Optimization problem are usually solved with linear programming optimization methods that have linear objective function and the constraints. The following is the general form of single-objective constrained optimization problem(Chapra & Canale, 2012).
Minimize / Maximize, Z t c1 x1 c x
2 ... cn xn .................(4.11) 2
subjected ..to, i1 x1 x
2 ... in xn bi .....................(4.12) i2
Where: Zt- is the objective function to be minimized or maximized, Cj - is coefficient for jth activity undertaken, Xj - is magnitude of jth activity, aij - amount of the ith resource that is consumed for each unit of the jth activity, bi - amount of the ith resource that is available.
In the same manner, the new cost function expressed in terms of some sensitive design variables can be subjected to constrained optimization according to the actual site condition. Further, it will be able engineers to identify for the least cost river diversion site.
4.4.2 Design Variables and Quantities of Works
As mentioned in part of materials and methods, the main components of river diversion structure were first prioritized for planning and layout design of intended research purpose. These were the main canal and the headwork with components including; the weir body, downstream apron, cutoff depths, wing walls and divide walls as shown in figure (2.1). These were employed to account for the major volume of works and capture site variability with sample data.
It was desired to express the total cost in empirical function of some sensitive design variables of main canal and headwork. This is because the total cost of diversion structure likely depends on factors of some most sensitive design variables. Therefore, relevant design variables of main canal and headwork were tested for their empirical significance with regard to measured quantities of engineering works and total cost estimates.
As pointed before, measured quantity of engineering works refers to those civil works indicated in table (3.2). Furthermore, the matrix type scatter plot, graphical data summary and curve fitting
31 methods were used to give us an insight to identify the most characterizing design variables particularly with regard to measured quantity of works.
4.4.2.1 Main Canal
As pointed in literature review section (2.2) main canal is water conveyance structure that delivers water from diversion head work to the command peak field level. The main canal alignment is determined by the actual site topographic slope, distance and geologic conditions (Arora, 2002; Baban, 1995; Garg, 1987). It’s section dimension are usually designed by using the manning’s uniform flow equation (Chow, 1959).
Q K S ................................(4.13)
Where: Q (m3/s) is Discharge, k-conveyance, S-is longitudinal bed slope of canal.
As equation (4.13) indicates, the design discharge (Qd), the average canal bed slope (S) and the conveyance (K) can be regarded as the main canal relevant design variables. It is obvious that the discharge (Qd) and the canal bed slope (S) are site dependent. But the conveyance (K) is site independent section factor of canal (Te Chow, 1959). So that, the main canal quantities of civil works can be better characterized by the section factor, conveyance (K).
Regarding to the construction methods, main canal can be made of concrete, masonry, earth materials(Arora, 2002; Continental Consultants, 2002a). The study sampled shows majorities of main canals are rectangular and lined masonry type as shown in figure (2.4). The canal construction work usually comprises site clearing, bulk excavation, masonry, plastering and backfilling works.
4.4.2.2 Head Work Structures
River diversion head works are usually designed as gravity structure with structural safety requirement of overturning, sliding, over stress and material failures (Baban, 1995; Novák et al., 2007). The headwork component comprise structure like the weir body, downstream apron, cutoff depths, wing walls and divide walls as shown in figure (2.1) and (2.2). As partly discussed in the literature review, the headwork structure is desired to have adequate section to accommodate the hydrologic and hydraulic requirements. Moreover, the components should be designed safe against any hydraulic, structural and geotechnical failures.
32 The weir body, divide wall and wing walls are usually designed as simple gravity structure. Most of the time designers use Bligh’s formula to determine the tentative section of weir body as discussed in part of literature review. Similarly for stable elementary profile of simple gravity structure the height to base width ratio (H/B) approximated as value greater or equal to 2/3 (Arora, 2002). Eventually, these tentative dimensions need to be modified accordingly for stability requirements of structure against failure of overturning, sliding and tension under critical loading condition. While other headwork components like the apron floor and cutoff piles design requires subsurface hydraulic and geotechnical considerations. Mostly Bligh’s and Khosla’s creep theories are used to analyze seepage water uplift pressures and the safety of the structure against downstream scour and piping failures(Arora, 2002; Baban, 1995). As shown in figure (2.2) and (2.3), the common adopted diversion weir geometries are the broad crested and sometimes the ogee types. Similarly, the wing wall of diversion structure is mostly designed with trapezoidal section geometry(Baban, 1995; Garg, 1987).
Relevant design variables of the headwork account primarily parts of weir body, apron floor and wing walls. Design variables of headwork includes the weir body height (Hb), base width (B), top width (T); apron length (La), height of wing wall (Hw), length of Wing wall (Lw).
The combination of important design variables was tested to better characterize the section design work quantities. The stable section weir body can be characterized with approximate ratio of (H/B) and (T/B).
For the given foundation condition, the cutoff depth and thickness of apron are found geometrically proportional to the weir height (Hb). This is because of the total creep length and the apron floor thickness are designed based on the static head. This can be shown in figure 8.26, 8.27 and 8.28 at annex. So that, we can characterize the apron thickness and cutoff depths with weir height (Hb) that is jointly expressed (Hb*Lb) in the cost function. The downstream apron can be accounted to the cost function by the product of apron length (L a) and its width or weir crest length (Lb).
In accordance with the common construction methods adopted in the study area, the diversion weirs are usually made of cyclopean concrete core with reinforced capping concrete structure. Similarly, the apron floor and divide walls are usually constructed from reinforced concrete structures. While the wing walls are usually constructed from masonry materials. Beside to this,
33 the quantities of civil works of the head work are found dependent on the above mention design variables.
4.4.3 Unit rate trend function development
The unit price of civil works is usually used to obtain simple mathematical relation in the cost estimation procedure (Sedighizadeh et al., 2011). Moreover, practical engineering problems can be expressed in mathematical trend function of certain decision variables (Chapra & Canale, 2012; Chen et al., 2003; Sedighizadeh et al., 2011). The trend function is an empirical approximating function that fits the general pattern of the data without necessarily matching the individual points. This is because of observed data points are known only at set of discrete points and often exhibits significant variability and associated substantial error (Helsel & Hirsch, 1992). The trend function is usually formulated using least-squares regression method of curves fitting that minimizes the discrepancy between the data points and the trend curve. The general of least square regression analysis is expressed as follow; (Chen et al., 2003)
Where e2-is squares of the residual error, yi o- is observed values; yi p- is predicted values.
The predictive trend function usually can be of linear, exponential, polynomial and power type. The least square regression technique is important to readily estimate the coefficients of predetermined empirical equation that best fit the observed data (Helsel & Hirsch, 1992). The best fit of the curve can usually be expressed by coefficient of determination (R2) to be closer to one (Singh & Masuku, 2012).
The common civil work items or engineering works accounts for list of civil construction works indicated in table (3.2). The unit of works are usual expressed as; site clearing (birr/m2), Normal soil excavation (birr/m3), loose bed & soft formation excavation (birr/m3), hard rock excavation (birr/m3), back fill & compaction (birr/m3), concrete works (birr/m3), masonry work (birr/m3), reinforcement bar (birr/kg), finishing works like plastering work (birr/m 2) and pointing work (birr/m2). Unit rate trend function of civil works can be used to extrapolate for unit rate values to the future years.
34 5 RESULTS AND DISCUSSIONS
Unit Rate Trend Function
The unit rate trend function was found an important means to capture the effect of temporal and spatial variation of cost of diversion structures. As elaborated in methods, it was done by partly lumping the unit rate data for each of the seven zones of study area over period of 10 years (2006 to 2015). Accordingly, the unit rate trend function was developed for the common item of civil woks shown in table (3.2).
Figures (5.1), (5.2), and (5.3) shows the unit rate trend function formulated for earth works, concrete & masonry works, finishing and bar works in Awi and W/Gojjam spatial zone of study area.
Linear 5.0 2.0 (LbWREx) Nsx/10 = 0.420x - 841.2 Linear (HREx) Bfc/10 = 0.948x - 1903. R² = 0.843 0.0 R² = 0.898 2.5 Linear (BFC) 2006 2008 2010 2012 2014 2016 Time,T(year) Linear (SC)
Figure 5-1 Earth work unit rate trend function for Awi&W/Gojjam area
Where; Sc _site clearing, Nsx- normal soil excavation, Lbx,-loose bed & soft excavation, HRx-hard rock excavation, Bfc-back fill & compaction,
35 Figure 5.1 shows that the earth work unit rate trends fit best by linear equation which is confirmed by R2 values greater 0.75. The trend shows an increasing of unit rate for earth works. This can be because of increasing labor cost and machinery cost with years.
As written in trend equation, all earthwork items are expressed per 10s except that of hard rock excavation which is expressed per 100. This is important avoid an over fitting effect of trend equation.
Figure 5.2 concrete and masonry works unit rate trends fitted best with linear function having R2 value greater than 0.75. This indicates the statistical adequacy of trend curve fitting. The unit rates trend function is also expressed per 100 in order to avoid over fitting effect of trend equation.
Similarly figure 5.3 depicts unit rate trend function for that of reinforcement bars, plastering and pointing finishing works.
36 These work items unit rate data fitted best by linear equation. It was also confirmed by R 2 greater than 0.75 which is statistically adequate.
Figure 5-3 Finishing and bar work unit rate trend function for Awi&W/Gojjam area
Where; Bar- reinforcement bar, Pst - plastering work, Pt-pointing work,
The same was done for all of the seven clustered spatial zones of study areas including; E/Gojjam, N/Gondar, S/Gondar, N/Wollo &Waghimira, S/Wollo &Oromiya and N/Shewa. The unit rate trend function result summary graphs of these seven spatial zones are presented in figure (8.1 to 8.18) at annexes.
The overall least square regression result of unit rate trend function development was found good fit for the common construction work items adopted in the study area. The result is also confirmed by performance indicator statistics, coefficient of determination (R2) with 0.74 to 0.98 minimum and maximum values. Hence, it can be further applied to predict the future unit rate of common construction work items. Furthermore, it can be incorporated to the new cost function mathematical model so as to ensure reliable cost estimates.
37 Relevant Design Variables and Quantity of Works
There relevant design variables of main canal and headwork components were tested for their significance correlation with regard to measured quantities of engineering works, and the MINITAB data analysis result has been presented in different figures in this chapter.
5.2.1 Main Canal Design Variables
The canal design variables including design discharge (Qd), average canal bed slope (So) and conveyance (K) were tested for their correlation with the measured quantity of works. Here, the canal quantities of engineering works are considered geometrically proportional to its cross sectional area. So that, we can use the canal cross sectional area (A c) to represent the associated quantities of engineering works. With the help of the matrix type scatter plot, the above design variables were plotted against the respective canal cross sectional area and the result was given in figure (5.4) below.
Figure (5.4) shows that the canal bed slope makes poor correlation with the canal cross sectional area though these slopes are none scouring and none silting ranging in (0.001 to 0.002) values. In addition to this, the result shows that the canal cross sectional area (Ac) makes better empirical trend with conveyance (K) than it makes with the canal design discharge (Qd) and bed slope (So). This trend was also tested with curve fitting of data points as shown in figure (5.5).
Matrix Plot of canal section area, Ac vs Qd, So, K
Figure 5-4 Matrix plot showing trend of canal section area against its design variables
38 Fitted Line Plot canal section area, Ac vs K
1.6 Regression Ac = 0.1815 + 0.02464 K+ 0.000630 K**2 95% C I 1.4 S 0.0838046 Canal section area,Ac (m2)
R-Sq 89.0% 1.2 R-Sq(adj) 88.3%
1.0
0.8
0.6
0.4
0.2
0.0 0 5 10 15 20 25 30 Conveyance, K
Figure 5-5 Curve fitting of canal section area (Ax) against canal conveyance (K)
The trend of canal cross sectional area (Ac) against the conveyance (K) is found good fit with an R2 value of 0.89. In general the results show that conveyance of the canal (K) is found the most important design variable to account the main canal quantities of works. Conveyance (K) is a canal section factor that combines the design discharge and the bed slope(Te Chow, 1959).
Since the conveyance can represent the canal section narrowness and wideness, we can employ the combined factor (Kc*Lc) conveyance times canal length as explanatory variable in formulating cost function using MLR analysis.
In regarding to the main canal design consideration, figure (5.6) and (5.7) depicts important results about the irrigation duty, canal design discharge and command area. The irrigation duty graphical data summery for the sampled 35 diversion irrigation projects is presented in figure (5.6).
Figure (5.6) shows that, the whole sample irrigation duty lies in the range of 1.1 to 2.2 with the 1.55 l/s/ha medial value.
Moreover, about the 62.9 % of the samples irrigation duties lies in the inter quartile range of 1.37 to 1.80 l/s/ha.
39 Summary for Irrigation Duty,D (l/s/ha) A nderson-D arling N ormality Test A -S quared 0.32 P -V alue 0.519 M ean 1.5995 S tD ev 0.2439 V ariance 0.0595 Frequency
S kew ness 0.313951
Kurtosis -0.106421 N 35 M inimum 1.1000 1st Q uartile 1.3700 M edian 1.5500 3rd Q uartile 1.8000 1.2 1.5 1.8 2.1 M aximum 2.2000 95% C onfidence Interv al for M ean 1.5157 1.6832 95% C onfidence Interv al for M edian D,L/s/ha 1.4861 1.7209 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter vals 0.1973 0.3195 Mean
Median
1.50 1.55 1.60 1.65 1.70 1.75
Figure 5-6 Summary of sampled irrigation design Duty
Figure 5-7 Line plot showing Irrigation area against design discharge
Figure (5.7) depicts an empirical correlation between the irrigation command area and canal design discharge.
40 About 77% of sample projects have command sizes are in the range of 40ha to 125 ha. These correspond to the 0.08 m3/s to 0.23m3/s canal design discharge values.
Furthermore, the river base flow potential and the availability of irrigation command area are the most important limiting resources in planning and layout design of a river diversion structure (Arora, 2002; Baban, 1995; Garg, 1987). For a given diversion project the possible limiting factor can be either the command size or the river base flow. In case where the limiting factor is the river base flow, one can trace the corresponding irrigation command area from above empirical equation shown in figure (5.7). On other hands, we can also determine the corresponding diversion canal discharge if the command size is found to be a limiting factor.
5.2.2 Head Work Design Variables
Sort of design variables of head work have been already identified in table 3.1. These includes the weir body height (Hb), weir length (Lb), base width (B), top width (T); height of wing wall and divide (Hw), length of Wing wall (Lw), apron length (La). The graphical data summery presented in figure (8.26, 8.27, 8.28) shows the geometric proportionalities for ratio of apron thickness to the weir height (ta /Hb), downstream cutoff depth to the weir height (d2/Hb), cutoff depth ratio (d2/d1). So the variable weir height (Hb) expressed in the cost function can account for the apron thickness and cutoff depths. One can also initially deduce the corresponding apron length (L a) from the weir height (Hb) using figure 8.29. The graphical data summary result of some of the head work design variables is presented below.
Summary for weir height,H(m)
A nderson-D arling N ormality T est A -S quared 1.23 P -V alue < 0.005 M ean 2.1923 S tD ev 0.8899 V ariance 0.7919 S kew ness 1.15914 Kurtosis 0.96939 Frequency
N 35 M inimum 1.0000 1st Q uartile 1.5000 M edian 2.0000 3rd Q uartile 2.6600 1 2 3 4 M aximum 4.7000 95% C onfidence Interv al for M ean 1.8866 2.4980
H,m 95% C onfidence Interv al for M edian
1.6609 2.1939 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter v als 0.7198 1.1659 Mean
Median
1.6 1.8 2.0 2.2 2.4 2.6
Figure 5-8 Summary of diversion weirs heights
41 Figure (5.8) shows sampled diversion projects have median weir height about 2.0m with minimum and maximum values ranging 1.0 to 4.7m. In addition, 51.4% of the samples diversion height lies within 1.55m and 2.66m inter quartile ranges.
Summary for weir length, L
A nderson-D arling N ormality Test A -S quared 1.18 P -V alue < 0.005 M ean 24.142 S tD ev 11.129 V ariance 123.858 Frequency
S kew ness 1.30107
Kurtosis 1.98546 N 35 M inimum 8.200 1st Q uartile 18.000 M edian 22.000 3rd Q uartile 30.000 10 20 30 40 50 M aximum 56.000 95% C onfidence Interv al for M ean 20.319 27.965 95% C onfidence Interv al for M edian L,m 19.652 25.348 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter vals 9.002 14.581 Mean
Median
20 22 24 26 28
Figure 5-9 summary of diversion weirs length
Figure (5.9) indicates that sampled diversion projects have median weir length about 22m with minimum and maximum values ranging from 8.2 to 56m. In addition, 65.7% of the samples weir length lies within 18m and 30m inter quartile ranges.
Another important result depicted in figure (5.10) is the diversion weir overturning factor of safety (FOS). It shows that the whole sample FOS lies in the range of 1.69 to 2.48 and 54.3% of the samples lie in the inter quartile range of 1.82 to 2.15. Beside to this, 57.1% of samples with FOS values less than two can be regarded as conservative design or economical. Similarly, 42.9% of samples having FOS values greater than two (2.0) are likely regarded as over designed despite the value of FOS is relative.
These FOS values highlights about sampled diversion weirs are technically safe and economical. So, sampled diversion data are reliable enough to develop predictive cost function.
42 Summary for Weir body Overturning-FOS A nderson-Darling N ormality Test A -S quared 0.81 P -V alue 0.033 M ean 2.0053 S tDev 0.2384 V ariance 0.0568 Frequency
S kew ness 0.599104
Kurtosis -0.574729 N 35 M inimum 1.6899 1st Q uartile 1.8176 M edian 1.9825 3rd Q uartile 2.1503 1.8 2.0 2.2 2.4 M aximum 2.4811 95% C onfidence Interv al for M ean 1.9234 2.0872 95% C onfidence Interv al for M edian FOS 1.8905 2.0231 95% C onfidence Interv al for S tDev 9 5 % C onfidence Inter vals 0.1928 0.3123 Mean
Median
1.90 1.95 2.00 2.05 2.10
Figure 5-10 Summary of diversion weir overturning factor of safety
Summary for wing wall height, Hw
A nderson-D arling N ormality Test A -S quared 1.57 P -V alue < 0.005 M ean 4.0303 S tD ev 1.4606 V ariance 2.1333 Frequency
S kew ness 1.53190
Kurtosis 4.12753 N 70 M inimum 1.5000 1st Q uartile 3.0000 M edian 3.8400 3rd Q uartile 4.5250 2 4 6 8 10 M aximum 10.2000 95% C onfidence Interv al for M ean Hw,m 3.6820 4.3786 95% C onfidence Interv al for M edian 3.5000 4.2648 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter vals 1.2523 1.7526 Mean
Median
3.50 3.75 4.00 4.25 4.50
Figure 5-11 Summary of sampled headwork wing wall height
43 Figure (5.11) shows that the total diversion wing walls height lies in the range of 1.50 to 10.2 m and 68.6% of the samples are found in the inter quartile range of 3.0 to 4.52 m.
Likewise, figure (5.12) shows that diversion wing walls length lies in range (11.0 to 40.0) m with a median 11.0m. Likely, 62.9% of sample found in Inter quartile range (8.0 to 15.0) m unit.
Summary for Length of wingwall, Lw
A nderson-D arling N ormality Test A -S quared 1.36 P -V alue < 0.005 M ean 12.529 S tD ev 6.545 Frequency
V ariance 42.837 S kew ness 2.29040 Kurtosis 8.28257 N 35 M inimum 4.000 1st Q uartile 8.000 M edian 11.000 3rd Q uartile 15.000 10 20 30 40 M aximum 40.000 95% C onfidence Interv al for M ean Lw,m 10.280 14.777 95% C onfidence Interv al for M edian 10.000 13.000 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter vals 5.294 8.575 Mean
Median
10 11 12 13 14 15
Figure 5-12 Summary of sampled headwork wing wall length
The importance of each design variable was tested by plotting the respective cross sectional area of the structure against the associated quantities of engineering works. Figure (5.15) shows the plotted of weir body section area against its relevant design variables like weir height (H), base width (B) and top width (T). This was employed to test for the statistical tend existence between the section quantities and relevant design variables. The same was done for the wing wall structures as well shown in annex figure (8.20).
Furthermore, different scholars like (Novak, 2007 and Garg, 1987) pointed that the elementary profile of simple gravity structures have approximated 2/3 value of H/B ratio. Figure (5.13) and (5.14) also shows similar results for this research. Figure (5.13) shows the (H/B) ratio of sampled diversion weirs lies in the range of 0.34 to 0.75 with the median value of 0.633. About 57.1% of samples have H/B ratio values lies in 0.53 to 0.704 inter quartile range.
44 Similarly, figure (5.14) depicts that the (T/B) ratio of sampled diversion weirs lies in range 0.218 to 0.577 with medial value of 0.333. About the 60.0% of samples T/B ratio lies in 0.28 to 0.40 Inter quartile range. These results indicate diversion headwork gravity structures have a certain approximate value of geometric ratios.
Summary for Weir Body, H/B
A nderson-D arling N ormality Test A -S quared 1.05 P -V alue 0.008 M ean 0.61166 S tD ev 0.11335 V ariance 0.01285 Frequency
S kew ness -0.821441
Kurtosis -0.304255 N 35 M inimum 0.34286 1st Q uartile 0.53333 M edian 0.63333 3rd Q uartile 0.70370 0.4 0.5 0.6 0.7 M aximum 0.75000 95% C onfidence Interv al for M ean 0.57272 0.65060 95% C onfidence Interv al for M edian 0.60000 0.68265 H/B ratio 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter vals 0.09169 0.14852 Mean
Median
0.56 0.58 0.60 0.62 0.64 0.66 0.68
Figure 5-13 Summary of stable weir body (H/B) ratio
Summary for weir body,T/B
A nderson-D arling N ormality Test A -S quared 0.91 P -V alue 0.018 M ean 0.35858 S tD ev 0.08974 Frequency
V ariance 0.00805 S kew ness 0.680432 Kurtosis -0.294724 N 35 M inimum 0.21818 1st Q uartile 0.28571 M edian 0.33333 3rd Q uartile 0.40000 0.2 0.3 0.4 0.5 0.6 M aximum 0.57692 95% C onfidence Interv al for M ean 0.32776 0.38941 95% C onfidence Interv al for M edian T/B ratio 0.30234 0.38079 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter v als 0.07259 0.11758 Mean
Median
0.30 0.32 0.34 0.36 0.38 0.40
Figure 5-14 Summary of stable weir body (T/B) ratio
45 Matrix Plot of Weir body Section area,A vs H, B, T 3 6 9 25
20 Section Area,A(m3/m)
15
10
0 1 3 5 0.6 1.2 1.8 Height,H(m) Base width,B(m) Top width,T(m)
Figure 5-15 Matrix plot showing trend of Weir body stable section area against relevant design variables
Fitted Line Plot of weir body stable section area
25 Regression A =- 5.172 + 9.090 H-3.192H**2+0.5405H**3 95% C I
S 0.625163 20 R-Sq 98.2% Section Area,A(m3/m)
R-Sq(adj) 98.0%
15
10
0 1 2 3 4 5 Height,H(m)
Figure 5-16 Curve fitting of weir body stable section area (A) against height (H)
The matrix scatter plot result shown in figure (5.15) indicates that the weir body stable section area (A) makes smoother trend with height (H) than the rest of design variables. Its statistical trend curve fitting shown in figure (5.16) also indicates good fit with an R 2 value of 0.98.
46 Above all results show that height of weir (H) is found the most important design variable to account the weir body section quantities of work in cost estimation procedure. Therefore, we can use the combined factor (H*Lb) height times weir length in formulating empirical function with the cost of diversion structure. Likewise, we can apply a similar combined factor (Hw*Lw) height times wing wall length for the wing wall headwork component.
Cost Function Formulation
In previous section, we had already identified list of relevant design variables for the main canal, weir part and wing walls. And then, the combined cost of main canal and head work was mathematically aggregated with these sorts of important design variables which are considered as cost driver factors. The multiple linear regression (MLR) analysis of cost function formulation result is presented below.
Table 5-1 CT-MLR analysis result with sort of design variables for rocky class diversions in Awi &W/Gojjam area of year 2015
47 Where: CT - is total cost per thousands of Ethiopian birr, Qc-is canal discharge, Kc-is canal conveyance, Lc-is canal length, Hb- is weir height, Lb-is weir length, La- is apron length, Hw-is height of wing wall, Lw – is the length of a wing wall.
Residual frequancy Histogram of MLR analysis
(response is CT)
4 Frequency
0 -2 -1 0 1 2 Standardized Residual
Figure 5-17 Residual frequency histogram of CT-MLR analysis result for rocky class diversions in Awi &W/Gojjam area of year 2015
Table 5-2 CT- MLR analysis result with sort of design variables for alluvial class diversions in Awi &W/Gojjam area of year 2015
Where: Where: CT - is total cost per thousands of Ethiopian birr, Qc-is canal discharge, Kc-is canal conveyance, Lc-is canal length, Hb- is weir height, Lb-is weir length, La- is apron length, Hw- is height of wing wall, Lw – is the length of a wing wall.
Sample of total cost multiple linear regression (MLR) analysis result is presented in table (5.2) & (5.3) for both rocky and alluvial foundation class of diversion weirs found in Awi and W/Gojjam spatial zone.
Figure 5-18 Residual frequency histogram of CT- MLR analysis result for alluvial class diversions in Awi &W/Gojjam area of year 2015
Cost function results shown in table (5.1) and (5.2) were formulated based on equation (4.3). The result residual frequency histogram shows some values higher at -1.5 and 0.5 ends and relatively lower values near zero as shown in figure (5.17). This indicates the total cost MLR empirical truncation does have significant outliers.
Table (5.1) of the total cost MLR analysis result shows that only the (Lc, Lb, Lb, Lw) design variables are found sensitive with test statistics P-value less than 0.1 for the 90% confidence interval considered. On other hand, it makes insignificant for (Kc, Qc , La, Hw) design variables with P-value greater than 0.1.
49 Similarly, table (5.2) and figure (5.18) shows the total cost MLR analysis result for alluvial class of diversions. The residual histogram indicates frequency values are relative higher near zero. This implies that the total cost MLR truncation has significant outliers. Moreover, table (5.2) shows that the total cost-MLR analysis makes aggregate only for (Lc, Lb) design variables with P-value less than 0.1 for the 90% confidence interval considered. On the contrary, it makes insignificant for (Kc, Qc, Hb, La, Lw) design variables with P-value greater than 0.1. However, as shown in figure (5.4 & 5.5), the canal conveyance design variable (Kc) was found important that makes trend with variation of canal section quantity of works. Figure (5.15 & 5.16) also analogously indicates that wing wall height (Hw) can make an important trend with variation of wing wall section quantities of work.
As a result, these all important design variables which can make significant trend against the factor of section quantity of work were desired to be aggregated with in total cost MLR-eqn. This was done by applying a likely regarded combination of design variables as explanatory variable of MLR analysis as expressed in equation (4.4).
Sample of total cost MLR analysis result is presented below in table (5.3) & (5.4) for both rocky and alluvial foundation class of diversion weirs found in Awi and W/Gojjam spatial zone.
Table 5-3 CT- MLR analysis result with combined design variables for rocky class diversions in Awi &W/Gojjam area of year 2015
50 Where: CT - is total cost estimated per thousands of Ethiopian birr, (L c*Kc)-is product design variable of canal length and conveyance, (Hb*Lb)-is product design variable of weir height and length, (La*Lb)-is product of apron length and weir length, (Lw*Hw) – is product of wing wall length to its height.
Residual frequncy Histogram of CT-MLR analysis
(response is CT/1000-(AWG-15))
4 Frequency
0 -2 -1 0 1 2 Standardized Residual
Figure 5-19 Residual frequency histogram of CT- MLR analysis result for rocky class diversions in Awi &W/Gojjam area of year 2015
Table 5-4 CT- MLR analysis result with combined design variables for alluvial class diversions in Awi &W/Gojjam area of year 2015
Reg. Statistics S 778.153 2 R 0.99 R2(adj) 0.989 R2(pred) 0.62 N 13 Predictor Coef SE t Stat P-value constant 153.235 625.122 0.245 0.813 LcKc 0.143 0.006 23.426 0.0001 HbLb 39.233 5.774 6.795 0.0001 LaLb 3.886 1.858 2.091 0.070 LwHw -9.106 10.661 -0.854 0.418 Where: CT - is total cost estimated per thousands of Ethiopian birr,
51 (Lc*Kc)-is product design variable of canal length and conveyance, (H b*Lb)-is product design variable of weir height and length, (La*Lb)-is product of apron length and weir length, (Lw*Hw) – is product of wing wall length to its height.
Figure 5-20 Residual frequency histogram of CT- MLR analysis result for alluvial class diversions in Awi &W/Gojjam area of year 2015
As result shown in table (5.3) the total cost MLR analysis makes significant correlation with the pre-defined meaningful combined design variables for rocky class of diversion weirs. The total cost MLR analysis makes aggregate for (LcKc, HbLb, LwHb), combined design variables, with P- value less than 0.1 for the 90% confidence interval considered. But the apron variable (LaLb) is found insignificant with p-value greater than 0.1 for rocky diversions. In addition, the result residual frequency histogram with higher values near zero and relatively lower values to the other ends as shown in figure (5.19). This indicates the total cost MLR function has less significant error. The (R2) and (R2adj) values greater than 0.75 are also indicates that the model fits the data well.
Similarly, table (5.4) and figure (5.20) shows the total cost MLR analysis result for alluvial class of diversions. Table (5.4) shows that the total cost MLR analysis makes aggregate only for (Lc*Kc, Hb*Lb, La*Lb) combined design variables with P-value less than 0.1. On other hand it makes insignificant for (Lw*Hw) an explanatory variable with P-value greater than 0.1 for the 90% confidence interval considered. The residual frequency histogram also indicates that total cost
52 function has a relatively lower residual error compared to figure (5.18). In addition to this the coefficient of determination (R2) and (R2adj) with 0.99 and 0.98 respective values are also another indicator that the model fits the data well.
Finally, the total cost can be empirically truncated using Multiple Linear Regression (MLR) analysis with important combined design variables considered as explanatory variables. The general cost function equation (4.5) can be rewritten as follow.
C L k H L L L . . L H T 0 1 c c 2 b b 3 a b 4 w w .............(5.1)
0 m0 t b0
1 m1 t b1
2 m2 t b2
3 m3 t b3
4 m4 t b4 Where; CT - is total cost estimated per thousands of Ethiopian birr,
Coefficients; a0, a1, a2, a3, a4- are parameters of Lc*Kc, Hb*Lb, La*Lb, Hw*Lw respectively,
(LC*Kc)-is product design variable of canal length and conveyance, (Hb*Lb)-is product design variable of weir height and length, (La*Lb)-is product of apron length and weir length, (Lw*Hw) – is product of wing wall length to its height. (mo, m1,m2,m3,bo,b1,b2,b3) are coefficients of parametric equation defined for each spatial zones, (t)-is time variant determined by calendar year minus 2000,as the base year.
The parameters or coefficient of explanatory variables shown in equation (5.1) are found important to account the spatial and temporal variation of unit rate in conjunction with the unit rate trend function developed. As a result, the cost function can possibly aggregate the spatial and temporal variation of cost. Beside to this, the cost function equation was developed for the rocky and alluvial foundation scenarios. It was formulated by using 22 rocky and 13 alluvial foundation geology classes of sampled diversion weir projects.
Table (5.5) shows the summarized results of cost function parameters defined for each of the seven study area spatial zones with the two foundations class of headwork as per equation 5.1.
53 Table 5-5 Summary of cost function parameters for calibrated equations
No Spatial zone Foundation scenario Parameter Equations R2
Where; a0, a1, a2, a3, a4- are coefficients of Lc*Kc, Hb*Lb, La*Lb, Hw*Lw respectively written in equation 5.1 and (t)-is time variant given by calendar year minus 2000,as the base year.
The result of the total cost function was found statistically sound result with a coefficient of determination (R2) ranging from 0.90 to 0.91 and 0.98 to 0.99 minimum to maximum values for rocky and alluvial foundation scenarios respectively. Generally results shows that the total cost estimate of river diversion structure depends on some factors of sensitive design variables
55 combination including canal length (Lc), canal conveyance (Kc), weir height (Hb), weir length (Lb), apron length (La), wing wall length (Lw) and height (Hw). The variation of these sensitive design variables has the significant implication on the total cost estimate. Moreover, the total cost estimate can be truncated in terms of the above sensitive design variables combinations to give us the total cost function.
Hence, the new cost function can then be further applied to quantitatively evaluate and optimize for alternative sites of river diversion structures.
Validation
Validation of the newly developed cost function was carried out to evaluate its overall performance in predicting observed costs. For the good predictive ability of the model the predicted costs should approach the measured costs with statistical acceptable performance indicator values.
The validation was done using randomly picked data from sample projects. The raw unit rate data in combination with spatially clustered quantity of engineering work was used to give us the observed cost estimates for each of the seven study area spatial zones.
56 Figure 5-21 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in Awi & W/Gojjam spatial zone
Figure 5-22 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in N/Gondar spatial zone
Figure 5-23 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in S/Wollo &Oromiya spatial zone
As shown in the above figures, the validation result indicates that the total cost function is found adequately fitting with observed total cost for each of 7 study area spatial zones over the years
57 2010 to 2015. The validation performance was indicated with coefficient of determination (R 2) ranging from 0.813 to 0.908 and 0.97 to 0.99 minimum and maximum values for rocky and alluvial foundation scenarios respectively in the entire study area.
For specified spatial zones, the cost function validation result shows slight variations within years as seen from the (R2) values changing after the first decimal place.
Hence, this developed cost function can be applied as a regional planning tool to predict total cost estimates for river diversion design works with a 90% of confidence interval or 10% cost variation allowed.
Application of Cost Function
The new cost function can be applied to solve practical diversion site feasibility evaluation problems. Since it was tested and found to give us statistically valid results. So it can be used to predict the cost estimates of river diversion structures. This is so important to ensure reliable decision making during diversion site feasibility study.
With provision of the cost function expressed in terms of sensitive design variables, one can quantitatively evaluate for feasibility of candidate river diversion sites by comparing the costs of individual sites. This is possible when the designer has collected relevant data on the variables of the cost function either from reconnaissance study or office desk work.
In addition to this, the designer can possibly optimize the desired for the new mathematical planning model in order to select the least cost river diversion layout in the presence of several alternative sites.
Therefore, we can employ the above derived cost function as an objective function of least cost river diversion layout design, or site selection optimization problem under set of constraints. Optimization constraints can be set by the user to account for the actual sites feasibility condition described in terms of decision variables.
So it can be rewritten as linear programming optimization problem as follow;
C T L k H L L L . . L H 0 1 c c 2 b b 3 a b 4 w w .............(5.2)
Possible Constraints:
58 K c C1 H H H C 0 A ; D0 Lc b D 1 E 0 b E 1 F 0 b F 1 ; L b ; L w ; Where: CT - is total cost function estimated in thousands of Ethiopian birr,
Coefficients; a0, a1, a2, a3, a4- are parameters of Lc*Kc, Hb*Lb, La*Lb, Hw*Lw respectively,
(LC*Kc)-is product design variable of canal length and conveyance, (Hb*Lb)-is product design variable of weir height and length, (La*Lb)-is product of apron length and weir length, (Lw*Hw) – is product of wing wall length to its height, Lw – is the longitudinal length of wing walls, A-is irrigation command area,
(C0 &C1)-refers to the (Kc/A) lower and upper boundary numerical values, (D0 & D1)- refers to the (Hb/Lc) lower and upper boundary numerical values, (E0&E1)- refers to the (Hb/Lb) lower and upper boundary numerical values, (F0&F1)- refers to the (Hb/Lw) lower and upper boundary numerical values. It can be also remarked that the set constraints should be readjusted with unit conversion factors so as to give us relevant numerical figures. The above constrained optimization problem can now be solved with the help of excel solver (Chapra & Canale, 2012).
Moreover, the optimization result can give us the combination of theoretical optimum decision variables corresponding to the least cost river diversion layout. These theoretical decision variables can now be approximated with one that qualitatively identified alternative sites. So we will be able to select the most feasible site among candidates. Hence, the cost function can be used as quantitative planning model for river diversion layout design and site selection procedure.
59 6 SUMMARY AND CONCLUSIONS
River water resource development is widely interfered with the development river diversion infrastructure projects. This can be done for different scale of irrigation projects. However, River diversion layout design and site selection problem is still found an important water resource system with some degree of complexity and alternatives. So it was desired to be quantified and optimized in order to improve the overall efficiency of river water resource development. This can be provided by formulating the total cost function. The total cost of river diversion structure can empirically be correlated with some sensitive design variables of headwork and main canal including; canal length (Lc), canal conveyance (Kc), weir height (Hb), weir length (Lb), apron length (La), wing wall length (Lw) and wing wall height (Hw) regardless of others variables. This can gives us the predictive cost function expressed in terms of the above sensitive design variables relevant combination in order to give us the total cost. The cost function makes aggregate the unit rate trend function and measured quantities of engineering works together. Its parameters were defined for each of the seven clustered study area spatial zones with the two general foundation conditions of headwork.
The cost function overall validation results show that there is an agreement between the predicted and observed costs with an R2 ranging 0.813 to 0.908 and 0.97 to 0.99 minimum and maximum values for rocky and alluvial foundations respectively. The new developed cost function can be applied to predict the cost for planning and evaluation in river diversion site selection problems. By now, the research problem that was the existing qualitative based river diversion site selection approaches can be improved to the new quantitative planning tool. The research specific objectives were also addressed well with statistically sounding results of cost function and the unit rate trend function formulation.
The new quantitative planning model can be employed to identify the least cost feasible river diversion layout and site selection procedure. Furthermore, it can allow us to consider a wide range of alternative sites to be evaluated quantitatively. However, the cost function need to be reformulated in need of updating the unit rate trend function after several years when unit prices of common construction work items are expected outrange. Therefore, having reconnaissance or desk work survey data and some hydrologic data, we can apply the cost function for the
60 desired spatial zone to identify the least cost feasible river diversion site among several alternatives.
It is possible to recommend that the science of the river water resource development, in particular the efficiency of diversion weir site selection procedure, can be better improved with development of total cost function. If the future researchers can avail adequate data of diversion structures, it is possible to build an advanced quantitative planning model in the study area, Amhara region.
61 7 REFERENCES
Arora, KR. (2002). Irrigation, water power and water resources engineering: Standard Publisher Distributors. Awulachew, Seleshi Bekele, Yilma, Aster Denekew, Loulseged, Makonnen, Loiskandl, Willibald, Ayana, Mekonnen, & Alamirew, Tena. (2007). Water resources and irrigation development in Ethiopia (Vol. 123): Iwmi. Baban, Rozgar. (1995). Design of diversion weirs: small scale irrigation in hot climates: John Wiley & Sons. BoWRD, ANRS. (2009_2015). IFAD & MDG Irrigation projects in Amhara Region, Ethiopia. C.Mitchell, Sherri Anderson and. (2015). Cost Function. from http://www.wisegeek.org/what-is- a-cost-function.htm#didyouknowout, Chapra, Steven C, & Canale, Raymond P. (2012). Numerical methods for engineers (Vol. 2): McGraw-Hill. Chen, Wenfang, de Kee Fong, D, & Kaloni, Purna N. (2003). Advanced mathematics for engineering and science: World Scientific. Chow, Vent. (1959). Open Channel Hydraulics. New York: McGraw-Hill. Continental Consultants, Concert Engineering and Consulting Enterprise P.L.C(CECE). (2002a). Design guideline on diversion structure, guideline, manuals & standard design of small & medium scale irrigation projects, Ethiopia. Continental Consultants, Concert Engineering and Consulting Enterprise P.L.C(CECE). (2002b). Study guideline on irrigation agronomy, procedural guideline for study of small and medium scale irrigation projects, Ethiopia. Elshorbagy, A, Corzo, G, Srinivasulu, S, & Solomatine, DP. (2009). Experimental investigation of the predictive capabilities of data driven modeling techniques in hydrology–Part 1: Concepts and methodology. Hydrology and Earth System Sciences Discussions, 6(6), 7055-7093. Garg, Santosh Kumar. (1987). Irrigation engineering and hydraulic structures: Khanna publishers. Girma, Michael M Awulachew, & Seleshi, B. (2007). Irrigation practices in Ethiopia: Characteristics of selected irrigation schemes (Vol. 124): IWMI. Helsel, Dennis R, & Hirsch, Robert M. (1992). Statistical methods in water resources (Vol. 49): Elsevier. Leopold, Luna Bergere, & Maddock Jr, Thomas. (1953). The hydraulic geometry of stream channels and some physiographic implications. Loucks, Daniel P, Van Beek, Eelco, Stedinger, Jery R, Dijkman, Jozef PM, & Villars, Monique T. (2005). Water resources systems planning and management: an introduction to methods, models and applications: Paris: UNESCO.
62 McCartney, Matthew P. (2007). Decision support systems for large dam planning and operation in Africa (Vol. 119): IWMI. MoWE. (2013). Irrigation scheme report. from http://www.mowr.gov.et/index.php MoWR. (1998). Abbay river basin integrated development master plan study, vol.II (Addis Ababa, Ethiopia). Novák, Pavel, Moffat, AIB, Nalluri, Chandra, & Narayanan, R. (2007). Hydraulic structures: CRC Press. Reale, S. (1993). Design and analysis of experiments: By Douglas C. Montgomery. John Wiley, Chichester, 1991. 649 pp. ISBN 0-471-52000-4 (HB), 0-471-52994-X (PB). Price£ 51· 00 (HB),£ 19· 95 (PB): Elsevier. Saliha, Alemayehu Habte. (2012). Decision support tool to optimize the operation of multi- purpose reservoirs: a case study in the lake tana catchment, Ethiopia. Sedighizadeh, Saeed, Mansoori, Abbas, Pirestani, Mohammad Reza, & Sedighizadeh, Davoud. (2011). A new model for economic optimization of water diversion system during dam construction using PSO algorithm. 5. Singh, Ajay S, & Masuku, Micah B. (2012). An insight in statistical techniques and design in agricultural and applied research. World Journal of Agricultural Sciences, 6, 568-584. Te Chow, Ven. (1959). Open channel hydraulics. TURAN, KAMİL HAKAN. (2004). Development of a computer program for optimum design of diversion weirs. Middle East Technical University. Vedula, S, & Mujumdar, PP. (2005). Water resources systems: modelling techniques and analysis: Tata McGraw-Hill. Yazdandoost, Farhad, & Attari, Jalal. (2004). Hydraulics of dam and river structures: Taylor & Francis.
63 8 ANNEXES Table 8-1 Summary of table for sample diversion weirs used in the research
No Diversion weir Geological Woreda North,m East,m Elv,m CT/1000
20.0 mas Linear (Cyclpn) 15.0 Linear (C-10 ) 10.0 Linear (C-20) Cypn/100 = 1.509x - 3019. Mas/100 = 1.187x - 2374. 5.0 R² = 0.986 Linear (C-25) R² = 0.944 Linear (mas) 0.0 2006 2008 2010 2012 2014 2016 Time,T,year
Figure 8-17 Concrete and masonry works unit rate trend function for N/Shewa spatial Zone
Bar&Finishings UnitRate Trend Function
20.0 0.0
18.0 Bar/10 = 0.379x - 759.0 1.0 16.0 R² = 0.939
14.0 Plst/10 = 1.346x - 2697.
R² = 0.890 2.0 Plast 12.0 Pt ETB/m2
ETB/Kg
10.0 3.0 Bar-Ø
8.0 Linear (Plast) 4.0 Linear (Pt) 6.0 Linear (Bar-Ø) 4.0 5.0 Pt/10 = 0.966x - 1938. 2.0 R² = 0.891 0.0 6.0 2006 2008 2010 2012 2014 2016 Time,T,year
Figure 8-18 Bar and finishing works unit rate trend function for N/Shewa spatial Zone
73 Summary for canal b/d ratio A nderson-D arling N ormality Test A -S quared 1.59 P -V alue < 0.005 M ean 0.85087 Frequency S tD ev 0.18929 V ariance 0.03583 S kew ness 1.44809 Kurtosis 2.10538 N 35 M inimum 0.62500 1st Q uartile 0.70000 M edian 0.83333 3rd Q uartile 0.88889 0.6 0.8 1.0 1.2 1.4 M aximum 1.45000 95% C onfidence Interv al for M ean 0.78585 0.91589 95% C onfidence Interv al for M edian b/d ratio 0.75000 0.84989 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter vals 0.15311 0.24801 Mean
Median
0.75 0.80 0.85 0.90
Figure 8-19 summary of main canal (b/d) ratio
Matrix Plot of Wing Wall section area,Aw vs H, B, T
0 5 10 50
40 Section area,A(m3/m)
30
20
10
0 3 6 9 0.50 0.55 0.60 Height,H(m) Base width,B(m) Top width,T(m)
Figure 8-20 Matrix plot showing trend of wing wall stable section area against relevant design variables
74 Summary for wing wall (Hu/Hd) ratio A nderson-Darling N ormality Test A -S quared 0.33 P -V alue 0.497 M ean 1.3393 S tDev 0.2966 V ariance 0.0880 Frequency
S kew ness 0.0562262
Kurtosis -0.0404272 N 35 M inimum 0.6308 1st Q uartile 1.1026 M edian 1.3611 3rd Q uartile 1.5000 0.6 0.9 1.2 1.5 1.8 M aximum 1.9615 95% C onfidence Interv al for M ean 1.2375 1.4412 95% C onfidence Interv al for M edian b/d ratio 1.1405 1.4565 95% C onfidence Interv al for S tDev 9 5 % C onfidence Inter vals 0.2399 0.3886 Mean
Median
1.1 1.2 1.3 1.4 1.5
Figure 8-21 summary of masonry wing wall (b/d) ratio
Figure 8-22 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in S/Gondar spatial zone
75 Figure 8-23 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in E/Gojjam spatial zone
Figure 8-24 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in N/Wollo& Waghimira spatial zone
76 Figure 8-25 Cost validation of year 2015 for Rocky &Alluvial foundation scenarios in N/shewa spatial zone
Summary for ta/Hb
A nderson-D arling N ormality Test A -S quared 0.48 P -V alue 0.189 M ean 0.43109 S tD ev 0.21977 V ariance 0.04830 Frequency
S kew ness 0.95265
Kurtosis 1.17574 N 13 M inimum 0.12000 1st Q uartile 0.27767 M edian 0.40000 3rd Q uartile 0.55833 0.2 0.4 0.6 0.8 1.0 M aximum 0.93750 95% C onfidence Interv al for M ean 0.29829 0.56390 95% C onfidence Interv al for M edian ta/Hb ratio 0.28789 0.51833 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter vals 0.15760 0.36279 Mean
Median
0.30 0.35 0.40 0.45 0.50 0.55 0.60
Figure 8-26 graphical summery of ratio of apron thickness (ta) to weir height (Hb) for alluvial foundation diversions
77 Summary for d2/Hb A nderson-D arling N ormality Test A -S quared 0.33 P -V alue 0.462 M ean 1.8270 S tD ev 0.7111 V ariance 0.5056 S kew ness 0.914174 Frequency Kurtosis 0.653539 N 13 M inimum 1.0000 1st Q uartile 1.1936 M edian 1.6800 3rd Q uartile 2.2813 1.0 1.5 2.0 2.5 3.0 3.5 M aximum 3.4351 95% C onfidence Interv al for M ean 1.3973 2.2567 95% C onfidence Interv al for M edian d2/Hb ratio 1.2387 2.2697 95% C onfidence Interv al for S tD ev 9 5 % C onfidence Inter vals 0.5099 1.1738 Mean
Median
1.2 1.4 1.6 1.8 2.0 2.2 2.4
Figure 8-27 graphical summery of ratio of d/s cutoff depth (d2) to weir height (Hb) for alluvial foundation diversions
Summary for d2/d1
A nderson-Darling N ormality Test A -S quared 0.79 P -V alue 0.030 M ean 1.8876 S tDev 0.5064 V ariance 0.2564 Frequency
S kew ness -1.06201
Kurtosis 0.63272 N 13 M inimum 0.8333 1st Q uartile 1.6500 M edian 2.0000 3rd Q uartile 2.2193 0.8 1.2 1.6 2.0 2.4 M aximum 2.5000 95% C onfidence Interv al for M ean 1.5816 2.1936 95% C onfidence Interv al for M edian d2/d1 ratio 1.7054 2.1772 95% C onfidence Interv al for S tDev 9 5 % C onfidence Inter vals 0.3631 0.8359 Mean
Median
1.6 1.7 1.8 1.9 2.0 2.1 2.2
Figure 8-28 graphical summery of ratio of cutoff depths (d2) to (d1) for alluvial foundation diversions
78 Summary for Hb/La A nderson-Darling N ormality Test A -Squared 0.80 P -V alue 0.028 M ean 0.18363 StDev 0.09898 V ariance 0.00980 Skew ness 1.17113 Frequency
Kurtosis 0.37934 N 13 M inimum 0.07895 1st Q uartile 0.11526 M edian 0.16667 3rd Q uartile 0.23289 0.10 0.15 0.20 0.25 0.30 0.35 0.40 M aximum 0.38462 95% C onfidence Interv al for M ean 0.12382 0.24345 95% C onfidence Interv al for M edian Hb/La ratio 0.11646 0.21631 95% C onfidence Interv al for StDev 9 5 % C onfidence Inter vals 0.07098 0.16339 Mean
Median
0.100 0.125 0.150 0.175 0.200 0.225 0.250
Figure 8-29 graphical summery of ratio of weir height (Hb) to apron length (La) for alluvial foundation diversion
