<%BANNER%>

Corporate-Level Management System for Critical Strategic Resources for Large Construction Program Managers

Permanent Link: http://ufdc.ufl.edu/UFE0022846/00001

Material Information

Title: Corporate-Level Management System for Critical Strategic Resources for Large Construction Program Managers
Physical Description: 1 online resource (188 p.)
Language: english
Creator: Agdas, Duzgun
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: geneticalgorithms, neuralnetworks, resourcemanagement, supplychainmanagement, transportationlogistics
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Resource management has long been one of the rather popular research topics in construction management research. The construction industry is heavily material oriented and better management can provide significant competitive advantages to the parties. The researchers focused on resource management at project level and provided useful methodologies to reduce the effects of resources over the construction schedule and project cost. Yet, this shortsighted approach does not resolve the root cause of the problem most of the large construction portfolio managers face, having the resources available at the construction site. The study aimed to develop a corporate level resource management system for large work programs. The study uses a prediction model that has been successfully used in construction research to determine the future resource requirements and uses the supply chain management principles and logistics to enhance the traditional resource handling processes. Proposed benefits include but are not limited to, improved supply chain integrity, reduced resource related delays, reduced inventory and transportation costs and ultimately improved customer satisfaction.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Duzgun Agdas.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Ellis, Ralph D.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022846:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022846/00001

Material Information

Title: Corporate-Level Management System for Critical Strategic Resources for Large Construction Program Managers
Physical Description: 1 online resource (188 p.)
Language: english
Creator: Agdas, Duzgun
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: geneticalgorithms, neuralnetworks, resourcemanagement, supplychainmanagement, transportationlogistics
Civil and Coastal Engineering -- Dissertations, Academic -- UF
Genre: Civil Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Resource management has long been one of the rather popular research topics in construction management research. The construction industry is heavily material oriented and better management can provide significant competitive advantages to the parties. The researchers focused on resource management at project level and provided useful methodologies to reduce the effects of resources over the construction schedule and project cost. Yet, this shortsighted approach does not resolve the root cause of the problem most of the large construction portfolio managers face, having the resources available at the construction site. The study aimed to develop a corporate level resource management system for large work programs. The study uses a prediction model that has been successfully used in construction research to determine the future resource requirements and uses the supply chain management principles and logistics to enhance the traditional resource handling processes. Proposed benefits include but are not limited to, improved supply chain integrity, reduced resource related delays, reduced inventory and transportation costs and ultimately improved customer satisfaction.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Duzgun Agdas.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Ellis, Ralph D.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022846:00001


This item has the following downloads:


Full Text

PAGE 1

1 CORPORATE-LEVEL MANAGE MENT SYSTEM FOR CRITICAL STRATEGIC RESOURCES FOR LARGE CONSTR UCTION PROGRAM MANAGERS By DUZGUN AGDAS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Duzgun Agdas

PAGE 3

3 To my family for their continuous support

PAGE 4

4 ACKNOWLEDGMENTS I thank m y supervision committee members (Dr. Charles Glagola, Dr. Michael McVay and Dr. Raymond Issa) for their invaluable time, their patience and willingness to share their immense academic and professional experience. I am deeply indebted to my committee chair and mentor Dr. Ralph Ellis. He led me with his knowledge, and also helped me to grow as a person and a researcher. I cannot adequately thank him for his positive influence. I am grateful to my family for bearing the stress of hard times with me and giving me direction when I needed it. Last but not least I thank all my friends at home and in Gainesville for making this a memorable and enjoyable time.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4LIST OF TABLES................................................................................................................. ..........9LIST OF FIGURES.......................................................................................................................10ABSTRACT...................................................................................................................................11CHAPTER 1 INTRODUCTION..................................................................................................................12Construction Industry Overview............................................................................................. 12Discussions about Resource Leveling and Allocation............................................................ 13Problem Definition............................................................................................................. ....13Overview of Solution..............................................................................................................142 LITERATURE REVIEW.......................................................................................................17Resource Management............................................................................................................17Resource Allocation........................................................................................................17Resource Leveling........................................................................................................... 18Prediction Model Selection..................................................................................................... 19Neural Networks..............................................................................................................19Genetic Algorithms......................................................................................................... 20Hybrid Models.................................................................................................................20Supply Chain Management and Cons truction Industry Applications.................................... 20Supply Chain Integrity.................................................................................................... 21Justification of Necessary Changes in the Traditional Construction Supply Chain........ 22Material Flow..................................................................................................................23Significance of Material Flow......................................................................................... 23Material Delivery Systems and Their Effects on SCM...................................................24Push vs. Pull Systems...................................................................................................... 253 SELECTION OF CONTEXTUAL DOMAIN.......................................................................26Necessity of Developing the Proposal around a Case Study.................................................. 26Organizational Details of the Florid a Department of Transportation..................................... 27Justification of Having a New Resource Management Initiative.................................... 28Proposed Benefits............................................................................................................ 29Supply Chain Indications of the Problem............................................................................... 30Supply Chain Integrity.................................................................................................... 31Feasibility of Implementing Suppl y Chain Principles Analysis...................................... 31Prediction Model............................................................................................................. 32

PAGE 6

6 4 PREDICTION MODEL DEVELOPMENT........................................................................... 34Neural Networks.....................................................................................................................34Introduction................................................................................................................... ..34Overview of Topology....................................................................................................35Structure of Neural Networks.................................................................................................36Number of Hidden Layers............................................................................................... 37Number of Hidden Nodes................................................................................................ 38Training....................................................................................................................... ....39Supervised training................................................................................................... 39Unsupervised training.............................................................................................. 40Training Rules.................................................................................................................40Generalized delta rule............................................................................................... 40The shortcomings of GDR and alternatives.............................................................43Summary..........................................................................................................................45Incorporating NNs with other Algorithms......................................................................45Genetic Algorithms............................................................................................................. ....46GA Mechanics.................................................................................................................46Details of the operations........................................................................................... 48Optimization of a simple function............................................................................ 49Genetic Algorithms in Civil Engineering........................................................................ 52Hybrid NN-GA Models..........................................................................................................53Hybrid NN-GA Model Applications in Construction..................................................... 53Conclusion about Predicti on Model Development.......................................................... 54Platform for Application.................................................................................................. 55Comparative Study.......................................................................................................... 565 DATA COLLECTION AND ANALYSIS ............................................................................. 61Data Structure Overview........................................................................................................61Identifying the Strategic Resources........................................................................................ 61FHWA Highway Construction Material Cost Study....................................................... 62FDOT Transportation C onstruction Cost Study.............................................................. 62Discussions on Findings of Two Studies......................................................................... 63Earthwork.................................................................................................................63Labor........................................................................................................................63Justifying the Critical Resource Selections............................................................................64Details of Justification Procedure.................................................................................... 65Relative costs of the chosen resources.....................................................................65Historic cost analysis................................................................................................66Checking the local availability of the resources.......................................................67Contractor surveys....................................................................................................68Conclusion of Justification of the Critical Resources.....................................................68Data Mapping.........................................................................................................................69Filtering the Data............................................................................................................. 69Summary and Conclusion of Data Processing................................................................ 71

PAGE 7

7 6 PREDICTION RESULTS AND INTERPRETATION.......................................................... 75Introduction................................................................................................................... ..........75Details of the Variables................................................................................................... 75Variables to be Determined.............................................................................................76Final Adjustments to the Historic Data........................................................................... 76Eliminating erroneous data points............................................................................ 77Adjusting the project costs with respect to consumer price index...........................77Model Development...............................................................................................................78Measuring Accuracy........................................................................................................ 78Error Term Limitations.................................................................................................... 78Prediction Model Development De tails for Different Resources.................................... 79Prediction Results............................................................................................................. ......79Adjustments to the Future Program Budgets................................................................... 79Adjustments for Substitute Materials.............................................................................. 80Discussions on Prediction Results................................................................................... 80Asphalt.....................................................................................................................80Concrete................................................................................................................... 80Reinforcing and structural steel................................................................................ 81Optional base............................................................................................................81Limestone base......................................................................................................... 81Limitations of the Prediction Model....................................................................................... 82Bid Item-Resource Conversion Issues............................................................................. 82Reliability of WMC as a Project Identifier...................................................................... 82Compatibility of Different Databases.............................................................................. 82User Defined Errors......................................................................................................... 83Resource Leveling/Allocation for Work Program Resource Requirement............................ 83Conclusions and Recommendations.......................................................................................84Conclusions.....................................................................................................................84Recommendations........................................................................................................... 857 APPLICATION OF SUPPLY CHAIN CONTEXT TO FDOT PROGRAM ........................ 92Aggregates..................................................................................................................... .........92Florida Aggregate Overview...........................................................................................92Recycled Asphaltic Product Usage.................................................................................94Problems..........................................................................................................................94Diminishing local supply......................................................................................... 94Importing the aggregates.......................................................................................... 95Transportation constraints........................................................................................95Supply chain issues.................................................................................................. 95Discussion of Limestone as Base Material...................................................................... 96Bituminous Materials........................................................................................................... ...96Problems..........................................................................................................................97Oil price fluctuations and availability of oil supply................................................. 97Transportation constraints........................................................................................97Supply chain issues.................................................................................................. 97

PAGE 8

8 Cement....................................................................................................................................98Problems..........................................................................................................................99Transportation constraints........................................................................................99Supply chain issues.................................................................................................. 99Steel......................................................................................................................................100Problems........................................................................................................................100Iron ore and other raw ma terial availability........................................................... 100Transportation constraints......................................................................................101Supply chain issues................................................................................................ 101Recommendations................................................................................................................ .101Strategic Planning and Positioning of the Firm to Minimize Resource Related Damages.....................................................................................................................101Improved supply chain integrity.................................................................................... 102Improved supplier relations.................................................................................... 102Alternative contractual agreements........................................................................ 103Information technology and its benefits over integrity.......................................... 104Logistics of Transportation Sy stems and Innovative Approaches................................ 106Push-Pull System Synergies and Their Effect on Inventory Costs............................... 107APPENDIX A DATA ANALYSIS AND MAPPING ADDIT IONAL DOCUMENTATION.................... 110B PREDICTION MODEL DOCUMENTATION................................................................... 113LIST OF REFERENCES.............................................................................................................184BIOGRAPHICAL SKETCH.......................................................................................................188

PAGE 9

9 LIST OF TABLES Table page 4-1 Results of the comparative study....................................................................................... 585-1 Sample FDOT bid item descrip tions and corresponding resource....................................725-2 FDOT Estimates Office cost study.................................................................................... 725-3 ARTBA cost study results.................................................................................................725-4 FDOT WMC definition listing........................................................................................... 725-5 WMC composition of FDOT work program..................................................................... 746-1 Data definitions and examples to be used in prediction modeling.................................... 866-2 Results of comparative correlati on analysis of resources and cost....................................866-3 Accuracy of prediction models.......................................................................................... 866-4 Prediction analysis results................................................................................................ ..876-5 Raw resources conversion matrix...................................................................................... 876-6 Program budget with respect to pr oject including different resources.............................. 876-7 Program project counts with respect to project including di fferent resources................... 876-8 Program adjustment factor s for different resources...........................................................876-9 Factored resource requi rements for work program............................................................886-10 Adjusted factored resource requirements for work program............................................. 886-11 Raw resource requirements for work program (all in tons)............................................... 886-12 Total raw resource requirements for work program.......................................................... 887-1 Historic crude oil prices...................................................................................................108

PAGE 10

10 LIST OF FIGURES Figure page 1-1 Solution proposal overview...............................................................................................163-1 Ultimate supply chain ..................................................................................................... ..333-2 FDOT supply chain............................................................................................................334-1 Typical node............................................................................................................... ........584-2 Feed-forward neural network.............................................................................................584-3 Fully connected neural network......................................................................................... 594-4 Supervised training........................................................................................................ ....594-5 Unsupervised training...................................................................................................... ..594-6 GA optimization mechanics............................................................................................... 605-1 Highway construction material cost pie ............................................................................ 746-1 Total aggregate requirement per year................................................................................ 896-2 Total bituminous material requirement per year................................................................ 896-3 Total cement requ irement per year.................................................................................... 906-4 Total steel requirement per year........................................................................................906-5 Total base material (lim estone) requirement per year....................................................... 917-1 Miami-Dade County location..........................................................................................1087-2 Highly populated counties of the stat e and major means of transportation..................... 109

PAGE 11

11 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CORPORATE-LEVEL MANAGE MENT SYSTEM FOR CRITICAL STRATEGIC RESOURCES FOR LARGE CONSTR UCTION PROGRAM MANAGERS By Duzgun Agdas December 2008 Chair: Ralph Ellis Major: Civil Engineering Resource management has long been one of the rather popular re search topics in construction management research. The constructi on industry is heavily material oriented and better management can provide significant competi tive advantages to the parties. The researchers focused on resource management at project leve l and provided useful methodologies to reduce the effects of resources over the construction sc hedule and project cost. Ye t, this shortsighted approach does not resolve the ro ot cause of the problem most of the large construction portfolio managers face, having the resources available at the construction site. The study aimed to develop a corporate level resource management system for large work programs. The study uses a prediction model that has been successfully used in construction research to determine the future resource requi rements and uses the supply chain management principles and logistics to enha nce the traditional resource hand ling processes. Proposed benefits include but are not limited to, improved supply ch ain integrity, reduced resource related delays, reduced inventory and transportation costs a nd ultimately improved customer satisfaction.

PAGE 12

12 CHAPTER 1 INTRODUCTION Construction Industry Overview Construction industry is one of the most crucial industries contributing to the overall welfare of any country. Not only is the m onetary value of construction a la rge percentage of the respective countries gross domes tic product (GDP), but also the industry provides crucial services to people; dams, commercial/residentia l buildings, roads and ai rports. In addition, the industry provides major employment and investment opportunities. Among the highly industrialized countries, includi ng USA, UK, Denmark and Australi a, the value of construction industry varied from 5 to 11.3% of the resp ective countries GDP (through 1997 to 1999). The industry is also a major source of employment and entrepreneurship. For the aforementioned countries, the construction industry made up for 4.5-11% of total employment while 8.9-13.7% of the total registered firms inve sted in construction (Bennett 2003). Characteristically, constructi on is a highly resource-intens ive industry in terms of equipment, labor and material. The material an d equipment costs account for around 60% of the project costs (Akintoye 1995). Thus, resource management is one of the most important issues to be addressed. The biggest challenge for managers in terms of managing scarce resources is the trade-off between minimizing cost s and finishing the project on time while meeting the quality requirements. Not surprisingly, methods have been developed to address the resource management issues at the project level. There have been two main techniques to handle the resource management: resource allocation, whic h is also known as resource-constrained scheduling, and resources leveling (Leu and Yang 1998, Hegazy 1999, Senouci and Adeli 2001).

PAGE 13

13 Discussions about Resource Leveling and Allocation The common trend in resource m anagement is to a pproach the issues at the project level. In resource allocation procedure, either an optimi zation algorithm or a heuristic method is adapted after the resource demands for tasks of a give n project are determine d. Together with the prioritization method, resource demands and the cap acity at hand, the alloca tion carried out for a given project. On the other hand, in resource leveling, once the re source demand and the available resources are determined for a projec t, the heuristic model within the scheduling software allocates the resources and then levels them to reduce the peaks, an attempt to optimize the cost, while maintaining the critical path method (CPM) logic. Flaws of CPM scheduling are apparent and these principles are based on CPM logic, so they do also have flaws but they provide better management abilit y to the project at the project level. The methods not only focus on preventing possible resource caus ed delays in completion time of the projects, but also reduce the resource related costs. Problem Definition Although these approaches provide efficient re sults at the project level, large-scale construction portfolios m ay suffer from not ha ving corporate level resource management system/principles available. The cu rrent practices assume certain levels of resources available for a given project and calculate the resource demand for the individual tasks within the project and proceed according to that very comparison. In othe r words, the methods assume that there will be some adequate resources available on the site to complete the construc tion, there are no methods to optimize the process of getting resources to the construction site and providing common sense solutions to the problems may be encountered. This is a corporate level management problem and needs to be addressed at the corporate level before the project level re source management can be applicable. There have been a number

PAGE 14

14 of studies in literature appro aching the resource management from a broader, business approach to address the management related problems. Yet, they have been on an ad hoc basis and nonconsistent in terms of the methodology as they fa il to provide a broad and comprehensive generic solution applicable to multiple firms. There was a need for a methodology to explore the nature of the re source relation of the projects and the suppliers and crea te a generic solution that can be applied to different cases. And a need for a case study to explor e the details of the supply chain relations of a construction company and the suppliers; not only to observe th e applicability of the theoretical supply chain relations developed in mainly the manufacturi ng industry to the constr uction industry, but also have a concrete example of the problems that ar e being faced and what can be gained by having a corporate level resource management syst em. Although the aim was to develop a methodology that is generic, a relatively larg e construction portfolio is to be studied as the solution procedure is more likely to be a learning process as there has been little effort in li terature approaching the issue from the perspective proposed in this repo rt. The literature review in the following chapter has been useful in terms of collecting the differe nt pieces from different studies that are to contribute to the completeness of the methodology to be developed; however, there is no single study whose methodology can be adapted to solve th e resource management problem analyzed in this report. Overview of Solution The solution proposed in this report is to a ddress the corporate level resource requirements for large scale construction portfolios. The soluti on can be thought as a tw o stage application. In the firs t stage, the resource requirements for a work program was to be predicted and the resource allocation and leveling synergies be used to minimize the damages caused by resource related issues. Then the supply chai n relations of the firm needed to be analyzed to identify the

PAGE 15

15 problems the firm is facing and these was used to develop generic solutions to corporate level resource management problems. As per the prediction model, neural networks mathematical models inspired by the human nerve system was used as the prediction models as they are the better suited for the purposed of this report than the parame ter based prediction models. Developing a model to predict the future reso urce requirements for different projects was to make it possible to perform a resource leveling/a llocation approach at th e corporate level. The relation of individual ta sks with a project and the relations of projects with a construction portfolio are essentially the sa me. The individual tasks have re source needs and are scheduled on the time scale to be completed at certain times to have the project to be completed successfully. Similarly, a construction portfolio, valid for rela tively larger corpora tions undertaking a number of different studies, is made up of a number projects that have certain resource requirements and to be completed at specific time frames. Thus, as per the discussion about resource leveling and allocation, knowing the resource requirements for a future construction portfolio, is to enable the managers to have better resource management reduced costs, better quality and overall improved customer satisfaction. In addition to the th eoretical principles to be applied to make the resource demands smoother, havi ng a successful prediction model answers to one of the rather big disadvantages of the cons truction industry preventing having solid supplier relations, uncertainty (Nicolini et al. 2001). The second step was analyzing the supply chain and logistics related issues, point out the problems and propose alternative solutions to the problems identified. Below is the flowchart of the solution methodology.

PAGE 16

16 Figure 1-1. Solution proposal overview

PAGE 17

17 CHAPTER 2 LITERATURE REVIEW Resource Management Resource managem ent in construction literatur e has had two main approaches; the rather traditional resource leveling and allocation, and ma terial flow utilizing the logistics and supply chain principles derived from different industr ies, mainly manufacturing and retail industry. Resource Allocation Resource allocation can be desc ribed as a m ethod to minimize the unavoidable delays in the completion time of a project due to scarce resources. There have b een a large number of studies to address the resource a llocation problem. The main focus of the research has two main solution procedures: analytical (numeric) solutio ns and heuristic solutions. The analytical solutions are to produce better results than the he uristics. The analytical procedures convert the real life problems into functions whose paramete rs are the factors that affect the outcome. Once the problem is configured and transformed to a function, optimization methods are used to maximize the performance. However, the comple xity of the algorithms and the questionable performance of the models under large scale projects reduces th eir reliability. The time required to produce a solution with complex projects can cr eate serious issue in add ition to the possibility of not converging to a solution. Yet, there have been numerous articles ab out the exact solutions to resource-constrained schedul ing problems using various diffe rent methodologies. Jiang and Shi (2005) proposed an algorithm called enumer ative branch-and-cut procedure (EBAC) for solving multiple-resource constrained schedulin g problems. Ammar and Mohieldin (2002) also proposed a methodology for resource constraine d project scheduling using simulation. The complexity of the exact solutions to the resource-constrained scheduling makes heuristic approaches more feas ible solution alternative to th e problem. These methods are to

PAGE 18

18 mimic the behavior of a person who would be d ealing with a similar problem. For instance, a very common heuristic rule is to give priority with respect to the float the activities have. Accordingly, the critical path activities given pr iority during resource assignment, as a delay on the critical path delays the project comple tion. Although these solutions do not guarantee an optimum solution to the resource constraint problems, they provide acceptable/good solutions; the non-convergence problem of the analytical solu tions is not a problem with heuristic solution approaches, to the problem with much less effo rt than the exact solution procedures. As an example Tsai and Chui (1996) proposed two heur istic algorithms to sc hedule projects with multiple resource constraints. Resource Leveling Resource leveling, on the other hand, focuses on reducing the peak and low de mands to reduce the variations in resource requirements which reduce costs and improve the productivity while keeping the project duration as close to th e original length as possi ble. The importance of resource leveling is clear to every researcher and developer in the industry. To automate the resource leveling while keeping the CPM logic within the schedul e, the solution to the leveling was encoded into planning and scheduling softwa re. Software developers have come up with heuristic algorithms to properly level the available resources tasks and to avoid peak resource demands while trying to maintain the original proj ect duration. Scheduling so ftware have tools to assign resources to individual tasks and calculate the demand for these by using productivity rates. Then reports are produced to enable users to visually see the demands that are over the available resources and users are given the opt ion to eliminate demands over the available resources. The critical path method logic is main tained as much as possible to minimize and/or eliminate the possible delays in completi on time of the tasks on critical path.

PAGE 19

19 Resource leveling and resource a llocation together manage th e resource requirements of a construction project sequentially rather than si multaneously. To address these demands, Hegazy (1999) proposed a method to optimize the resour ce allocation and leveling for a project using genetic algorithms while searching for the op timum solution for both methods simultaneously In this study however, the a pplication of the re source allocation and leveling is to be simpler than the construction literature methods as the relation of projects with each other and the overall portfolio is not very clear. In a project the activities on the cri tical path are to have privilege over the ones that have slack in te rms of resource allocation. Resource leveling, again, is carried out by keeping the CPM schedule logic as the reference. Prediction Model Selection The m ost common prediction models used in construction research have been; parameter based regression based regression models and neural network base d prediction models. The latter traditionally took longer to develop and more co stly; however, has proven to be more accurate (Hua 1996). In addition with the recent adva ncements in processing power of personal computers, especially the increase random access memory, frequency of processors and the introduction of multi-number cells within the co re processors enabling better parallel processing performance, makes the neural network modeling more feasible. Neural Networks The use and the details of the neural networks and the genetic algorith m s to this study is explained in detail in chapter 4, but simply a hybrid genetic al gorithm and neural network model was used to determine the resource demands of a short term construction portfolio. Although the neural networks have been developed in 1950s the algorithm made them more feasible and suitable for construction research, generalized de lta rule, was developed by Rumelhart et al. (1998) and the actual publication of the algorithm was used to provide better understanding of

PAGE 20

20 the neural network fundamentals and for the civil engineering applications there was no shortage of research papers using neural networks (F lood 2006) as a method to predict and resolve a number of different problems. Genetic Algorithms In a sim ilar manner the fundamentals of the genetic algorithms were covered by a number of books written in different industries and differe nt applications than the construction related activities. Yet like neural networks, the gene tic algorithms and optimization techniques were popular amongst the construction re search community and the publis hed papers provided a clear picture of the uses and limitations of the genetic algorithms. Hybrid Models Although the m ethodology to be used in the pr ediction model was a hybrid model and to be executed by commercial software it was necessary to address the basics of both to better understand the limitations of each and the possible be nefits of the hybrid model, which is more costly to construct than regular neural network models. In addition to what the literature review about each methodology indicated, a num ber of studies comparing the hybrid models to not only regular neural network models but to the parameter based prediction models, multivariate linear/non-linear regression models and hybrid model has been found to be more superior in terms of accuracy to any other predicti on model in each one of the studies. Supply Chain Management and Construction Industry Applications The resource m anagement problem to be addres sed in this report is essentially due to the lack of integrity within the construction suppl y chain Supply chain management has long been successfully applied to a number of industries to improve the efficiency, reduce the costs and ultimately improve the customer satisfaction. The adaptation of the supply chain principles has been slow in construction indus try. Vrijhoef and Koskela (1999) provides a generic supply chain

PAGE 21

21 model that can be applied to almost all industr ies in addition to providing a summary of the concepts applicability and issues related to it in construction. The main problems the authors pointed out resulted from lack of integrity within the supply chai n and shortsighted approach to the problems that are at larger scale. Thes e problems can be attributed to the lack of communication between parties in upstream and downstream end of the supply chain. Supply Chain Integrity Supply chain integrity and transparency are two of the m ay factor that will contribute to the success in any industry and IT tools ar e indispensible elements for these purposes. Manufacturing industry has long been using state of the art technol ogic devices to handle almost all the business related activities. A number of different reasons may constitute a valid discussion for slow responsiveness of construction industry to technological advancements. These factors include but not limited to (Agdas, 2006); Traditional industry and traditional wo rkforce being resistant to change Slim profit margins and tight cash-flow sche dules preventing investment in innovation Shortsighted/short term profit oriented industry Although the above proposed fact ors are valid, it is obvious th at there is a need for the industry to bring the standards to the scale in terms of innovation as even a slight enhancement in the processes will improve the profitability si gnificantly due to the high turnover value of construction products. In manuf acturing industry where the in novation and the competitive advantage are indispensible for success latest technology is used to improve the costs, response time and ultimately customer satisfaction. It is unlikely to expect the construction industry to achieve the levels of complexity of the manufact uring industry; yet, this should not be a reason to overlook a number of principles developed in that industry. The construction industry is dubbed as underperforming, inefficient, unproduc tive, fragmented and wasteful by London and

PAGE 22

22 Kenley (2001). And the solution is to have a more integrated industry with better relations with the other parties involved in the process. In an attempt to im prove the communication with the suppliers Akintoye (1995) mentions the use of EDI technology as a way. EDI technology has long been used in a number of industries. The discussion on this issue can be found in later chapters. In a highly competitive and material or iented industry like construction, carrying high levels of inventory can be disast rous to the firm due to the loss es to occur because of inventory carrying and material waste costs. Justification of Necessary Changes in th e Tra ditional Construction Supply Chain Although the concept of lean construction is not very related to th is report, the lean construction analysis provides insight to a numbe r of problems that is to be addressed via the methodology proposed in this study. Studying the lean construction studies pr ovided insight as to what are the roots of the problems related to resource management. Naim and Barlow (2003); the authors proposed a agile and lean system for housing construction in UK to better handle the problems they have observed. The lean and agile construction is a set of ideas and a philosophy aimed at reducing the waste, the non value adding activities, and thus, improving the schedule, productivity and ultimately the customer experien ce. Agile, by definition is referred to as the ability to adapt to changes, within this terminology is used as a term to define a system that will enable the party to exploit the profitability chances in a vola tile industry. The following are the foreseen problems in housing industry that are relevant to the civil/h eavy engineering field. The shortsightedness at regiona l and site levels caused by price based agreements and suppliers inability to fore see the market conditions Putting too much power at site managers hands without any stra tegic decision making framework or leadership from top management Weak supplier relations, the s hort notice of supply demands fro m site, lack of mid-term demand calculation ability combined with the poo r delivery by the supplier in addition to the confrontational contractual terms.

PAGE 23

23 Problems in acquiring the sub-contractors, due to lack of medium term planning, subcontractors commit more than what they can do in order to make sure they will get some work to keep the business running. Excessive supply inventory causing major da mages as the materials are stored in relatively large quantities at sites, which not only increases the inventory carrying cost, but also the wastage of the materials. In a ddition the materials are handled on an ad hoc basis resulting in inefficien cies in material handling. The short term cost only oriented approach prevents level schedules and create excessive and unexpected delays, causing change ine fficiencies and poor customer satisfaction. In addition Nicolini et al. (2001) proposed cl ustering in order to reduce the complexity of the interdependencies to reduce the uncertainties and enjoying better collaboration within the supply chain and better supply chain management. The authors proposal is in agreement with the lean construction principles suggested by Th omas et al. (2002) that aimed at reducing the number of activities to improve the speed and eliminating waste. Material Flow In addition to the supply chain m anagement related construction rese arch literature whose pieces can be of use developing the methodology to address the resource management problem, there has also been a number of studies ai med to improve the material flow through the construction projects which is very similar to resource management context. Significance of Material Flow It is im portant to understand the importance of the materials for construction, as the numbers are to indicate the signi ficance of them and create more incentive for researchers to invest in the subject. Jang et al. (2003) suggest that the c ontinuous improvement and customer satisfaction are the core of constr uction logistics and they carried out a detailed survey to figure out the factors to that are most relevant to th e project managers level of satisfaction with the construction logistics the importance is of the study is because a project manager is the lowest management level in a constructi on industry that will be directly affected by the decision making

PAGE 24

24 of the higher level managers and/or corporate procedures. Of the five factors determined, (i) personnel, (ii) material flow, (iii) schedule adhe rence, (iv) contractors organization and (v) information flow, material and information flow ar e determined to be most material flow to the site and. The result is not surprising as material flow and the information flow are directional trends from one end of the supply chain to the other as suggested by (Men tzer 2001) and they are interrelated too. Better material flow can be maintained via be tter information management resulting in supply chain integrit y. In addition, the surv ey participants poin ted out some other issues within construction that needed to be addressed; reducing the inventory, improving the communication and the collaboration with the pr oject participants, improving information flow and enhanced e-commerce within the supply chain. These are cures to the problems identified in construction by a number of different researchers. Material Delivery Systems and Their Effects on SCM In a precedent case to what Jeng argued, Ag apiu et al. (1998) poi nted out the reduced productivity caused by not optim ized material supply to the constr uction site and they proposed a logistics approach to the material handling issu e to enhance the productivity and as a case study they have analyzed a Danish house building project. Unlike the common material handling procedures that are handled on an ad hoc basis and far from being effici ent, the authors have proposed a methodology to handle the material handling more efficiently by reducing the production, transportation and waste. This required a strategic pl anning of the material handling prior to construction and improved communi cation between project participants. The methodology employed some of the methodologies from manufacturing industry, just-in-time (JIT) delivery which is derived from lean production and aimed at improving the delivery timing of the materials to the site to reduce the i nventory holding costs and the waste by reducing the batch sizes and increasing the number of deliver ies. A total of 5% total saving observed by the

PAGE 25

25 methodology imposed on the project. The conclusive results were that th e logistics should be focusing on exchange of information and better collaboration between parties. Push vs. Pull Systems In a sim ilar manner, Akintoye (1995) proposed a JIT delivery system for construction material delivery to maximize the efficiency and quality, and minimize the waste. The JIT is referred to as delivering the goods to the site just in time the time th ey are to be used. The author identified the necessary steps to implement the JIT system; (i) what material to be used, (ii) who is the supplier (iii) what is the best distribu tion system. The successful implementation criteria was deemed to be depending on the improved co mmunication and participation of the parties involved, training and level schedul es, that require better planning. Tommelein and En Yi Li (1999) also commented on the possible JIT benefits construction may enjoy if applied. They made their case by using ready mix concrete that is more like to be used in JIT delivery as it has a short delivery time and is made to the order. The authors have identified the synergies of a pull system within the ready mix concrete de liverythe construction industry is a typical example of a push system. A push system is an analogy base on the possible future needs and the production is done accordingly. Due to the high levels of uncertainty, delays between the order and the actual delivery time of the material and very complex relation between parties. However, such a system is tend to incr ease the amount of inventor y carried, waste and is slow to adapt to the changes. As each party the ordering done before the actual construction started, not only there are larger than required inventory amounts, every party accounts for the uncertainty by pushing up the amount to be ordered, and if the act ual quantities are not what is needed to be used in actual pr oject, a push system cannot adapt as t industry is positioned to react to sudden changes. On the other hand, a pull syst em is more adapt than a push system; however, it requires better supply chain integration and management.

PAGE 26

26 CHAPTER 3 SELECTION OF CONTEXTUAL DOMAIN Necessity of Developing the Proposal around a Case Study The solution was to be structured around a case study, as there havent been sim ilar studies in the literature this far. In addition, the claim that not having a higher level resource management system would result in inefficienci es and delays in completion of projects was intuitive at the earlier stages of research. And there was a needed more concrete examples to constitute a valid case. In orde r the proposal to be convergent with an industry, the subject firm/institution needed to be functioning over geog raphically diverse territories, working with a number of different suppliers/sub contractors and large enough in si ze to afford strategic planning tools to identify the types of th e projects it would be undertaking in near future. In addition to these, there was a definite need for solid data ma nagement by the firm, as healthy data as crucial for successful prediction model development. The best candidate has determined to be the Florida Department of Transportation (FDOT). Not only the institution fits in the a bove criterion, but also it has been undertaking projects in the region researchers were familiar w ith which is to make the supply chain analysis more efficient. The study was to overlook the five year work plan FDOT uses was a good tool to be used for the analysis. As a strategic tool, FDOT Office of Work Program develops a short term plan called five year work plan (FYWP) every five years to better manage the projects to be undertaken next five years. Ex amples of FYWP data entries can be seen in Appendix A. A committee works with the input from a number of district offices, city councils, governors office in order to decide what type of construction projects needed to be completed at what time and where these projects should be performed. After that, the five year work plan is published on FDOTs website ( http://www2.dot.state.fl.us/fmsupporta pps/workprogram /WorkProgram.aspx )

PAGE 27

27 and used as a strategic tool the institution abides by for the next five years. Recently, FDOT has started to fail to meet the project milest ones planned on the five year work plan. The main problem was the unavailability of the resources for cont ractors/subcontractors use, rather than the budget constr aints. In addition to the afore told reasons to choose the FDOT to be the case study subject, the resource relate d problems institution is facing is a striking example of the problems resulted from short sighte d approach to the resource management issues within construction industry. The situation faced by the institution was a prime example of proposed problems that might be faced by large construction program manage rs. Analysis of the issue within the context of the resource management proposed in this stu dy will provide a generic solution applicable to other large scale programs and firms as well. Th e problems institution is facing will be addressed and the implications of the problems faced by the case subject and other firms will be discussed. Organizational Details of the Florid a Department of Transportation The initial stage to begin the proposal proces s was to identify the or ganizational structure of the construction operation of the case firm FDOT uses contractors to carry out the construction projects included in the FYWP. After the design is completed and bid item quantities are determined by exte rnal consultants, FDOT announces the projects and qualified contractors bid on unit price for bid items to be co mpleted in that project. Total project bid is calculated as the sum of all the bid item quantities multiplied with respective unit prices and. as common to the transportation construction, the lo west bidder is selected for the job. FDOT maintains a list of bid items to be used in bidding procedures. The recent problems FDOT is facing in successful completion of the projects they have undertaken can almost all the time be attributed to contractor failures in predicting/foreseeing and/or carrying out some risk analysis for the resources they are using for construction

PAGE 28

28 operations. However, due to the nature and the tr aditional nature of the contractors, expecting such measures and reflecting the effects of possibl e market fluctuations to their bid is rather unrealistic. In addition to that as the instituti on does not account for the supply availability during the planning phase of the FYWP, FDOT suffers significant damages. The level of complexity and uncertainty in constructi on industry has long been known and different approaches, lean construction principles, applied to reduce the level of complexity and resultantly reduce the uncertainty. The struct ure of FDOT work program and the way the construction projects completed make the level of uncertainty even higher for the organization than it is for the regular constr uction firms. Figure 3-1 is a generic fully integrated supply chain model developed by Mentzer (2001) and it is applicable to almo st industries, including the construction. Figure 3-1 repres ents the most complex supply chain illustration the authors proposed. Figure 3-2 shows the simplified version of FD OT supply chain. As can be seen FDOT has more complex supply chain relations than regula r construction firms. Also attention should be give to the fact that, there is neither a logistics supplier nor a market research firm employed by the institution. It is important in todays global market structure to account for the market changes and be able to foresee th ese disruptions and act accordingly. Justification of Having a New Resource Management Initiative The success ful completion of the FDOT projects is crucial for the safety and welfare of the state. And for the FDOT to be able to finish th eir already-planned projects, they needed a model that will help them to predict the requirements of the projects ahead of time with certain amount of accuracy. Having such a prediction model at ha nd, the institution will be able to allocate the cumulative resource requirements of these projects in a way to avoid incu rring resource related delays. Due to shortsightedness of industry, the executors of th e projects, contractors, do not

PAGE 29

29 have capabilities to foresee possible market react ions in terms of price/availability of the construction materials. Although this case seems not to be a problem of FDOTs; but rather the contractors problem as they fail to meet their contractual du ties and in any of the courts decision FDOT will be awarded for damages it incurred due to the contractors failure Already occurred delays will be added to the time required for such litigations to be settled can be disastrous for institution as well as the citizens of the state. And the institu tion is to make sure the planned progress of the projects and the actual progress is to be as consistent as possibl e as the plan is developed to enhance the well being of the citizens of the stat e. Thus, any delays caused by the incompetency of the executive contractors of the projects undertaken should be prevented/reduced by the institution as the results of their actions will affect the welfare of citizens of the state. The importance of the transportation projects are far more than the monetary value they represent. The traditional loss prevention measur es will not be sufficient enough in this case and there is a need for a be tter innovative solution. Proposed Benefits The m odel is to predict the resource require ments, discover the supply relations of the resources and, resultantly, is to provide bett er resource management abilities to the upper management. After the prediction model created, project information obtained from the FYWP was used to forecast the demand requirements fo r those projects. Once the demand was obtained, the applicability of resource le veling and allocation to the work program resource requirements was analyzed. After this step was completed, The obvious benefits of such an appl ication include but not limited to: Reducing the possibility resource related delays in project completion time.

PAGE 30

30 Reducing the overhead expenses that may be caused by the idle time during construction caused by the resource related problems. Preventing possible litigations with contractors due to reso urce related problems, which in return hurts the FDOT most. Improving the logistics cap abilities of the firm, that is to reduce the inventory costs and improve efficiency Improving the supply chain management and ultim ately attain a fully integrated relations with the parties involved Supply Chain Indications of the Problem In-depth analysis of the s ituation revealed that the problems being faced were resulted from supply chain related issues. Supply chain management (SCM) is a widely accepted and applied business strategy aimed at reducing the supplier-owner rela tion related dela ys in projects or deficiencies in products. A lthough the system is more app licable to the manufacturing industry, the principles are appli cable to other industries as well. The formal of SCM definition varies among different authoritie s; yet, the common theme in the definition is the coordination of various activities that add value to production of any goods during the process, taken as raw materials from suppliers and conveyed to the end users as finished products. The FDOT case is a rather complex supply chain case when the variety of the projects undertaken, the dispersed geogra phic activity areas, the differing level of specialization for different projects requiring and the respective workforce, the discontinuity of the resource inflow, uncertainty resulted from the complex supply chain relations and uniqueness of the projects are all considered. However, the research indicated SC synergies that the institution can benefit significantly without implementing a fully integrated SCM system that will require significant investment, time and staff training.

PAGE 31

31 Supply Chain Integrity The problem s FDOT facing in meeting the de mand requirements of their projects is ultimately the result of lack of supplier chai n integrity. The main focus of the supply chain descriptions discussed above is coordination and improving the relationship of parties that participate in a business and having a seamless integrated chain. And had there been a fully engaged supply chain, officials would have s een the possible issues that may arise from excessive demand and proceed accordingly. Or, even better, they could have been working in accordance with the suppliers while preparing the work plan so that the problem they are facing right now wouldnt have taken. But as explaine d earlier, the program a doption process is not under the control of the c onstruction division of the institution, but rath er the third parties that is not participating in actual cons truction related activities. Feasibility of Implementing Suppl y Chain Principles Analysis It m ay be argued that the demand requirement of the institution for such a large scale plan over five years may be hard to foresee and the sa me argument is valid with the supplier side of the equation. However, not all th e resources used in constructi on are of the same importance. Differing types of projects requir e different types of materials th at are important for successful completion and the same rule applies to the crit ical resource requirement s of different projects too. This lean construction principle was adapte d for determining the resources and types of projects to be analyzed. Arguably the most apparent distinction between the manufacturing industry and the construction industry is the demand side of the supply chain issues. In manufacturing the demand is up to the amount that the firm wants to pr oduce and the mixes of each material are set and there is a continuous inflow of the materials from suppliers. In contrast, in the case of construction industry the product, a construction project, is uni que in terms of size, location,

PAGE 32

32 environment, end users needs. In addition the lack of IT integration, not optimized construction methods, rather primitive data storage and analysis for quality purposes makes the case harder for construction firms to have as clear reso urce requirements as manu facturing industry. However, there are synergies of SCM systems a pplicable to this very specific FDOT case. In addition to better managing resource requirem ents as a corporation, depending on the past data and experience, better strategies can be ad apted by FDOT to optimize the efficiency of the work program. For instance, a vulnerable supply w ith large price fluctuations in recent history and varying and large demand (to be determined by the model developed for this very study) requirements indicate a require ment for strong supplier relati ons. And in case the analysis indicates inconsistent trends, it may be necessa ry to adopt a SCM system different than the current regulation of the instit ution. Although the peak demand requi rements can be foreseen and eliminated after the prediction model is used, th e supply side cannot be le ft completely volatile and precautions for possible market disturbances must be taken. Detailed analysis of each resource with respect to price vulnerability and supplier relations is provided in following chapters. Prediction Model Due to the nature of the construction and unique products to be deliver ed, it is challenging to find a forecast m odel to predict the resource requi rements of the projects that will be used to determine which resources will be more critical th an others. However, the unpredictability of the resource requirements for different projects requires a somewhat dynami c model to capture the trends in demand fluctuations. The most promising methodology s eems to be neural networks (NN) because of their dynamic structure and adaptability. There have been numerous applications of neural networ ks; from cognitive science rese arch to artific ial intelligence

PAGE 33

33 applications to recognition of handwritten zip c odes by computers, due to their ability to learn from example, to be useful with even not comple te data and generalize from what is available. Figure 3-1. Ultimate supply chain (Ada pted from Mentzer, J. T. (2001) Supply Chain Management, Sage Publications, Thousand Oaks, CA). Figure 3-2. FDOT supply chain

PAGE 34

34 CHAPTER 4 PREDICTION MODEL DEVELOPMENT Neural Networks Introduction Neural networks (NNs) are m athematical mode ls inspired by the human neural system. Studies initially started to mimic the neural system of humans but failed to do so because of the level of complexity of the biol ogical system when compared to the mathematical one. However, these studies led to using the NNs in various applications in numerous industries. Artificial neural networks are computa tional models that are compos ed of connected adaptive computational units (Vonk et al. 1997). As stated earlier the biggest advantage of the NNs is their ability to learn by the exam ple and adapt to pattern within the data. Ability to learn makes NNs very useful tools for complex problems that cannot be modeled and/or solved via traditional mathematical models (Adeli and Karim 2001). In a ddition to the ability to learn and generalize, NNs are efficient tools when the data is incomplete or includes error, adaptability to changes in circumstances of the problem that is being anal yzed, and ability to pro cess data rapidly (Flood and Kartam 1994). NNs have also proven to be more efficient th an regular parameter based prediction models. Hua (1996) has shown that NNs are better predic tion models than the conventional multivariate linear regression models. He carried out his research on residential construction demand forecasting in Singapore. He selected 12 econom ic variables to be the indicators of the residential demand forecast, and Mean Absolute Percentage Error was the relative measurement for the performance of the two models. NN model has produced one fifth of the error that the comparable regression model produced. In addi tion, the regression model suffered from the autocorrelation of the socalled independent variab les. That is, the independent variables, as the

PAGE 35

35 name implies, are to be independent of each other and otherwise the R2 and statistical significance values will be biased and will not be a good indicator of the actual explanatory values of these. In Huas research this issue need ed to be tackled separately by another statistical analysis, which reduced the efficiency of the mo del. In a similar way, Garza and Rouhana (1995) undertook a comparative study to compare a NN mode l to a multivariate linear regression model and a multivariate non-linear regression model in a cost estimating problem. In the analysis the error term was the mean square error, n xx MSEn i actual i 1 2)( 4-1 Where n was the number of observations. The NN model produced error terms nearly one third of those produced by the multivariate non-linear model and one-fifth of those produced by the multivariate linear regression model. Overview of Topology NNs are designed to work in three layers. Th e in put layer is the la yer where inputs are transmitted to the second (hidden) layer(s). Ther e can be more than one layer depending on the pattern of the data range, where the actual ma thematical operations occurs thorough multiplying the input data transmitted to th e layer by the input la yer with the weights of the corresponding neuron. After inputs are multiplied with the weights they are summed and passed through the third layer. The third layer contai ns a activation function that take s the result of the mathematical equation and uses the result in a function and shows the result of that operation. Activation functions are step function, linear function, ramp function and sigmoid function. Figure 4-1 shows the NN activity: The operation at the hidden layer(s ) can be summarized as follows:

PAGE 36

36 )(1 ii n ixwy 4-2 Where y is the activity le vel created at a neuron, w is the weights on the connections to the node and x is the input values on the connections to the node. Once the value of y is determined, it is used in activation function to display the result. One of the most common activation functions is sigmoid function whose formula is as follows: ye yf 1 1 )( 4-3 There are a number of other activation functions that can be used in NN applications including but not limited to: Step function: The outcome is either -1 or 1, depending on whether the y value is positive or negative Linear function: The output y is multiplied with a constant. Ramp function: Is a combination of Linear and Step functions. The output is the multiplication of y with a constant within a maximum and a minimum cap at the boundaries. Structure of Neural Networks The structure of a NN is one of the most im portant decisions to be made as it is a major factor in determining the applic ability of the network to a give n problem set. Depending on the problem, the type of the structure of the NN (n umber of hidden layer, number of nodes) plays a significant role in performance of the prediction model. NNs ar e classified into two depending on the directions of the connections of the laye rs: feed-forward and fee dback (fully connected) networks. In the feed-forward configura tion, the layers are interconnected in one direction, while in the feedback configuration the data flow is circ ulative in the neural sy stem and the circulation continues until the required results are obtained (Figure 4-3). As the pattern of connections imply

PAGE 37

37 the fully connected networks are much harder to develop and used for a ddressing rather complex problems. For the purposes of this research, the NNs will be used for prediction; feed forward networks should provide sufficient accuracy. This argument is also confirmed by the fact that almost all of the NN based research for variou s applications in construction management has provided results with enough accuracy. Number of Hidden Layers Flood and Kartam (1994) discusses the earlier findings of other researchers to determine the optimum number of hidden layers for a gi ven problem. They discuss two different prior academic studies that claim one or two hidden layers (using the sigmoidal activation at the hidden layer) are to be enough for any practical problem solutions. Hecht-Nielsen (1989) argued one hidden layer should be enough to provide soluti ons to any practical pr oblems. However, this proposal does not show the way the existing tr aining algorithms operate. On the other had, Lapedes and Farber (1988) have proven that a combination of a sigmoidal steps generated at the nodes in the first hidden layer and the bump shaped features of the nodes at the second hidden layer can provide solutions to a ny given problem can be achieved by proper combination of these bump shaped features at the output nodes. N one of the proposals assu me sigmoidal activation functions at the output node as the function provides values only between 0 and 1 and using the function will result in values falling between these two. Flood and Kartam further argue that despite the proof by Hecht-Nielsen, there are many problems that are extremely difficult to model by using the single hidden layer approach. Tw o layers provide much better flexibility for complex solution sets and theref ore should be the starting poin t for new model development. However, the tradeoff between accuracy and the time required to process the model should be kept in mind when approaching any given problem. For this research, it is safer to use a two hidden layers as higher degrees of accuracy is more desirable.

PAGE 38

38 Number of Hidden Nodes Determ ining the number of hidden nodes is a ra ther challenging prope rty of an NN model to be determined as there are no rules to dete rmine the number of nodes in the hidden layer(s). Kamarthi et al. (1992) describe d a trial and error methodology to determine the number of nodes in the hidden layer of the neural network model th ey created. They have described the goal as to develop a model with the fewest hidden nodes as to have an efficient model (the higher the number of nodes the longer it will take to train and use the neural network model). However, it should be noted that too few nodes will prevent the model from converging to a solution. The research team started with 20 nodes and incr eased the number by 10 until they reached an acceptable solution. The desired solution was achieved at the 50 hidden nodes. Similarly, Flood and Kartam (1994) have discusse d the same issue. They have argued that there is no certain way to determine the number of nodes to in the hidden layers. Increasing the number of hidden nodes seemed to have increased the fit of the solution set created by the neural network. However, a large number of hidden nodes can deviate from the trend at the intermediate points while providing a good fit to the training points. In addition large number of hidden nodes can lead to longer training times and can slow down the processing speed in using the model. On the other hand few hidden nodes can prevent the system from producing acceptable results. In order to handle the situa tion a range of combinations is to be made available and the one with the best performan ce is to be chosen. Flood (1991) developed the radial-Gaussian system that determines the number of hidden nodes while training the network. The system adds nodes to the system in a seque ntial manner while training each on the error produced by its predecessors. This allows the nu mber of hidden nodes to be determined while the training procedure takes place (Gagarin et al. 1994).

PAGE 39

39 Training Before neural network models can be used for any purpose, the weights of the nodes within th e model need to be ad justed to show the characteristics of the data set. This is accomplished via training procedure. Once the mode l is properly trained, it is to produce the desired output for a given pr oblem provided it is given the input. The performance of an NN model is highly dependent on the training met hod applied to it. There are two major training approaches: supervised trai ning and unsupervised training. Supervised training It is the m ost popular means of training an NN m odel. In such training the target values are present within the data set ava ilable for training. The neural netw ork model is used to predict some results which are compared to target output values. The typical le arning procedure starts with a neural network model of arbitrary numbe r of nodes, a fixed topology for connections and random weights for each node (Flood and Kartam 1994). The network is then presented with the set of training patterns includi ng the inputs and the actual outpu ts. The output of the neural network is then compared to the actual outpu ts of the given problem And depending on the governing training rule, an error is calculated using the differe nce between the target output values and the actual output values from the neur al network. This difference between the desired output and the actual output is used to adjust the weights of th e nodes and to reduce the error. The process is repeated until the desired error level or the lowest possible error is obtained. Although there are a number of ways algorithms that can be used in supervised learning method, the most popular methods used is the generalized delta rule (GDR) developed by Rumelhart et al. (1986).

PAGE 40

40 Unsupervised training Rather unpopular but equally as im portant as the supervised training, unsupervised training do not feed the network with the desired outputs. For give n inputs the network is left to itself to configure the solutions to the given problem. Th e training set is applied to the network until a stable output is obtained. Clustering algorithms are common training tools for this type of training. The clustering procedure adjusts the weights of each node in a way to force each output node to respond to as many training sets as possible. The output nodes are not to have overlapping outputs. The result of the algorithm is the training set clusters each of which is presented by a different output node. The problem falls into a class determined by the most active output node. In many algorithms the output nodes are designed to enhance their output while surpassing those of the other nodes. Training Rules As m entioned above the supervised learning rule is the more popular one and for the purpose of this research, the NN will be fed with a set of historic input/out put data and the future output data will be predicted by using the input about future data available, the supervised learning is more suitable, as the required output will be available during the training procedure. Although there have been a number of training rules and corresponding weight adjustments discussed in literature, the generali zed delta rule has been the predominantly used one due to its simplicity and applicability to mu ltilayer feed forward networks, which has been the focal point of interest for construction research. Generalized delta rule The m ost common type of supervised learni ng algorithms is the generalized delta rule (GDR) developed by Rumelhart et al. (1984). Th e rule is quite common as it is easy to implement and applicable to multilayer feed-forward networks. The GDR is a supervised

PAGE 41

41 learning algorithm modifying the initial wei ghts assigned to the nodes depending on the error term produced using the difference between the ac tual outputs of the network and the target outputs. Once the network is run and the outputs and the error te rms are produced, the modifications depending on the error terms are ba ck propagated to the network to adjust the weights, the procedure is repeated until the erro rs produced are within the acceptable limits. The common terminology back propagation feed forward multilayer neural networks are called as such as they use the back propagation to kj enhance the reliability of the networks. The error that starts the w hole procedure is calculated as (Rumelhart et al. 1984): 4-4 Where, Ek is the error term corresponding to the training pattern k, tkj is the is the target output for the jth element of the output pattern and okj is the actual output produced by the same component of the output pattern for the input pattern k. GDR rule is based on the logic to implement a gradient descent to total error produc ed by the network thats the sum of all error for different training patterns, The GDR is derived from the delta rule to be applied to the networks with semi-linear activation functions, which are non-decreasing and differentiable functions, i.e. sigmoid function, of the net output y. Where, 4-5 And ykj is the total output for the set k and wji is the weights of the nodes and oki=iki if i is an input unit. 4-6 Here the function f needs to be non-decreasing, differentiable and semi-linear in order the GDR to be applicable. Otherwise, the derivativ e would be infinity at the threshold and zero elsewhere.

PAGE 42

42 The derivative of the of the e rror term with respect to the weights of the nodes are to be used to adjust the weights of the nodes. 4-7 Where, 4-8 After defining the second term in above equation, the first part is defined as, 4-9 Where, is defined as the error signal. Thus the equation 4-4 can be rewritten as, 4-10 And in order to implement the gradient descen t approach to E, the weight adjustments are to be done as following, 4-11 Where, is defined as the learning rate that will increase the ch ange in the weights without causing oscillations. Keeping that in mind, the learni ng rate can be selected as large as possible for practical purposes. In calculating the error signal th e chain rule for derivatives is used and the equation 4-9 can be rewritten as, 4-12 And from equation 4-6, 4-13

PAGE 43

43 There are two possible combinations for the first term in the error signal formula, depending on the node being an output node or not. In case the node is an output node, 4-14= Substituting the above two equations in to the error signal formula, 4-15 If the node is not an output node, 4-16 As mentioned earlier the term is a term introduced to enlarge the weight adjustments in order to make the gradient descent approach feasible. Although a true gradie nt descent approach will require infinitesimal modifications to the we ights of the nodes, for practical purposes it is not done accordingly. One way to improve the learning rate wit hout causing the oscillation is introducing a momentum term to the in addition to the learning rate. The resulting formula is as follows, 4-17 Here, is a constant which determines the effect of the past wei ght adjustments. The momentum term filters the high-frequency vari ations of the error surface in the weight adjustment process. The shortcomings of GDR and alternatives Gagarin et al. (1994), A deli and Karim (2001), and Flood (1991) noted that despite its popularity, generalized delta rule (GDR) used in BP methodology suffers from a number of significant drawbacks: It does not specify the number of nodes to be used to solve a given problem.

PAGE 44

44 The solution, even for simple problems, takes a long time to converge The networks tend to be entr apped at local-error minima The second problem is not very crucial due to the advancements in processing capabilities. However, the other two are still valid and need to be addressed. Adeli and Hung (1994) have developed an adaptive conjugate gradient lear ning algorithm for feed forward networks to overcome the shortcomings of the GDR approach. They have used Powells modified conjugate gradient algorithm with an approximate line sear ch to minimize the system error. In order to overcome the trial and error selection of the c onstant learning and momentum ratios issue, the step length in the inexact line research is adapted. This mathematical approach provides a more solid theoretical foundation to th e learning and it has proven that this approach has better convergence ratios than the GDR algorithm. Adeli and Karim (2001) mentioned radial ba sis function neural networks (RBFFN) to overcome the shortcomings of GDR. An RBFFN uses a radial function, an example of which is a Gaussian function, to cover the input space that tr ansforms a vector from input set to output. Flood (1991) has developed the radial-Gaussian incrementallearning network (RGIN) to overcome the shortcomings of the GDR algorith ms. The RGIN approach has been shown to overcome these issues by providing the number of nodes to be used for a given problem, faster and generally more accurate results and certainty in achieving convergence. The model takes its name from the bump like solution surfaces it pro duces that can be more efficient in modeling higher degree function surfaces where the sigm oid function proves insufficient due to its simplicity. Unlike sigmoidal networks GDR algorith m, RGIN uses two different functions for hidden nodes and output nodes (Gagarin et al. 1994). This network type consists only one hidden layer, in which the hidden nodes us es a Gaussian like radial function.

PAGE 45

45 Summary Neural Networks have been widely used in various as pects of civil eng ineering applications and in construction area. The publications about NN us age in civil engineering have varied from vibration control of multidegree-of -freedom structures (Madan 2006) to measuring organizational effectiveness (Sinha and McKi m 2000). The vast number of publications and research led to books and journal papers that are published to c over the NN applications within the civil engineering/construction management. Construction was one of the more prolific divisions of civil engineering when it comes to the NN publications. Not only there have been a large number of studies proposing alternative uses of NN in construction, but Boussabaine (1996) published a comprehensive review of the applications of NNs in construction management as did Adeli and Karim (2001). The NN applications to the construction management include but are not limited to predicting the project costs and cash flows, risk analysis, decision making, resource optimi zation, and predicting bidding results. Incorporating NNs with other Algorithms Although NNs have successfully been applie d to significant com putational methods to address various problems in a number of disciplin es, the architecture of NNs are generally found by trial and error (Vonk et al. 1997). This method is not only time consuming but also, not efficient as it might and most of the time will not provide the optimal structure to provide the best solutions to a given problem. In addition to that, the learning algori thms, playing a crucial role in performance of NNs, are dependent on the topology of the network. There have been a number of proposals suggesting methods to ove rcome the arbitrary t opology selection and provide a more general approach in literature. There have been great interest a nd a number of successful applications of gene tic algorithms (GAs) to optimize various properties of a network system to optimize the ability of the networ k to produce solutions. There have been some

PAGE 46

46 proposals to determine the number of hidden node s in a network, as in the RGIN example discussed in previous sections, but generalization of the determin ing structural properties was not clarified by those methods. Genetic Algorithms GAs are stochastic search algorithm s whose logi c is inspired by the Darwinian survival of the fittest and geneti c inheritance theories (Goldberg 1989, Michalewicz 1994). The theories suggest that the fittest individuals will survive and the make up of a population and their genetics will be inherited by the following generations. GAs were developed by Josh Holland and his peers in University of Michigan in an atte mpt to understand and explain the natural adaptive selection processes and mimics those with artificial systems. The algorithms work the same way that nature does, the more fit individuals have higher chances of survival and their genetics will lead to more fit offspring than their parents. There are also slight ch ances of mutation of the offspring. The individuals (genotypes) that are to be subjected GA meth odology are essentially solutions to any given problem and the goal is using the solution pool to achieve an optimum solution to the given problem. GAs differ from other optimization methods in a variety of aspects. GAs are probabilistic algorithms, yet they differ from random algorithms as they combine the direct and stochastic research elements. GAs use a population of possible soluti ons rather a single point solution. GAs work with the coding of parameters rather than the parameters themselves GA Mechanics GAs are essentially sim ilar to other evolu tionary algorithms in terms of structural composition. There is a pool of solutions to a given problem at any iteration t t txxtS2 1,........, )( and each member of the pool, t ix is evaluated for its fitness, a subjective

PAGE 47

47 criteria determined by the nature of the pr oblem and user preference. The new population (iteration t+1) is formed by the more fit individu als of the previous gene ration (iteration t). This phenomenon is defined as reproduction. A new and a better solution set is created for a given problem using the fitness levels of the individuals in the previous population. The members of the new population undergo changes by two ope rations of the GA system that are: crossover and mutation. The purposes of these operations are further enhancing the members of the new generation. The crossover operation is similar to mating two strong individuals to form a fitter offspring. In the algorithm world, two acceptable solu tions are used to create a new solution. It is done by swapping the corresponding segments of the parent chromosomes. The purpose of the operation is to provide information exchange between solutions of a given problem. Another important operator is the mutation that adds more variability to the population. Mutation changes components (one or more genes) of a chromoso me (solution set) in a random manner. The mutation rate, a probability number chosen by the user s, is the factor that is used to determine how it will be implemented. For GAs to be applicable to a certai n problem, five criteria must be met: The solution surface to the problem should be genetically represented There should be an initial set of solutions available There should be a function to measure the fitness of the members of the population There should be genetic operations to be applied to the chromosomes (crossover, reproduction) There should be values for various paramete rs of GAs use (probability of mutation, population size)

PAGE 48

48 Details of the operations Although there are various differe nt rules and selection crit eria depending on the problem type and requirements for the genetic optimization rules, below are the most fundamental description of the genetic optimization procedures. Reproduction: As stated earlier, the natural selec tion theory of Darwin requires the stronger members of a generation to have better chances of survival and mating with other strong individuals to continue their existence. In order to mimic this, a biased roulette wheel method is used. Once an initial population of possible solutions is created, th eir fitness is measured using a specific function depending on the needs of the user and the nature of the problem. Depending on their fitness, the i ndividuals are given slot s in a roulette wheel. Essentially, the stronger individuals have better chances of survival, but it is not guaranteed. Thus, the biased roulette wheel is a good way to mimic the phenomenon as th e fitter chromosomes o ccupy larger areas in the roulette wheel and as the selection is at rand om in the wheel, they have a better chance of being selected to be included in the next generation and procreat e new offspring with other fit chromosomes. Whenever a new offspring is required, the roulette wheel spins and parent chromosomes are randomly selected to be used in crossover opera tion. The exact replicas of the two parents are used. Crossover: Once the parent chromosomes are selected and placed in a mating pool, they are matched in pairs at random. And a random number is generated between 1 and length of the chromosome string minus 1, [1, m-1]. For example, if there are 10 strings in a chromosome, a random number is to be generate d between 1 and 9 inclusive and th e crossover is executed at the characters between random number plus one an d the length of the string [k+1, m]. Suppose X1 and X2 two possible solutions to a given problem. X1=0001101010

PAGE 49

49 X2=1000010111 And assume the randomly selected number between 1 and 9 is 4. The cross over operation is executed between the 5th and the 10th components of the chromosome strings, by simply swapping the string components of the other string. X1=0001|101010 X2=1000|010111 X 1=0001010111 X 2=1000101010 The two offspring are incl uded in the new population. Mutation: Regardless of the success of the reproduc tion and the crossover operations, in an attempt to optimize a solution to a given problem, they may lose some significant improvements especially when the input is in bi nary format. The mutation operator is aimed at eliminating such losses. The mutation operato r works by changing a string components sign, form 0 to 1 and vice versa. The mutation rate and the randomization determine which components are to be mutated. An independent mutation procedure of the reproduction and crossover procedures provides a benchmark for obtaining important notions that the other two might be losing during process. However, as the mutation rates are generally very small, it is considered to be a secondary adaptation procedure for the GAs. Optimization of a simple function Michalewicz (1994) showd how the GAs work by optim izing a simple function with one variable. The function was defined as: 1)10sin()( xxxf And the problem is to find an x in the interval [-1, 2] that will maximize the value of f(x). In order to be able use the GA methodology, the solution domain is to be represented in binary

PAGE 50

50 format. In the example the accuracy requirement is 6 digits after the after the decimal point requiring the interval [-1, 2] be divided by at least 3000000 and as the binary system is to be used, 221<3000000<222 22 bit strings are to be used in GA analysis. As the strings are to be in binary format, the binary string is to be converte d to base 10 from base 2 after the analysis and a proper function required giving the boundary domain values for the above conversion. In this example, 12 3 122 xx Note that for x value of zero, x will be -1 and for x value of 222-1 (the maximum value 22 bit binary string can have), x will be 2. The chromosomes 0000000000000000000000 and 1111111111111111111111 are the boundary for the domain and any value in between can be converted to base 10 by simple al gebra. For instance chromosome 10001001110110101000111 is equivalent to number 0.637197 as x=(10001001110110101000111)2=2288967 and x(2288967)=-1+2288967.3/4194303=.637197. As mentioned above, there is a need for a fitness function to compare the fitness of different x values in this case. As the objective of the optimization is to maximize the value of a single value function that very func tion is to be used as the evaluation functi on and the fitness of the individual strings will be ra nked with respect to the greatness of the output of the function. The author determined three chromosomes as the starting point, x1=1000101110110101000111 x2=0000001110000000010000 x3=1110000000111111000101 and the corresponding x values are as following

PAGE 51

51 x1=0.637197 f(x1) = 1.586345 x2=-0.958973 f(x2) = 0.078878 x3=1.627888 f(x3) = 2.250650 Evidently, x3 is the fittest of all chromosomes as the evaluation function returns the highest value for it. In order to illustrate the eff ects of the genetic operations applied to the chromosomes, the author demonstrated the impacts of mutation a nd crossover over the chromosomes. The first operation applied was the mutation to the x3. The fifth gene in the chromosome was mutated, its value was changed from 0 to one, resulting in the new string 1110100000111111000101 which is in base 2 and equa ls to 1.721638 and the resulting evaluation function value of -0.082257 is a significant reduction fr om the original chromosome. On the other hand had the tenth chromosome se lected, the resulting string would have been 1110000001111111000101 that has a value of 1.630818 in base 10 with an evaluation function value of 2.343555, better than the orig inal result. As can be seen the mutation operation can be quite significant, although the chan ces of its occurrences are rare. The other illustration was about crossover operation. Chromosomes 2 and 3 were chosen to be used and the cross over point was the fifth gene, the operation will be carried out between the sixth and the twenty second genes. x2=00000|01110000000010000 x3=11100|00000111111000101 x4=0000000000111111000101 x5=1110001110000000010000 x4=-0.998113 f(x4) = 0.940865

PAGE 52

52 x5=1.666028 f(x5) = 2.459245 As can be observed the second offspring has better fit than both of the parents. During the analysis, a population size of 50, crossover probability of 0.25 and mutation probability of 0.01 is used. After 150 generations, the GA optimizat ion found the optimum solution to the problem that is the maximum value of the function of discussion. The only apparent problem with GA optimization is that the genetic optimization parameters, mutation and cross-over rate, are determined by trial and error as there is no spec ified generalization rule for determining these parameters but for trial and error (Kim et al. 2005). Genetic Algorithms in Civil Engineering The genetic algorithm optimization has been on e of the most successful and widely used methods to address some very complicated constr uction/civil engineering problems. Due to their ability to address complicated problems, without getting stuck at local minimums/maximums as they evaluate a population set of solutions with randomized yet structured search algorithm rather than focusing on a single point solution, they have been applied to qu ite different types of problems. Li and Love (1998) used GAs to allocate the necessary tem porary structures for construction at construction site with thei r respective spaces. Marzouk and Moselhi (2002) developed a methodology to optimize a simulation de veloped for estimating the cost and time for earth moving operations. The most popular appli cation of GA optimization in construction has been resource management. Kandil and Raye s (2006) developed a pa rallel GA system to optimize the resource utilization for large sc ale projects that enables users to combine computational abilities of multiple computers to reduce the processing time of the optimization procedure. Leu and Yang (1999) proposed a GA based resource constrained construction scheduling process to optimize the construction duration while introducing new crossover and mutation operators. Hegazy (1999) proposed an algorithm to optimize both resource allocation

PAGE 53

53 and leveling using GAs. The traditional approach, heuristic algorithms, failed to provide an optimum solution under both the resource leveli ng and allocation priorities. Leu and Hung (2001) proposed a GA based optimal resource constrained scheduling simulation model that combines the uncertainty in activity dur ation with the resource constraints. Hybrid NN-GA Models As discussed in previously, although NNs have proven to be useful and easy to im plement tools, they are not fully regulated as the most common NN models, multilayer feed forward back-propagation, do not have a universal method to determine the structure of the network. Although, some methods developed pr oposes partial solutions to so me of the problems, radial basis function neural network (RBFNN) and the radial-Gaussian incremental-learning network (RGIN) model discussed previ ously provides a framework for number of hidden node; none of these models addressed the parameter selection cr iteria question for NNs. The performance of an NN is highly dependent on its structure and the best structure has proven to be varying for problems of different nature. To address this issue GA optimization has been used to optimize the neural network structure as well as the we ights of the nodes in order to optimize the performance the NN. The focus has been on th e optimization of the topology and the weights and there have been some research about the hyb rid models. The main advantage of these models is that the individual NNs are eval uated against each other with re gard to the error margin they are producing for a given data set to minimize the error produced by the pred icted output values. Hybrid NN-GA Model Applications in Construction Hua (2000) developed a hybrid NN-GA m odel for a study to compare the performance of the proposed model to an NN in predicting th e construction demand for Singapore residential construction industry. Hua developed an NN mode l as well as a NN-GA model to perform his analysis. The NN model was a three layer BP netw ork with 2 input nodes, 5 hidden nodes (single

PAGE 54

54 hidden layer) and an output layer with a singl e node. Hua has used some of the guidelines proposed in other publications as guidelines while developing his model. The results of the study indicated that the hybrid model wa s superior to the traditional NN model. In addition, the hybrid model reached a high level of accuracy after 15 000 iterations, whereas the traditional model needed to be trained for 1000000 training cycl es. However, the NN-GA model took longer to converge as a number of evolutiona ry cycles needed to be comple ted as part of the optimization procedure. However, the time requirement is not as important a factor it used to be as the improved processing speed of PCs with advancem ents in technology is far beyond that of the publication date. Similarly, Kim et al. (2005) proposed an analysis of comparison between hybrid NN-GA model and a traditional NN model in preliminary cost estimation analysis study. They have found that the hybrid model, using GAs to determine the neural network parameters, was the more accurate than a traditional model and they al so pointed out that a hybrid model using GAs to optimize the weights produced the worst results. The study concluded that using GA optimization solved the problem of determini ng the neural network parameters; however, the authors pointed out that when th e GA optimization is applied to the parameters for optimization, the genetic operators were found by trial and error. Regardless of that fact, the GA optimization has proven to be successful method to overcom e some of the shortfalls of the NN during development. Conclusion about Prediction Model Development An NN-GA model is the m ost pr omising tool to identify the project resource requirements for a large portfolio. Considering the dispersed and diversified natu re of the projects undertaken by FDOT as a part of their Five Year Work Plan, any other method is prone to not being successful in identifying the patterns in data The GA optimization is indispensable for the

PAGE 55

55 research as not only it has prove n to be an effective way of op timizing the network but also the lack of insight about data that will be us ed makes the benchmarks proposed about the NN parameters in literature futile. Although, GA op timization training takes considerably longer than the regular NN training, there is no need to continuous iterations as the only parameter that cannot be optimized is the number of hidden layers and as stated above there are some valid benchmarks with solid reasoning as to why to st art with two hidden layers. Once the model is developed and the accuracy is evaluated a more complex model can be developed; however, it is not very likely that a model with two hidden laye rs will prove to be insufficient for the purposes of this research. FDOT deals with transportation projects mainly and the important characteristic of the transportation is their linearity. Thus a two layer feed forward NN with genetic optimization should provide enough accuracy in de termining the project resource requirements for FDOT portfolio. Platform for Application In this research, NeuroS olutions version 5, (N euro Dimensions, Inc. Gainesville, FL) was used for model development. The software follows the same logic discussed in this section of the report. Users are required to input the number of hidden layers and the number of hidden nodes has initial values assigned by the software as well as the momentum and the learning rate. The genetic optimization can be applied to number of nodes, learning rate and momentum in hidden layers 1-2 and the output layer. The software does not allow the users to change the genetic optimization variables such as, mutation probability and selection method for individuals for cross-over operations, but the default is the roulette wheel method. It is obvious that the gene tic optimization methods can be better organized and better methods can be developed to address the issue of genetic optimization tools being determined by

PAGE 56

56 deterministic methods. Yet even with the imperf ections genetic optimization has proven to be more accurate than plain neural network models and the multivariate regression analyses. Comparative Study In order to prove that a hybrid NN-GA m odel would be superior to any other prediction methods, a comparative study performed to witness what number of researchers pointed out. The data was collected from a projec t undertaken to develop maintena nce of traffic (MOT) quantities prediction models for FDOT. The data was in the form at to be used in this very research and the data source was the same as well. The original project goal was to de velop regression models using the past data for MOT bid item quantities in order to predict the future quantities. MOT items are used to maintain the existing traffic in case of construction. Th ey include but are not limited to signs, barriers, and barricades. In order to test whether the assumption neural networks being better prediction models than the parame ter based modeling, a small data set was chosenaround 20 data points, as regression models tend to perform better for sm aller data sets and neural networks perform rather poorly when the tr aining set is not large. The bid item quantities (dependent variable) were pred icted using project characteristic s, length, days used location (rural/urban) and final ex penditure (independent variables). Th e regression analysis had an R2 value of .84, which is quite high and statistically significant. The error term chosen to evaluate the fitness of different modeli ng techniques was mean absolute percent error (MAPE), which is calculated as; 100 11 n i i iiA PA n 4-17 Where, Ai is the actual quantity value for a given project data and Pi is the prediction value for the same data point. The results were as pr edicted; the neural ne twork modeling has proven

PAGE 57

57 to be more accurate than the regression model. The studies in the litera ture about the positive effects of genetic optimization a nd increased number of hidden laye r were proven to be right in this analysis (Table 4-1).

PAGE 58

58 Table 4-1. Results of the comparative study Linear Regression NN 1 Hidden Layer NN-GA 1 Hidden Layer NN-GA 2 Hidden Layers MAPE(%) 32.81 18.69 4.46 2.86 Figure 4-1. Typical node Figure 4-2. Feed-forwa rd neural network

PAGE 59

59 Figure 4-3. Fully connected neural network Figure 4-4. Supervised training Figure 4-5. Unsupervised training

PAGE 60

60 Figure 4-6. GA optimization mechanics

PAGE 61

61 CHAPTER 5 DATA COLLECTION AND ANALYSIS Data Structure Overview FDOT outsources the design for the projec ts they are to complete. Once the design is complete, the bids are announced and the contra ctors bid on the jobs as a lump sum and the lowest bidder is selected to do the job. Different from the trad itional regulation, the contractors bid for bid items. FDOT develope d and maintains a master pay ite m list (MPIL). The job to be completed is presented as sum of bid items of various types and quanti ties and the contractors submit unit prices bids for each item and production of these unit prices w ith the quantities are then summed over all the bid items included in the job and the lump sum amount is determined. Essentially the data base FDOT maintains is in terms of bid item summaries submitted for a number of jobs. The main database samples can be seen in Appendix A. The master data base documentation was over 1.2M lines of bid item data entry recorded for a number of different projects. The main reason for the ex cessively large data set was due to the fact that the institution recorded all the bids for comple ted projects. Some of the bid it ems were entered a number of times as multiple contractors bid for that job. The master data base documentation did not ha ve project characteristic information to be used as independent variables to determine the dependent variable obtained from the master file. In order to obtain this information, another fi le dubbed as contract hi story summary was used. This file did nto have any item specific inform ation but did have general project information. Examples of the data samples of this f ile also can be seen in Appendix A. Identifying the Strategic Resources Although a fram ework was developed to identify the critical resources, there was a need to have a benchmark as a starting point.

PAGE 62

62 FHWA Highway Construction Ma terial Cost Study For this purpose the studies of US Depa rtment of Transportation Federal Highway Administration (FHWA) were used as the reference ( http://fhwa.dot.gov ). F HWA had been collecting data over the federally funded transportation projects that have a budget that is over a million dollars. And the institution publishes th e findings as construction economics. According to their findings the relative cost fractions of the construction resources used in federal funded transportation projects were determined (Figure 5-1). Although the cost is not the main interest in of this research, it is more likely that the resources that make up the largest fractions of th e overall cost are the ones that are used most and are more critical. HWA study of cost fractions of different resources over the total project spending was valid for projects with budgets of one million dollar and above and this was not of a big concern as most of the projects FDOT unde rtake are quite large in volume and spending. The FHWA study narrowed down the resources of interest to 6; aggregates, portland cement, bituminous materials, steel, wages and others. In addition to taking these resources into consideration, with committee advising, limestone is included in the list. Limestone has long been used as a base material in FDOT constr uction project in addition to bitumens. Thus, the final list to be analyzed was reduced down to aggregates, portland cement, bituminous material (both for road construction and as base material, steel and labor. FDOT Transportation Construction Cost Study In order to see the app licability of this analysis to FDOT a reference literature was sought and a study by FDOT Specifications and Estim ates ha s revealed that all th ese resources, except for the labor component were analyzed in deta il to identify the cost trends. Although the study seems as only a mere cost analysis study, th e basic micro economic supply demand relation (Baye 2006) will be valid for this case too. The sudden and sharp price fluctuations can be

PAGE 63

63 associated to problems in obtaini ng these resources as well as the fact that these resources were the once studied is another indicat or of the importance of them. Discussions on Findings of Two Studies Before starting with the m ethodology to just ify the importance of these resources, the difference between two studies, FHWA a nd FDOT, needed to be addressed. Earthwork The first item was the earthwork; it was include d in FDOT s tudy but was not an identified major component of FHWA study. Ea rthwork was excluded as it is not the cost of resources or availability that is driving f actor behind it, but rather the e quipment and energy costs and the earthwork was not included in the list of strategic resources. Labor The largest direct resource cost to the highway cons truction was determ ined to be the labor costs by the FHWA study. Equipment overhead, is not only indirect cost items but also nonrelevant to the objectives of this report. Howeve r, a deeper analysis of the labor costs over the years at both national scale and within the state of Fl orida revealed that the la bor is not to be of a big concern to the managers at any level of planning for state of Florida anytime soon. First of all, on the contrary to the price change trend of ot her construction resources, the labor costs have marginally increased over the last ten years and have been dwindling last couple of years. This can be attributed partly to the slow market activ ity and abundance of labor. Below are the seasonally adjusted weekly earnings of construction workers for Florida and US. The rate of change for FL labor wages over last 7 year s has been around, 2.9% and state wide has been around .4%, the relatively high increase in the la bor wages can be attributed to the excessive labor requirement s of the company per heavy inve stment in infrastructure, and yet lower than 3.6% of the US average increase over the consumer price index over last 7 years. This is a rather

PAGE 64

64 striking example as almost all of the construction related costs have exceeded the inflation rate by quite large amounts over the last decade. Th e Engineering News Record (ENR) (McGrawHill 2008) construction cost index, which takes into a number of c onstruction related costs into account to calculate an index to represent the effect of all the co st factors have increased by 4.1% on average for the last 7 years. Tables 5-1 and 5-2 provide average weekly earnings of construction workers for Fl orida and US respectively. In addition a study by Eisenhauer et al. (2007) indicated that one third of the construction workers in state of Florida are immigrants a nd their number has increased more than 40% in construction industry last three year s. This is by far the highest rate of increase of all industries in terms of change in volume and second in term s of total number of im migrant employees. And again according to the same st udy, the immigrants demand less than what the native workers demand per hour because of a number of reasons including but not limited to being nonunionized. The immigration trend to the state of Florida tends not to be slowing down and the construction labor seems the last of the concerns in terms of availability for strategic planning purposes. Justifying the Critical Resource Selections As briefly discussed earlier, the biggest cha llenge was to determ ine which resource will become critical and after which amounts, and, unfo rtunately, there were no benchmarks that can be used to determine what those resources are. Thats why the FHWA and FDOT studies used as the benchmarks and the list of resources was narrowed down to some of the most important commodities for construction, cement, aggregates, structural and reinforcing steel, bituminous materials and limestone as a base material. A four level benchmark is cons idered to be sufficient to justify the importance of previously identified the critical resources for the portfolio.

PAGE 65

65 The first step was to find the ratio of the total material expense over the total project budget to determine the cost fractio ns of the resources selected. The second step was to look at the price indi ces of the resources an alyzed in previous step. Frequent and significant fl uctuations in price of the ma terial is either excess demand for those at certain times and/or unavailability due to ot her problems. A good indicator was to compare the price increase of these resources over the consumer price index over the last decade. The local availability of the resources is anot her concern that will a ffect the availability of those through market fluctuations. Commod ities are more vulnerable to be affected by the changes in global economy, which is much more volatile for unexpected disturbances unlike rather mature US economy. The last step was to send out some surveys to the contractors thr oughout Florida to obtain the information about the problematic resources. Details of Justification Procedure Relative costs of the chosen resources The first s tage was to create ratios of total e xpenditure on the resources in a given projects to the total project expenditure The purpose of having these ratio s was to have an understanding of what are the quantitative relatio ns of individual resources with the overall project costs. This study can be put in the same context as the FHWA study. This analysis provided some insight to which types of materials are more significant for which projects in terms of total expenditure on these. And, intuitively, it is more logical to pu t more emphasis on resources are larger in volume and/or cost more. Once the data was reduced to the construction projects with budgets of over than a million dollars, the details of these project s studied and a necessity to carry out an analysis to convert the bid items to the raw materials was observed. The FDOT master pay item lis t and the presentation estimates office published were used as the tool s to determine the bid items that include the resources of the interest (Table 5-1).

PAGE 66

66 After the bid items were reduced to the ones related to the fundamental resources necessary to complete the transportation projects, a ratio analysis was carried out to see whether these resources were as crucial as independent studie s proposed. In order to do so, the overall spending over the projects was compared to the spending over these 5 resources and the ratio was around 50%. FDOT estimates office publishes the average bid item prices for FDOT projects and the average of three years was used as the average pr ice for the bid item. The average price was then multiplied with the item quantity to obta in approximate cost of these items. Relative cost fractions of these resources are more than what the FHWA study proposed but as the analysis was carried out with the bid items, the aver age price includes the overhead, equipment and other indirect expenses too. However, consider ing the variety of resources involved in construction projects, the ratio found was significant in terms of defining the level of criticality for the resources. Note that the eff ects of the inflation over the resource prices was ignored, as the purpose at th is stage of the overall me thodology was to obtain a rough approximation of the cost fraction of the selected resources and al so the total expenditure of the projects were not adjusted for inflation either. Historic cost analysis An ef ficient way to assess the price/quantity volatility of the selected resources is to check the price fluctuations of the resources agai nst the consumer price index. The open market economy, by definition, enables free price settings for goods depending on the demand level for these goods. There are no quotes, tariffs or min/max pr ice restrictions to the resources to be used in construction. It is necessary to carry out this analysis at local as well as at international levels as in construction both local material, limestone in FL, and commodities are common resources, steel. Almost all of the countri es of the world have suffered from excess demand for steel from China during 2004 and the situation is not getting a ny better as their demand is not getting lower.

PAGE 67

67 Not only the price of raw material s increase, but also the freight cost increased significantly due to excess demand China and India (Ambrogi 2007 and Mathews 2007). Thus the impact of international commerce and their relation to international commodities are never to be undermined in any analysis. There were numerous resources to prove the claim that the price of construction materials have been rising at a greater rate than the inflation rate. The more relevant studies were the ones carried out by FDOT Specifications and Estimat es Office, The American Road & Transportation Builders Association (ARTBA), and Portland Ceme nt Association. Table 52 provides the results of an FDOT study aimed at identifying the recen t cost trends of the construction materials. The study results are rather shocking when considering the stable and mature economy of the US, and the inflation rate thats around 3% over last 10 years. However, this study had only FDOT data and specific to the state of Florida. The ARTBA carries out the cost analysis studies at the national level. The results are as pres ented through April 2004 to April 2008 (Table 5-3). The results of ARTBA study were much more striking than the FDOT study as the sharp increase trend seems to be continuing rather th an leveling off. The two studies not only support the findings of each other, but also the similari ty in the numbers indicates that most of the construction resources are commodities at nati onal and probably at the international level. Checking the local availability of the resources It was im portant to identify the difference among locally available resources; available within the country resources; and international commodities as the greater the distance gets the more the chances of unexpected problems to ta ke place. The current example of which can be observed in truck load transportation fueled with rapid increase in oil prices which climbed to unprecedented numbers. As per discussions in the prev ious chapters, most of the resources to be analyzed in this study are commodities; therefore, are riskier than locally available resources.

PAGE 68

68 Except for the limestone and some aggregates (t he details of the local availability of the aggregates will be discussed in detail in followi ng chapters), the resources are all commodities, which increase the necessity of paying attention to these resources in term s of logistics and supply chain relations. Contractor surveys The last step in determining what the critic al resources is sendi ng out surveys to the contractors working with FDOT in order to both confirm the findings in previous stages and/or explaining/further explaining the unc lear spots/inconsistencies. Th e size of the data, the variety of projects and the vast amount of resources to be utilized during construction may lead to overlooking some of the relations that is crucial for the success of the proposed analysis and asking the people who are carrying out the construction at the first hand and facing the resource related problems in their daily operations is arguably the best way of obtaining information about the problems. In the development stages of this report; how ever, a piece of literature was found to have addressed the needs of the contractors survey. The study was carried out by Florida Transportation Builders Association and was pr esented after the FDOT study about the price changes of construction materials. The study not only confirmed the findi ngs about the critical resources acknowledged in this report, but also have proven the apparent lack of supply chain integrity as one of the main problems they ar e facing in completing the projects they have undertaken. Thus the necessity to have a contractor survey was removed. Conclusion of Justification of the Critical Resources The initially selected resources using the FHWA and FDOT Specifications and Estimates Office were confirmed to be stra tegically important by the justific ation process, as not only they are commodities but also their pr ices fluctuate frequently and in greater amounts. In order to

PAGE 69

69 make the analysis more feasible it was necessary to reduce the number of re sources in order to improve the efficiency of the analysis. The sp ending on these selected resources make up more than half or the FDOTs spending on the projects and they are also f ound to be strategically vulnerable in terms of availability. Data Mapping Following the m aster pay item list and descriptions of the items, the data including items of interest were selected. In or der to further narrow down the data to be analyzed, the data was sorted with respected the work mix code (WMC ) definitions. WMC is defined as a way devised within the organization to classify the construction description of the project. The FDOT Specifications and Estimates Office publishes the list of the WMC definiti ons. In order to see what is the cost fractions of different projects types, WMC, the past data and the five year work plan was sorted with respect to the WMC definiti ons. The results were rath er interesting as more than half of the projects carried out were re surfacing and new road construction. it is more reasonable to focus the interest on the more freque nt jobs that are more expensive to carry out. A benchmark of 80% of the portfolio spending was chos en to be the point for cut. In other words, the historic and planned portfolio data was to be sorted with respect to the WMC definitions, high to low, and whenever the sum of the top WMC definitions reaches over 80% of the overall spending, those WMC definitions were chosen to be included in the analysis. FDOT has defined 103 project definitions and it is very likely that some of these projects occur much more frequently than the others. The comple te WMC listing can be found in Table 5-4. Filtering the Data In order to narrow down the list of the W MC to be analyzed, the five year work plan as well as the historic project data was used. The projects worth over $1,000,000 was chosen to constitute the population of the projects to be used for a number of reasons. First of all there are a

PAGE 70

70 number of small projects and/or parts of small projects included in the portfolios that will not make any impact on the overall institution functioning and they needed to be discarded to better focus on more expensive and important projects. In addition the FHWA study that is used as the basis for resources to be selected was va lid for projects that have budgets over $ 1,000,000. Moreover, the work program FDOT publishes incl udes some dummy cost information for some of the projects. These projects are the ones that ar e not finalized by the dist rict offices at present time but are to be included in the program. Thus, instead of includi ng the actual budget of the project, the managers put a dummy place holder, i. e. $2 dollar resurfacing project, in order to have a valid entry fo r the fiscal year. After eliminating the projects whose budget were less than $1,000,000 and eliminating the non construction WMC definitions, the frequencies of the different WMC projects have been studied. The results were rather interesting as only five types of construction projects, i.e. only 5 WMC definitions; add lanes and re construct, add lanes and rehabi litate pavement, interchange construction, new road construction and resurf acing, made up more than 80% of the FDOT construction portfolio (Table 5-5). The idea of focusing on the rather expensiv e and more frequent projects improved the efficiency of the research significantly. The work program was divided into fiscal years in order to predict the resource re quirements of the program per year. The independent variables provided in the five year work plan were, WMC, transpor tation system definition (TS), total cost of the projects, length of the project a nd the number of lanes to be c onstructed. The dependent variables to be predicted were; structural concrete, steel (structural/reinforcing), bituminous material (for paving and as base material), aggregates a nd limestone (as base material) quantities.

PAGE 71

71 Summary and Conclusion of Data Processing The resources to be analyzed were determ ined by extens ive literature review and a research methodology developed for this very rese arch. And at the spending on these resources was more than 50% of the project spending. This analysis carried on ove r the program portfolio institution undertakes. As a result of an attempt to make the resear ch feasible and efficient this sort of a study was necessary and has proven its value by reducing the number of resources to be studied significantly. As the predic tion is only first part of the overall solution, focusing on the most valued resources is to improve the efficiency. The data collection, filtering and gathering wa s quite a tedious activity mainly due to the size of the data to be analyzed and the way the database was constr ucted. The bid items are grouped under the corresponding res ources to make up the population to the testing to be done. The bid items are collected form a bid item data base FDOT maintains. This data base did not have any project specific data, and thus needed to be cross referenced with another database used to store the project characteristics of the past projects FDOT completed. The five year work plan project data filtered in multiple steps to make the analysis more feasible and efficient. The da ta was filtered with respect to cost and WMC definition, and divided into consecutive fiscal years. The resu lts were efficient and satisfactory. The filtered project count was less than half of the original population; yet, the tota l expenditures of these projects were about 80% of the construction rela ted spending. This way it was possible to focus on the projects that cost more a nd use greater amounts of resources.

PAGE 72

72 Table 5-1. Sample FDOT bid item de scriptions and co rresponding resource Bid Item Number Item Description Corresponding Resource 2331 2 Asph conc type S Bit. material, aggregates 2285701 Base optional (base group 01) Bit. Material 2285702 Base optional (base group 02) Bit. material 2285703 Base optional (base group 03) Bit. material 2400 2 25 Conc class ii (mass) (substruc.) Cement, aggregate 2400 4 25 Conc class iv (mass) (substruc.) Cement, aggregate 2415 1 9 Reinf steel (a pproach slabs) Steel 2415 1 2 Reinf steel (bridge) Steel 0285709990 Base optional limestone (10") Limestone Table 5-2. FDOT Estimat es Office cost study Resource Unit 2005 2006 2005-2006 2007 2006-2007 2005-2007 Asphalt TN $68.49 $90.75 +32.5% $104.44 +15.1% +52.5% Str. Steel LB $1.34 $1.68 +25.4% $2.07 +23.2% +54.5% Rein. Steel LB $0.86 $.96 +11.6% $.99 +3.1% +15.1% Concrete CY $653.43 $892.89 +36.6% $913.49 +2.3% +39.8% Table 5-3. ARTBA cost study results Resource 2004-2005 2005-2006 2006-2007 2007-2008 2004-2008 Asphalt +5.8% +22.7% +14.6% +5.2% +56.5% Aggregates +6.3% +8.9% +9.5% +6.5% +35.0% Steel Scrap +5% +8.3% +31.1% +57.3 +234.5% Cement +12.3% +15.1% +6.2% -0.2% +37.0% Ready Mix Concrete +12.4% +13.0% +5.1% +1.6% +35.6% CPI +3.5% +3.5% +2.6% +3.9% +14.2% Table 5-4. FDOT WM C definition listing Work mix code Work mix name Work mix code Work mix name 0008 Access improvement 0112 Mcco weigh station static/wim 0213 Add lanes & reconstruct 0224 Mill and resurface 0218 Add lanes & rehab. Pvmnt 9924 Miscellaneous construction 0549 Add left turn lane(s) 0029 Miscellaneous structure 0550 Add right turn lane(s) 0121 Multi-lane reconstruction 0547 Add thru lane(s) 0020 New bridge construction 0551 Add turn lane(s) 0002 New road construction 0754 Adv traveler info. sys. 0775 Overhead signing 0106 Bike path/trail 0206 Parking facility 0925 Bridge painting 0543 Pave shoulders 7093 Bridge operations 0328 Pedestrian/wildlife overpass 0429 Bridge rehabilitation 9980 Preliminary engineering

PAGE 73

73 Table 5-4. Continued Work mix code Work mix name Work mix code Work mix name 0022 Bridge replacement 8330 Rail crossing improvements 0024 Bridge-repair/rehabilitation 8301 Rail improvement 0023 Bridge-rep. and add la nes 0768 Railroad crossing 8250 Clear zone clear & grub 0767 Railroad signal 0321 Const. Bridge low level 0423 Replace high level bridge 0324 Const. Bridge movable span 0421 Replace low level bridge 8262 Const/relo. security fence 0422 Replace medium level bridge 0323 Construct bridge high level 0424 Replace movable span bridge 0327 Construct bridge culvert 0427 Replace or widen br culvert 0330 Construct culvert 0331 Replace or widen culvert 0310 Construct special structure 0425 Replace railroad bridge 9919 Construct/recons. Median 0109 Rest area 9983 Corridor improvement 0110 Rest area (dual) 9915 Drainage improvements 0012 Resurfacing 9916 Dredge 0217 Rigid pavement reconstruction 0061 Emergency operations 0227 Rigid pavement rehabilitation 1039 Environmental permits acq 0102 Road reconstruction 2 lane 0046 Environmental test/mitigate 6060 Routine maintenance 0222 Farp-pave should. & resurf. 9917 Safety project 0220 Federal aid resurface/repave 0205 Sidewalk 0315 Fender work 0774 Signing/pavement markings 0552 Fix horiz. Or vert. Curve 0216 Skid hazard overlay 0005 Flex. pavement reconstruct. 9981 Special surveys 0117 Frontage road 0225 State pave shoulders & resurf. 0041 Funding action 0215 State resurface/repave 0541 Guardrail 0223 State widen and resurface 0014 Hwy-enhancement 0718 T.o.p.i.c.s. 0004 Hwy-reconstruction 0123 Toll plaza 0103 Interchange (major) 0717 Tra ffic control devices/system 0104 Interchange (minor) 0010 Traffic ops improvement 0231 Interchange (modify) 0714 Traffic signal update 0719 Interconnection 0716 Traffic signals 0118 Intersection (major) 0720 Upgrade exist.traffic signals 0119 Intersection (minor) 0113 Weigh station (single) 0756 Its freeway management 0122 Welcome station 0752 Its surveillance system 1053 Wetland mitigation/restoration 1070 Landscaping 0544 Widen bridge 0777 Lighting 0542 Widen road 0226 Maint. resurfacing (flex) 0221 Widen/resurface exist lanes 0111 Mcco weigh station static

PAGE 74

74 Table 5-5. WMC composition of FDOT work program Fiscal year Program Budget ($) Budget of screened projects(>1M) ($) Construction budget ($) Selected WMC budget ($) Total freq. of selected WMC ($) 2009 7,246,848,634 7,719,886,917 6,163,556,344 5,140,559,817 .834 2010 5,286,220,958 5,117,442,942 4,330,807,461 3,407,381,562 .787 2011 4,259,600,329 4,157,187,729 3,526,703,402 2,639,564,559 .748 2012 3,804,057,122 3,737,763,211 3,478,315,284 2,685,043,273 .772 2013 2,385,361,921 2,348,026,211 2,014,834,096 1,676,400,100 .832 Total 23,455,127,247 22,607,268,72719,514,216,58715,548,949,311 .797 Steel 5/ 4.26% Bitumens 4/ 7.6% Equipment, Overhead & Profit 36.66% Wages 19.14% Other 15.46% Aggregates 13.24% Portland Cement 3.6% Figure 5-1. Highway constructi on material cost pie (Adapted from FHWA cost studies, http://www.fhwa.dot.gov/ last accessed July 15, 2008)

PAGE 75

75 CHAPTER 6 PREDICTION RESULTS AND INTERPRETATION Introduction After the data collected and m a pped according to the needs of the research, the past data was further analyzed in order to see its complian ce with the work program data that was to be used in prediction modeling. The historic databa ses available provided a ll the column heading related to developing a prediction model that future work program had. Details of the Variables The independent variables to be used in the prediction m odel were; WMC, TS, length of the project, number of lanes and the total expe nditure. Table 6-1 shows the examples of each variable to be included in the analysis. The initia l list of variables needed to be changed due to a number of reasons. Project length was taken out, as the future data included somehow unreasonably higher values as project length. The past data indicat ed that the average pr oject budget per mile of construction was around $3.6M/mile, whereas it is $2.2M/mile for future data. Such a larger price change per mile will result in inflated prediction quantities a nd length is the cause of this problem. As the results clearly indicate there is a problem in length as an independent variable as the project budgets seem unreasonably low when co mpared to the project length. The causes of the discrepancy is not clear but when the analysis is run and the networks trained, the prediction results are far from being close to the average quan tities. The market value of a ton of asphalt is used as the basis of this analysis and in some of the prediction results the asphalt cost was much higher than the overall project budget In order to eliminate the effect of the unreasonable values of project length included in fu ture work program, the independe nt variable was taken out. The remaining variables were to provide a good estima tion of resources as most of the construction

PAGE 76

76 expenses is resource expenses and the other e xpenses are highly related to the amount of resources to be used. Variables to be Determined Independent variables compiled from the historic project data used to pred ict the yearly requirements of concrete, structural/reinforci ng steel, asphalt, bituminous material (base material), and limestone. Note that concrete a nd asphalt are combinations of raw materials and they need to be disintegrated to those. In order to do so, two re search study findings were used. The concrete composition was determined by th e concrete mix design study carried out by fellow researchers at University of Florida fo r FDOT. The weighted average of concrete mix design of this study was used as the basis for de termining the material ratios for structural concrete. The main components ar e assumed to be cement, water and aggregates and admixtures are assumed to be negligible. The concrete mix design sheet can be seen in Appendix B. Similar to structural concrete, the asphalt is a combination different raw materials; aggregates and bituminous materials. In order to get an average asphalt mix design value, the experiments carried out by FDOT State Materials Office laborator y tests were used. The office collects random data from all 7 districts of state of Florida and the compositions of these samples are recorded and published. Random counties from each district were chosen and the average mix values are calculated. The mix quantities of di fferent counties and distri cts vary little and the average mix ratio was determined to be good indicat or of what the actual design values will be like. The asphalt design spreadsheet as well as th e FDOT laboratory test samples can also be seen in Appendix B. Final Adjustments to the Historic Data The data set was com piled in 6 different subset s, each for a different resource to be used; asphalt, concrete, structural steel, reinforcing steel, bituminous ba se material and limestone base

PAGE 77

77 material. Before using the data in analysis, a fe w more modifications to the historic data was necessary. Eliminating erroneous data points The historic data to be included in the anal ysis contained some out liers, unreasonably low quantities at som e of the data points. i.e. 7 t ons of asphalt for a 2 mile new road construction project. There might be a number of explanations to this phenomenon a nd it is not possible to identify any erroneous entries except for the significantly low and higher quantity entries. And as there were no benchmarks as to how to eliminate those entries, an ad hoc approach for each data set for different resources was adapted and the data was filtered accordingly. Adjusting the project costs with respect to con sumer price index The project data consisting of the data set to be used in prediction model development starts from 1993 to 2003 and the predicted quantiti es are going to be valid from 2009. In order to make up for the increase in prices of material and equipment, 4% average inflation was assumed to be a good representative of the change in prices of constructi on materials. Note that the time frame of these projects exclude the 2004-2006 pe riod, through which the construction material prices rocketed to unprecedented amounts in a 3year span. After the adjustments were made, a comparative analysis of correlation values betw een the historic material quantities and the project cost was carried out. The study aimed at both finding whether the project cost and the resource requirements are correlated and whet her the inflation adjustment has improved the correlation values. Table 6-2 displays the result of this analysis. The results were rather interesting as elimin ating the outliers did not significantly improve the correlation of project cost and resources ne eded. However, the high values of correlation values, except for the asphalt, were good indicators that the pr ediction model be precise. For asphalt case, the correlation between cost and the quantity is relative low, yet there are 4 more

PAGE 78

78 variables and 3 of them are categori cal variables. It is very likely that different types of projects have significantly different resour ce requirements than each other. Model Development Measuring Accuracy There was a need to m easure the accuracy of the developed method over the historic data set, and the best representative of the error determined to be mean absolute percent error (MAPE). The main attraction of the formula was due to the representati on of the error per fraction of the population mean. As the quantities of different re sources varied significantly it was necessary to have an error term like MAPE However, there are some problems with error term to be identified. Error Term Limitations The erro r term tends to inflate the over pr edictions significantly and the case becomes rather severe when data range gets wider. Calculating the weighted MAPE will provide a better understanding of this flaw. Another point to be noticed is that, the total resource require ments are sums of individual predictions over a data set and the over and underestimates are to cancel the effect of each other. The MAPE term cannot show this; ra ther another analysis will be useful as to how the predicted totals relate to th e actual sum of the data quantities. Although the MAPE has some disadvantages it is to indicate some insight as to how accurate the prediction model is. In order to get a better idea about how the prediction model compares to the historic data totals, absolute percent error (APE) wa s calculated. APE is found by a similar formula as the MAPE, but as it is applied to only one hi storic sum and a single predicted sum, there is no need to sum the erro r terms and average over the data set. A detailed analysis related to the accu racy of the predicted mode l is given in Table 6-3.

PAGE 79

79 Prediction Model Development Details for Different Reso urces The analysis of different resources has been similar except for the steel prediction models for which the data needed to be further filtered.. All the data were sorted out with respect to the input data restrictions of the NeuroSolutions software with mi nor modifications depending on the data combination. Prediction Results The hybrid model developed using the geneti c algorithm optimization tool embedded in NeuroSolutions version 5. The genetic optimi zation used to optimize the neural network structure. The maximum epoch number was set to 1000 for each neural network model. For genetic optimization, 50 chromosomes were crea ted for 100 generations. The prediction results for the FDOT work program is provided in Tabl e 6-4. In addition, as was mentioned before the raw material quantities needed to be calculated from the results of the ne ural network analysis. Table 6-5 shows the numbers of conversion to raw materials. Note that the limestone aggregates assumed to have been used only as base material and not used in concrete or asphalt. This is not the case as some of the limestone aggregates are us ed in concrete and asphalt; yet, the relative amounts of these are not known precisely. The limes tone was be analyzed together with the aggregates in the next chapter. The difference is to be stressed as limestone is locally available and some of the aggregates to be used in other app lications are not. Adjustments to the Future Program Budgets FYWP is updated at the beginning of every fiscal year and becau se of that the total number of projects and the tota l project budgets decreases as time a dvances. Table 6-6 and 6-7 gives the details of the work program content for different resources for the next fi ve years. In order to have the resource requirements for the upcoming years, it is necessary to adjust for the discrepancy resulted from ever decreasing pr ogram budget resulted from the dynamic structure

PAGE 80

80 of the work program. These adjustment factors can be seen in Table 6-8. Adjustment factors have been applied to years 2010-2013 using the to tal program budget as the basis of determining the factors. Adjusted material qua ntities are given in Table 6-9. Adjustments for Substitute Materials Lim estone base and optional base materials are substitutes for each other; however, in prediction modeling these items are considered as independent of each other which is not correct. In order to have this fact take n into account, past data was an alyzed to observe the relative quantities of these resources. The analysis indicated that 56% of th e base material was bituminous, and 44% was limestone based. The updated resource requirements can be seen in Table 6-10. Table 6-10 shows the total quantities, in tons, of raw materials derive d from the quantities predicted earlier. Figures 6-1 to 6-5 represent the hi stograms of the predicted raw material requirements for future work program. Discussions on Prediction Results Asphalt The asphalt requirem ent for next five years has been predicte d to be about 60M tons/year. Although, the number seems relatively high, consid ering the size of the program budget and asphalt being the overwhelmingly most common resource to be used, the quantities are justifiable. The predicted asphalt quantities we re later divided into its main components, bituminous asphalt binder and the aggregates. Th e FDOT study was used to calculate the relative weights of each element. Concrete The concrete requirem ent for next five years has been predicted to be about 500K CY/year. The results would have been highe r had there been more bridge construction projects. However,

PAGE 81

81 they were not common in the work program and th e results were affected accordingly. Similar to asphalt the concrete is composed of other materi als and needed to be disintegrated according to the typical proportions. In addition the concrete prediction results were in cubic yards and needed to be converted to the tons and the conversion was completed by using the same mix design used to determine the component proportions. Reinforcing and structural steel The requirem ents for structural and reinforc ing steel have determined to be around 40k tons for each per year. The predictions for these ha ve been rather challenging as the scarcity of the data and problems with the categorical data. As the structural steel data did not have a lot of exemplars, TS was taken out from the list of the independent variables. In a similar way some of the WMC definitions were excluded as there were no exemplars for both reinforcing and structural steel. These resources were considered as raw resources ; yet, they were bundled under the same category as steel. Optional base As per earlier discussions, th e option al base material is nothing but hot asphalt mix and was treated accordingly. The analysis results in dicated 20M CY of optional base usage but as optional base can be substituted with limestone base the quantity was factored by 56%, which was determined by analyzing the past data. In disi ntegrating the material into its components, the asphalt mix used earlier was determined to be va lid and the quantities needed to be converted from cubic yards to tons. Limestone base The average requirem ents has been predicte d to be around 20M CY and factored down with .46 as can be used interchangeably with optional base. In calcula ting the total aggregate requirements the limestone base requirements were a dded to the total and used as it is after being

PAGE 82

82 converted to tons from cubic yards. All the composite materials have been disintegrated to the raw resources as these are substitutes of each ot her and overall requirement need to reflect that fact. Limitations of the Prediction Model Bid Item-Resource Conversion Issues The resource quantities used in th e prediction model development were determined by grouping the related bid item quantities. The bid it ems to be considered was determined by an FDOT cost study. The main issue with the bid items results from the fact that the list is dynamic and changes constantly. Although it is not possi ble to justify, the ev er changing bid item descriptions can disrupt the data. Reliability of WMC as a Project I dentifier WMC definitions have determined to include work items that are not related to the work mix definition. For instance, add lanes and reco nstruct bid item includes both structural and reinforcing steel bid elements in it. So it is mo re than likely that this WMC will include some sort of bridge construction but it is not disclose d by the WMC. In addition the severe difference among the total material requirements of differe nt projects within the same WMC definition. Compatibility of Different Databases The network training data was gathered from two databas es that are maintained but different offices and there have been some discrepancies between two. A single source database(s) would have been easie r in terms of tracking irregula rities. In addition there is no benchmark to measure the correct ness of the databases used. For instance the work program indicates a budget of 6B for 2009 and from th e database records 2002 spending was 800M. this is a clear indicator that the da ta management is not comprehensive and consistent with the projects completed.

PAGE 83

83 In addition to the unusual tre nds in the past data, the future data (FYWP) had some discrepancies as well. The unusual project budgets were explained by the fact that the most of those projects were at the conceptual stage and the numbers were randomly chosen so as to have the project included in the work program. However, the unusual length data trend was not explicable and the officials communicated c ouldnt provide a sufficient answer either. User Defined Errors These are th e errors introduced to the databa ses in data recording and/or the errors occurred during data gathering and processing. Th ere is no method to completely eliminate the former except for looking out for outlier data poin ts. The latter can also reduce the reliability of the model as there are no proposed methods to account for these and/or observe these as the work progresses. Resource Leveling/Allocation for Work Prog ram Resource Requirement The resource loads clearly indicate that th e resources requirements are not smooth but rather fluctuating significantly. These results were expected as no resource management principles were involved in work program deve lopment stage. As per the case at the project level, this will create problems in both obtai ning and optimizing the price of these resources. The resource requirement for each year can be flattened by moving some of the projects on the time scale. As the prediction model develope d in this study is not very suitable for point predictions, using the budget allo cated as the benchmark for leve ling can be used effectively. The leveling here needs to in clude the overall program budget for the fiscal year as per dollar resource spending as the source of leveling is more accurate. In addition some allocation principles can be adopted and applied. These are likely to be more efficient when combined with the leveling, than trying to level the resources as not only the model is not very suitable for point predictions but also there are factors to be considered. The

PAGE 84

84 optimal case is to determine set of heuristic rules to be applied to the work program in resource assignment and categorize the projects according to that reasoning. Different project batches can be prepared using the heuristics as the basis and developed mode ls can be used to predict the requirements of different batches. Some heuristic rules to be adapted can be but nut limited to following: Giving priority to the already star ted projects in resource allocation Using the budget and/or size of the projec ts as the basis for resource allocation Determining strategically more import ant projects, i.e. adding lanes over resurfacing, as the basi s of resource allocation. Either one or multiple rules can be used to optimize the project combination and count of work program to smoothen the resource demand for different years. Leveling the resource demand is not necessarily need to be requiring sa me level of resources ev ery year; as the overall capacity is the total supp lier capacity for different resources. Better supplier relations and input will be required to carry out such an analysis. Conclusions and Recommendations Conclusions The corporate resource requirem ent prediction is feasible and affordable provided there is data available to carry out analysis. The FDOT case methodology developed in this report can be used in other large programs as the methodology is generic and exce pt for the details of data acquisition and analysis, the procedures are universal. One of the basic assumptions of the analysis in this report has b een the importance of smooth resource demands and it has been proven that it might be more significant than the project level resource management case. The level of fluctuations in resource requirements can be in millions of tons and considering the stage of economy, these disturbances are to be avoided as supplier cap acities are likely to easily adjust to these abrupt changes in demand.

PAGE 85

85 Resource leveling and allocation synergies at the corporate level resource related issues are essentially the same in terms of the rela tion to the work program. And it is reasonable to pay attention to the resource requirements and their fluctuations from year to year as these kind of disruptions may affect the av ailability and/or price of the resources. Recommendations Better d ata management will enable better pr ediction results for work program and better point prediction ability. Review of the work program development structure may result in significant improvements to the data stability. Input from suppliers in terms of their pr edicting capacity will enable better resource leveling/allocation capacity as th e supplier capacity is the determinant factor in resource leveling/allocation process.

PAGE 86

86 Table 6-1. Data definitions and examples to be used in prediction modeling Variable Name Notation Code (if nonnumeric) Types (Non-numeric data) Unit (Numeric data) Work Mix Code WMC 213 Add lanes & reconstruct NA Work Mix Code WMC 218 Add lanes & rehab. pavement NA Work Mix Code WMC 103 Interchange NA Work Mix Code WMC 002 New road construction NA Work Mix Code WMC 012 Resurfacing NA Trans. System TS 1 Intrastate interstate NA Trans. System TS 2 Intrastate turnpike NA Trans. System TS 3 Intrastate state highway NA Trans. System TS 5 Non-intrastate state highway NA Trans. System TS 6 Non-intrastate off-state highway NA Project Length Length NA NA Miles Project Budget Cost NA NA Dollars Number of Lanes Lanes NA NA Each Table 6-2. Results of comparative corre lation analysis of resources and cost Resource Correlation with unadjusted co st Correlation with adjusted cost Asphalt 0.34 0.39 Concrete 0.78 0.77 Str. Steel 0.81 0.77 Rein. Steel 0.78 0.77 Optional base 0.68 0.69 Limestone base 0.77 0.77 Table 6-3. Accuracy of prediction models Resource MAPE (%) Weighted MAPE (%) Actual total qty. Predicted total qty. APE (%) Asphalt 223.42 69.68 95,262,474 TN 95,610,677TN 0.37 Concrete 153.90 59.61 228,186 CY 228,887 CY 0.31 Limestone base 106.90 34.52 14,770,906 CY 14,764,529 CY 0.04 Optional base 152.57 45.94 29,872,063 TN 29,082,984 TN 2.64 Rein. Steel 166.13 64.14 55,747,030 LB 57,829,774 LB 3.74 Str. Steel 212.82 26.70 20,199,200 LB 20,645,244 LB 2.21

PAGE 87

87 Table 6-4. Predicti on analysis results Year Asphalt (TN) Concrete (CY) Limestone base (CY) Opt. base (TN) Rein. Steel(LB) Str. Steel (LB) 2009 68,752,575 536,378 28,584,471 29,067,962 85,056,888 72,047,680 2010 39,266,066 292,568 11,993,075 12,740,937 45,558,684 31,592,962 2011 31,873,347 241,467 9,743,893 10,107,069 33,047,278 19,494,266 2012 32,840,099 248,235 13,194,231 13,394,009 43,903,382 34,489,327 2013 12,926,962 142,592 8,254,367 8,413,431 34,864,691 29,744,630 Table 6-5. Raw resour ces conversion matrix Resource Aggregate Bit. Materi al Steel Limestone Cement Asphalt 93.13% 6.87% NA NA NA Concrete 74.8% N A NA NA 18.5% Str. Steel NA NA 100% NA NA Rein. Steel NA NA 100% NA NA Base Limestone NA NA NA 100% NA Base Optional 90% 10% NA NA NA Table 6-6. Program budget w ith respect to project including different resources Year Asphalt (1000$) Concrete (1000$) Limestone base(1000$) Opt. base (1000$) Rein. Steel (1000$) Str. Steel (1000$) 2009 3,799,576 3,788,438 3,265,636 3,799,576 3,788,438 2,178,009 2010 2,367,550 2,367,550 1,654,720 2,367,550 2,367,550 1,329,115 2011 1,825,882 1,825,882 1,568,631 1,825,882 1,825,882 790,632 2012 1,875,214 1,875,214 1,711,792 1,875,214 1,875,214 995,953 2013 1,064,273 1,064,273 915,571 1,064,273 1,064,273 1,012,915 Table 6-7. Program project c ounts with respect to project including different resources Year Asphalt Concrete Limestone base Opt. base Rein. steel Str. steel 2009 430 425 389 430 425 190 2010 235 235 221 235 235 64 2011 168 168 162 168 168 40 2012 168 168 162 168 168 57 2013 60 60 55 60 60 48 Table 6-8. Program adjustment factors for different resources Year Asphalt Concrete Limestone base Opt. base Rein. steel Str. steel 2009 1.00 1.00 1.00 1.00 1.00 1.00 2010 1.60 1.60 1.97 1.60 1.60 1.64 2011 2.08 2.07 2.08 2.08 2.07 2.75 2012 2.03 2.02 1.91 2.03 2.02 2.19 2013 3.57 3.56 3.57 3.57 3.56 2.15

PAGE 88

88 Table 6-9. Factored resource re quirements for work program Year Asphalt (TN) Concrete (CY) Limestone base (CY) Opt. base (TN) Rein. Steel(LB) Str. Steel (LB) 2009 68,752,575 536,378 28,584,471 29,067,962 85,056,888 72,047,680 2010 63,016,368 468,153 23,668,667 20,447,365 72,900,783 51,771,108 2011 66,326,961 501,009 20,285,209 21,032,343 68,568,267 53,702,211 2012 66,540,913 501,502 25,171,023 27,139,065 88,696,672 75,423,303 2013 46,150,729 507,577 29,441,472 30,036,908 124,106,052 63,958,054 Table 6-10. Adjusted factored res ource requirements for work program Year Asphalt (TN) Concrete (CY) Limestone base (CY) Opt. base (TN) Rein. Steel(LB) Str. Steel (LB) 2009 68,752,575 536,378 12,577,167 16,278,059 85,056,888 72,047,680 2010 60,592,649 450,148 10,414,214 11,450,524 70,096,889 49,779,896 2011 61,322,999 463,211 8,925,492 11,778,112 63,395,211 49,650,708 2012 59,154,615 445,833 11,075,250 15,197,877 78,850,996 67,051,064 2013 39,449,823 433,880 12,954,248 16,820,669 106,086,330 54,671,603 Table 6-11. Raw resource requirement s for work program (all in tons) Year Agg. Asph. Agg. Conc. Agg. Opt. Base Cement Bit mat. Rein. Steel Str. Steel Base Limestone 2009 64,029,273 794,397 19,777,841 196,4756,351,10842,528 36,024 12,577,167 2010 58,687,143 693,353 13,912,387 171,4845,474,27736,450 25,886 10,414,214 2011 61,770,299 742,014 14,310,406 183,5195,734,47334,284 26,851 8,925,492 2012 61,969,552 742,744 18,465,420 183,7006,091,14844,348 37,712 11,075,250 2013 42,980,174 751,742 20,437,112 185,9264,852,62262,053 31,979 12,954,248 Table 6-12. Total raw resource re quirements for work program Year Aggregate Cement Bit material Steel Base Limestone 2009 84,601,512 196,475 6,351,108 78,552 16,350,317 2010 73,292,884 171,484 5,474,277 62,336 13,538,478 2011 76,822,719 183,519 5,734,473 61,135 11,603,140 2012 81,177,716 183,700 6,091,148 82,060 14,397,825 2013 64,169,029 185,926 4,852,622 94,032 16,840,522

PAGE 89

89 Figure 6-1. Total aggregat e requirement per year Figure 6-2. Total bituminous material requirement per year

PAGE 90

90 Figure 6-3. Total cement requirement per year Figure 6-4. Total steel requirement per year

PAGE 91

91 Figure 6-5. Total base material (limestone) requirement per year

PAGE 92

92 CHAPTER 7 APPLICATION OF SUPPLY CHAIN CONTEXT TO FDOT PROGRAM The analys is carried in the previous chapter was the first step of the two step solution proposed in this report. The quant ities obtained in step one can be used to flatten the resource requirements and thus, optimize the price and the po ssibility of delays w ithin the existing supply chain. In this section each resource will be eval uated with respect to FDOT and their supplier relations. Looking into the details of each resource and the supply chain issues is to reveal the problems the institution facing and the more gene ralized solutions can be devised using this analysis. Aggregates Aggregates are one of the m ost important resources for transp ortation construction, as it is the main component for asphalt and structural conc rete as well as being the base material for a number of transportation projects. Aggregates for FDOT projects are mainly supplied from Florida, Georgia and Alabama; although, internat ional imports have started to become more significant. State of Florida has a lot of limestone mines and the oolitic limestone is the main type of local aggregate used in transportation co nstruction; however, limestone is in relatively lower levels of the hardness scale for the rocks. The importance of hardness is clear as it is the major determinant of strength and durability of the material and in case of asphaltic and concrete mixes the aggregates add to strength and durabil ity of the mixture and overall structure. Florida Aggregate Overview Of asphalt and concrete, asphalt is m ore relia nt on the aggregate mix for strength and durability than the concrete and recently superpav e asphalt has determined to be the standard for transportation projects. With the difference in th e matrix composition of the material, superpave asphalt requires stronger aggregates than the re gular asphalt and most local Florida aggregates

PAGE 93

93 are not suitable for this purpose. Thus, the aggr egates are either importe d via ports from other countries or provided fr om Alabama and Georgia. Both stat es have granite mines that are stronger and more durable than the oolitic limestone of Florida and more suitable for superpave asphaltic mixes. The census information about the state indicates that only 20 counties of total 67 consists of more than 80% of the residents and inevitab ly there is a relation between the population and the transportation needs of the ar ea. Although the relation may not be linear, it is unreasonable to overlook the fact that highly popul ated areas have higher transpor tation loads and requires more construction work than the relatively less populated regions. Because of these reasons it is safe to assume that the majority of the work FDOT is carrying out is concentr ated in South Florida (Roughly is the region south of Orlando). And this imposes a great risk for contractors doing business for FDOT as not only th ere is a significant transporta tion cost involved, but also Georgia granite has quotes in terms of quantities to be imported. According to a study published in Aggregates manager in august 2006, the transporta tion cost of aggregates can be more than the cost of the material if the di stance travelled is more than 20 miles (Dunphy 2006). The study was completed when the diesel fuel was about 3$ per gallon, which is about 5$ by 2008. If the model in the above study is still valid the cost of tran sporting aggregates from GA to South Florida will be many times more costly than the actual price of the material. However, due to its geography south Florida is home to a number of major ports and importing the aggregates from other states or co untries via sea freight may prove to be more efficient than the traditional truck hauling. Yet, as expected due to historic all time high pieces for crude oil and diesel, the shi pping costs have been at all time highs. In addition to the ever

PAGE 94

94 increasing petroleum prices, Chinese and Indian interest in steel not only pushed the steel prices up, but also increased demand for sea freight an d resulted in all time high shipping costs. Recycled Asphaltic Product Usage State of Florida has been invest ing in m ethods to be able to use recycled asphaltic products (RAP) in asphaltic mixes. The design specificati ons limit the use of recycled material to 15%, but when the amount of asphaltic mix requirement is considered (both as pavement and base material), the savings can be si gnificant. In the analysis in pr evious section this fact was discarded as the quantity reduction in aggregat e and asphaltic binder requirement would have been the same for different years. Problems Diminishing local supply A very large sum of the overall construction aggregates and nearly a ll of the relatively harder, suitable to be used in concrete and aspha lt mixes, aggregates (local) are produced in one single region, Lake Belt region located in Miami-Dade County (F igure 7-1). In addition to the apparent transportation problems, Miami-Dade County is the most heavily populated region in Florida and traffic jam is correlated with th e population density. Moreover, 30% of the total production at the region was halte d by federal court due to environmental concerns. The environmental concerns not only affect the ex isting aggregate production but also affect the possibility of opening new mines. Aggregates are probably the only resource that can all be locally produced and it is necessary to be able to do so as aggregates are by far the most widely used in transportation construc tion material and transportation and handling costs are to be minimized.

PAGE 95

95 Importing the aggregates It is a common practice f or contractors to import granite from Alabama and Georgia; however, these are a far cry from resolving the problems as truck freight is far from being feasible due to high transporta tion cost, rail and port capacity are not large enough to meet the needs of the work program. In a ddition there are limitations to how much the states can import to the Florida, as well as the environmental constrai nts they too will be facing if the capacity of those mines to be improved. There is a definite ne ed for an alternative than the current practice as the judicial effect of closing the mines will be much severe as the safety inventory soon will be depleted. Transportation constraints As the truck freight is not feasible, the rem ain ing options are trains and sea freight. It is obvious these are better options, in terms of economy, than the tr uck freight; however, both train and seaborne transportation requ ires much more complex trans portation logistics planning and information exchange. It is not likely to have a ship or barge available for transportation of aggregates in short notice. These agreements are tend to have been arranged in advance. Similarly, the trains require booki ngs in advance. In addition both of trains and ships require docking stations with a lot of expensive, hea vy and specialized equipm ent requiring skilled workers. Supply chain issues The core of the aggregate related p roblem is the discrepancy between the FDOT and the suppliers. As the study by the Florida Trans portation Builders Association indicated, the suppliers have the ability to dictate the contr actors ability to perform and, thus the overall project performance. The fact that FDOT has no control over the aggregates supplier is a severe problem in the very complicated supply construc tion chain. The relation to the suppliers is way

PAGE 96

96 more important for aggregates than any other re source and also the dwind ling local supply paired with increased competition for the materials adds to the criticality of the resource. Discussion of Limeston e as Base Materia l Limestone is widely used as the base ma terial for FDOT transportation construction projects and the limestone used is all locally provided. Florida limestone is more suitable to be used as base material than being used in structural c oncrete and asphalt. This is due to relatively high strength and hardness required in asphalt and concrete mixes, and as discussed in previous chapters, limestone is not a very hard stone a nd limestone aggregates ar e rarely hard enough not be used in concrete and/or asphalt. Special attent ion and strategic planning is to be applied to limestone. This is the sole transportation constr uction material the state can self sustain and keeping it that way is crucial. Bituminous Materials Bitum inous materials are used extensively as not only they are the binder to the asphalt mixes FDOT uses to resurface/pave the roads but also different specialized asphaltic mixes are used as alternative base material to the limes tone very frequently. Un like the aggregates, the bituminous materials are international commodities as the raw material they are processed form is crude oil. Asphalt binders are petroleum by products and all the complexity of purchasing, transporting, refining, and deliveri ng the gas to the end users appl y fully to the asphalt binders with a couple of extra links in the supply chain making it more complicated. Asphalt binders can be either processed from crude oil in refineries or can be im ported processed and ready to be used in asphalt mixes. However, in both cases there is a need for a plant to prepare the hot asphalt mixes before transporting it to the construction site where it will be applied. In former the binder will be moved from a refinery within US to the asphalt mixing plant or a distribut ion center by different

PAGE 97

97 means of transportation. Should the binder be imported directly, the material will be delivered to a port seaborne and is to be transported to the mixing plant from port and/or a distribution center. In addition, the aggregates are to be readily available at th e mixing plant to have timely preparation of the hot mix that can be applied. Problems Oil price fluctuations and availability of o il supply Oil still is the majo r energy source for every other nati on and the supply is not evenly distributed, the supply availabil ity dictates the prices much more significan tly than other resources. As explained above, the crude oil and binder itself need to be transported significant distances before being applied at the construction site. Thus, the cr ude oil price is by far the most significant determinant of the availability and th e overall cost of the resource. Table 7-1 shows the per barrel cost of crude oil recently and as can be seen, the oil prices do not follow certain trends, making it extremely hard to have c ontingencies against th e price changes. Transportation constraints Asphalt binder needs to be transported and is to be transported in larger am ounts. The only feasible way to do so is using seaborne freight as the tool due to the geography of the country. And shipping costs have been to astronomic values tri ggered by the ever increasing diesel prices accompanied with demands from China and India caused by their unprecedented growth observed recently. The growth trend of China and India is not very likely to be slowing anytime soon, and the seaborne freight is becoming an international commodity as well and proper strategic transportation pla nning is crucial for successful delivery of the goods. Supply chain issues The bitum inous material processing, purchasi ng, delivering is alrea dy a very multi party oriented and complex supply chain; actual as phalt mix processing and delivery to the actual

PAGE 98

98 construction site only makes a complicated case worse. And the fact that FDOT is not involved in any stage, upstream or downstream, of the whole process is making the situation hard to handle for contractors. Although, th e contractors have escalators to make up for the changes in the price of crude oil and by produc ts, these cannot be more than contingencies added to the total project bid. Static contingencie s not only are far from providing enough protection to contractors in severe cases, but also can be used to infl ate the profits if the predicted escalations in price/availability does not occur. Improved supply chain integrity can reduce the risk both parties carrying resulted from the nature of the supply chain. Cement Ce ment is the binder used in concrete mixe s. Cement is produces rather simple, as powdered clay and limestone mixes are heated in la rge kiln to change the crystal structure of the minerals to give them stickiness and binding ab ility when mixed with water. The chemical reaction taken place when the concrete mix is dryi ng, it creates a paste that hardens and keeps the aggregates and reinforcing bars together, which are the components that give the mix its strength. Water, cement and aggregates (fine and coarse ) are the main components of a concrete mix. Depending on the specific needs of the projects some admixtures (po zzolan, volcanic ash, air entraining admixtures) can be a dded for various purposes. Yet, these admixtures will be in relatively smaller amounts and the main mixture will be the same. Florida has enough limestone and clay res ources to produce cement. The major cost components in cement production, unlike other cons truction resources, are equipment and energy costs. Raw materials used in cement production are cheap and easy to process before cement production. Cement is also easier to be store and transport, reasons to have importing cement as an alternative to producing it. Cement as a raw mate rial is rather less problematic than the rest of

PAGE 99

99 the resources used in construction due to ease of production, cheap and abundant raw materials and ease in transportation and storage. Problems Transportation constraints As per earlier discussions about transportation related concer ns, applies to the cem ent as well, the soaring shipping costs accompanie d by the excess demand from flourishing Chinese and Indian economies and increase s in diesel costs. However, concrete transportation imposes unique problem other resources will not be subjected to. After cem ent is being mixed aggregates and water, it will be plastic and workable for the next 2 hours. After two hours it will start to harden and wont be workable. Th is is a unique problem and need s to be handles with care as concrete is generally transported after being mixe d via specially designed rotary trucks. Special attention needed to arrival time of thes e trucks as their capacity is only 7-9 m3 (9.15-11.76 CY) and quite a number of these are required for large scale projects. The time at the queue for these trucks may be significant due to the limited time the mix will be useful. Supply chain issues Sim ilar to the hot mix asphalt, cement is not to be used as it is in construction, and again like asphalt mixes, aggregates are crucial for tim ely processing of concrete. Any delay in the delivery of the aggregates to the site will delay the processing of ready mix concrete, will disturb the delivery schedule, and resultantly will result in delays in completion of project. As the ready mix concrete is to be delivered and place in a ti mely manner, the delivery schedule will dictate the site lay out and some of the activities precedence relations. Delivery problems will cause the failure of the whole sequence and will result in de lays. The necessity of having a central control of the chain is obvious and it s hould be undertaken by FDOT as they have the centralized capacity to ensure the integrity.

PAGE 100

100 Steel FDOT uses structural and rein forcing steel for variety of pur poses. Reinforcing steel is the m aterial to make up for the lack of tensile strength of the concrete and is used whenever concrete is used as the structural materi al. Reinforcing steel is processe d form iron ore by adding zinc and coal to the mixture. It is then shaped into bars with certain textures to it in order to improve the friction between concrete and the reinforcing bar. On the other hand, structural steel is used as the structural mate rial as it is. Although, concrete is still the most popular construction material for most of the construction jobs, steel is becoming a solid contender as a substitute. Structur al steel can replace the concrete as beams, girders and even columns for some of the engineer ing projects. The production of structural steel is essentially similar to that of reinforcing steel up to the shaping part. At that very last stage the structural steel is molded into different shapes and sizes depending on the needs of the industry. Problems Iron ore and other raw material availability Due to the changing nature of the econom y, the effects of the booming economy in 3rd world countries have significant impact on the re st of the world. It has been argued in various parts how construction resources have become commodities and the how the excess demand from China and India changed the economics of sh ipping. The case is much severe for steel as it is the most demanded resource of all by Chinese. 1/3 of the world steel production is used by China and with the steady and ever increasing gr owth numbers they are pl anning to achieve; any reduction in demand is unlikely to occur anytime soon. The case is so severe that the Sydney ports, Australia is the major iron ore exporter to China, are conge sted and Chine has started to import iron ore from Brazil. The distance Brazil (Rio de Janeiro) to China (Beijing) (10.764 miles) is almost twice that of from Sydney (5,559 miles). And actually the transportation cost of

PAGE 101

101 iron ore from Brazil China well exceeds the materi al cost Ambrogi (2008) It is not surprising that such high demands for iron ore will also cr eate shortages of admixtures used in steel production. Transportation constraints As per discu ssion in other resources the excess demand for international commodities increased significantly the shippi ng costs and the case is not di fferent for steel. However, because of the geometrical design and durability of the materials, steel can be transported rather efficiently and easily than other materials. Supply chain issues Unlike other commoditized resources, steel supply chain is relati vely s traight forward. Yet, steel is an international commodity that is very high on demand in a relatively low price sensitive market. Strong supplier and transportation agen ts relations may prove to be crucial for successful completion of the projects. The high demand for the good and the suppliers ability to change the prices without much resistance due to the scarcity and lack of alternatives makes the case severe on contractor side and gives suppliers/distributers too much power. Some of this power can be reduced by improving the integr ity of the supply chain and increasing the involvement level of the inst itution into the process. Recommendations Strategic Planning and Positioning of the Firm to Minimize Resource Related Damages Resources a re a large part of construction costs and they do cause delays and the firms will bear significant losses. This can be avoidable by adapting stra tegic management principles, coupled with the innovative and dynamic company positioning. The methodology will be valid for large scale programs, and the initial stage wi ll be predicting the resource requirements in

PAGE 102

102 order to foresee possible problems related to those and position the firm in a way to reduce the damages. In this report a case study was studied and th e resource requirements were determined with reasonable efficiency for a future program with limited time and somewhat limited data. Had there been a more comprehensive study as to de termine the materials requirements with better data feeding, the accuracy of the model can be improved. Once the resource requirements for the plan are obtained the resource leveling and/or alloca tion (if applicable) can be carried out to have smoother resource requirements over the timeline. The input from suppliers/producers and a market analysis can indicate the levels of availability of each resource compared to previous years, and this information should be enough to ha ve a relatively smoother resource load curve and reduce the damages. Improved supply chain integrity It is apparent by the analysis ca rried out and the literature re view that it is necessary to have an integrated supply chain in order to achieve successful completion of the projects. The owners need to get more involved in supplier selection process. It is not common for traditional construction contracts to have an integral structure for damages a nd rewards. In traditional view every party is to complete what they have been assigned to and the contra ctual agreements are to keep them safe from any damages from th e change of contractual agreement clause. Improved supplier relations The suppliers are crucial for successful co m pletion of any construction project. The amounts of the resources, the non-traditional layout of transportation projects, length and the duration of the projects make it impossible to store the resources to be used in a project near the construction site in a temporary structure. St oring these in warehouse is not a economically feasible option as the inventory carrying costs will be too high to bear. The optimal case is to

PAGE 103

103 implement a combination of a pull-push system to have the resources in re quired quantities at the site at the required time and this will requi re improved supplier-contractor-owner relations. In the case study FDOT is to get hands on suppl ier selection process for their contractors. It is quite likely two contractors working for FD OT projects will plan to use the same supplier for construction resources they will be requiri ng for the construction project they need to complete for FDOT, and they will suffer from lack of supplies because the supplier will not have enough for both. Central planning ability is not being fully utilized by the FDOT. Although the concepts seems very unfamiliar to construction, the synergies exist. In heavy construction (factories, power plants) the equipment is purchased by the owner, and/or the seller is determined by the owner. There are no restrictions as to why the same principle should not be applicable to the case of FDOT. It is necessary for the institu tion to force a fully inte grated supply chain to reduce/minimize the resource related losses and improve the dynamic structure of the supply chain which will help the system to shift from a push system to a pull system. A pull system is more agile, and responsive to th e changes in environment. Alternative contractual agreements Two of the reasons preventing the construction f irms achievi ng fully integrated systems are current contractual agreements and shortsightedness of the pa rties involved in construction. In fact these two reasons are highly related and feed on each other and makes effects of each other worse. The parties approach the contract as a document to minimize their losses resulted from problems faced during construction. Essentiall y this results in inefficiencies and reduced collaboration between parties as the overall picture is not consid ered by the project participants. The synergy can be explained as the bullwhip effect observed in manufacturing industry. The manufacturers suffered greatly as the parties in different levels of supply chain adjusted their ordering time and quantities in order to have contingencies for them. And the results were

PAGE 104

104 disastrous as these contingencies accumulated and their cumulative effect at the last link of the supply chain was significantly hi gher than what the cumulative contingencies applied at the downstream of the chain. Similarly, the contractual agreements in cons truction industry are set up to protect the parties, not in order to maximize the efficiency of the construction and improve the integrity of the supply chain. Agreements to force the parties involved in collaboration with other parties is to improve th is situation and improve the ove rall project performance. Information technology and its benefits over integrity The supply chain integrity has long been an indispensible elem ent for the successful functioning of the manufacturing industry and one of the main difference s between the rather innovative industries and construction industries is the level of IT diffusion. Enterprise resource planning (ERP) systems have long provided manuf acturers/retailers competitive advantage. However, the ERP systems are not likely to be a part of construction industry anytime soon, as not only they require great capital investment, also rigorous training of skilled workers is required. In addition the ever changing parties an d locations make it near impossible to have such a system even the first two hurdles are pa ssed. ERP systems require all the parties involved in a supply chain to be a part of an integral system, and the business environmental and the geology of the activity area constantly changes in construction. Another alternative to provide data integrity ha s long been thought to be the electronic data interchange (EDI) platforms. These systems enable the parties to exchange business related information seamlessly and instantly. However, similar to ERP systems, EDI systems are very expensive to implement complicated to use and requires broadband networks as the files produced are relatively larger. Alth ough it is undeniable that improved data interchange abilities via an EDI system will improve the overall integr ity of the supply chain, construction industry as a characteristics, requires signifi cant amount of data to be inte rchanged between various parties

PAGE 105

105 mainly due to the laws and regulations, it will not be a very feasib le option mainly because of the capital investment required (Agdas 2006). An alternative data exchange platform can be developed using a current advancement in computer literacy. Extensible mark-up language (XML) provides free, practical and easy to implement data exchange platforms. A recent re search project completed by Ellis et al. (2008) proved that XML technology can be incorporated in to software programs used on a daily basis by novice computer users and these simple daily applications can provide quite powerful data exchanging abilities. In the study it was shown that Microsoft excel spread sheets, that are used by the contractors working for FD OT to record a number of progr ess related quantities, can be coded into data exchange platforms that can in teract with any form of database in both directions. This is a quite si gnificant advancement as the applications do not require any additional hardware or software. An internet browser and internet connection are the only prerequisites to have such a functioning system. In addition the produced files sizes are relatively insignificant as they are in plain text format. With the 3G technology even at the remote sites, a laptop with a 3G adaptor should provide enough ne tworking power to submit/r eceive the files to be exchanged. The technology also can be used to unify the data standards, which are the main boundaries between the data interc hange among different software ap plications. In other words, XML technology can be used to crea te interfaces that will work as adaptors of different files types and will enable users to freely use output of different software platforms with their application platform. Such application diffusion through the levels of supply chain will improve the responsiveness of the system as well as reducing the complexity a nd level of uncertainty. As the need for users in data exchange is to be elimin ated, user defined errors are to be reduced and

PAGE 106

106 better data management will be achieved w ithout significant investment. By reducing the complexity of the supply chain, the system will be better adaptive and agile than the traditional systems. Agile and adaptive systems will better adjust to the discrepancies in supply demand related issues. Logistics of Transportation Sy stems and Innov ative Approaches Undoubtedly transportation costs are of major in terest in any heavy construction firm as transportation constraints present a great deal of problems. Atte ntion should be paid to the logistics transportation planning. Concepts like, logistic network planning can be adapted and with the advancements in geographic informati on systems (GIS) technology the supply networks can be retrieved for any of the resources to be used. Once the possible rout es to the suppliers and their capacity constraints are iden tified optimization methods can be applied to minimize either total cost and/or delivery time. Figure 7-2 shows some of the GIS abilities that can be easily applied to this problem. In the figure a ll the major transportation means and major ports/exchange stations are highli ghted. This study can be used to find the optimal route, and/or capacity optimizing for resource requirements. Th e GIS applications allow users to customize the maps; so, actual road location and length can be inserted in to the picture to have an accurate picture of the situation. The GIS technology enables the users to have maps with spatial data; FDOT can devise state maps with s upplier capacity and location identified. In addition to the network optimization problems, a methodology can be adapted to maximize the efficiency of distribution models There have been a number of different methodologies to optimize the distribution netw orks including but not limited to travelling salesman problem (TSP) and nearest neighbor (NN) algorithms. The logic behind these algorithms can prove to be beneficial in te rms of enhancing the transportation spending.

PAGE 107

107 Push-Pull System Synergies and Th eir Effect on Inventory Costs The trad itional construction supply chain system push system, due to its nature is slow, not agile and non-adaptive. One of the major cause s of the resource related problems stems from this very fact and IT diffusion should be of the highest priority as IT implementation can improve the supply chain integrity and a hybrid pu sh-pull system can and should be attained. Not only the system will be able to adjust to the changes in demand and supply better, the reduced need for inventory will reduce the losses and inve ntory carrying costs sign ificantly. Considering the cost fraction of the resources in construction projects, such a system adaption will improve the profits significantly.

PAGE 108

108 Table 7-1. Historic crude oil prices Year Inflation Adjusted Price 1990 $38.02 1991 $31.86 1992 $29.47 1993 $24.92 1994 $22.69 1995 $23.62 1996 $28.01 1997 $24.95 1998 $15.70 1999 $21.30 2000 $34.16 2001 $27.92 2002 $27.22 2003 $32.34 2004 $42.80 2005 $54.99 2006 $62.11 2007 $66.40 2008 $98.66 May 2008 (highest in 2008) $117.40 Figure 7-1. Miami-Dade County location

PAGE 109

109 Figure 7-2. Highly populated c ounties of the state and major means of transportation

PAGE 110

110 APPENDIX A DATA ANALYSIS AND MAPPI NG ADDIT IONAL DOCUMENTATION Table A-1. Neural network comparison set Linear Regression 1 Layer NN 1 layer NN-GA 2 layer NN-GA Actual Quantity Out Qty MAPE Out Qty MAPE Out Qty MAPE Out Qty MAPE 33,321 123,12 63.05 35,499 6.54 41,739 25.26 40,898 22.74 39,680 906,62 128.48 61,704 55.50 40,401 1.82 39,811 0.33 40,670 393,66 3.21 40,568 0.25 38,028 6.50 39,193 3.63 46,154 5,270 88.58 33,572 27.26 42,014 8.97 41,540 10.00 53,672 56,997 6.20 53,074 1.11 55,021 2.51 53,506 0.31 60,156 74,050 23.10 70,220 16.73 52,836 12.17 57,562 4.31 81,530 130,182 59.67 119,996 47.18 89,650 9.96 84,089 3.14 121,311 101,183 16.59 81,483 32.83 121,181 0.11 121,261 0.04 135,380 196,409 45.08 181,210 33.85 135,415 0.03 135,283 0.07 140,654 110,585 21.38 111,725 20.57 138,840 1.29 140,177 0.34 145,080 162,030 11.68 149,916 3.33 143,241 1.27 144,974 0.07 163,626 190,715 16.56 188,181 15.01 163,404 0.14 163,697 0.04 184,160 178,460 3.10 170,459 7.44 183,071 0.59 183,461 0.38 185,979 161,029 13.42 155,567 16.35 186,493 0.28 186,349 0.20 248,255 228,319 8.03 229,030 7.74 248,471 0.09 248,325 0.03 344,426 286,483 16.82 318,687 7.47 345,766 0.39 344,142 0.08 Average 32.81 Average 18.70 Average 4.46 Average 2.86

PAGE 111

111 Table A-2. Sample FDOT bi d item data base entries Cont ID Vendor ID Pay item code Item description Original qty Unit price 20808 VF592319514002 2730 76216 Casing steel 24.5 820 20817 VF363285002001 2730 76216 Casing steel 308 800 20817 VF590594298007 2730 76216 Casing steel 308 910 20817 VF590767981001 2730 76216 Casing steel 308 288 20817 VF591104806003 2730 76216 Casing steel 308 946 20817 VF592011933001 2730 76216 Casing steel 308 400 20817 VF592339951001 2730 76216 Casing steel 308 9.84 20817 VF592508000001 2730 76216 Casing steel 308 1100 20817 VF592023298001 2730 76216 Casing steel 308 800 20817 VF592098662003 2730 76216 Casing steel 308 1000 20940 VF351035114001 2730 76216 Casing steel 327 683.78 20757 VF592905929001 2647 11 25 Mast arm comb 1 24977.71 20757 VF840915605002 2647 11 25 Mast arm comb 1 22000 20559 VF363285002001 2647 11 26 Mast arm comb 1 20000 20559 VF590594298007 2647 11 26 Mast arm comb 1 10000 20559 VF591104806003 2647 11 26 Mast arm comb 1 17798.4 20559 VF592023298001 2647 11 26 Mast arm comb 1 27273 20559 VF592098662003 2647 11 26 Mast arm comb 1 16733.92 20559 VF592629362001 2647 11 26 Mast arm comb 1 15800 21227 VF592319514002 2300 1 1 Bit. mat. 5789 0.4 21232 VF351035114001 2300 1 1 Bit. mat. 725 0.35 21232 VF581913649001 2300 1 1 Bit. mat. 725 1 21232 VF591166537001 2300 1 1 Bit. mat. 725 1 21232 VF592098662012 2300 1 1 Bit. mat. 725 1 21232 VF593666875001 2300 1 1 Bit. mat. 725 1.05 21232 VF381650909002 2300 1 1 Bit. mat. 725 0.63 21233 VF351035114001 2300 1 1 Bit. mat. 2944 1 21233 VF590582156001 2300 1 1 Bit. mat. 2944 0.8 18980 VF650240625001 0536 2 Guardrail 40 33 Table A-3. Sample FDOT contract history summary data base entries Cont_I D Begin Point End Point Lengt h RCI Begin RCI End Rural/Urb an No. Lanes Transp.Syst em 21396 0.000 1.000 1.000 0.000 11.065 1 0 01 18797 9.239 10.177 0.938 0.000 10.385 3 2 05 21046 3.707 6.178 2.471 0.977 6.848 3 2 03 19054 2.237 4.044 1.807 2.237 7.469 3 2 05 20757 9.117 13.138 4.021 7.469 9.395 1 2 05 20063 3.991 6.390 2.399 2.237 7.469 3 2 05 19805 13.138 16.868 3.730 12.358 15.792 1 2 05 21140 6.390 9.154 2.764 2.237 7.469 3 2 05 21141 7.716 11.831 4.115 5.558 11.891 3 4 05 19660 11.900 12.300 0.400 0.000 18.337 1 2 05 21450 11.842 14.165 2.323 5.558 11.891 3 2 05

PAGE 112

112 Table A-4. Sample FYWP data base entries Dist Seg System Work Year Cost 01 1 Non-intrastate off state highwBike path/trail 2009 11,004 01 1 Non-intrastate off state highwSidewalk 2009 210,743 01 1 Off state hwy sys/off fed sys Landscaping 2009 225,000 03 1 Non-intrastate off state hi ghwMiscellaneous construction 2009 2,138 04 1 Intrastate interstate Rest area 2009 1,464 04 1 Intrastate interstate Routine maintenance 2009 1,592,472 04 1 Intrastate interstate Routine maintenance 2010 2,964,000 04 1 Intrastate interstate Routine maintenance 2009 747,913 05 1 Non-intrastate state highway A dv traveler information systm 2009 36,400 05 1 Non-intrastate state highway A dv traveler information systm 2009 15,960 05 1 Off state hwy sys/off fed sys Bridge replacement 2012 1,645,115 05 1 Off state hwy sys/off fed sys Bridge replacement 2009 135,105 05 1 Off state hwy sys/off fed sys Bridge replacement 2010 111,909 05 1 Off state hwy sys/off fed sys Bridge replacement 2011 647,460 05 1 Off state hwy sys/off fed sys Bridge replacement 2012 338,299 06 1 Non-intrastate state highway Routine maintenance 2009 6,201 06 1 Non-intrastate state highway Routine maintenance 2010 6,201 06 1 Non-intrastate state highway Routine maintenance 2011 6,201 06 1 Non-intrastate state highway Routine maintenance 2012 6,201 06 1 Non-intrastate state highway Routine maintenance 2013 6,201 07 1 Intrastate interstate In spect construction projs. 2009 1,000,000 07 1 Intrastate interstate In spect construction projs. 2010 1,000,000 07 1 Intrastate interstate In spect construction projs. 2011 1,000,000 07 1 Intrastate interstate In spect construction projs. 2012 1,000,000 02 1 Non-intrastate state highway Add left turn lane(s) 2011 230,637 03 1 Off state hwy sys/off fed sy s Miscellaneous construction 2009 343,890

PAGE 113

113 APPENDIX B PREDICTION MODEL DOCUMENTATION Figure B-1. FDOT State Material s Office asphalt mix design test result spreadsheet (Source: FDOT State Materials Office, http://www.dot.state.fl.us/statematerialsoffice/laboratory/asphalt/centr allaboratory/compo sition s%5C/index.htm last accessed July 10, 2008 Table B-1. Asphalt mix de sign-district averages District Average bituminous ma terial content by weight (%) 1 7.64 2 7.58 3 7.42 4 6.34 5 6.60 6 6.12 7 7.03 Turnpike 6.20 The statistics of the above values were as following; the average is 6.87, the standard deviation is 0.63, and co-effi cient of variation is 0.09.

PAGE 114

114 Table B-2. Sample concrete mix design Coarse agg. W/C Cement (lbs/yd3) Fly ash (lbs/yd3) Slag (lbs/yd3) Water (lbs/yd3) FA (lbs/yd3) CA (lbs/yd3) Miami oolite 0.24 800 200 236 931 1679 Miami oolite 0.33 656 144 265.6 905 1740 Miami oolite 0.41 494 123 254 1175 1747 Miami oolite 0.37 600 152 278 1000 1774 Miami oolite 0.33 400 400 262 1062 1750 Miami oolite 0.36 380 380 270 1049 1736 Miami oolite 0.41 197 461 267 1121 1750 Miami oolite 0.44 306 306 269 1206 1710 Georgia granite 0.24 800 200 236 931 1948 Georgia granite 0.33 656 144 265.6 909 1981 Georgia granite 0.41 494 123 254 1176 2027 Georgia granite 0.37 600 152 278 1000 2056 Georgia granite 0.33 400 400 262 1066 2045 Georgia granite 0.41 197 461 267 1125 2045 Table B-3. Data input sample with symbolic variables translated WMC(21 8) WMC(21 3) WMC(12 ) WMC(10 3) WMC( 2) TS(6) TS(1) TS(5) TS(2) TS(3) 0 0 0 0 1 0 0 0 0 1 0 0 0 1 0 0 0 0 1 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 1 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0

PAGE 115

115 Prediction data sets and predicted quantities: The prediction data sets was obtained from the five year work plan for various resour ces and the models trained with above data were used to predict the future wo rk program resource requirements Table B-4. Prediction data and results for asphalt WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 1 1 1176984 236696 218 4 5 2653300 147283 213 1 1 3966168 263107 218 4 5 9000000 187104 213 1 1 9660518 312349 218 4 5 16639131 245891 213 1 1 15152448 352049 103 6 5 5165562 30158 213 1 1 20157318 380589 12 1 5 1081000 130931 213 1 1 64798281 382047 12 2 5 8922328 166540 213 1 1 68804811 375871 12 2 5 6801311 144767 213 2 1 4378142 262029 12 2 5 5388453 131507 213 2 1 5990520 279855 12 2 5 3633106 116356 213 2 1 8350269 304881 12 2 5 4070526 119998 213 2 1 172605686 909107 12 2 5 5014437 128158 213 3 1 1010681 181093 12 2 5 5669009 134063 213 3 1 4461728 223562 12 2 5 3429408 114689 213 4 1 1705100 129580 12 2 5 2206832 105072 213 4 1 2000000 132627 12 2 5 2613927 108202 213 4 1 2225000 134970 12 2 5 1992999 103456 213 4 1 2253922 135272 12 2 5 8080552 157626 213 4 1 2800000 141022 12 2 5 3408976 114523 213 4 1 3235799 145670 12 2 5 2427113 106757 213 4 1 3250000 145822 12 2 5 6100975 138073 213 4 1 3590000 149484 12 2 5 6285013 139809 213 4 1 4672090 161326 12 2 5 7775748 154487 213 4 1 5702552 172836 12 2 5 6106969 138129 213 4 1 6161370 178024 12 2 5 6576341 142590 213 4 1 14824949 277744 12 2 5 2012039 103599 213 4 1 15640539 286772 12 2 5 18154133 288165 213 4 1 107560004 580807 12 2 5 2325152 105974 213 6 1 1172671 44365 12 2 5 7306531 149747 213 6 1 1219682 44571 12 2 5 2689540 108791 213 6 1 1394389 45337 12 2 5 7077000 147469 213 6 1 1890036 47536 12 2 5 4658421 125032 213 6 1 2882629 52051 12 2 5 1852829 102407

PAGE 116

116 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 6 1 3350924 54234 12 2 5 19031224 301845 213 6 1 4679205 60608 12 2 5 7186507 148553 213 6 1 7189047 73404 12 2 5 16362915 261271 213 6 1 22329791 172097 12 2 5 5848411 135717 213 8 1 1100000 19071 12 2 5 2874078 110240 213 8 1 1210300 19350 12 2 5 1364005 98814 213 8 1 1295000 19564 12 2 5 2733693 109136 213 8 1 2822338 23525 12 2 5 5303453 130740 213 8 1 5060804 29634 12 2 5 3031048 111484 218 4 1 1247256 265193 12 2 5 7028045 146986 218 4 1 1324179 266030 12 2 5 2121546 104425 218 4 1 1492789 267868 12 2 5 4811940 126373 218 4 1 1770818 270913 12 2 5 2029521 103730 218 4 1 2114386 274696 12 2 5 6813118 144882 218 4 1 2821967 282562 12 2 5 7024236 146949 218 4 1 3274450 287643 12 2 5 1217713 97758 218 4 1 3919380 294950 12 2 5 1044000 96516 218 4 1 35395422 676848 12 2 5 1108000 96972 218 4 1 113915158 967879 12 2 5 1447000 99417 103 2 1 2667200 108352 12 2 5 1101000 96922 103 4 1 1607592 127431 12 3 5 2481464 84164 103 4 1 1654618 127565 12 3 5 7271988 115718 103 4 1 3408770 132387 12 3 5 15207399 191562 103 4 1 3705105 133168 12 3 5 23734302 315040 103 6 1 6169033 122229 12 3 5 4381233 95626 103 8 1 3644675 83832 12 3 5 2366012 83508 103 8 1 3981900 84267 12 3 5 4803267 98351 103 6 1 15061018 131567 12 3 5 1121820 76710 103 3 1 1677174 119053 12 3 5 2277784 83009 103 3 1 2301657 121139 12 3 5 2304994 83163 103 3 1 2648762 122286 12 3 5 3785185 91891 103 3 1 2654793 122305 12 3 5 1742625 80040 103 3 1 5969089 132718 12 3 5 4585727 96938 103 4 1 1380000 126780 12 4 5 4655524 89312 103 4 1 5312360 137226 12 4 5 3034535 81508 103 4 1 6400000 139794 12 4 5 3513724 83747 103 4 1 12459068 151249 12 4 5 2318643 78264 103 4 1 49778910 134038 12 4 5 2703891 79995 103 6 1 2879312 117003 12 4 5 1631462 75259 103 6 1 2027599 115505 12 4 5 4488534 88477 103 6 1 26137824 133568 12 4 5 6141414 97070

PAGE 117

117 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 103 2 1 13567530 147199 12 4 5 1263699 73693 103 4 1 1191408 126237 12 4 5 4182438 86966 103 4 1 2035680 128640 12 4 5 2125444 77409 12 1 1 1212202 118215 12 4 5 1640535 75298 12 1 1 1270370 118788 12 4 5 5159874 91877 12 1 1 7004322 181949 12 4 5 1419851 74355 12 1 1 12300966 251135 12 4 5 3596110 84138 12 1 1 16348584 309374 12 4 5 4485447 88462 12 1 1 258704303 1060267 12 4 5 1515518 74763 12 2 1 5126542 198218 12 4 5 3642920 84361 12 2 1 7870386 241937 12 4 5 9699792 118362 12 2 1 19398408 450763 12 4 5 6393357 98447 12 4 1 1138747 125417 12 4 5 3016096 81423 12 4 1 1349709 128911 12 4 5 2662205 79806 12 4 1 1858138 137549 12 4 5 1398242 74263 12 4 1 3506058 167641 12 4 5 1784897 75921 12 4 1 5731556 213232 12 4 5 6737032 100356 12 4 1 6871638 238652 12 4 5 3526796 83809 12 4 1 10145137 318144 12 4 5 5583750 94087 12 4 1 17501485 513266 12 4 5 2144332 77492 12 6 1 1001235 68638 12 4 5 1035883 72738 12 6 1 2943374 91738 12 4 5 3854711 85375 12 6 1 3112200 93929 12 4 5 6803161 100727 12 6 1 5456007 127601 12 4 5 4160640 86859 12 6 1 5949853 135502 12 4 5 1991058 76819 12 6 1 6251345 140468 12 4 5 3962485 85895 12 6 1 7646571 164883 12 4 5 2460320 78897 12 6 1 9218116 195245 12 4 5 2177070 77637 12 6 1 12343896 264529 12 4 5 1559720 74952 12 8 1 1745365 77576 12 4 5 7323437 103695 213 2 3 1938884 54666 12 4 5 2533434 79225 213 2 3 2000000 55152 12 4 5 2917849 80972 213 2 3 3186742 64959 12 4 5 1127000 73119 213 2 3 3805883 70357 12 5 5 3916620 95334 213 2 3 4000000 72090 12 5 5 4794072 99258 213 2 3 5197159 83191 12 5 5 1837340 86583 213 2 3 6075993 91795 12 5 5 7242670 111010 213 2 3 7423000 105709 12 5 5 1343071 84609 213 2 3 8861138 121494 12 5 5 15367120 160474 213 2 3 9118903 124419 12 5 5 2041000 87408 213 2 3 12728717 168099 12 6 5 12073870 162781

PAGE 118

118 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 2 3 15605648 205724 12 6 5 3926664 119247 213 2 3 18981143 251494 12 6 5 3714290 118261 213 2 3 19593720 259856 12 6 5 3793119 118626 213 4 3 1301978 10353 12 6 5 7647389 137676 213 4 3 2217939 13684 12 6 5 1874348 109996 213 4 3 3917194 20419 12 6 5 2537133 112917 213 4 3 4011524 20815 12 6 5 3698696 118189 213 4 3 8343814 41835 12 6 5 2030390 110678 213 4 3 24640000 183148 12 6 5 1512707 108428 213 6 3 1365500 2868 12 6 5 8567578 142593 213 6 3 2944473 6123 12 6 5 5883715 128660 213 6 3 3447554 7203 12 6 5 5218444 125392 213 6 3 4804195 10223 12 6 5 4879735 123756 213 6 3 23527259 71507 12 6 5 8263001 140949 213 6 3 26958663 87965 12 6 5 1732500 109379 218 2 3 14659298 254661 12 6 5 2822474 114194 218 2 3 23023341 369355 12 6 5 18684310 207953 218 4 3 1362051 144435 12 6 5 17573144 199646 218 4 3 1687559 146693 12 6 5 8824585 143994 218 4 3 17869800 284250 12 8 5 5910101 218640 103 6 3 11261935 108868 213 4 1 36353090 439174 103 8 3 3850821 53279 213 2 1 6800000 288597 103 8 3 10544390 63509 213 6 1 2530000 50430 103 8 3 14014605 69050 213 6 1 1650000 46466 103 8 3 30532009 97036 213 8 1 5450000 30734 103 8 3 221412954 1100904 213 6 1 249217770 1093471 103 1 3 4987446 130315 213 6 1 1197858 44475 103 6 3 34108627 145688 218 4 1 1651718 269607 103 4 3 1396528 121659 218 6 1 4100000 318787 2 2 3 1250000 44353 218 4 1 30708210 624468 2 2 3 1576356 46898 218 4 1 2000000 273434 2 2 3 2310271 52913 218 4 1 39962233 723506 12 2 3 1182762 179410 218 4 1 3418477 289268 12 2 3 1320990 182159 218 4 1 1254161 265268 12 2 3 1882863 193583 218 4 1 1776062 270970 12 4 3 1121933 119451 218 4 1 1720000 270355 12 4 3 1499066 125726 103 4 1 9232000 145767 12 4 3 1836873 131530 103 1 1 48476309 289775 12 4 3 2175141 137519 103 1 1 41327005 267793 12 4 3 2306544 139894 103 4 1 25519263 158854 12 4 3 3352874 159786 12 6 1 36945764 871770

PAGE 119

119 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 4 3 3755425 167913 12 4 1 21917476 624231 12 4 3 4067811 174404 12 4 1 17868268 522886 12 4 3 4459035 182764 12 6 1 25091836 617175 12 4 3 4843570 191230 12 4 1 9910141 312168 12 4 3 4847885 191327 12 4 1 12466090 378716 12 4 3 5095252 196907 12 6 1 22023453 530832 12 4 3 5117504 197414 213 2 3 4000000 72090 12 4 3 13882517 454356 213 2 3 8929852 122271 12 4 3 21814357 715252 213 4 3 22071672 154050 12 5 3 15154163 415717 213 2 3 3043078 63734 213 1 5 2022003 136783 213 2 3 7012906 101381 213 2 5 1007367 111020 218 2 3 6349622 165720 213 2 5 1067184 111434 218 4 3 24186687 346797 213 2 5 1197962 112343 218 2 3 49251652 756118 213 2 5 1286567 112963 218 2 3 1250000 123572 213 2 5 1345065 113374 103 8 3 53396930 136818 213 2 5 1382505 113637 103 6 3 79299607 188610 213 2 5 1423370 113926 12 4 3 4769972 189591 213 2 5 1467914 114241 12 4 3 10268701 336427 213 2 5 1581871 115051 12 4 3 6726481 236283 213 2 5 1856688 117025 12 4 3 12434258 405762 213 2 5 1870704 117126 12 2 3 5624233 279720 213 2 5 1930200 117558 12 4 3 2503280 143500 213 2 5 2514154 121872 12 4 3 11268263 367818 213 2 5 2846411 124390 12 2 3 5987393 288956 213 2 5 2964262 125294 12 4 3 1546637 126533 213 2 5 2989999 125493 12 4 3 1039549 118108 213 2 5 3749812 131476 213 4 5 6634965 167776 213 2 5 3948064 133080 213 4 5 31292523 378313 213 2 5 4000815 133509 213 4 5 12397812 204555 213 2 5 4382030 136652 213 4 5 2000000 142460 213 2 5 4428194 137037 213 2 5 6000000 150752 213 2 5 5093984 142701 213 4 5 1460000 139730 213 2 5 5514468 146387 213 4 5 10276666 190275 213 2 5 5771377 148682 213 2 5 2700000 123275 213 2 5 6835002 158535 213 2 5 3470000 129243 213 2 5 8642282 176643 213 2 5 2798600 124025 213 2 5 9018000 180634 213 2 5 5421129 145562 213 2 5 9110435 181628 213 4 5 17276584 241031 213 2 5 9771581 188882 213 4 5 3342114 149436 213 2 5 15034729 256163 213 4 5 1591340 140390

PAGE 120

120 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 2 5 17170414 288536 213 4 5 1264550 138753 213 2 5 20314536 341463 213 2 5 2807739 124095 213 2 5 21444057 361906 213 2 5 1371870 113562 213 2 5 22927312 389767 213 2 5 2867000 124548 213 2 5 28107048 493642 213 2 5 29482249 522098 213 2 5 29549112 523483 213 2 5 5879000 149653 213 2 5 34124268 616694 213 6 5 32989965 370768 213 3 5 18154370 257080 213 2 5 1801000 116622 213 4 5 1008236 137479 218 4 5 30522524 378326 213 4 5 1031645 137595 218 4 5 9000000 187104 213 4 5 1041888 137646 218 4 5 52019051 580703 213 4 5 1050001 137686 218 4 5 10000000 194118 213 4 5 1107309 137970 218 4 5 1136429 138915 213 4 5 1150000 138182 218 4 5 5238917 162546 213 4 5 1326789 139063 12 2 5 5670187 134074 213 4 5 1353628 139198 12 3 5 4886020 98893 213 4 5 1411139 139485 12 4 5 3151826 82051 213 4 5 1458760 139724 12 2 5 4056243 119878 213 4 5 1464803 139754 12 2 5 12574133 209473 213 4 5 1585790 140362 12 6 5 2678900 113550 213 4 5 1637464 140623 12 4 5 3963163 85898 213 4 5 1729804 141089 12 4 5 2573947 79408 213 4 5 1935954 142134 12 3 5 3582103 90648 213 4 5 1993452 142427 12 4 5 3506780 83715 213 4 5 1998870 142454 12 3 5 11036450 147511 213 4 5 2047680 142703 12 2 5 12026411 202590 213 4 5 2069499 142815 12 4 5 1450000 74483 213 4 5 2095207 142946 12 5 5 4654347 98624 213 4 5 2268244 143832 12 4 5 3242621 82474 213 4 5 2272080 143852 12 4 5 15838532 166408 213 4 5 2700000 146064 12 3 5 18265045 230536 213 4 5 3462362 150075 12 2 5 3014988 111356 213 4 5 3934828 152606 12 2 5 4264549 121642 213 4 5 4418140 155233 12 2 5 2653938 108513 213 4 5 5569697 161645 12 4 5 18120206 188808 213 4 5 6339884 166059 12 4 5 7409819 104196 213 4 5 6684233 168065 12 2 5 11891327 200917 213 4 5 7266605 171505 12 2 5 1095064 96880 213 4 5 7403865 172325 12 4 5 17255530 179993 213 4 5 9239147 183613 12 2 5 5678555 134151 213 4 5 10049427 188798 12 4 5 3530091 83825

PAGE 121

121 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 4 5 10560260 192132 12 4 5 8642835 111604 213 4 5 12350000 204223 12 2 5 3749467 117316 213 4 5 14034501 216209 12 5 5 15014658 157929 213 4 5 14512760 219723 12 3 5 1276958 77531 213 4 5 14663196 220839 12 4 5 5974488 96168 213 4 5 14847933 222215 12 3 5 1655339 79565 213 4 5 15520418 227291 12 2 5 4427366 123034 213 4 5 19820798 262224 12 5 5 1304168 84456 213 4 5 22572609 286944 12 2 5 9453764 172356 213 4 5 27901688 340395 12 6 5 1465903 108227 213 4 5 34099959 411942 12 5 5 6182568 105771 213 4 5 36831121 446453 12 2 5 1880918 102617 213 4 5 154568188 1038514 12 6 5 2582481 113119 213 6 5 3243778 214794 12 6 5 5458885 126565 218 2 5 3589000 96298 12 4 5 3043333 81549 218 2 5 5720000 107075 12 6 5 7118087 134915 218 2 5 7483865 116822 12 6 5 6036580 129420 218 2 5 7933333 119432 12 5 5 2363588 88728 218 2 5 11091826 139344 12 1 5 19976496 359396 218 2 5 63137848 801119 12 6 5 9062784 145303 218 4 5 1013299 138254 12 6 5 3299408 116354 218 4 5 2178065 144615 12 2 5 2425750 106746 218 4 5 2500000 146418 12 4 5 2785860 80368 218 4 5 3223705 150540 12 6 5 3093632 115418 218 4 5 3615000 152810 12 6 5 3970908 119453 218 4 5 3825000 154040 12 6 5 2870358 114409 218 4 5 4815000 159955 12 8 5 3292112 201658 218 4 5 5010000 161142 12 2 5 4030182 119659 218 4 5 5352000 163243 12 2 5 8048887 157298 218 4 5 5752350 165731 12 4 5 5936356 95963 218 4 5 5857787 166392 12 2 5 12142356 204034 218 4 5 8271598 182124 12 6 5 8732398 143490 218 4 5 8683500 184927 12 4 5 12925322 141581 218 4 5 22880098 302230 12 6 5 5123326 124931 218 4 5 36115970 435599 12 5 5 1188531 84000 218 5 5 4500000 203409 12 6 5 11668806 160325 103 4 5 4702760 7762 12 2 5 3514007 115379 103 6 5 5668351 31511 12 4 5 6684147 100060 2 2 5 3646098 22972 12 6 5 4801228 123379 2 4 5 2355424 42971 12 6 5 3916572 119200 12 2 5 1044000 96516 12 6 5 2557786 113009

PAGE 122

122 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 2 5 1160595 97349 12 4 5 2109345 77338 12 2 5 1165841 97386 12 6 5 4457620 121741 12 2 5 1178085 97474 12 4 5 4281816 87454 12 2 5 1216766 97752 12 4 5 3506513 83713 12 2 5 1366215 98830 12 2 5 1538183 100083 12 2 5 1372181 98874 12 6 5 2935859 114705 12 2 5 1383606 98957 12 6 5 5035671 124507 12 2 5 1699586 101270 12 3 5 1425784 78325 12 2 5 1705962 101317 12 4 5 1387705 74218 12 2 5 1760281 101719 12 6 5 4487702 121884 12 2 5 1775052 101829 12 4 5 4520277 88635 12 2 5 1971052 103291 12 3 5 5697556 104353 12 2 5 2016282 103631 12 6 5 4687110 122833 12 2 5 2140886 104571 12 5 5 4846549 99498 12 2 5 2189904 104943 12 2 5 7114726 147842 12 2 5 2274299 105586 12 6 5 10427553 153006 12 2 5 2305694 105825 12 4 5 5289555 92548 12 2 5 2363287 106266 12 2 5 10626696 185712 12 2 5 2411198 106634 12 6 5 11870935 161546 12 2 5 3126788 112248 12 2 5 2776038 109468 12 2 5 3199216 112828 12 2 5 2529045 107543 12 2 5 3437350 114754 12 2 5 1985404 103399 12 2 5 3452769 114880 12 6 5 20018605 218340 12 2 5 3708246 116975 12 4 5 10006799 120399 12 2 5 4084554 120116 213 4 1 13100142 258243 12 2 5 4558109 124161 213 4 1 4659746 161189 12 2 5 4674890 125175 213 2 1 6800000 288597 12 2 5 4710678 125487 213 6 1 1655000 46489 12 2 5 4724476 125607 213 6 1 1900000 47580 12 2 5 4831632 126545 213 8 1 1658000 20491 12 2 5 5022711 128231 213 8 1 1610000 20368 12 2 5 5118497 129083 213 8 1 5725000 31519 12 2 5 5319569 130885 213 3 1 107722494 534758 12 2 5 5475387 132295 213 4 1 25671605 381131 12 2 5 5612662 133547 218 4 1 116245584 972076 12 2 5 5818455 135440 218 4 1 8029224 343163 12 2 5 5826748 135517 218 4 1 10195255 369538 12 2 5 5854406 135773 218 4 1 1656480 269659 12 2 5 6091274 137982 103 6 1 19643263 133609 12 2 5 6525644 142103 103 4 1 3300000 132098 12 2 5 7103062 147726 213 2 3 4000000 72090

PAGE 123

123 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 2 5 7281472 149497 213 2 3 1205000 48976 12 2 5 8210198 158975 213 2 3 8227929 114429 12 2 5 10561682 184953 213 4 3 1593100 11390 12 2 5 11609772 197459 213 4 3 16207313 96870 12 2 5 12660016 210566 213 2 3 1500000 51231 12 2 5 13625430 223120 213 2 3 1500000 51231 12 2 5 15701745 251719 213 2 3 54238235 576422 12 3 5 1359060 77968 213 2 3 77981699 666697 12 3 5 1371589 78035 213 2 3 6550000 96592 12 3 5 1463514 78528 213 2 3 2208874 56827 12 3 5 1465874 78541 213 2 3 1366415 50205 12 3 5 3099236 87752 213 2 3 5143562 82679 12 3 5 4461558 96139 213 6 3 7159000 15851 12 3 5 4946931 99294 213 4 3 5278613 26378 12 3 5 4960900 99386 213 4 3 2760000 15752 12 3 5 6016860 106574 218 2 3 5973235 162299 12 3 5 6057037 106856 218 4 3 58011014 656249 12 3 5 8207008 122980 103 4 3 2005000 124240 12 3 5 9858625 136813 103 8 3 5700000 56043 12 4 5 1079000 72918 103 8 3 128773840 431610 12 4 5 1245992 73619 103 4 3 4000000 132613 12 4 5 1259861 73677 12 3 3 3894753 211931 12 4 5 1268622 73714 12 4 3 13035225 425767 12 4 5 1276000 73745 12 4 3 3611991 164986 12 4 5 1298638 73841 12 4 3 14152837 463549 12 4 5 1300826 73850 12 4 3 17235573 568888 12 4 5 1351702 74066 12 6 3 20581900 492458 12 4 5 1390469 74230 12 2 3 13808116 511597 12 4 5 1396946 74257 12 2 3 10708107 419724 12 4 5 1466441 74553 12 4 3 10062782 330111 12 4 5 1537000 74854 12 2 3 2762067 212272 12 4 5 1627371 75242 12 2 3 13451516 500988 12 4 5 1735858 75709 12 4 3 2727098 147677 12 4 5 1786269 75927 12 4 3 10773729 352142 12 4 5 1826410 76101 12 4 3 18241179 602645 12 4 5 1845560 76184 12 4 3 12255163 399854 12 4 5 1900308 76423 12 4 3 15362918 504942 12 4 5 1965000 76705 12 3 3 3168899 195207 12 4 5 2046056 77060 213 2 5 31792782 569733 12 4 5 2153278 77532 213 4 5 15834468 229697 12 4 5 2216121 77809 213 2 5 17259040 289943

PAGE 124

124 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 4 5 2268020 78039 213 4 5 3415392 149825 12 4 5 2385575 78563 213 4 5 6435287 166612 12 4 5 2466182 78923 213 2 5 9000000 180441 12 4 5 2503795 79092 213 4 5 34886044 421702 12 4 5 2579040 79431 213 2 5 54451658 892843 12 4 5 2590500 79482 213 4 5 17176264 240227 12 4 5 2708251 80015 213 4 5 81580986 911606 12 4 5 2779589 80339 213 4 5 4666666 156598 12 4 5 2784589 80362 213 4 5 3443000 149972 12 4 5 3549995 83919 213 4 5 10000000 188478 12 4 5 3801364 85118 213 2 5 30384990 540776 12 4 5 3807021 85145 213 2 5 21307263 359393 12 4 5 3829577 85254 213 2 5 3080000 126188 12 4 5 4034516 86245 213 4 5 8118107 176644 12 4 5 4464075 88356 213 2 5 49201128 847382 12 4 5 5094596 91541 213 4 5 2561964 145347 12 4 5 5642793 94399 213 4 5 35536766 429889 12 4 5 6464367 98839 213 2 5 1371870 113562 12 4 5 6878027 101149 213 4 5 3300000 149213 12 4 5 7311251 103624 213 4 5 9269484 183805 12 4 5 7613931 105388 213 4 5 4000000 152958 12 4 5 8458572 110466 213 2 5 7036116 160464 12 4 5 11074754 127759 218 2 5 5000000 103315 12 4 5 11639122 131826 218 4 5 9000000 187104 12 4 5 12969867 141932 218 2 5 1237030 85561 12 4 5 13301908 144571 218 4 5 9599233 191283 12 5 5 1353112 84649 218 2 5 3000000 93500 12 5 5 1369000 84712 12 4 5 10045743 120660 12 5 5 1838048 86586 12 2 5 4658578 125033 12 5 5 2088477 87601 12 2 5 12160562 204261 12 5 5 2438951 89039 12 3 5 6976833 113506 12 5 5 2551034 89504 12 2 5 5049679 128471 12 5 5 2675634 90022 12 8 5 3438059 202584 12 5 5 2994084 91360 12 4 5 9948188 120008 12 5 5 3559814 93779 12 4 5 4632350 89196 12 5 5 4712443 98887 12 2 5 9541401 173329 12 5 5 5236308 101291 12 2 5 8635874 163466 12 5 5 6355652 106610 12 2 5 1326196 98541 12 5 5 10244042 127224 12 2 5 1541191 100105 12 6 5 1059299 106488 12 2 5 6238811 139371 12 6 5 1193000 107057 12 2 5 2170201 104794

PAGE 125

125 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 6 5 1285293 107451 12 2 5 9875095 177071 12 6 5 1656000 109047 12 2 5 2443824 106885 12 6 5 1681500 109158 12 2 5 3343393 113991 12 6 5 1773294 109556 12 3 5 3699005 91362 12 6 5 1801397 109678 12 2 5 2565163 107823 12 6 5 1983602 110474 12 2 5 3221396 113007 12 6 5 2244528 111620 12 2 5 7553487 152228 12 6 5 2295081 111843 12 2 5 8090665 157731 12 6 5 2613474 113258 12 4 5 12192709 135940 12 6 5 2896351 114527 12 2 5 8566395 162727 12 6 5 3434068 116970 12 6 5 4882596 123769 12 6 5 3788041 118603 12 4 5 8580317 111217 12 6 5 4123070 120165 12 4 5 7224194 103122 12 6 5 4622137 122523 12 2 5 9590353 173875 12 6 5 4651284 122662 12 4 5 2190546 77696 12 6 5 4939781 124044 12 2 5 1650825 100910 12 6 5 5003451 124351 12 4 5 3407383 83246 12 6 5 5141718 125020 12 4 5 3858722 85394 12 6 5 5380545 126182 12 2 5 3798992 117727 12 6 5 5788574 128188 12 5 5 10881497 130954 12 6 5 5955972 129019 12 2 5 7778322 154513 12 6 5 6038066 129428 12 6 5 2550001 112975 12 6 5 6117293 129824 12 4 5 7110781 102472 12 6 5 7204215 135361 12 5 5 4214954 96652 12 6 5 7231792 135504 12 3 5 17557184 220986 12 6 5 7980439 139438 12 4 5 9301238 115768 12 6 5 9035815 145154 12 3 5 7991840 121273 12 8 5 7638989 230272 12 2 5 3359771 114124 12 2 5 1632193 100773 12 5 5 4948107 99962 213 2 1 6800000 288597 12 5 5 7680992 113247 213 2 1 6507000 285450 12 4 5 10308880 122437 213 4 1 1631931 128829 12 6 5 1472917 108257 213 4 1 3818700 151964 12 4 5 3469388 83538 213 4 1 3848040 152283 12 4 5 1463715 74542 213 4 1 207267948 1017727 12 4 5 5631627 94340 213 4 1 2800000 141022 12 2 5 2238403 105312 213 4 1 1500000 127477 12 3 5 2212977 82645 213 6 1 1654208 46485 12 4 5 3807129 85146 213 8 1 4900000 29183 12 4 5 1047006 72784 218 4 1 95820463 943371 12 2 5 2351420 106175 218 4 1 11000000 379473 12 4 5 1805057 76009

PAGE 126

126 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 218 4 1 5000000 307361 12 2 5 3525944 115477 218 4 1 4396089 300400 12 6 5 2418583 112390 218 4 1 12248906 395011 12 6 5 2007775 110579 218 4 1 1101920 263615 12 5 5 2049415 87442 218 4 1 5517567 313376 12 2 5 1567142 100295 218 4 1 5362223 311566 12 4 5 1914598 76485 103 2 1 6612212 123140 12 4 5 8562635 111107 103 4 1 1220082 126320 12 3 5 10310008 140828 103 4 1 285652252 1117872 12 3 5 1461621 78518 103 8 1 14624049 96314 12 5 5 32146974 341768 103 6 1 192815313 944304 12 2 5 3445881 114823 103 6 1 4451377 119614 12 2 5 8359161 160536 103 6 1 4440000 119596 12 4 5 2436333 78789 103 6 1 76572470 98028 12 4 5 11895284 133714 12 1 1 11700000 242836 12 4 5 8488886 110652 12 4 1 9971149 313716 12 2 5 3577976 115903 12 4 1 2121593 142145 12 2 5 1143486 97226 12 4 1 4155416 180368 12 2 5 1415341 99187 12 4 1 7876752 262113 12 2 5 9309746 170766 12 6 1 5729692 131944 12 3 5 5519522 103133 12 6 1 6892822 151399 12 2 5 1041410 96497 12 6 1 16611070 375639 12 2 5 8474952 161757 12 6 1 5323553 125531 12 3 5 1270504 77497 12 8 1 6723957 128101 12 1 5 12709323 259174 213 2 3 4000000 72090 12 8 5 1438210 190104 213 2 3 4000000 72090 12 2 5 3205411 112878 213 2 3 1027640 47641 12 2 5 2301684 105795 213 2 3 27782525 367105 12 2 5 1201000 97638 213 2 3 20130222 267171 213 2 1 6800000 288597 213 6 3 3030000 6305 213 3 1 4000000 217843 218 2 3 2300000 131528 213 4 1 29847551 409032 218 2 3 10224603 203966 213 6 1 10100000 89497 218 4 3 12315464 232166 213 6 1 200200313 995954 103 8 3 15000000 70651 213 6 1 1550000 46023 103 8 3 23505773 84888 213 6 1 10900000 94159 103 8 3 2645500 51504 213 6 1 4000000 57314 103 6 3 76026825 183821 213 8 1 94937616 669755 103 6 3 4098500 93978 213 8 1 14575435 59939 12 2 3 6851870 311497 213 8 1 6400000 33467 12 2 3 5799141 284151 213 8 1 31429350 135295 12 2 3 3055488 218728 218 4 1 3917121 294924

PAGE 127

127 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 4 3 4288545 179089 218 4 1 4956852 306862 12 4 3 2969782 152298 218 4 1 2231848 275995 12 4 3 10730063 350771 218 4 1 1500000 267947 12 4 3 15776518 519122 103 2 1 100606191 346752 12 4 3 10158057 333026 103 1 1 1924480 93123 12 4 3 5302399 201659 213 2 3 4000000 72090 12 4 3 15106943 496167 213 2 3 42589140 510290 12 4 3 1933328 133220 213 2 3 3927175 71437 12 4 3 3121365 155232 213 2 3 56687098 587007 12 4 3 6899607 240723 213 2 3 4895000 80321 12 4 3 7561050 258139 213 2 3 4647500 78005 12 4 3 18425428 608758 213 2 3 2600000 60022 12 5 3 11614275 309453 213 2 3 6050000 91535 213 2 5 6000000 150752 213 2 3 40045877 491043 213 2 5 12338463 219521 213 4 3 5000000 25116 213 2 5 8802104 178331 213 6 3 2000000 4151 213 2 5 6273557 153263 213 6 3 15000500 38511 213 2 5 6000000 150752 103 4 3 2448336 126114 213 2 5 5157479 143252 103 8 3 30997927 97848 213 2 5 8171987 171760 213 2 5 20000000 335899 213 2 5 10763667 200248 213 2 5 8000000 170005 213 2 5 13126302 229745 213 2 5 1900000 117339 213 2 5 1371870 113562 213 2 5 27353088 478106 213 2 5 1247154 112687 213 2 5 1371870 113562 213 2 5 5000000 141889 213 2 5 37620116 683013 213 3 5 1100000 117237 213 3 5 12004509 194558 213 4 5 15584468 227780 213 3 5 8000000 161832 213 4 5 6795601 168718 213 4 5 1994727 142433 213 4 5 16480483 234715 213 4 5 3825238 152016 213 4 5 1525150 140057 213 4 5 2000000 142460 213 4 5 2309806 144046 213 4 5 3518000 150371 213 4 5 2000000 142460 213 4 5 30800000 372619 213 4 5 55149076 696337 213 4 5 10000000 188478 213 4 5 6800000 168744 213 4 5 42598665 524001 213 4 5 3693703 151310 213 4 5 8000000 175924 213 4 5 23833362 298915 213 4 5 12715195 206771 213 4 5 21500000 277081 213 4 5 5330000 160292 213 4 5 18565708 251575 213 6 5 25000000 326028 213 4 5 2000000 142460 218 2 5 5000000 103315 213 4 5 3821184 151994 218 2 5 1280326 85749 213 4 5 3209591 148735 218 2 5 5000000 103315

PAGE 128

128 Table B-4. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 4 5 9921126 187969 218 4 5 9000000 187104 213 4 5 10072973 188951 218 4 5 2000000 143627 213 4 5 1500000 139931 218 5 5 5568318 210760 213 4 5 1200000 138431 103 6 5 12724724 52866 213 4 5 22558049 286808 12 2 5 6197850 138984 213 6 5 1715000 207652 12 4 5 4705619 89564 218 2 5 8866515 125025 Table B-5. Prediction data and results for concrete WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 1 1 1212202 -156 12 4 5 4485447 420 12 1 1 1270370 -152 12 4 5 1515518 278 12 1 1 7004322 191 12 4 5 3642920 378 12 1 1 12300966 603 12 4 5 9699792 704 12 1 1 16348584 993 12 4 5 6393357 519 12 1 1 258704303 25782 12 4 5 3016096 348 12 2 1 5126542 229 12 4 5 2662205 331 12 2 1 7870386 434 12 4 5 1398242 273 12 2 1 19398408 1619 12 4 5 1784897 290 12 4 1 1138747 216 12 4 5 6737032 537 12 4 1 1349709 230 12 4 5 3526796 373 12 4 1 1858138 265 12 4 5 5583750 476 12 4 1 3506058 385 12 4 5 2144332 307 12 4 1 5731556 560 12 4 5 1035883 256 12 4 1 6871638 657 12 4 5 3854711 389 12 4 1 10145137 961 12 4 5 6803161 540 12 4 1 17501485 1803 12 4 5 4160640 404 12 6 1 1001235 302 12 4 5 1991058 300 12 6 1 2943374 443 12 4 5 3962485 394 12 6 1 3112200 456 12 4 5 2460320 322 12 6 1 5456007 642 12 4 5 2177070 309 12 6 1 5949853 684 12 4 5 1559720 280 12 6 1 6251345 709 12 4 5 7323437 569 12 6 1 7646571 832 12 4 5 2533434 325 12 6 1 9218116 979 12 4 5 2917849 344 12 6 1 12343896 1297 12 4 5 1127000 260 12 8 1 1745365 221 12 5 5 3916620 341 12 2 3 1182762 737 12 5 5 4794072 384 12 2 3 1320990 749 12 5 5 1837340 242 12 2 3 1882863 794 12 5 5 7242670 513 12 4 3 1121933 93 12 5 5 1343071 220

PAGE 129

129 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 4 3 1499066 114 12 5 5 15367120 1019 12 4 3 1836873 132 12 5 5 2041000 252 12 4 3 2175141 151 12 6 5 12073870 738 12 4 3 2306544 158 12 6 5 3926664 283 12 4 3 3352874 219 12 6 5 3714290 273 12 4 3 3755425 242 12 6 5 3793119 276 12 4 3 4067811 261 12 6 5 7647389 476 12 4 3 4459035 285 12 6 5 1874348 186 12 4 3 4843570 309 12 6 5 2537133 217 12 4 3 4847885 309 12 6 5 3698696 272 12 4 3 5095252 324 12 6 5 2030390 193 12 4 3 5117504 326 12 6 5 1512707 170 12 4 3 13882517 973 12 6 5 8567578 527 12 4 3 21814357 1753 12 6 5 5883715 381 12 5 3 15154163 735 12 6 5 5218444 347 12 2 5 1044000 359 12 6 5 4879735 330 12 2 5 1160595 364 12 6 5 8263001 510 12 2 5 1165841 364 12 6 5 1732500 180 12 2 5 1178085 365 12 6 5 2822474 230 12 2 5 1216766 367 12 6 5 18684310 1211 12 2 5 1366215 374 12 6 5 17573144 1124 12 2 5 1372181 374 12 6 5 8824585 542 12 2 5 1383606 375 12 8 5 5910101 234 12 2 5 1699586 389 103 2 1 6612212 2351 12 2 5 1705962 390 103 4 1 1220082 848 12 2 5 1760281 392 103 4 1 285652252 25903 12 2 5 1775052 393 103 8 1 14624049 441 12 2 5 1971052 402 103 6 1 192815313 25793 12 2 5 2016282 404 103 6 1 4451377 298 12 2 5 2140886 410 103 6 1 4440000 297 12 2 5 2189904 413 103 6 1 76572470 18948 12 2 5 2274299 417 103 8 3 15000000 2499 12 2 5 2305694 418 103 8 3 23505773 3777 12 2 5 2363287 421 103 8 3 2645500 1145 12 2 5 2411198 423 103 6 3 76026825 14295 12 2 5 3126788 458 103 6 3 4098500 700 12 2 5 3199216 462 103 6 5 5165562 1084 12 2 5 3437350 473 213 2 1 6800000 216 12 2 5 3452769 474 213 2 1 6507000 197 12 2 5 3708246 487 213 4 1 1631931 -293 12 2 5 4084554 506 213 4 1 3818700 -194

PAGE 130

130 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 2 5 4558109 530 213 4 1 3848040 -193 12 2 5 4674890 536 213 4 1 207267948 25245 12 2 5 4710678 538 213 4 1 2800000 -241 12 2 5 4724476 538 213 4 1 1500000 -299 12 2 5 4831632 544 213 6 1 1654208 -596 12 2 5 5022711 554 213 8 1 4900000 -770 12 2 5 5118497 559 213 2 3 4000000 261 12 2 5 5319569 569 213 2 3 4000000 261 12 2 5 5475387 577 213 2 3 1027640 92 12 2 5 5612662 585 213 2 3 27782525 2492 12 2 5 5818455 596 213 2 3 20130222 1576 12 2 5 5826748 596 213 6 3 3030000 210 12 2 5 5854406 598 213 2 5 6000000 1448 12 2 5 6091274 610 213 2 5 12338463 1960 12 2 5 6525644 634 213 2 5 8802104 1664 12 2 5 7103062 665 213 2 5 6273557 1468 12 2 5 7281472 675 213 2 5 6000000 1448 12 2 5 8210198 727 213 2 5 5157479 1386 12 2 5 10561682 866 213 2 5 8171987 1614 12 2 5 11609772 931 213 2 5 10763667 1825 12 2 5 12660016 998 213 2 5 13126302 2030 12 2 5 13625430 1062 213 2 5 1371870 1124 12 2 5 15701745 1205 213 2 5 1247154 1115 12 3 5 1359060 320 213 2 5 5000000 1375 12 3 5 1371589 320 213 3 5 1100000 687 12 3 5 1463514 325 213 4 5 15584468 1126 12 3 5 1465874 325 213 4 5 6795601 614 12 3 5 3099236 402 213 4 5 16480483 1185 12 3 5 4461558 469 213 4 5 1525150 360 12 3 5 4946931 494 213 4 5 2309806 396 12 3 5 4960900 494 213 4 5 2000000 382 12 3 5 6016860 549 213 4 5 55149076 5391 12 3 5 6057037 551 213 4 5 6800000 614 12 3 5 8207008 669 213 4 5 3693703 460 12 3 5 9858625 764 213 4 5 23833362 1725 12 4 5 1079000 258 213 4 5 21500000 1542 12 4 5 1245992 266 213 4 5 18565708 1328 12 4 5 1259861 266 213 4 5 2000000 382 12 4 5 1268622 267 213 4 5 3821184 466 12 4 5 1276000 267 213 4 5 3209591 438 12 4 5 1298638 268 213 4 5 9921126 783

PAGE 131

131 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 4 5 1300826 268 213 4 5 10072973 791 12 4 5 1351702 271 213 4 5 1500000 359 12 4 5 1390469 272 213 4 5 1200000 346 12 4 5 1396946 273 213 4 5 22558049 1624 12 4 5 1466441 276 213 6 5 1715000 -177 12 4 5 1537000 279 218 4 1 95820463 23322 12 4 5 1627371 283 218 4 1 11000000 1303 12 4 5 1735858 288 218 4 1 5000000 595 12 4 5 1786269 291 218 4 1 4396089 535 12 4 5 1826410 292 218 4 1 12248906 1480 12 4 5 1845560 293 218 4 1 1101920 241 12 4 5 1900308 296 218 4 1 5517567 648 12 4 5 1965000 299 218 4 1 5362223 632 12 4 5 2046056 303 218 2 3 2300000 322 12 4 5 2153278 308 218 2 3 10224603 1033 12 4 5 2216121 310 218 4 3 12315464 618 12 4 5 2268020 313 218 2 5 8866515 5650 12 4 5 2385575 318 218 4 5 2653300 2578 12 4 5 2466182 322 218 4 5 9000000 3388 12 4 5 2503795 324 218 4 5 16639131 4531 12 4 5 2579040 327 12 6 1 36945764 5146 12 4 5 2590500 328 12 4 1 21917476 2426 12 4 5 2708251 334 12 4 1 17868268 1851 12 4 5 2779589 337 12 6 1 25091836 2988 12 4 5 2784589 337 12 4 1 9910141 937 12 4 5 3549995 374 12 4 1 12466090 1201 12 4 5 3801364 386 12 6 1 22023453 2521 12 4 5 3807021 387 12 4 3 4769972 304 12 4 5 3829577 388 12 4 3 10268701 681 12 4 5 4034516 398 12 4 3 6726481 430 12 4 5 4464075 419 12 4 3 12434258 851 12 4 5 5094596 451 12 2 3 5624233 1124 12 4 5 5642793 479 12 4 3 2503280 170 12 4 5 6464367 522 12 4 3 11268263 758 12 4 5 6878027 544 12 2 3 5987393 1159 12 4 5 7311251 568 12 4 3 1546637 116 12 4 5 7613931 585 12 4 3 1039549 89 12 4 5 8458572 632 12 2 5 5670187 588 12 4 5 11074754 786 12 3 5 4886020 490 12 4 5 11639122 821 12 4 5 3151826 355 12 4 5 12969867 906 12 2 5 4056243 504

PAGE 132

132 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 4 5 13301908 928 12 2 5 12574133 993 12 5 5 1353112 220 12 6 5 2678900 223 12 5 5 1369000 221 12 4 5 3963163 394 12 5 5 1838048 242 12 4 5 2573947 327 12 5 5 2088477 254 12 3 5 3582103 425 12 5 5 2438951 270 12 4 5 3506780 372 12 5 5 2551034 275 12 3 5 11036450 835 12 5 5 2675634 281 12 2 5 12026411 958 12 5 5 2994084 296 12 4 5 1450000 275 12 5 5 3559814 323 12 5 5 4654347 377 12 5 5 4712443 380 12 4 5 3242621 359 12 5 5 5236308 407 12 4 5 15838532 1102 12 5 5 6355652 465 12 3 5 18265045 1331 12 5 5 10244042 685 12 2 5 3014988 453 12 6 5 1059299 150 12 2 5 4264549 515 12 6 5 1193000 156 12 2 5 2653938 435 12 6 5 1285293 160 12 4 5 18120206 1271 12 6 5 1656000 176 12 4 5 7409819 573 12 6 5 1681500 178 12 2 5 11891327 949 12 6 5 1773294 182 12 2 5 1095064 361 12 6 5 1801397 183 12 4 5 17255530 1206 12 6 5 1983602 191 12 2 5 5678555 588 12 6 5 2244528 203 12 4 5 3530091 373 12 6 5 2295081 206 12 4 5 8642835 642 12 6 5 2613474 220 12 2 5 3749467 489 12 6 5 2896351 234 12 5 5 15014658 994 12 6 5 3434068 259 12 3 5 1276958 316 12 6 5 3788041 276 12 4 5 5974488 496 12 6 5 4123070 293 12 3 5 1655339 334 12 6 5 4622137 317 12 2 5 4427366 523 12 6 5 4651284 319 12 5 5 1304168 218 12 6 5 4939781 333 12 2 5 9453764 800 12 6 5 5003451 336 12 6 5 1465903 168 12 6 5 5141718 343 12 5 5 6182568 456 12 6 5 5380545 355 12 2 5 1880918 398 12 6 5 5788574 376 12 6 5 2582481 219 12 6 5 5955972 385 12 6 5 5458885 359 12 6 5 6038066 389 12 4 5 3043333 350 12 6 5 6117293 394 12 6 5 7118087 447 12 6 5 7204215 451 12 6 5 6036580 389 12 6 5 7231792 453 12 5 5 2363588 266

PAGE 133

133 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 6 5 7980439 494 12 1 5 19976496 1619 12 6 5 9035815 554 12 6 5 9062784 555 12 8 5 7638989 323 12 6 5 3299408 253 12 2 5 1632193 386 12 2 5 2425750 424 103 2 1 2667200 1853 12 4 5 2785860 337 103 4 1 1607592 882 12 6 5 3093632 243 103 4 1 1654618 886 12 6 5 3970908 285 103 4 1 3408770 1049 12 6 5 2870358 232 103 4 1 3705105 1078 12 8 5 3292112 109 103 6 1 6169033 419 12 2 5 4030182 503 103 8 1 3644675 -207 12 2 5 8048887 718 103 8 1 3981900 -192 12 4 5 5936356 494 103 6 1 15061018 1227 12 2 5 12142356 965 103 3 1 1677174 1339 12 6 5 8732398 537 103 3 1 2301657 1405 12 4 5 12925322 903 103 3 1 2648762 1442 12 6 5 5123326 342 103 3 1 2654793 1442 12 5 5 1188531 213 103 3 1 5969089 1825 12 6 5 11668806 713 103 4 1 1380000 862 12 2 5 3514007 477 103 4 1 5312360 1241 12 4 5 6684147 534 103 4 1 6400000 1358 12 6 5 4801228 326 103 4 1 12459068 2120 12 6 5 3916572 282 103 4 1 49778910 12464 12 6 5 2557786 218 103 6 1 2879312 195 12 4 5 2109345 305 103 6 1 2027599 143 12 6 5 4457620 309 103 6 1 26137824 2810 12 4 5 4281816 410 103 2 1 13567530 3432 12 4 5 3506513 372 103 4 1 1191408 846 12 2 5 1538183 382 103 4 1 2035680 921 12 6 5 2935859 235 103 6 3 11261935 1253 12 6 5 5035671 338 103 8 3 3850821 1254 12 3 5 1425784 323 103 8 3 10544390 1947 12 4 5 1387705 272 103 8 3 14014605 2370 12 6 5 4487702 310 103 8 3 30532009 5067 12 4 5 4520277 422 103 8 3 221412954 25767 12 3 5 5697556 532 103 1 3 4987446 1799 12 6 5 4687110 320 103 6 3 34108627 4165 12 5 5 4846549 387 103 4 3 1396528 470 12 2 5 7114726 666 103 4 5 4702760 1923 12 6 5 10427553 636 103 6 5 5668351 1119 12 4 5 5289555 461 213 1 1 1176984 -92 12 2 5 10626696 870

PAGE 134

134 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 213 1 1 3966168 62 12 6 5 11870935 725 213 1 1 9660518 440 12 2 5 2776038 441 213 1 1 15152448 898 12 2 5 2529045 429 213 1 1 20157318 1409 12 2 5 1985404 403 213 1 1 64798281 10750 12 6 5 20018605 1320 213 1 1 68804811 11859 12 4 5 10006799 722 213 2 1 4378142 66 103 4 1 9232000 1689 213 2 1 5990520 164 103 1 1 48476309 13590 213 2 1 8350269 320 103 1 1 41327005 11107 213 2 1 172605686 24864 103 4 1 25519263 4569 213 3 1 1010681 -196 103 8 3 53396930 10525 213 3 1 4461728 -20 103 6 3 79299607 15158 213 4 1 1705100 -290 213 4 1 36353090 2875 213 4 1 2000000 -277 213 2 1 6800000 216 213 4 1 2225000 -267 213 6 1 2530000 -570 213 4 1 2253922 -266 213 6 1 1650000 -596 213 4 1 2800000 -241 213 8 1 5450000 -758 213 4 1 3235799 -221 213 6 1 249217770 25442 213 4 1 3250000 -221 213 6 1 1197858 -610 213 4 1 3590000 -205 213 2 3 4000000 261 213 4 1 4672090 -153 213 2 3 8929852 587 213 4 1 5702552 -101 213 4 3 22071672 1615 213 4 1 6161370 -77 213 2 3 3043078 204 213 4 1 14824949 468 213 2 3 7012906 453 213 4 1 15640539 529 213 4 5 6634965 606 213 4 1 107560004 18778 213 4 5 31292523 2382 213 6 1 1172671 -610 213 4 5 12397812 927 213 6 1 1219682 -609 213 4 5 2000000 382 213 6 1 1394389 -604 213 2 5 6000000 1448 213 6 1 1890036 -589 213 4 5 1460000 357 213 6 1 2882629 -559 213 4 5 10276666 803 213 6 1 3350924 -544 213 2 5 2700000 1213 213 6 1 4679205 -500 213 2 5 3470000 1266 213 6 1 7189047 -410 213 2 5 2798600 1219 213 6 1 22329791 383 213 2 5 5421129 1405 213 8 1 1100000 -845 213 4 5 17276584 1239 213 8 1 1210300 -843 213 4 5 3342114 444 213 8 1 1295000 -841 213 4 5 1591340 363 213 8 1 2822338 -812 213 4 5 1264550 349 213 8 1 5060804 -766 213 2 5 2807739 1220 213 2 3 1938884 142 213 2 5 1371870 1124

PAGE 135

135 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 213 2 3 2000000 145 213 2 5 2867000 1224 213 2 3 3186742 213 213 2 5 29482249 3801 213 2 3 3805883 249 213 2 5 5879000 1439 213 2 3 4000000 261 213 6 5 32989965 1371 213 2 3 5197159 335 213 2 5 1801000 1152 213 2 3 6075993 391 218 4 1 1651718 287 213 2 3 7423000 481 218 6 1 4100000 294 213 2 3 8861138 582 218 4 1 30708210 5590 213 2 3 9118903 600 218 4 1 2000000 316 213 2 3 12728717 881 218 4 1 39962233 8721 213 2 3 15605648 1131 218 4 1 3418477 442 213 2 3 18981143 1457 218 4 1 1254161 254 213 2 3 19593720 1520 218 4 1 1776062 297 213 4 3 1301978 28 218 4 1 1720000 293 213 4 3 2217939 76 218 2 3 6349622 653 213 4 3 3917194 170 218 4 3 24186687 1884 213 4 3 4011524 176 218 2 3 49251652 9364 213 4 3 8343814 444 218 2 3 1250000 245 213 4 3 24640000 1897 218 4 5 30522524 7080 213 6 3 1365500 116 218 4 5 9000000 3388 213 6 3 2944473 205 218 4 5 52019051 11920 213 6 3 3447554 235 218 4 5 10000000 3527 213 6 3 4804195 317 218 4 5 1136429 2403 213 6 3 23527259 1921 218 4 5 5238917 2893 213 6 3 26958663 2321 12 3 3 3894753 559 213 1 5 2022003 1675 12 4 3 13035225 901 213 2 5 1007367 1100 12 4 3 3611991 234 213 2 5 1067184 1103 12 4 3 14152837 996 213 2 5 1197962 1112 12 4 3 17235573 1278 213 2 5 1286567 1118 12 6 3 20581900 894 213 2 5 1345065 1122 12 2 3 13808116 2017 213 2 5 1382505 1124 12 2 3 10708107 1649 213 2 5 1423370 1127 12 4 3 10062782 666 213 2 5 1467914 1130 12 2 3 2762067 868 213 2 5 1581871 1137 12 2 3 13451516 1973 213 2 5 1856688 1156 12 4 3 2727098 182 213 2 5 1870704 1157 12 4 3 10773729 720 213 2 5 1930200 1161 12 4 3 18241179 1376 213 2 5 2514154 1200 12 4 3 12255163 837 213 2 5 2846411 1223 12 4 3 15362918 1103 213 2 5 2964262 1231 12 3 3 3168899 507

PAGE 136

136 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 213 2 5 2989999 1232 12 4 5 10045743 724 213 2 5 3749812 1285 12 2 5 4658578 535 213 2 5 3948064 1299 12 2 5 12160562 966 213 2 5 4000815 1303 12 3 5 6976833 601 213 2 5 4382030 1330 12 2 5 5049679 555 213 2 5 4428194 1333 12 8 5 3438059 116 213 2 5 5093984 1381 12 4 5 9948188 718 213 2 5 5514468 1412 12 4 5 4632350 427 213 2 5 5771377 1431 12 2 5 9541401 805 213 2 5 6835002 1511 12 2 5 8635874 752 213 2 5 8642282 1652 12 2 5 1326196 372 213 2 5 9018000 1682 12 2 5 1541191 382 213 2 5 9110435 1689 12 2 5 6238811 618 213 2 5 9771581 1743 12 2 5 2170201 412 213 2 5 15034729 2204 12 2 5 9875095 825 213 2 5 17170414 2408 12 2 5 2443824 425 213 2 5 20314536 2729 12 2 5 3343393 469 213 2 5 21444057 2849 12 3 5 3699005 431 213 2 5 22927312 3013 12 2 5 2565163 431 213 2 5 28107048 3626 12 2 5 3221396 463 213 2 5 29549112 3809 12 2 5 7553487 690 213 2 5 34124268 4426 12 2 5 8090665 721 213 3 5 18154370 1846 12 4 5 12192709 856 213 4 5 1008236 337 12 2 5 8566395 748 213 4 5 1031645 338 12 6 5 4882596 330 213 4 5 1041888 339 12 4 5 8580317 639 213 4 5 1050001 339 12 4 5 7224194 563 213 4 5 1107309 342 12 2 5 9590353 808 213 4 5 1150000 344 12 4 5 2190546 309 213 4 5 1326789 351 12 2 5 1650825 387 213 4 5 1353628 353 12 4 5 3407383 367 213 4 5 1411139 355 12 4 5 3858722 389 213 4 5 1458760 357 12 2 5 3798992 491 213 4 5 1464803 358 12 5 5 10881497 723 213 4 5 1585790 363 12 2 5 7778322 703 213 4 5 1637464 365 12 6 5 2550001 217 213 4 5 1729804 369 12 4 5 7110781 557 213 4 5 1935954 379 12 5 5 4214954 355 213 4 5 1993452 381 12 3 5 17557184 1278 213 4 5 1998870 382 12 4 5 9301238 680 213 4 5 2047680 384 12 3 5 7991840 656

PAGE 137

137 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 213 4 5 2069499 385 12 2 5 3359771 470 213 4 5 2095207 386 12 5 5 4948107 392 213 4 5 2268244 394 12 5 5 7680992 537 213 4 5 2272080 394 12 4 5 10308880 740 213 4 5 2700000 414 12 6 5 1472917 168 213 4 5 3462362 449 12 4 5 3469388 370 213 4 5 3934828 472 12 4 5 1463715 276 213 4 5 4418140 495 12 4 5 5631627 479 213 4 5 5569697 552 12 2 5 2238403 415 213 4 5 6339884 591 12 3 5 2212977 360 213 4 5 6684233 608 12 4 5 3807129 387 213 4 5 7266605 639 12 4 5 1047006 257 213 4 5 7403865 646 12 2 5 2351420 420 213 4 5 9239147 745 12 4 5 1805057 291 213 4 5 10049427 790 12 2 5 3525944 478 213 4 5 10560260 819 12 6 5 2418583 211 213 4 5 12350000 924 12 6 5 2007775 192 213 4 5 14034501 1027 12 5 5 2049415 252 213 4 5 14512760 1057 12 2 5 1567142 383 213 4 5 14663196 1067 12 4 5 1914598 296 213 4 5 14847933 1078 12 4 5 8562635 638 213 4 5 15520418 1122 12 3 5 10310008 791 213 4 5 19820798 1418 12 3 5 1461621 325 213 4 5 22572609 1625 12 5 5 32146974 2598 213 4 5 27901688 2068 12 2 5 3445881 474 213 4 5 34099959 2661 12 2 5 8359161 736 213 4 5 36831121 2950 12 4 5 2436333 321 213 4 5 154568188 23766 12 4 5 11895284 837 213 6 5 3243778 -129 12 4 5 8488886 634 218 4 1 1247256 253 12 2 5 3577976 480 218 4 1 1324179 260 12 2 5 1143486 363 218 4 1 1492789 274 12 2 5 1415341 376 218 4 1 1770818 297 12 2 5 9309746 791 218 4 1 2114386 326 12 3 5 5519522 523 218 4 1 2821967 388 12 2 5 1041410 359 218 4 1 3274450 429 12 2 5 8474952 743 218 4 1 3919380 489 12 3 5 1270504 316 218 4 1 35395422 7103 12 1 5 12709323 1083 218 4 1 113915158 24537 12 8 5 1438210 27 218 2 3 14659298 1552 12 2 5 3205411 462 218 2 3 23023341 2818 12 2 5 2301684 418

PAGE 138

138 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 218 4 3 1362051 -121 12 2 5 1201000 366 218 4 3 1687559 -104 103 6 1 19643263 1790 218 4 3 17869800 1141 103 4 1 3300000 1039 218 2 5 3589000 4736 103 4 3 2005000 504 218 2 5 5720000 5095 103 8 3 5700000 1430 218 2 5 7483865 5403 103 8 3 128773840 24006 218 2 5 7933333 5483 103 4 3 4000000 621 218 2 5 11091826 6060 213 4 1 13100142 344 218 2 5 63137848 17107 213 4 1 4659746 -153 218 4 5 1013299 2389 213 2 1 6800000 216 218 4 5 2178065 2522 213 6 1 1655000 -596 218 4 5 2500000 2560 213 6 1 1900000 -589 218 4 5 3223705 2646 213 8 1 1658000 -835 218 4 5 3615000 2693 213 8 1 1610000 -835 218 4 5 3825000 2718 213 8 1 5725000 -752 218 4 5 4815000 2840 213 3 1 107722494 19629 218 4 5 5010000 2864 213 4 1 25671605 1463 218 4 5 5352000 2907 213 2 3 4000000 261 218 4 5 5752350 2958 213 2 3 1205000 101 218 4 5 5857787 2971 213 2 3 8227929 537 218 4 5 8271598 3289 213 4 3 1593100 43 218 4 5 8683500 3345 213 4 3 16207313 1050 218 4 5 22880098 5604 213 2 3 1500000 117 218 4 5 36115970 8260 213 2 3 1500000 117 218 5 5 4500000 2068 213 2 3 54238235 7388 12 1 1 11700000 551 213 2 3 77981699 13321 12 4 1 9971149 943 213 2 3 6550000 422 12 4 1 2121593 284 213 2 3 2208874 157 12 4 1 4155416 434 213 2 3 1366415 110 12 4 1 7876752 746 213 2 3 5143562 331 12 6 1 5729692 665 213 6 3 7159000 470 12 6 1 6892822 765 213 4 3 5278613 250 12 6 1 16611070 1790 213 4 3 2760000 106 12 6 1 5323553 631 213 2 5 31792782 4105 12 8 1 6723957 572 213 4 5 15834468 1142 12 2 3 6851870 1243 213 2 5 17259040 2417 12 2 3 5799141 1141 213 4 5 3415392 447 12 2 3 3055488 893 213 4 5 6435287 596 12 4 3 4288545 275 213 2 5 9000000 1680 12 4 3 2969782 196 213 4 5 34886044 2742 12 4 3 10730063 716 213 2 5 54451658 7813

PAGE 139

139 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 4 3 15776518 1141 213 4 5 17176264 1232 12 4 3 10158057 673 213 4 5 81580986 10526 12 4 3 5302399 337 213 4 5 4666666 507 12 4 3 15106943 1080 213 4 5 3443000 448 12 4 3 1933328 137 213 4 5 10000000 787 12 4 3 3121365 205 213 2 5 30384990 3918 12 4 3 6899607 441 213 2 5 21307263 2835 12 4 3 7561050 486 213 2 5 3080000 1239 12 4 3 18425428 1394 213 4 5 8118107 684 12 5 3 11614275 484 213 2 5 49201128 6842 12 1 5 1081000 425 213 4 5 2561964 407 12 2 5 8922328 768 213 4 5 35536766 2811 12 2 5 6801311 649 213 2 5 1371870 1124 12 2 5 5388453 573 213 4 5 3300000 442 12 2 5 3633106 483 213 4 5 9269484 746 12 2 5 4070526 505 213 4 5 4000000 475 12 2 5 5014437 553 213 2 5 7036116 1526 12 2 5 5669009 588 218 4 1 116245584 24637 12 2 5 3429408 473 218 4 1 8029224 925 12 2 5 2206832 413 218 4 1 10195255 1195 12 2 5 2613927 433 218 4 1 1656480 287 12 2 5 1992999 403 218 2 3 5973235 620 12 2 5 8080552 720 218 4 3 58011014 8903 12 2 5 3408976 472 218 2 5 5000000 4972 12 2 5 2427113 424 218 4 5 9000000 3388 12 2 5 6100975 611 218 2 5 1237030 4355 12 2 5 6285013 621 218 4 5 9599233 3471 12 2 5 7775748 703 218 2 5 3000000 4639 12 2 5 6106969 611 12 2 5 6197850 616 12 2 5 6576341 636 12 4 5 4705619 431 12 2 5 2012039 404 103 2 1 100606191 24630 12 2 5 18154133 1386 103 1 1 1924480 2066 12 2 5 2325152 419 103 4 3 2448336 530 12 2 5 7306531 677 103 8 3 30997927 5160 12 2 5 2689540 437 103 6 5 12724724 1683 12 2 5 7077000 664 213 2 1 6800000 216 12 2 5 4658421 535 213 3 1 4000000 -45 12 2 5 1852829 397 213 4 1 29847551 1959 12 2 5 19031224 1454 213 6 1 10100000 -293 12 2 5 7186507 670 213 6 1 200200313 24765 12 2 5 16362915 1253 213 6 1 1550000 -599

PAGE 140

140 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 2 5 5848411 597 213 6 1 10900000 -258 12 2 5 2874078 446 213 6 1 4000000 -523 12 2 5 1364005 374 213 8 1 94937616 9783 12 2 5 2733693 439 213 8 1 14575435 -513 12 2 5 5303453 568 213 8 1 6400000 -736 12 2 5 3031048 453 213 8 1 31429350 252 12 2 5 7028045 661 213 2 3 4000000 261 12 2 5 2121546 409 213 2 3 42589140 4905 12 2 5 4811940 543 213 2 3 3927175 257 12 2 5 2029521 405 213 2 3 56687098 7964 12 2 5 6813118 649 213 2 3 4895000 316 12 2 5 7024236 661 213 2 3 4647500 301 12 2 5 1217713 367 213 2 3 2600000 179 12 2 5 1044000 359 213 2 3 6050000 389 12 2 5 1108000 362 213 2 3 40045877 4429 12 2 5 1447000 378 213 4 3 5000000 233 12 2 5 1101000 361 213 6 3 2000000 151 12 3 5 2481464 372 213 6 3 15000500 1076 12 3 5 7271988 617 213 2 5 20000000 2695 12 3 5 15207399 1108 213 2 5 8000000 1601 12 3 5 23734302 1786 213 2 5 1900000 1159 12 3 5 4381233 465 213 2 5 27353088 3533 12 3 5 2366012 367 213 2 5 1371870 1124 12 3 5 4803267 486 213 2 5 37620116 4934 12 3 5 1121820 309 213 3 5 12004509 1368 12 3 5 2277784 363 213 3 5 8000000 1095 12 3 5 2304994 364 213 4 5 1994727 381 12 3 5 3785185 435 213 4 5 3825238 467 12 3 5 1742625 338 213 4 5 2000000 382 12 3 5 4585727 475 213 4 5 3518000 452 12 4 5 4655524 429 213 4 5 30800000 2335 12 4 5 3034535 349 213 4 5 10000000 787 12 4 5 3513724 372 213 4 5 42598665 3622 12 4 5 2318643 315 213 4 5 8000000 677 12 4 5 2703891 333 213 4 5 12715195 946 12 4 5 1631462 283 213 4 5 5330000 540 12 4 5 4488534 420 213 6 5 25000000 842 12 4 5 6141414 505 218 4 1 3917121 489 12 4 5 1263699 267 218 4 1 4956852 591 12 4 5 4182438 405 218 4 1 2231848 336 12 4 5 2125444 306 218 4 1 1500000 274

PAGE 141

141 Table B-5. Continued WMC Lanes TS Cost Pred. qty WMCLanesTS Cost Pred. qty 12 4 5 1640535 284 218 2 5 5000000 4972 12 4 5 5159874 454 218 2 5 1280326 4362 12 4 5 1419851 274 218 2 5 5000000 4972 12 4 5 3596110 376 218 4 5 9000000 3388 218 5 5 5568318 2182 218 4 5 2000000 2502 Table B-6. Prediction data and results for structural steel WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 103 2 2667200 985128 103 8 2645500 5730 103 4 1607592 756675 103 6 76026825 3711600 103 4 1654618 758056 103 6 4098500 167481 103 4 3408770 809377 103 6 5165562 186735 103 4 3705105 818003 213 2 6800000 27598 103 6 6169033 205697 213 2 6507000 21826 103 8 3644675 11498 213 4 1631931 29898 103 8 3981900 13505 213 4 3818700 62642 103 6 15061018 415189 213 4 3848040 63103 103 3 1677174 967693 213 4 2072679484182364 103 3 2301657 979954 213 4 2800000 46998 103 3 2648762 986671 213 4 1500000 28023 103 3 2654793 986787 213 6 1654208 295968 103 3 5969089 1047174 213 8 4900000 567720 103 4 1380000 749991 213 2 4000000 -22817 103 4 5312360 864477 213 2 4000000 -22817 103 4 6400000 895565 213 2 1027640 -65933 103 4 12459068 1061063 213 2 27782525 848100 103 4 49778910 1796043 213 2 20130222 447419 103 6 2879312 146581 213 6 3030000 334858 103 6 2027599 132648 213 2 6000000 12124 103 6 26137824 805659 213 2 12338463 161917 103 2 13567530 1023103 213 2 8802104 70418 103 4 1191408 744450 213 2 6273557 17315 103 4 2035680 769235 213 2 6000000 12124 103 6 11261935 315906 213 2 5157479 -3227 103 8 3850821 12721 213 2 8171987 56286 103 8 10544390 59912 213 2 10763667 118519 103 8 14014605 91647 213 2 13126302 185319 103 8 30532009 371337 213 2 1371870 -61428 103 8 221412954 4609936 213 2 1247154 -63074 103 1 4987446 791560 213 2 5000000 -5992 103 6 34108627 1192081 213 3 1100000 -50285

PAGE 142

142 Table B-6. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 103 4 1396528 750477 213 4 15584468 293986 103 4 4702760 846918 213 4 6795601 112339 103 6 5668351 196130 213 4 16480483 315352 213 1 1176984 -26515 213 4 1525150 28379 213 1 3966168 29297 213 4 2309806 39714 213 1 9660518 186522 213 4 2000000 35191 213 1 15152448 404214 213 4 55149076 1545849 213 1 20157318 663469 213 4 6800000 112417 213 1 64798281 3199670 213 4 3693703 60685 213 1 68804811 3309313 213 4 23833362 508557 213 2 4378142 -16602 213 4 21500000 443979 213 2 5990520 11946 213 4 18565708 366991 213 2 8350269 60222 213 4 2000000 35191 213 2 172605686 4044145 213 4 3821184 62681 213 3 1010681 -51297 213 4 3209591 53206 213 3 4461728 -7849 213 4 9921126 170990 213 4 1705100 30944 213 4 10072973 174009 213 4 2000000 35191 213 4 1500000 28023 213 4 2225000 38469 213 4 1200000 23799 213 4 2253922 38893 213 4 22558049 472909 213 4 2800000 46998 213 6 1715000 297643 213 4 3235799 53607 103 4 9232000 974736 213 4 3250000 53824 103 1 48476309 444552 213 4 3590000 59070 103 1 41327005 486556 213 4 4672090 76279 103 4 25519263 1358965 213 4 5702552 93396 103 8 53396930 1601695 213 4 6161370 101249 103 6 79299607 3840693 213 4 14824949 276272 213 4 36353090 895440 213 4 15640539 295308 213 2 6800000 27598 213 4 107560004 3302452 213 6 2530000 320491 213 6 1172671 282837 213 6 1650000 295852 213 6 1219682 284108 213 8 5450000 587019 213 6 1394389 288852 213 6 2492177704133406 213 6 1890036 302490 213 6 1197858 283518 213 6 2882629 330596 213 2 4000000 -22817 213 6 3350924 344218 213 2 8929852 73360 213 6 4679205 384087 213 4 22071672 459536 213 6 7189047 464101 213 2 3043078 -37778 213 6 22329791 1009485 213 2 7012906 31868 213 8 1100000 444018 213 4 6634965 109504 213 8 1210300 447366 213 4 31292523 732063

PAGE 143

143 Table B-6. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 213 8 1295000 449948 213 4 12397812 222168 213 8 2822338 497970 213 4 2000000 35191 213 8 5060804 573327 213 2 6000000 12124 213 2 1938884 -53742 213 4 1460000 27456 213 2 2000000 -52893 213 4 10276666 178083 213 2 3186742 -35600 213 2 2700000 -42883 213 2 3805883 -25940 213 2 3470000 -31237 213 2 4000000 -22817 213 2 2798600 -41429 213 2 5197159 -2525 213 2 5421129 1475 213 2 6075993 13555 213 4 17276584 334755 213 2 7423000 40279 213 4 3342114 55238 213 2 8861138 71774 213 4 1591340 29320 213 2 9118903 77760 213 4 1264550 24703 213 2 12728717 173365 213 2 2807739 -41294 213 2 15605648 266722 213 2 1371870 -61428 213 2 18981143 397485 213 2 2867000 -40414 213 2 19593720 423760 213 2 29482249 951435 213 4 1301978 25228 213 2 5879000 9861 213 4 2217939 38366 213 6 32989965 1330529 213 4 3917194 64191 213 2 1801000 -55642 213 4 4011524 65681 103 6 19643263 557243 213 4 8343814 140560 103 4 3300000 806207 213 4 24640000 531527 103 4 2005000 768336 213 6 1365500 288065 103 8 5700000 24242 213 6 2944473 332381 103 8 1287738404615271 213 6 3447554 347057 103 4 4000000 826571 213 6 4804195 387931 213 4 13100142 237426 213 6 23527259 1050929 213 4 4659746 76078 213 6 26958663 1163120 213 2 6800000 27598 213 1 2022003 -10887 213 6 1655000 295990 213 2 1007367 -66194 213 6 1900000 302767 213 2 1067184 -65422 213 8 1658000 461109 213 2 1197962 -63719 213 8 1610000 459624 213 2 1286567 -62556 213 8 5725000 596797 213 2 1345065 -61783 213 3 1077224943708948 213 2 1382505 -61287 213 4 25671605 561363 213 2 1423370 -60744 213 2 4000000 -22817 213 2 1467914 -60150 213 2 1205000 -63627 213 2 1581871 -58621 213 2 8227929 57516 213 2 1856688 -54877 213 4 1593100 29345 213 2 1870704 -54684 213 4 16207313 308785

PAGE 144

144 Table B-6. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 213 2 1930200 -53862 213 2 1500000 -59721 213 2 2514154 -45593 213 2 1500000 -59721 213 2 2846411 -40720 213 2 54238235 2540693 213 2 2964262 -38962 213 2 77981699 3414106 213 2 2989999 -38576 213 2 6550000 22666 213 2 3749812 -26833 213 2 2208874 -49962 213 2 3948064 -23657 213 2 1366415 -61501 213 2 4000815 -22804 213 2 5143562 -3473 213 2 4382030 -16537 213 6 7159000 463110 213 2 4428194 -15766 213 4 5278613 86267 213 2 5093984 -4346 213 4 2760000 46398 213 2 5514468 3162 213 2 31792782 1098349 213 2 5771377 7864 213 4 15834468 299896 213 2 6835002 28295 213 2 17259040 327829 213 2 8642282 66775 213 4 3415392 56366 213 2 9018000 75404 213 4 6435287 106005 213 2 9110435 77561 213 2 9000000 74986 213 2 9771581 93395 213 4 34886044 847288 213 2 15034729 246907 213 2 54451658 2552043 213 2 17170414 324411 213 4 17176264 332288 213 2 20314536 455686 213 4 81580986 2491773 213 2 21444057 507904 213 4 4666666 76190 213 2 22927312 580490 213 4 3443000 56793 213 2 28107048 867483 213 4 10000000 172556 213 2 29549112 955588 213 2 30384990 1008027 213 2 34124268 1252458 213 2 21307263 501439 213 3 18154370 279344 213 2 3080000 -37220 213 4 1008236 21129 213 4 8118107 136344 213 4 1031645 21454 213 2 49201128 2253515 213 4 1041888 21596 213 4 2561964 43441 213 4 1050001 21709 213 4 35536766 868574 213 4 1107309 22506 213 2 1371870 -61428 213 4 1150000 23101 213 4 3300000 54591 213 4 1326789 25577 213 4 9269484 158213 213 4 1353628 25955 213 4 4000000 65498 213 4 1411139 26766 213 2 7036116 32338 213 4 1458760 27439 103 2 1006061911466193 213 4 1464803 27524 103 1 1924480 800649 213 4 1585790 29241 103 4 2448336 781325 213 4 1637464 29977 103 8 30997927 383987 213 4 1729804 31297 103 6 12724724 352280

PAGE 145

145 Table B-6. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 213 4 1935954 34264 213 2 6800000 27598 213 4 1993452 35096 213 3 4000000 -14206 213 4 1998870 35175 213 4 29847551 686971 213 4 2047680 35883 213 6 10100000 563538 213 4 2069499 36200 213 6 2002003134029611 213 4 2095207 36574 213 6 1550000 293105 213 4 2268244 39103 213 6 10900000 591886 213 4 2272080 39160 213 6 4000000 363476 213 4 2700000 45499 213 8 94937616 2389301 213 4 3462362 57092 213 8 14575435 950537 213 4 3934828 64469 213 8 6400000 621155 213 4 4418140 72170 213 8 31429350 1674790 213 4 5569697 91149 213 2 4000000 -22817 213 4 6339884 104343 213 2 42589140 1828890 213 4 6684233 110372 213 2 3927175 -23994 213 4 7266605 120752 213 2 56687098 2666774 213 4 7403865 123232 213 2 4895000 -7818 213 4 9239147 157626 213 2 4647500 -12066 213 4 10049427 173540 213 2 2600000 -44346 213 4 10560260 183802 213 2 6050000 13065 213 4 12350000 221141 213 2 40045877 1656326 213 4 14034501 258229 213 4 5000000 81648 213 4 14512760 269098 213 6 2000000 305552 213 4 14663196 272547 213 6 15000500 741651 213 4 14847933 276802 213 2 20000000 441621 213 4 15520418 292478 213 2 8000000 52536 213 4 19820798 399322 213 2 1900000 -54280 213 4 22572609 473311 213 2 27353088 822716 213 4 27901688 627523 213 2 1371870 -61428 213 4 34099959 821737 213 2 37620116 1490101 213 4 36831121 911253 213 3 12004509 124318 213 4 154568188 4031281 213 3 8000000 47118 213 6 3243778 341081 213 4 1994727 35115 103 2 6612212 1008821 213 4 3825238 62745 103 4 1220082 745293 213 4 2000000 35191 103 4 285652252 4247545 213 4 3518000 57953 103 8 14624049 97870 213 4 30800000 716608 103 6 192815313 4543269 213 4 10000000 172556 103 6 4451377 173747 213 4 42598665 1106032 103 6 4440000 173544 213 4 8000000 134152 103 6 76572470 3734258 213 4 12715195 229023

PAGE 146

146 Table B-6. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 103 8 15000000 101817 213 4 5330000 87125 103 8 23505773 218396 213 6 25000000 1100373 Table B-7. Prediction data and results for reinforcing steel WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 12 1 1212202 -67662 12 4 1515518 -64688 12 1 1270370 -67094 12 4 3642920 -42859 12 1 7004322 -4660 12 4 9699792 29562 12 1 12300966 65910 12 4 6393357 -11959 12 1 16348584 129621 12 4 3016096 -49472 12 1 258704303 6621421 12 4 2662205 -53137 12 2 5126542 -26582 12 4 1398242 -65842 12 2 7870386 5971 12 4 1784897 -62019 12 2 19398408 183958 12 4 6737032 -7873 12 4 1138747 -68377 12 4 3526796 -44096 12 4 1349709 -66318 12 4 5583750 -21383 12 4 1858138 -61289 12 4 2144332 -58415 12 4 3506058 -44316 12 4 1035883 -69375 12 4 5731556 -19683 12 4 3854711 -40590 12 4 6871638 -6259 12 4 6803161 -7081 12 4 10145137 35546 12 4 4160640 -37280 12 4 17501485 149489 12 4 1991058 -59958 12 6 1001235 -69710 12 4 3962485 -39428 12 6 2943374 -50229 12 4 2460320 -55207 12 6 3112200 -48468 12 4 2177070 -58085 12 6 5456007 -22844 12 4 1559720 -64252 12 6 5949853 -17156 12 4 7323437 -782 12 6 6251345 -13632 12 4 2533434 -54459 12 6 7646571 3192 12 4 2917849 -50494 12 6 9218116 23198 12 4 1127000 -68491 12 6 12343896 66539 12 5 3916620 -39923 12 8 1745365 -62413 12 5 4794072 -30307 12 2 1182762 -67949 12 5 1837340 -61496 12 2 1320990 -66599 12 5 7242670 -1768 12 2 1882863 -61042 12 5 1343071 -66383 12 4 1121933 -68541 12 5 15367120 113326 12 4 1499066 -64851 12 5 2041000 -59456 12 4 1836873 -61501 12 6 12073870 62599 12 4 2175141 -58104 12 6 3926664 -39815 12 4 2306544 -56773 12 6 3714290 -42096 12 4 3352874 -45938 12 6 3793119 -41252

PAGE 147

147 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 12 4 3755425 -41656 12 6 7647389 3202 12 4 4067811 -38288 12 6 1874348 -61127 12 4 4459035 -34016 12 6 2537133 -54421 12 4 4843570 -29756 12 6 3698696 -42263 12 4 4847885 -29707 12 6 2030390 -59563 12 4 5095252 -26934 12 6 1512707 -64716 12 4 5117504 -26683 12 6 8567578 14777 12 4 13882517 89723 12 6 5883715 -17924 12 4 21814357 231227 12 6 5218444 -25544 12 5 15154163 109864 12 6 4879735 -29352 12 2 1044000 -69297 12 6 8263001 10903 12 2 1160595 -68165 12 6 1732500 -62540 12 2 1165841 -68114 12 6 2822474 -51483 12 2 1178085 -67994 12 6 18684310 170717 12 2 1216766 -67618 12 6 17573144 150751 12 2 1366215 -66156 12 6 8824585 18080 12 2 1372181 -66098 12 8 5910101 -17618 12 2 1383606 -65986 103 2 6612212 253149 12 2 1699586 -62867 103 4 1220082 148309 12 2 1705962 -62804 103 4 2856522526637811 12 2 1760281 -62264 103 8 14624049 452519 12 2 1775052 -62117 103 6 1928153136611045 12 2 1971052 -60159 103 6 4451377 208602 12 2 2016282 -59705 103 6 4440000 208376 12 2 2140886 -58450 103 6 76572470 4293042 12 2 2189904 -57955 103 8 15000000 463345 12 2 2274299 -57100 103 8 23505773 749269 12 2 2305694 -56782 103 8 2645500 174013 12 2 2363287 -56196 103 6 76026825 4253564 12 2 2411198 -55708 103 6 4098500 201659 12 2 3126788 -48315 103 6 5165562 222934 12 2 3199216 -47556 213 2 6800000 164340 12 2 3437350 -45045 213 2 6507000 159603 12 2 3452769 -44881 213 4 1631931 87325 12 2 3708246 -42161 213 4 3818700 118268 12 2 4084554 -38107 213 4 3848040 118699 12 2 4558109 -32924 213 4 2072679486601743 12 2 4674890 -31632 213 4 2800000 103566 12 2 4710678 -31235 213 4 1500000 85531 12 2 4724476 -31081 213 6 1654208 87629 12 2 4831632 -29889 213 8 4900000 134442

PAGE 148

148 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 12 2 5022711 -27750 213 2 4000000 120939 12 2 5118497 -26672 213 2 4000000 120939 12 2 5319569 -24397 213 2 1027640 79174 12 2 5475387 -22623 213 2 27782525 655506 12 2 5612662 -21051 213 2 20130222 437064 12 2 5818455 -18680 213 6 3030000 106841 12 2 5826748 -18584 213 2 6000000 151516 12 2 5854406 -18264 213 2 12338463 263264 12 2 6091274 -15508 213 2 8802104 197995 12 2 6525644 -10393 213 2 6273557 155862 12 2 7103062 -3465 213 2 6000000 151516 12 2 7281472 -1294 213 2 5157479 138382 12 2 8210198 10235 213 2 8171987 187157 12 2 10561682 41230 213 2 10763667 233243 12 2 11609772 55917 213 2 13126302 278885 12 2 12660016 71199 213 2 1371870 83797 12 2 13625430 85761 213 2 1247154 82116 12 2 15701745 118819 213 2 5000000 135968 12 3 1359060 -66227 213 3 1100000 80141 12 3 1371589 -66104 213 4 15584468 330325 12 3 1463514 -65201 213 4 6795601 164269 12 3 1465874 -65177 213 4 16480483 350135 12 3 3099236 -48603 213 4 1525150 85873 12 3 4461558 -33988 213 4 2309806 96672 12 3 4946931 -28600 213 4 2000000 92373 12 3 4960900 -28443 213 4 55149076 1939821 12 3 6016860 -16376 213 4 6800000 164340 12 3 6057037 -15908 213 4 3693703 116437 12 3 8207008 10195 213 4 23833362 536283 12 3 9858625 31685 213 4 21500000 472416 12 4 1079000 -68957 213 4 18565708 398540 12 4 1245992 -67332 213 4 2000000 92373 12 4 1259861 -67197 213 4 3821184 118305 12 4 1268622 -67111 213 4 3209591 109416 12 4 1276000 -67039 213 4 9921126 217818 12 4 1298638 -66818 213 4 10072973 220566 12 4 1300826 -66797 213 4 1500000 85531 12 4 1351702 -66299 213 4 1200000 81482 12 4 1390469 -65918 213 4 22558049 500795 12 4 1396946 -65855 213 6 1715000 88459 12 4 1466441 -65172 218 4 95820463 4190446

PAGE 149

149 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 12 4 1537000 -64477 218 4 11000000 188645 12 4 1627371 -63583 218 4 5000000 99471 12 4 1735858 -62507 218 4 4396089 91384 12 4 1786269 -62005 218 4 12248906 209379 12 4 1826410 -61605 218 4 1101920 49875 12 4 1845560 -61414 218 4 5517567 106524 12 4 1900308 -60867 218 4 5362223 104395 12 4 1965000 -60219 218 2 2300000 64477 12 4 2046056 -59406 218 2 10224603 176167 12 4 2153278 -58325 218 4 12315464 210506 12 4 2216121 -57690 218 2 8866515 155019 12 4 2268020 -57164 218 4 2653300 68889 12 4 2385575 -55969 218 4 9000000 157058 12 4 2466182 -55147 218 4 16639131 288867 12 4 2503795 -54763 12 6 36945764 628507 12 4 2579040 -53992 12 4 21917476 233332 12 4 2590500 -53874 12 4 17868268 155979 12 4 2708251 -52663 12 6 25091836 301806 12 4 2779589 -51927 12 4 9910141 32376 12 4 2784589 -51875 12 4 12466090 68334 12 4 3549995 -43849 12 6 22023453 235503 12 4 3801364 -41163 12 4 4769972 -30576 12 4 3807021 -41102 12 4 10268701 37223 12 4 3829577 -40860 12 4 6726481 -8000 12 4 4034516 -38649 12 4 12434258 67865 12 4 4464075 -33960 12 2 5624233 -20918 12 4 5094596 -26942 12 4 2503280 -54768 12 4 5642793 -20705 12 4 11268263 51071 12 4 6464367 -11119 12 2 5987393 -16720 12 4 6878027 -6182 12 4 1546637 -64381 12 4 7311251 -931 12 4 1039549 -69340 12 4 7613931 2788 12 2 5670187 -20390 12 4 8458572 13385 12 3 4886020 -29281 12 4 11074754 48351 12 4 3151826 -48053 12 4 11639122 56337 12 2 4056243 -38414 12 4 12969867 75818 12 2 12574133 69928 12 4 13301908 80825 12 6 2678900 -52966 12 5 1353112 -66285 12 4 3963163 -39421 12 5 1369000 -66129 12 4 2573947 -54044 12 5 1838048 -61489 12 3 3582103 -43507 12 5 2088477 -58979 12 4 3506780 -44308

PAGE 150

150 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 12 5 2438951 -55425 12 3 11036450 47815 12 5 2551034 -54279 12 2 12026411 61911 12 5 2675634 -52999 12 4 1450000 -65334 12 5 2994084 -49701 12 5 4654347 -31859 12 5 3559814 -43745 12 4 3242621 -47100 12 5 4712443 -31215 12 4 15838532 121083 12 5 5236308 -25341 12 3 18265045 163093 12 5 6355652 -12404 12 2 3014988 -49483 12 5 10244042 36888 12 2 4264549 -36148 12 6 1059299 -69148 12 2 2653938 -53222 12 6 1193000 -67849 12 4 18120206 160485 12 6 1285293 -66949 12 4 7409819 276 12 6 1656000 -63300 12 2 11891327 59958 12 6 1681500 -63047 12 2 1095064 -68801 12 6 1773294 -62135 12 4 17255530 145184 12 6 1801397 -61855 12 2 5678555 -20294 12 6 1983602 -60033 12 4 3530091 -44061 12 6 2244528 -57402 12 4 8642835 15741 12 6 2295081 -56889 12 2 3749467 -41720 12 6 2613474 -53638 12 5 15014658 107610 12 6 2896351 -50717 12 3 1276958 -67030 12 6 3434068 -45079 12 4 5974488 -16870 12 6 3788041 -41306 12 3 1655339 -63306 12 6 4123070 -37689 12 2 4427366 -34364 12 6 4622137 -32216 12 5 1304168 -66764 12 6 4651284 -31893 12 2 9453764 26297 12 6 4939781 -28680 12 6 1465903 -65177 12 6 5003451 -27966 12 5 6182568 -14440 12 6 5141718 -26410 12 2 1880918 -61061 12 6 5380545 -23704 12 6 2582481 -53957 12 6 5788574 -19025 12 6 5458885 -22811 12 6 5955972 -17085 12 4 3043333 -49187 12 6 6038066 -16129 12 6 7118087 -3282 12 6 6117293 -15204 12 6 6036580 -16147 12 6 7204215 -2236 12 5 2363588 -56193 12 6 7231792 -1900 12 1 19976496 194917 12 6 7980439 7346 12 6 9062784 21169 12 6 9035815 20818 12 6 3299408 -46502 12 8 7638989 3098 12 2 2425750 -55560 12 2 1632193 -63536 12 4 2785860 -51862 103 2 2667200 174414 12 6 3093632 -48662

PAGE 151

151 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 103 4 1607592 155161 12 6 3970908 -39337 103 4 1654618 156000 12 6 2870358 -50987 103 4 3408770 188349 12 8 3292112 -46579 103 4 3705105 194026 12 2 4030182 -38696 103 6 6169033 243721 12 2 8048887 8205 103 8 3644675 192863 12 4 5936356 -17313 103 8 3981900 199385 12 2 12142356 63595 103 6 15061018 465115 12 6 8732398 16892 103 3 1677174 156403 12 4 12925322 75151 103 3 2301657 167686 12 6 5123326 -26618 103 3 2648762 174073 12 5 1188531 -67893 103 3 2654793 174185 12 6 11668806 56761 103 3 5969089 239518 12 2 3514007 -44232 103 4 1380000 151125 12 4 6684147 -8505 103 4 5312360 225927 12 6 4801228 -30228 103 4 6400000 248616 12 6 3916572 -39924 103 4 12459068 392887 12 6 2557786 -54210 103 4 49778910 2221896 12 4 2109345 -58768 103 6 2879312 178361 12 6 4457620 -34031 103 6 2027599 162702 12 4 4281816 -35959 103 6 26137824 855108 12 4 3506513 -44311 103 2 13567530 422849 12 2 1538183 -64465 103 4 1191408 147806 12 6 2935859 -50307 103 4 2035680 162848 12 6 5035671 -27604 103 6 11261935 361828 12 3 1425784 -65572 103 8 3850821 196840 12 4 1387705 -65946 103 8 10544390 343840 12 6 4487702 -33700 103 8 14014605 435269 12 4 4520277 -33341 103 8 30532009 1051899 12 3 5697556 -20075 103 8 221412954 6628326 12 6 4687110 -31496 103 1 4987446 219324 12 5 4846549 -29722 103 6 34108627 1231440 12 2 7114726 -3323 103 4 1396528 151417 12 6 10427553 39391 103 4 4702760 213603 12 4 5289555 -24738 103 6 5668351 233253 12 2 10626696 42125 213 1 1176984 81173 12 6 11870935 59664 213 1 3966168 120439 12 2 2776038 -51963 213 1 9660518 213135 12 2 2529045 -54504 213 1 15152448 320980 12 2 1985404 -60015 213 1 20157318 437749 12 6 20018605 195724 213 1 64798281 2580652 12 4 10006799 33677

PAGE 152

152 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 213 1 68804811 2865050 103 4 9232000 312119 213 2 4378142 126562 103 1 48476309 2128019 213 2 5990520 151366 103 1 41327005 1647872 213 2 8350269 190200 103 4 25519263 829445 213 2 172605686 6513373 103 8 53396930 2491198 213 3 1010681 78948 103 6 79299607 4485394 213 3 4461728 127815 213 4 36353090 967603 213 4 1705100 88324 213 2 6800000 164340 213 4 2000000 92373 213 6 2530000 99754 213 4 2225000 95491 213 6 1650000 87572 213 4 2253922 95893 213 8 5450000 142899 213 4 2800000 103566 213 6 2492177706629658 213 4 3235799 109793 213 6 1197858 81453 213 4 3250000 109998 213 2 4000000 120939 213 4 3590000 114923 213 2 8929852 200221 213 4 4672090 130983 213 4 22071672 487631 213 4 5702552 146836 213 2 3043078 107028 213 4 6161370 154075 213 2 7012906 167812 213 4 14824949 313984 213 4 6634965 161666 213 4 15640539 331548 213 4 31292523 774080 213 4 107560004 5317385 213 4 12397812 264427 213 6 1172671 81115 213 4 2000000 92373 213 6 1219682 81746 213 2 6000000 151516 213 6 1394389 84101 213 4 1460000 84989 213 6 1890036 90858 213 4 10276666 224274 213 6 2882629 104740 213 2 2700000 102150 213 6 3350924 111454 213 2 3470000 113178 213 6 4679205 131091 213 2 2798600 103546 213 6 7189047 170703 213 2 5421129 142452 213 6 22329791 494592 213 4 17276584 368229 213 8 1100000 80141 213 4 3342114 111327 213 8 1210300 81620 213 4 1591340 86772 213 8 1295000 82760 213 4 1264550 82350 213 8 2822338 103883 213 2 2807739 103676 213 8 5060804 136898 213 2 1371870 83797 213 2 1938884 91531 213 2 2867000 104518 213 2 2000000 92373 213 2 29482249 711387 213 2 3186742 109088 213 2 5879000 149607 213 2 3805883 118080 213 6 32989965 835937 213 2 4000000 120939 213 2 1801000 89636 213 2 5197159 138992 218 4 1651718 56507

PAGE 153

153 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 213 2 6075993 152720 218 6 4100000 87475 213 2 7423000 174570 218 4 30708210 625706 213 2 8861138 199022 218 4 2000000 60768 213 2 9118903 203532 218 4 39962233 930892 213 2 12728717 270951 218 4 3418477 78614 213 2 15605648 330787 218 4 1254161 51700 213 2 18981143 408581 218 4 1776062 58023 213 2 19593720 423635 218 4 1720000 57339 213 4 1301978 82854 218 2 6349622 118105 213 4 2217939 95392 218 4 24186687 452489 213 4 3917194 119717 218 2 49251652 1316512 213 4 4011524 121109 218 2 1250000 51650 213 4 8343814 190090 218 4 30522524 620323 213 4 24640000 559470 218 4 9000000 157058 213 6 1365500 83711 218 4 52019051 1447661 213 6 2944473 105620 218 4 10000000 172608 213 6 3447554 112853 218 4 1136429 50288 213 6 4804195 132985 218 4 5238917 102713 213 6 23527259 527635 12 3 3894753 -40159 213 6 26958663 629430 12 4 13035225 76799 213 1 2022003 92677 12 4 3611991 -43189 213 2 1007367 78903 12 4 14152837 93929 213 2 1067184 79702 12 4 17235573 144836 213 2 1197962 81454 12 6 20581900 206629 213 2 1286567 82646 12 2 13808116 88573 213 2 1345065 83435 12 2 10708107 43249 213 2 1382505 83940 12 4 10062782 34432 213 2 1423370 84493 12 2 2762067 -52108 213 2 1467914 85096 12 2 13451516 83100 213 2 1581871 86644 12 4 2727098 -52469 213 2 1856688 90400 12 4 10773729 44157 213 2 1870704 90593 12 4 18241179 162663 213 2 1930200 91411 12 4 12255163 65240 213 2 2514154 99532 12 4 15362918 113258 213 2 2846411 104225 12 3 3168899 -47874 213 2 2964262 105903 12 4 10045743 34202 213 2 2989999 106270 12 2 4658578 -31813 213 2 3749812 117258 12 2 12160562 63860 213 2 3948064 120172 12 3 6976833 -4992 213 2 4000815 120951 12 2 5049679 -27447 213 2 4382030 126620 12 8 3438059 -45037

PAGE 154

154 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 213 2 4428194 127312 12 4 9948188 32888 213 2 5093984 137407 12 4 4632350 -32103 213 2 5514468 143901 12 2 9541401 27457 213 2 5771377 147915 12 2 8635874 15652 213 2 6835002 164909 12 2 1326196 -66548 213 2 8642282 195224 12 2 1541191 -64435 213 2 9018000 201762 12 2 6238811 -13780 213 2 9110435 203383 12 2 2170201 -58154 213 2 9771581 215126 12 2 9875095 31906 213 2 15034729 318456 12 2 2443824 -55375 213 2 17170414 365788 12 2 3343393 -46038 213 2 20314536 441731 12 3 3699005 -42260 213 2 21444057 470942 12 2 2565163 -54134 213 2 22927312 510925 12 2 3221396 -47323 213 2 28107048 665956 12 2 7553487 2042 213 2 29549112 713643 12 2 8090665 8729 213 2 34124268 878973 12 4 12192709 64328 213 3 18154370 388730 12 2 8566395 14762 213 4 1008236 78915 12 6 4882596 -29320 213 4 1031645 79227 12 4 8580317 14940 213 4 1041888 79364 12 4 7224194 -1993 213 4 1050001 79473 12 2 9590353 28106 213 4 1107309 80239 12 4 2190546 -57949 213 4 1150000 80811 12 2 1650825 -63351 213 4 1326789 83188 12 4 3407383 -45362 213 4 1353628 83550 12 4 3858722 -40547 213 4 1411139 84327 12 2 3798992 -41188 213 4 1458760 84972 12 5 10881497 45653 213 4 1464803 85054 12 2 7778322 4825 213 4 1585790 86697 12 6 2550001 -54290 213 4 1637464 87401 12 4 7110781 -3371 213 4 1729804 88661 12 5 4214954 -36689 213 4 1935954 91490 12 3 17557184 150470 213 4 1993452 92283 12 4 9301238 24288 213 4 1998870 92358 12 3 7991840 7489 213 4 2047680 93032 12 2 3359771 -45865 213 4 2069499 93334 12 5 4948107 -28587 213 4 2095207 93689 12 5 7680992 3618 213 4 2268244 96092 12 4 10308880 37770 213 4 2272080 96146 12 6 1472917 -65108 213 4 2700000 102150 12 4 3469388 -44705

PAGE 155

155 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 213 4 3462362 113068 12 4 1463715 -65199 213 4 3934828 119977 12 4 5631627 -20833 213 4 4418140 127161 12 2 2238403 -57464 213 4 5569697 144761 12 3 2212977 -57722 213 4 6339884 156922 12 4 3807129 -41101 213 4 6684233 162463 12 4 1047006 -69267 213 4 7266605 171982 12 2 2351420 -56317 213 4 7403865 174252 12 4 1805057 -61818 213 4 9239147 205649 12 2 3525944 -44105 213 4 10049427 220139 12 6 2418583 -55633 213 4 10560260 229479 12 6 2007775 -59790 213 4 12350000 263490 12 5 2049415 -59372 213 4 14034501 297405 12 2 1567142 -64179 213 4 14512760 307384 12 4 1914598 -60724 213 4 14663196 310556 12 4 8562635 14714 213 4 14847933 314472 12 3 10310008 37786 213 4 15520418 328931 12 3 1461621 -65219 213 4 19820798 429291 12 5 32146974 481744 213 4 22572609 501192 12 2 3445881 -44954 213 4 27901688 659331 12 2 8359161 12121 213 4 34099959 878036 12 4 2436333 -55452 213 4 36831121 987326 12 4 11895284 60015 213 4 154568188 6396646 12 4 8488886 13772 213 6 3243778 109908 12 2 3577976 -43551 218 4 1247256 51617 12 2 1143486 -68331 218 4 1324179 52542 12 2 1415341 -65674 218 4 1492789 54578 12 2 9309746 24400 218 4 1770818 57959 12 3 5519522 -22118 218 4 2114386 62178 12 2 1041410 -69322 218 4 2821967 71012 12 2 8474952 13594 218 4 3274450 76765 12 3 1270504 -67093 218 4 3919380 85108 12 1 12709323 71931 218 4 35395422 770995 12 8 1438210 -65450 218 4 113915158 5199354 12 2 3205411 -47491 218 2 14659298 251706 12 2 2301684 -56822 218 2 23023341 424842 12 2 1201000 -67771 218 4 1362051 52999 103 6 19643263 609297 218 4 1687559 56944 103 4 3300000 186281 218 4 17869800 313109 103 4 2005000 162293 218 2 3589000 80813 103 8 5700000 233909 218 2 5720000 109314 103 8 1287738406320048

PAGE 156

156 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 218 2 7483865 134383 103 4 4000000 199737 218 2 7933333 140994 213 4 13100142 278359 218 2 11091826 190142 213 4 4659746 130797 218 2 63137848 2048353 213 2 6800000 164340 218 4 1013299 48817 213 6 1655000 87640 218 4 2178065 62965 213 6 1900000 90996 218 4 2500000 66968 213 8 1658000 87681 218 4 3223705 76116 213 8 1610000 87027 218 4 3615000 81150 213 8 5725000 147188 218 4 3825000 83877 213 3 1077224945324568 218 4 4815000 96978 213 4 25671605 589982 218 4 5010000 99606 213 2 4000000 120939 218 4 5352000 104255 213 2 1205000 81549 218 4 5752350 109761 213 2 8227929 188110 218 4 5857787 111223 213 4 1593100 86796 218 4 8271598 146031 213 4 16207313 344033 218 4 8683500 152236 213 2 1500000 85531 218 4 22880098 421502 213 2 1500000 85531 218 4 36115970 794987 213 2 54238235 1883630 218 5 4500000 92765 213 2 77981699 3529599 12 1 11700000 57208 213 2 6550000 160295 12 4 9971149 33197 213 2 2208874 95266 12 4 2121593 -58645 213 2 1366415 83723 12 4 4155416 -37337 213 2 5143562 138168 12 4 7876752 6051 213 6 7159000 170209 12 6 5729692 -19705 213 4 5278613 140247 12 6 6892822 -6004 213 4 2760000 102999 12 6 16611070 134074 213 2 31792782 791997 12 6 5323553 -24352 213 4 15834468 335794 12 8 6723957 -8030 213 2 17259040 367825 12 2 6851870 -6497 213 4 3415392 112387 12 2 5799141 -18903 213 4 6435287 158451 12 2 3055488 -49061 213 2 9000000 201447 12 4 4288545 -35885 213 4 34886044 908654 12 4 2969782 -49954 213 2 54451658 1896720 12 4 10730063 43553 213 4 17176264 365922 12 4 15776518 120055 213 4 81580986 3786636 12 4 10158057 35721 213 4 4666666 130901 12 4 5302399 -24592 213 4 3443000 112787 12 4 15106943 109100 213 4 10000000 219244 12 4 1933328 -60537 213 2 30384990 742234

PAGE 157

157 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 12 4 3121365 -48372 213 2 21307263 467348 12 4 6899607 -5923 213 2 3080000 107557 12 4 7561050 2135 213 4 8118107 186241 12 4 18425428 165997 213 2 49201128 1588790 12 5 11614275 55982 213 4 2561964 100204 12 1 1081000 -68938 213 4 35536766 934509 12 2 8922328 19344 213 2 1371870 83797 12 2 6801311 -7103 213 4 3300000 110719 12 2 5388453 -23614 213 4 9269484 206184 12 2 3633106 -42964 213 4 4000000 120939 12 2 4070526 -38259 213 2 7036116 168192 12 2 5014437 -27843 218 4 1162455845305279 12 2 5669009 -20403 218 4 8029224 142417 12 2 3429408 -45129 218 4 10195255 175700 12 2 2206832 -57784 218 4 1656480 56565 12 2 2613927 -53634 218 2 5973235 112829 12 2 1992999 -59939 218 4 58011014 1757148 12 2 8080552 8602 218 2 5000000 99471 12 2 3408976 -45345 218 4 9000000 157058 12 2 2427113 -55546 218 2 1237030 51494 12 2 6100975 -15395 218 4 9599233 166318 12 2 6285013 -13237 218 2 3000000 73266 12 2 7775748 4793 12 2 6197850 -14261 12 2 6106969 -15325 12 4 4705619 -31291 12 2 6576341 -9790 103 2 1006061915639541 12 2 2012039 -59747 103 1 1924480 160840 12 2 18154133 161095 103 4 2448336 170375 12 2 2325152 -56584 103 8 30997927 1074289 12 2 7306531 -988 103 6 12724724 399961 12 2 2689540 -52856 213 2 6800000 164340 12 2 7077000 -3781 213 3 4000000 120939 12 2 4658421 -31814 213 4 29847551 723770 12 2 1852829 -61342 213 6 10100000 221057 12 2 19031224 177109 213 6 2002003136591540 12 2 7186507 -2451 213 6 1550000 86210 12 2 16362915 129863 213 6 10900000 235780 12 2 5848411 -18333 213 6 4000000 120939 12 2 2874078 -50948 213 8 94937616 4662794 12 2 1364005 -66178 213 8 14575435 308704 12 2 2733693 -52401 213 8 6400000 157885 12 2 5303453 -24580 213 8 31429350 778954

PAGE 158

158 Table B-7. Continued WMC Lanes Cost Pred. qty. WMCLanesCost Pred. qty. 12 2 3031048 -49316 213 2 4000000 120939 12 2 7028045 -4373 213 2 42589140 1245279 12 2 2121546 -58645 213 2 3927175 119864 12 2 4811940 -30108 213 2 56687098 2036581 12 2 2029521 -59572 213 2 4895000 134366 12 2 6813118 -6962 213 2 4647500 130612 12 2 7024236 -4419 213 2 2600000 100739 12 2 1217713 -67608 213 2 6050000 152308 12 2 1044000 -69297 213 2 40045877 1126657 12 2 1108000 -68676 213 4 5000000 135968 12 2 1447000 -65363 213 6 2000000 92373 12 2 1101000 -68744 213 6 15000500 317725 12 3 2481464 -54991 213 2 20000000 433783 12 3 7271988 -1410 213 2 8000000 184238 12 3 15207399 110727 213 2 1900000 90996 12 3 23734302 271637 213 2 27353088 641833 12 3 4381233 -34870 213 2 1371870 83797 12 3 2366012 -56169 213 2 37620116 1020438 12 3 4803267 -30205 213 3 12004509 256766 12 3 1121820 -68542 213 3 8000000 184238 12 3 2277784 -57065 213 4 1994727 92301 12 3 2304994 -56789 213 4 3825238 118364 12 3 3785185 -41337 213 4 2000000 92373 12 3 1742625 -62440 213 4 3518000 113875 12 3 4585727 -32619 213 4 30800000 756692 12 4 4655524 -31846 213 4 10000000 219244 12 4 3034535 -49279 213 4 42598665 1245738 12 4 3513724 -44235 213 4 8000000 184238 12 4 2318643 -56650 213 4 12715195 270683 12 4 2703891 -52708 213 4 5330000 141041 12 4 1631462 -63543 213 6 25000000 570007 12 4 4488534 -33691 218 4 3917121 85079 12 4 6141414 -14922 218 4 4956852 98888 12 4 1263699 -67160 218 4 2231848 63631 12 4 4182438 -37043 218 4 1500000 54666 12 4 2125444 -58606 218 2 5000000 99471 12 4 1640535 -63453 218 2 1280326 52015 12 4 5159874 -26205 218 2 5000000 99471 12 4 1419851 -65630 218 4 9000000 157058 12 4 3596110 -43358 218 4 2000000 60768 12 4 4485447 -33725 218 5 5568318 107222

PAGE 159

159 Table B-8. Prediction data a nd results for optional base WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 1 1 1176984 94266 12 2 5 1852829 26501 213 1 1 3966168 110066 12 2 5 19031224 91401 213 1 1 9660518 147509 12 2 5 7186507 43089 213 1 1 15152448 190011 12 2 5 16362915 79170 213 1 1 20157318 233324 12 2 5 5848411 38628 213 1 1 64798281 558528 12 2 5 2874078 29433 213 1 1 68804811 571886 12 2 5 1364005 25137 213 2 1 4378142 97795 12 2 5 2733693 29023 213 2 1 5990520 107184 12 2 5 5303453 36869 213 2 1 8350269 121943 12 2 5 3031048 29894 213 2 1 172605686 646728 12 2 5 7028045 42550 213 3 1 1010681 68646 12 2 5 2121546 27261 213 3 1 4461728 85538 12 2 5 4811940 35312 213 4 1 1705100 63331 12 2 5 2029521 27000 213 4 1 2000000 64622 12 2 5 6813118 41823 213 4 1 2225000 65618 12 2 5 7024236 42537 213 4 1 2253922 65747 12 2 5 1217713 24734 213 4 1 2800000 68207 12 2 5 1044000 24258 213 4 1 3235799 70212 12 2 5 1108000 24433 213 4 1 3250000 70277 12 2 5 1447000 25367 213 4 1 3590000 71867 12 2 5 1101000 24414 213 4 1 4672090 77078 12 3 5 2481464 25146 213 4 1 5702552 82257 12 3 5 7271988 39755 213 4 1 6161370 84631 12 3 5 15207399 69684 213 4 1 14824949 137608 12 3 5 23734302 109452 213 4 1 15640539 143384 12 3 5 4381233 30635 213 4 1 107560004 612784 12 3 5 2366012 24825 213 6 1 1172671 55154 12 3 5 4803267 31908 213 6 1 1219682 55339 12 3 5 1121820 21457 213 6 1 1394389 56031 12 3 5 2277784 24581 213 6 1 1890036 58020 12 3 5 2304994 24656 213 6 1 2882629 62130 12 3 5 3785185 28870 213 6 1 3350924 64127 12 3 5 1742625 23117 213 6 1 4679205 70002 12 3 5 4585727 31249 213 6 1 7189047 81965 12 4 5 4655524 23970 213 6 1 22329791 178080 12 4 5 3034535 19732 213 8 1 1100000 64584 12 4 5 3513724 20958 213 8 1 1210300 65041 12 4 5 2318643 17945 213 8 1 1295000 65394 12 4 5 2703891 18900 213 8 1 2822338 71944 12 4 5 1631462 16276 213 8 1 5060804 82215 12 4 5 4488534 23522

PAGE 160

160 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 218 4 1 1247256 48230 12 4 5 6141414 28092 218 4 1 1324179 48472 12 4 5 1263699 15402 218 4 1 1492789 49006 12 4 5 4182438 22706 218 4 1 1770818 49893 12 4 5 2125444 17471 218 4 1 2114386 51002 12 4 5 1640535 16298 218 4 1 2821967 53331 12 4 5 5159874 25344 218 4 1 3274450 54851 12 4 5 1419851 15771 218 4 1 3919380 57062 12 4 5 3596110 21171 218 4 1 35395422 235201 12 4 5 4485447 23513 218 4 1 113915158 611805 12 4 5 1515518 15999 103 2 1 2667200 99421 12 4 5 3642920 21292 103 4 1 1607592 88018 12 4 5 9699792 38908 103 4 1 1654618 88158 12 4 5 6393357 28813 103 4 1 3408770 93421 12 4 5 3016096 19686 103 4 1 3705105 94318 12 4 5 2662205 18796 103 6 1 6169033 90741 12 4 5 1398242 15720 103 8 1 3644675 82459 12 4 5 1784897 16645 103 8 1 3981900 82822 12 4 5 6737032 29808 103 6 1 15061018 106953 12 4 5 3526796 20991 103 3 1 1677174 91775 12 4 5 5583750 26518 103 3 1 2301657 94056 12 4 5 2144332 17517 103 3 1 2648762 95334 12 4 5 1035883 14867 103 3 1 2654793 95356 12 4 5 3854711 21844 103 3 1 5969089 107880 12 4 5 6803161 30001 103 4 1 1380000 87342 12 4 5 4160640 22648 103 4 1 5312360 99229 12 4 5 1991058 17143 103 4 1 6400000 102589 12 4 5 3962485 22126 103 4 1 12459068 121740 12 4 5 2460320 18294 103 4 1 49778910 237285 12 4 5 2177070 17597 103 6 1 2879312 84602 12 4 5 1559720 16105 103 6 1 2027599 83006 12 4 5 7323437 31535 103 6 1 26137824 125809 12 4 5 2533434 18476 103 2 1 13567530 152268 12 4 5 2917849 19438 103 4 1 1191408 86783 12 4 5 1127000 15080 103 4 1 2035680 89293 12 5 5 3916620 12380 12 1 1 1212202 -11890 12 5 5 4794072 14291 12 1 1 1270370 -11838 12 5 5 1837340 8110 12 1 1 7004322 -5975 12 5 5 7242670 19979 12 1 1 12300966 1051 12 5 5 1343071 7148 12 1 1 16348584 7714 12 5 5 15367120 42814 12 1 1 258704303 594461 12 5 5 2041000 8513

PAGE 161

161 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 2 1 5126542 -12268 12 6 5 12073870 17698 12 2 1 7870386 -9564 12 6 5 3926664 2391 12 2 1 19398408 6567 12 6 5 3714290 2052 12 4 1 1138747 -22341 12 6 5 3793119 2177 12 4 1 1349709 -22242 12 6 5 7647389 8811 12 4 1 1858138 -22001 12 6 5 1874348 -769 12 4 1 3506058 -21175 12 6 5 2537133 223 12 4 1 5731556 -19947 12 6 5 3698696 2027 12 4 1 6871638 -19265 12 6 5 2030390 -538 12 4 1 10145137 -17083 12 6 5 1512707 -1299 12 4 1 17501485 -10764 12 6 5 8567578 10543 12 6 1 1001235 -26401 12 6 5 5883715 5653 12 6 1 2943374 -25876 12 6 5 5218444 4516 12 6 1 3112200 -25828 12 6 5 4879735 3948 12 6 1 5456007 -25124 12 6 5 8263001 9963 12 6 1 5949853 -24966 12 6 5 1732500 -978 12 6 1 6251345 -24868 12 6 5 2822474 658 12 6 1 7646571 -24397 12 6 5 18684310 33702 12 6 1 9218116 -23830 12 6 5 17573144 30776 12 6 1 12343896 -22584 12 6 5 8824585 11038 12 8 1 1745365 -28058 12 8 5 5910101 -11711 213 2 3 1938884 83029 213 4 1 36353090 319046 213 2 3 2000000 83395 213 2 1 6800000 112110 213 2 3 3186742 90692 213 6 1 2530000 60651 213 2 3 3805883 94651 213 6 1 1650000 57052 213 2 3 4000000 95913 213 8 1 5450000 84082 213 2 3 5197159 103927 213 6 1 249217770 648516 213 2 3 6075993 110061 213 6 1 1197858 55253 213 2 3 7423000 119873 218 4 1 1651718 49512 213 2 3 8861138 130893 218 6 1 4100000 36639 213 2 3 9118903 132927 218 4 1 30708210 200321 213 2 3 12728717 163204 218 4 1 2000000 50631 213 2 3 15605648 189532 218 4 1 39962233 270835 213 2 3 18981143 222425 218 4 1 3418477 55340 213 2 3 19593720 228577 218 4 1 1254161 48252 213 4 3 1301978 41844 218 4 1 1776062 49910 213 4 3 2217939 45423 218 4 1 1720000 49730 213 4 3 3917194 52459 103 4 1 9232000 111459 213 4 3 4011524 52864 103 1 1 48476309 386185 213 4 3 8343814 73295 103 1 1 41327005 346337 213 4 3 24640000 179878 103 4 1 25519263 164062

PAGE 162

162 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 6 3 1365500 21489 12 6 1 36945764 -5470 213 6 3 2944473 25439 12 4 1 21917476 -5834 213 6 3 3447554 26746 12 4 1 17868268 -10390 213 6 3 4804195 30390 12 6 1 25091836 -15527 213 6 3 23527259 98122 12 4 1 9910141 -17251 213 6 3 26958663 113657 12 4 1 12466090 -15317 218 2 3 14659298 92626 12 6 1 22023453 -17550 218 2 3 23023341 112187 213 2 3 4000000 95913 218 4 3 1362051 56539 213 2 3 8929852 131433 218 4 3 1687559 57154 213 4 3 22071672 160477 218 4 3 17869800 97710 213 2 3 3043078 89789 103 6 3 11261935 29197 213 2 3 7012906 116833 103 8 3 3850821 17234 218 2 3 6349622 77479 103 8 3 10544390 27931 218 4 3 24186687 120822 103 8 3 14014605 34049 218 2 3 49251652 223808 103 8 3 30532009 68929 218 2 3 1250000 69635 103 8 3 221412954 646279 103 8 3 53396930 136461 103 1 3 4987446 61846 103 6 3 79299607 306199 103 6 3 34108627 93757 12 4 3 4769972 12153 103 4 3 1396528 14946 12 4 3 10268701 24132 2 2 3 1250000 240218 12 4 3 6726481 16164 2 2 3 1576356 242336 12 4 3 12434258 29459 2 2 3 2310271 247135 12 2 3 5624233 32390 12 2 3 1182762 21631 12 4 3 2503280 7845 12 2 3 1320990 21941 12 4 3 11268263 26548 12 2 3 1882863 23215 12 2 3 5987393 33346 12 4 3 1121933 5393 12 4 3 1546637 6133 12 4 3 1499066 6050 12 4 3 1039549 5251 12 4 3 1836873 6646 213 4 5 6634965 54350 12 4 3 2175141 7251 213 4 5 31292523 220270 12 4 3 2306544 7488 213 4 5 12397812 82030 12 4 3 3352874 9418 213 4 5 2000000 36612 12 4 3 3755425 10181 213 2 5 6000000 86448 12 4 3 4067811 10781 213 4 5 1460000 34782 12 4 3 4459035 11541 213 4 5 10276666 71068 12 4 3 4843570 12299 213 2 5 2700000 67147 12 4 3 4847885 12308 213 2 5 3470000 71394 12 4 3 5095252 12801 213 2 5 2798600 67682 12 4 3 5117504 12846 213 2 5 5421129 82852 12 4 3 13882517 33214 213 4 5 17276584 110845 12 4 3 21814357 56434 213 4 5 3342114 41367

PAGE 163

163 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 5 3 15154163 16746 213 4 5 1591340 35223 213 1 5 2022003 84265 213 4 5 1264550 34131 213 2 5 1007367 58342 213 2 5 2807739 67732 213 2 5 1067184 58641 213 2 5 1371870 60177 213 2 5 1197962 59297 213 2 5 2867000 68055 213 2 5 1286567 59745 213 2 5 29482249 299767 213 2 5 1345065 60041 213 2 5 5879000 85689 213 2 5 1382505 60231 213 6 5 32989965 152424 213 2 5 1423370 60439 213 2 5 1801000 62381 213 2 5 1467914 60666 218 4 5 30522524 269157 213 2 5 1581871 61250 218 4 5 9000000 112469 213 2 5 1856688 62670 218 4 5 52019051 412805 213 2 5 1870704 62743 218 4 5 10000000 118663 213 2 5 1930200 63053 218 4 5 1136429 69668 213 2 5 2514154 66145 218 4 5 5238917 90662 213 2 5 2846411 67943 12 2 5 5670187 38049 213 2 5 2964262 68587 12 3 5 4886020 32160 213 2 5 2989999 68728 12 4 5 3151826 20030 213 2 5 3749812 72975 12 2 5 4056243 32970 213 2 5 3948064 74108 12 2 5 12574133 63119 213 2 5 4000815 74411 12 6 5 2678900 439 213 2 5 4382030 76624 12 4 5 3963163 22128 213 2 5 4428194 76894 12 4 5 2573947 18576 213 2 5 5093984 80860 12 3 5 3582103 28278 213 2 5 5514468 83426 12 4 5 3506780 20940 213 2 5 5771377 85017 12 3 5 11036450 53057 213 2 5 6835002 91795 12 2 5 12026411 60932 213 2 5 8642282 104026 12 4 5 1450000 15843 213 2 5 9018000 106683 12 5 5 4654347 13982 213 2 5 9110435 107342 12 4 5 3242621 20262 213 2 5 9771581 112129 12 4 5 15838532 60786 213 2 5 15034729 154483 12 3 5 18265045 83097 213 2 5 17170414 173709 12 2 5 3014988 29847 213 2 5 20314536 203868 12 2 5 4264549 33609 213 2 5 21444057 215167 12 2 5 2653938 28791 213 2 5 22927312 230312 12 4 5 18120206 69950 213 2 5 28107048 285030 12 4 5 7409819 31793 213 2 5 29549112 300483 12 2 5 11891327 60397 213 2 5 34124268 348975 12 2 5 1095064 24398 213 3 5 18154370 146951 12 4 5 17255530 66413 213 4 5 1008236 33287 12 2 5 5678555 38076

PAGE 164

164 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 4 5 1031645 33363 12 4 5 3530091 21000 213 4 5 1041888 33397 12 4 5 8642835 35554 213 4 5 1050001 33423 12 2 5 3749467 32037 213 4 5 1107309 33612 12 5 5 15014658 41693 213 4 5 1150000 33752 12 3 5 1276958 21868 213 4 5 1326789 34338 12 4 5 5974488 27617 213 4 5 1353628 34427 12 3 5 1655339 22881 213 4 5 1411139 34619 12 2 5 4427366 34112 213 4 5 1458760 34778 12 5 5 1304168 7073 213 4 5 1464803 34798 12 2 5 9453764 51113 213 4 5 1585790 35204 12 6 5 1465903 -1367 213 4 5 1637464 35378 12 5 5 6182568 17451 213 4 5 1729804 35691 12 2 5 1880918 26580 213 4 5 1935954 36393 12 6 5 2582481 292 213 4 5 1993452 36590 12 6 5 5458885 4924 213 4 5 1998870 36608 12 4 5 3043333 19755 213 4 5 2047680 36776 12 6 5 7118087 7841 213 4 5 2069499 36851 12 6 5 6036580 5918 213 4 5 2095207 36940 12 5 5 2363588 9157 213 4 5 2268244 37539 12 1 5 19976496 92080 213 4 5 2272080 37552 12 6 5 9062784 11500 213 4 5 2700000 39055 12 6 5 3299408 1397 213 4 5 3462362 41807 12 2 5 2425750 28132 213 4 5 3934828 43563 12 4 5 2785860 19106 213 4 5 4418140 45398 12 6 5 3093632 1077 213 4 5 5569697 49937 12 6 5 3970908 2462 213 4 5 6339884 53107 12 6 5 2870358 732 213 4 5 6684233 54559 12 8 5 3292112 -13678 213 4 5 7266605 57066 12 2 5 4030182 32890 213 4 5 7403865 57666 12 2 5 8048887 46072 213 4 5 9239147 66040 12 4 5 5936356 27509 213 4 5 10049427 69948 12 2 5 12142356 61392 213 4 5 10560260 72480 12 6 5 8732398 10860 213 4 5 12350000 81772 12 4 5 12925322 49894 213 4 5 14034501 91130 12 6 5 5123326 4356 213 4 5 14512760 93896 12 5 5 1188531 6851 213 4 5 14663196 94777 12 6 5 11668806 16826 213 4 5 14847933 95864 12 2 5 3514007 31328 213 4 5 15520418 99885 12 4 5 6684147 29655 213 4 5 19820798 127894 12 6 5 4801228 3818 213 4 5 22572609 147861 12 6 5 3916572 2375

PAGE 165

165 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 4 5 27901688 190665 12 6 5 2557786 254 213 4 5 34099959 245787 12 4 5 2109345 17432 213 4 5 36831121 271165 12 6 5 4457620 3251 213 4 5 154568188 641934 12 4 5 4281816 22970 213 6 5 3243778 30700 12 4 5 3506513 20939 218 2 5 3589000 64573 12 2 5 1538183 25620 218 2 5 5720000 74049 12 6 5 2935859 833 218 2 5 7483865 82463 12 6 5 5035671 4209 218 2 5 7933333 84690 12 3 5 1425784 22264 218 2 5 11091826 101309 12 4 5 1387705 15695 218 2 5 63137848 453637 12 6 5 4487702 3301 218 4 5 1013299 69083 12 4 5 4520277 23607 218 4 5 2178065 74719 12 3 5 5697556 34671 218 4 5 2500000 76318 12 6 5 4687110 3629 218 4 5 3223705 79980 12 5 5 4846549 14407 218 4 5 3615000 81999 12 2 5 7114726 42844 218 4 5 3825000 83093 12 6 5 10427553 14228 218 4 5 4815000 88356 12 4 5 5289555 25701 218 4 5 5010000 89413 12 2 5 10626696 55496 218 4 5 5352000 91282 12 6 5 11870935 17260 218 4 5 5752350 93496 12 2 5 2776038 29147 218 4 5 5857787 94084 12 2 5 2529045 28430 218 4 5 8271598 108059 12 2 5 1985404 26875 218 4 5 8683500 110542 12 6 5 20018605 37345 218 4 5 22880098 210079 12 4 5 10006799 39905 218 4 5 36115970 311207 213 4 1 13100142 125829 218 5 5 4500000 97241 213 4 1 4659746 77018 103 4 5 4702760 64380 213 2 1 6800000 112110 103 6 5 5668351 45105 213 6 1 1655000 57072 2 2 5 3646098 116798 213 6 1 1900000 58061 2 4 5 2355424 109059 213 8 1 1658000 66918 12 2 5 1044000 24258 213 8 1 1610000 66715 12 2 5 1160595 24577 213 8 1 5725000 85417 12 2 5 1165841 24592 213 3 1 107722494 621337 12 2 5 1178085 24625 213 4 1 25671605 223723 12 2 5 1216766 24732 218 4 1 116245584 614452 12 2 5 1366215 25143 218 4 1 8029224 72397 12 2 5 1372181 25160 218 4 1 10195255 81390 12 2 5 1383606 25192 218 4 1 1656480 49527 12 2 5 1699586 26071 103 6 1 19643263 114968 12 2 5 1705962 26089 103 4 1 3300000 93092

PAGE 166

166 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 2 5 1760281 26241 213 2 3 4000000 95913 12 2 5 1775052 26282 213 2 3 1205000 78717 12 2 5 1971052 26835 213 2 3 8227929 125972 12 2 5 2016282 26963 213 4 3 1593100 42966 12 2 5 2140886 27317 213 4 3 16207313 119362 12 2 5 2189904 27456 213 2 3 1500000 80433 12 2 5 2274299 27697 213 2 3 1500000 80433 12 2 5 2305694 27787 213 2 3 54238235 521190 12 2 5 2363287 27952 213 2 3 77981699 594726 12 2 5 2411198 28090 213 2 3 6550000 113457 12 2 5 3126788 30176 213 2 3 2208874 84652 12 2 5 3199216 30390 213 2 3 1366415 79653 12 2 5 3437350 31099 213 2 3 5143562 103560 12 2 5 3452769 31145 213 6 3 7159000 37129 12 2 5 3708246 31913 213 4 3 5278613 58476 12 2 5 4084554 33056 213 4 3 2760000 47611 12 2 5 4558109 34518 218 2 3 5973235 76868 12 2 5 4674890 34882 218 4 3 58011014 340876 12 2 5 4710678 34994 103 4 3 2005000 16176 12 2 5 4724476 35037 103 8 3 5700000 20046 12 2 5 4831632 35374 103 8 3 128773840 496729 12 2 5 5022711 35976 103 4 3 4000000 20425 12 2 5 5118497 36280 12 3 3 3894753 21731 12 2 5 5319569 36921 12 4 3 13035225 30998 12 2 5 5475387 37420 12 4 3 3611991 9908 12 2 5 5612662 37863 12 4 3 14152837 33932 12 2 5 5818455 38530 12 4 3 17235573 42494 12 2 5 5826748 38557 12 6 3 20581900 7251 12 2 5 5854406 38647 12 2 3 13808116 56881 12 2 5 6091274 39422 12 2 3 10708107 46871 12 2 5 6525644 40860 12 4 3 10062782 23644 12 2 5 7103062 42804 12 2 3 2762067 25263 12 2 5 7281472 43413 12 2 3 13451516 55684 12 2 5 8210198 46639 12 4 3 2727098 8255 12 2 5 10561682 55249 12 4 3 10773729 25343 12 2 5 11609772 59290 12 4 3 18241179 45433 12 2 5 12660016 63465 12 4 3 12255163 29005 12 2 5 13625430 67413 12 4 3 15362918 37211 12 2 5 15701745 76256 12 3 3 3168899 20064 12 3 5 1359060 22086 213 2 5 31792782 324437 12 3 5 1371589 22120 213 4 5 15834468 101797

PAGE 167

167 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 3 5 1463514 22365 213 2 5 17259040 174530 12 3 5 1465874 22372 213 4 5 3415392 41635 12 3 5 3099236 26888 213 4 5 6435287 53507 12 3 5 4461558 30875 213 2 5 9000000 106555 12 3 5 4946931 32346 213 4 5 34886044 253049 12 3 5 4960900 32389 213 2 5 54451658 519004 12 3 5 6016860 35680 213 4 5 17176264 110201 12 3 5 6057037 35807 213 4 5 81580986 568438 12 3 5 8207008 42908 213 4 5 4666666 46358 12 3 5 9858625 48720 213 4 5 3443000 41736 12 4 5 1079000 14968 213 4 5 10000000 69706 12 4 5 1245992 15360 213 2 5 30384990 309430 12 4 5 1259861 15393 213 2 5 21307263 213787 12 4 5 1268622 15414 213 2 5 3080000 69224 12 4 5 1276000 15431 213 4 5 8118107 60847 12 4 5 1298638 15484 213 2 5 49201128 484542 12 4 5 1300826 15490 213 4 5 2561964 38567 12 4 5 1351702 15610 213 4 5 35536766 259090 12 4 5 1390469 15702 213 2 5 1371870 60177 12 4 5 1396946 15717 213 4 5 3300000 41213 12 4 5 1466441 15882 213 4 5 9269484 66184 12 4 5 1537000 16050 213 4 5 4000000 43808 12 4 5 1627371 16266 213 2 5 7036116 93112 12 4 5 1735858 16527 218 2 5 5000000 70764 12 4 5 1786269 16648 218 4 5 9000000 112469 12 4 5 1826410 16745 218 2 5 1237030 54964 12 4 5 1845560 16791 218 4 5 9599233 116161 12 4 5 1900308 16923 218 2 5 3000000 62084 12 4 5 1965000 17080 12 4 5 10045743 40033 12 4 5 2046056 17277 12 2 5 4658578 34831 12 4 5 2153278 17539 12 2 5 12160562 61464 12 4 5 2216121 17693 12 3 5 6976833 38781 12 4 5 2268020 17820 12 2 5 5049679 36062 12 4 5 2385575 18110 12 8 5 3438059 -13573 12 4 5 2466182 18309 12 4 5 9948188 39714 12 4 5 2503795 18402 12 4 5 4632350 23908 12 4 5 2579040 18589 12 2 5 9541401 51435 12 4 5 2590500 18618 12 2 5 8635874 48151 12 4 5 2708251 18911 12 2 5 1326196 25033 12 4 5 2779589 19090 12 2 5 1541191 25629 12 4 5 2784589 19102 12 2 5 6238811 39908

PAGE 168

168 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 4 5 3549995 21051 12 2 5 2170201 27400 12 4 5 3801364 21704 12 2 5 9875095 52669 12 4 5 3807021 21719 12 2 5 2443824 28184 12 4 5 3829577 21778 12 2 5 3343393 30819 12 4 5 4034516 22315 12 3 5 3699005 28618 12 4 5 4464075 23456 12 2 5 2565163 28534 12 4 5 5094596 25164 12 2 5 3221396 30456 12 4 5 5642793 26683 12 2 5 7553487 44348 12 4 5 6464367 29018 12 2 5 8090665 46218 12 4 5 6878027 30220 12 4 5 12192709 47299 12 4 5 7311251 31499 12 2 5 8566395 47902 12 4 5 7613931 32404 12 6 5 4882596 3953 12 4 5 8458572 34982 12 4 5 8580317 35359 12 4 5 11074754 43453 12 4 5 7224194 31240 12 4 5 11639122 45378 12 2 5 9590353 51616 12 4 5 12969867 50053 12 4 5 2190546 17630 12 4 5 13301908 51250 12 2 5 1650825 25934 12 5 5 1353112 7167 12 4 5 3407383 20684 12 5 5 1369000 7198 12 4 5 3858722 21854 12 5 5 1838048 8112 12 2 5 3798992 32187 12 5 5 2088477 8607 12 5 5 10881497 29436 12 5 5 2438951 9309 12 2 5 7778322 45127 12 5 5 2551034 9536 12 6 5 2550001 243 12 5 5 2675634 9789 12 4 5 7110781 30905 12 5 5 2994084 10441 12 5 5 4214954 13022 12 5 5 3559814 11622 12 3 5 17557184 79903 12 5 5 4712443 14110 12 4 5 9301238 37629 12 5 5 5236308 15279 12 3 5 7991840 42173 12 5 5 6355652 17857 12 2 5 3359771 30867 12 5 5 10244042 27690 12 5 5 4948107 14633 12 6 5 1059299 -1953 12 5 5 7680992 21054 12 6 5 1193000 -1761 12 4 5 10308880 40896 12 6 5 1285293 -1628 12 6 5 1472917 -1357 12 6 5 1656000 -1090 12 4 5 3469388 20843 12 6 5 1681500 -1053 12 4 5 1463715 15876 12 6 5 1773294 -918 12 4 5 5631627 26652 12 6 5 1801397 -876 12 2 5 2238403 27595 12 6 5 1983602 -607 12 3 5 2212977 24402 12 6 5 2244528 -218 12 4 5 3807129 21719 12 6 5 2295081 -142 12 4 5 1047006 14893 12 6 5 2613474 339 12 2 5 2351420 27918

PAGE 169

169 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 6 5 2896351 772 12 4 5 1805057 16693 12 6 5 3434068 1609 12 2 5 3525944 31364 12 6 5 3788041 2169 12 6 5 2418583 44 12 6 5 4123070 2707 12 6 5 2007775 -571 12 6 5 4622137 3522 12 5 5 2049415 8530 12 6 5 4651284 3570 12 2 5 1567142 25701 12 6 5 4939781 4048 12 4 5 1914598 16958 12 6 5 5003451 4155 12 4 5 8562635 35304 12 6 5 5141718 4387 12 3 5 10310008 50364 12 6 5 5380545 4790 12 3 5 1461621 22360 12 6 5 5788574 5489 12 5 5 32146974 109193 12 6 5 5955972 5778 12 2 5 3445881 31124 12 6 5 6038066 5921 12 2 5 8359161 47166 12 6 5 6117293 6059 12 4 5 2436333 18235 12 6 5 7204215 7997 12 4 5 11895284 46263 12 6 5 7231792 8048 12 4 5 8488886 35076 12 6 5 7980439 9431 12 2 5 3577976 31520 12 6 5 9035815 11448 12 2 5 1143486 24531 12 8 5 7638989 -10309 12 2 5 1415341 25279 12 2 5 1632193 25882 12 2 5 9309746 50586 213 2 1 6800000 112110 12 3 5 5519522 34114 213 2 1 6507000 110310 12 2 5 1041410 24251 213 4 1 1631931 63014 12 2 5 8474952 47577 213 4 1 3818700 72950 12 3 5 1270504 21851 213 4 1 3848040 73089 12 1 5 12709323 61138 213 4 1 207267948 646327 12 8 5 1438210 -14964 213 4 1 2800000 68207 12 2 5 3205411 30409 213 4 1 1500000 62443 12 2 5 2301684 27776 213 6 1 1654208 57069 12 2 5 1201000 24688 213 8 1 4900000 81450 213 2 1 6800000 112110 218 4 1 95820463 581473 213 3 1 4000000 83138 218 4 1 11000000 84899 213 4 1 29847551 260675 218 4 1 5000000 60882 213 6 1 10100000 97286 218 4 1 4396089 58729 213 6 1 200200313 638545 218 4 1 12248906 90528 213 6 1 1550000 56651 218 4 1 1101920 47774 213 6 1 10900000 101772 218 4 1 5517567 62765 213 6 1 4000000 66959 218 4 1 5362223 62196 213 8 1 94937616 533393 103 2 1 6612212 117515 213 8 1 14575435 134729 103 4 1 1220082 86868 213 8 1 6400000 88744 103 4 1 285652252 649737 213 8 1 31429350 253443

PAGE 170

170 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 103 8 1 14624049 93380 218 4 1 3917121 57054 103 6 1 192815313 530915 218 4 1 4956852 60727 103 6 1 4451377 87542 218 4 1 2231848 51385 103 6 1 4440000 87520 218 4 1 1500000 49029 103 6 1 76572470 190830 103 2 1 100606191 493464 12 1 1 11700000 164 103 1 1 1924480 98988 12 4 1 9971149 -17208 213 2 3 4000000 95913 12 4 1 2121593 -21873 213 2 3 42589140 448647 12 4 1 4155416 -20830 213 2 3 3927175 95438 12 4 1 7876752 -18631 213 2 3 56687098 532669 12 6 1 5729692 -25037 213 2 3 4895000 101868 12 6 1 6892822 -24655 213 2 3 4647500 100199 12 6 1 16611070 -20599 213 2 3 2600000 87037 12 6 1 5323553 -25166 213 2 3 6050000 109877 12 8 1 6723957 -27171 213 2 3 40045877 428529 213 2 3 4000000 95913 213 4 3 5000000 57217 213 2 3 4000000 95913 213 6 3 2000000 23048 213 2 3 1027640 77696 213 6 3 15000500 63372 213 2 3 27782525 313001 103 4 3 2448336 17092 213 2 3 20130222 234000 103 8 3 30997927 70063 213 6 3 3030000 25660 213 2 5 20000000 200761 218 2 3 2300000 71179 213 2 5 8000000 99576 218 2 3 10224603 84116 213 2 5 1900000 62896 218 4 3 12315464 81216 213 2 5 27353088 276961 103 8 3 15000000 35859 213 2 5 1371870 60177 103 8 3 23505773 52872 213 2 5 37620116 384532 103 8 3 2645500 15459 213 3 5 12004509 101461 103 6 3 76026825 288074 213 3 5 8000000 76952 103 6 3 4098500 15231 213 4 5 1994727 36594 12 2 3 6851870 35670 213 4 5 3825238 43152 12 2 3 5799141 32849 213 4 5 2000000 36612 12 2 3 3055488 25960 213 4 5 3518000 42012 12 4 3 4288545 11208 213 4 5 30800000 215877 12 4 3 2969782 8703 213 4 5 10000000 69706 12 4 3 10730063 25238 213 4 5 42598665 325117 12 4 3 15776518 38356 213 4 5 8000000 60315 12 4 3 10158057 23869 213 4 5 12715195 83750 12 4 3 5302399 13218 213 4 5 5330000 48973 12 4 3 15106943 36509 213 6 5 25000000 109527 12 4 3 1933328 6818 218 2 5 5000000 70764 12 4 3 3121365 8985 218 2 5 1280326 55133

PAGE 171

171 Table B-8. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 4 3 6899607 16532 218 2 5 5000000 70764 12 4 3 7561050 17959 218 4 5 9000000 112469 12 4 3 18425428 45979 218 4 5 2000000 73842 12 5 3 11614275 9972 218 5 5 5568318 103365 213 2 5 6000000 86448 103 6 5 12724724 59203 213 2 5 12338463 131855 12 2 5 6197850 39773 213 2 5 8802104 105152 12 4 5 4705619 24106 213 2 5 6273557 88179 213 2 5 5157479 81244 213 2 5 6000000 86448 213 2 5 8171987 100757 218 4 5 2653300 77086 213 2 5 10763667 119538 218 4 5 9000000 112469 213 2 5 13126302 138269 218 4 5 16639131 163511 213 2 5 1371870 60177 103 6 5 5165562 44169 213 2 5 1247154 59546 12 1 5 1081000 23470 213 2 5 5000000 80293 12 2 5 8922328 49180 213 3 5 1100000 43661 12 2 5 6801311 41784 213 4 5 15584468 100274 12 2 5 5388453 37141 213 4 5 6795601 55034 12 2 5 3633106 31686 213 4 5 16480483 105794 12 2 5 4070526 33013 213 4 5 1525150 35000 12 2 5 5014437 35950 213 4 5 2309806 37683 12 2 5 5669009 38045 213 4 5 2000000 36612 12 2 5 3429408 31075 213 4 5 55149076 432592 12 2 5 2206832 27505 213 4 5 6800000 55052 12 2 5 2613927 28675 213 4 5 3693703 42662 12 2 5 1992999 26897 213 4 5 23833362 157517 12 2 5 8080552 46183 213 4 5 21500000 139893 12 2 5 3408976 31014 213 4 5 18565708 119311 12 2 5 2427113 28136 213 4 5 2000000 36612 12 2 5 6100975 39454 213 4 5 3821184 43137 12 2 5 6285013 40061 213 4 5 3209591 40884 12 2 5 7775748 45118 213 4 5 9921126 69321 12 2 5 6106969 39474 213 4 5 10072973 70064 12 2 5 6576341 41029 213 4 5 1500000 34916 12 2 5 2012039 26951 213 4 5 1200000 33917 12 2 5 18154133 87299 213 4 5 22558049 147751 12 2 5 2325152 27843 213 6 5 1715000 27026 12 2 5 7306531 43499 218 2 5 8866515 89424 12 2 5 7077000 42716 12 2 5 2689540 28895 12 2 5 4658421 34831

PAGE 172

172 Table B-9. Prediction data a nd results for limestone base WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 1 1 1212202 83986 12 4 5 1784897 8453 12 1 1 1270370 84169 12 4 5 6737032 16401 12 1 1 7004322 103295 12 4 5 3526796 11047 12 1 1 12300966 123004 12 4 5 5583750 14386 12 1 1 16348584 139463 12 4 5 2144332 8972 12 1 1 258704303 654824 12 4 5 1035883 7399 12 2 1 5126542 77508 12 4 5 3854711 11558 12 2 1 7870386 86952 12 4 5 6803161 16519 12 2 1 19398408 133709 12 4 5 4160640 12043 12 4 1 1138747 34622 12 4 5 1991058 8750 12 4 1 1349709 35193 12 4 5 3962485 11728 12 4 1 1858138 36586 12 4 5 2460320 9435 12 4 1 3506058 41280 12 4 5 2177070 9019 12 4 1 5731556 48068 12 4 5 1559720 8133 12 4 1 6871638 51754 12 4 5 7323437 17466 12 4 1 10145137 63158 12 4 5 2533434 9543 12 4 1 17501485 93571 12 4 5 2917849 10117 12 6 1 1001235 13969 12 4 5 1127000 7525 12 6 1 2943374 18634 12 5 5 3916620 3198 12 6 1 3112200 19060 12 5 5 4794072 4316 12 6 1 5456007 25358 12 5 5 1837340 729 12 6 1 5949853 26777 12 5 5 7242670 7696 12 6 1 6251345 27660 12 5 5 1343071 177 12 6 1 7646571 31911 12 5 5 15367120 22315 12 6 1 9218116 37040 12 5 5 2041000 960 12 6 1 12343896 48387 12 6 5 12073870 9829 12 8 1 1745365 4274 12 6 5 3926664 -1018 12 2 3 1182762 10753 12 6 5 3714290 -1247 12 2 3 1320990 11058 12 6 5 3793119 -1162 12 2 3 1882863 12322 12 6 5 7647389 3393 12 4 3 1121933 18650 12 6 5 1874348 -3130 12 4 3 1499066 19646 12 6 5 2537133 -2471 12 4 3 1836873 20554 12 6 5 3698696 -1263 12 4 3 2175141 21479 12 6 5 2030390 -2977 12 4 3 2306544 21843 12 6 5 1512707 -3481 12 4 3 3352874 24827 12 6 5 8567578 4615 12 4 3 3755425 26017 12 6 5 5883715 1201 12 4 3 4067811 26958 12 6 5 5218444 422 12 4 3 4459035 28157 12 6 5 4879735 36 12 4 3 4843570 29358 12 6 5 8263001 4204 12 4 3 4847885 29372 12 6 5 1732500 -3268

PAGE 173

173 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 4 3 5095252 30157 12 6 5 2822474 -2180 12 4 3 5117504 30228 12 6 5 18684310 22622 12 4 3 13882517 64988 12 6 5 17573144 20153 12 4 3 21814357 109955 12 6 5 8824585 4966 12 5 3 15154163 76346 12 8 5 5910101 1358 12 2 5 1044000 36706 213 2 1 6800000 144010 12 2 5 1160595 36989 213 2 1 6507000 142001 12 2 5 1165841 37002 213 4 1 1631931 111755 12 2 5 1178085 37032 213 4 1 3818700 123404 12 2 5 1216766 37126 213 4 1 3848040 123566 12 2 5 1366215 37492 213 4 1 207267948 654893 12 2 5 1372181 37507 213 4 1 2800000 117876 12 2 5 1383606 37535 213 4 1 1500000 111078 12 2 5 1699586 38317 213 6 1 1654208 139113 12 2 5 1705962 38332 213 8 1 4900000 188664 12 2 5 1760281 38468 213 2 3 4000000 83508 12 2 5 1775052 38505 213 2 3 4000000 83508 12 2 5 1971052 38995 213 2 3 1027640 66106 12 2 5 2016282 39109 213 2 3 27782525 297256 12 2 5 2140886 39424 213 2 3 20130222 217569 12 2 5 2189904 39548 213 6 3 3030000 42385 12 2 5 2274299 39762 213 2 5 6000000 88202 12 2 5 2305694 39842 213 2 5 12338463 115972 12 2 5 2363287 39989 213 2 5 8802104 99848 12 2 5 2411198 40111 213 2 5 6273557 89296 12 2 5 3126788 41964 213 2 5 6000000 88202 12 2 5 3199216 42155 213 2 5 5157479 84889 12 2 5 3437350 42785 213 2 5 8171987 97143 12 2 5 3452769 42825 213 2 5 10763667 108590 12 2 5 3708246 43508 213 2 5 13126302 119788 12 2 5 4084554 44524 213 2 5 1371870 71025 12 2 5 4558109 45824 213 2 5 1247154 70596 12 2 5 4674890 46148 213 2 5 5000000 84279 12 2 5 4710678 46248 213 3 5 1100000 64735 12 2 5 4724476 46287 213 4 5 15584468 108037 12 2 5 4831632 46586 213 4 5 6795601 78388 12 2 5 5022711 47122 213 4 5 16480483 111472 12 2 5 5118497 47393 213 4 5 1525150 63770 12 2 5 5319569 47963 213 4 5 2309806 65811 12 2 5 5475387 48408 213 4 5 2000000 65000 12 2 5 5612662 48803 213 4 5 55149076 340628

PAGE 174

174 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 2 5 5818455 49398 213 4 5 6800000 78401 12 2 5 5826748 49422 213 4 5 3693703 69523 12 2 5 5854406 49502 213 4 5 23833362 142827 12 2 5 6091274 50193 213 4 5 21500000 132246 12 2 5 6525644 51477 213 4 5 18565708 119780 12 2 5 7103062 53215 213 4 5 2000000 65000 12 2 5 7281472 53760 213 4 5 3821184 69872 12 2 5 8210198 56653 213 4 5 3209591 68208 12 2 5 10561682 64427 213 4 5 9921126 88129 12 2 5 11609772 68108 213 4 5 10072973 88624 12 2 5 12660016 71936 213 4 5 1500000 63705 12 2 5 13625430 75581 213 4 5 1200000 62937 12 2 5 15701745 83842 213 4 5 22558049 136969 12 3 5 1359060 20277 213 6 5 1715000 73793 12 3 5 1371589 20300 218 4 1 95820463 612528 12 3 5 1463514 20473 218 4 1 11000000 120142 12 3 5 1465874 20477 218 4 1 5000000 80186 12 3 5 3099236 23652 218 4 1 4396089 76663 12 3 5 4461558 26473 218 4 1 12248906 129581 12 3 5 4946931 27517 218 4 1 1101920 59007 12 3 5 4960900 27547 218 4 1 5517567 83278 12 3 5 6016860 29896 218 4 1 5362223 82343 12 3 5 6057037 29988 218 2 3 2300000 39308 12 3 5 8207008 35111 218 2 3 10224603 61814 12 3 5 9858625 39372 218 4 3 12315464 50613 12 4 5 1079000 7459 218 2 5 8866515 83940 12 4 5 1245992 7691 218 4 5 2653300 31641 12 4 5 1259861 7711 218 4 5 9000000 50878 12 4 5 1268622 7723 218 4 5 16639131 80122 12 4 5 1276000 7733 12 6 1 36945764 201430 12 4 5 1298638 7765 12 4 1 21917476 115237 12 4 5 1300826 7768 12 4 1 17868268 95271 12 4 5 1351702 7839 12 6 1 25091836 112973 12 4 5 1390469 7894 12 4 1 9910141 62298 12 4 5 1396946 7903 12 4 1 12466090 72016 12 4 5 1466441 8001 12 6 1 22023453 94528 12 4 5 1537000 8100 12 4 3 4769972 29126 12 4 5 1627371 8229 12 4 3 10268701 48930 12 4 5 1735858 8383 12 4 3 6726481 35581 12 4 5 1786269 8455 12 4 3 12434258 58240 12 4 5 1826410 8513 12 2 3 5624233 21772

PAGE 175

175 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 4 5 1845560 8540 12 4 3 2503280 22392 12 4 5 1900308 8619 12 4 3 11268263 53114 12 4 5 1965000 8712 12 2 3 5987393 22792 12 4 5 2046056 8829 12 4 3 1546637 19773 12 4 5 2153278 8985 12 4 3 1039549 18435 12 4 5 2216121 9076 12 2 5 5670187 48969 12 4 5 2268020 9152 12 3 5 4886020 27385 12 4 5 2385575 9325 12 4 5 3151826 10471 12 4 5 2466182 9443 12 2 5 4056243 44447 12 4 5 2503795 9499 12 2 5 12574133 71618 12 4 5 2579040 9610 12 6 5 2678900 -2327 12 4 5 2590500 9627 12 4 5 3963163 11729 12 4 5 2708251 9802 12 4 5 2573947 9603 12 4 5 2779589 9909 12 3 5 3582103 24634 12 4 5 2784589 9917 12 4 5 3506780 11016 12 4 5 3549995 11083 12 3 5 11036450 42595 12 4 5 3801364 11475 12 2 5 12026411 69610 12 4 5 3807021 11483 12 4 5 1450000 7977 12 4 5 3829577 11519 12 5 5 4654347 4135 12 4 5 4034516 11842 12 4 5 3242621 10610 12 4 5 4464075 12530 12 4 5 15838532 36682 12 4 5 5094596 13563 12 3 5 18265045 66165 12 4 5 5642793 14487 12 2 5 3014988 41672 12 4 5 6464367 15915 12 2 5 4264549 45016 12 4 5 6878027 16654 12 2 5 2653938 40734 12 4 5 7311251 17444 12 4 5 18120206 43247 12 4 5 7613931 18005 12 4 5 7409819 17626 12 4 5 8458572 19613 12 2 5 11891327 69121 12 4 5 11074754 25011 12 2 5 1095064 36830 12 4 5 11639122 26263 12 4 5 17255530 40678 12 4 5 12969867 29349 12 2 5 5678555 48993 12 4 5 13301908 30149 12 4 5 3530091 11052 12 5 5 1353112 188 12 4 5 8642835 19972 12 5 5 1369000 205 12 2 5 3749467 43618 12 5 5 1838048 730 12 5 5 15014658 21551 12 5 5 2088477 1015 12 3 5 1276958 20124 12 5 5 2438951 1419 12 4 5 5974488 15057 12 5 5 2551034 1549 12 3 5 1655339 20834 12 5 5 2675634 1695 12 2 5 4427366 45463 12 5 5 2994084 2073 12 5 5 1304168 134 12 5 5 3559814 2757 12 2 5 9453764 60682

PAGE 176

176 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 5 5 4712443 4210 12 6 5 1465903 -3526 12 5 5 5236308 4897 12 5 5 6182568 6183 12 5 5 6355652 6425 12 2 5 1880918 38769 12 5 5 10244042 12427 12 6 5 2582481 -2425 12 6 5 1059299 -3913 12 6 5 5458885 701 12 6 5 1193000 -3786 12 4 5 3043333 10306 12 6 5 1285293 -3699 12 6 5 7118087 2715 12 6 5 1656000 -3343 12 6 5 6036580 1383 12 6 5 1681500 -3318 12 5 5 2363588 1331 12 6 5 1773294 -3229 12 1 5 19976496 134374 12 6 5 1801397 -3201 12 6 5 9062784 5296 12 6 5 1983602 -3023 12 6 5 3299408 -1686 12 6 5 2244528 -2764 12 2 5 2425750 40148 12 6 5 2295081 -2714 12 4 5 2785860 9918 12 6 5 2613474 -2393 12 6 5 3093632 -1901 12 6 5 2896351 -2105 12 6 5 3970908 -971 12 6 5 3434068 -1544 12 6 5 2870358 -2131 12 6 5 3788041 -1168 12 8 5 3292112 -1614 12 6 5 4123070 -805 12 2 5 4030182 44377 12 6 5 4622137 -254 12 2 5 8048887 56143 12 6 5 4651284 -221 12 4 5 5936356 14991 12 6 5 4939781 104 12 2 5 12142356 70032 12 6 5 5003451 176 12 6 5 8732398 4840 12 6 5 5141718 334 12 4 5 12925322 29242 12 6 5 5380545 610 12 6 5 5123326 313 12 6 5 5788574 1088 12 5 5 1188531 7 12 6 5 5955972 1287 12 6 5 11668806 9178 12 6 5 6038066 1385 12 2 5 3514007 42988 12 6 5 6117293 1480 12 4 5 6684147 16306 12 6 5 7204215 2824 12 6 5 4801228 -53 12 6 5 7231792 2859 12 6 5 3916572 -1029 12 6 5 7980439 3829 12 6 5 2557786 -2450 12 6 5 9035815 5259 12 4 5 2109345 8921 12 8 5 7638989 3535 12 6 5 4457620 -437 12 2 5 1632193 38149 12 4 5 4281816 12236 213 1 1 1176984 118190 12 4 5 3506513 11015 213 1 1 3966168 136097 12 2 5 1538183 37916 213 1 1 9660518 177864 12 6 5 2935859 -2064 213 1 1 15152448 224222 12 6 5 5035671 213 213 1 1 20157318 270409 12 3 5 1425784 20402 213 1 1 64798281 586385 12 4 5 1387705 7890

PAGE 177

177 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 1 1 68804811 597884 12 6 5 4487702 -404 213 2 1 4378142 127968 12 4 5 4520277 12621 213 2 1 8350269 154946 12 6 5 4687110 -181 213 2 1 172605686 654373 12 5 5 4846549 4384 213 3 1 1010681 104632 12 2 5 7114726 53251 213 3 1 4461728 123983 12 6 5 10427553 7265 213 4 1 1705100 112132 12 4 5 5289555 13889 213 4 1 2000000 113659 12 2 5 10626696 64651 213 4 1 2225000 114834 12 6 5 11870935 9501 213 4 1 2253922 114985 12 2 5 2776038 41050 213 4 1 2800000 117876 12 2 5 2529045 40413 213 4 1 3235799 120219 12 2 5 1985404 39032 213 4 1 3250000 120296 12 6 5 20018605 25783 213 4 1 3590000 122147 12 4 5 10006799 22728 213 4 1 4672090 128173 213 4 1 36353090 377278 213 4 1 5702552 134101 213 2 1 6800000 144010 213 4 1 6161370 136800 213 6 1 2530000 143506 213 4 1 14824949 194672 213 6 1 1650000 139092 213 4 1 15640539 200772 213 8 1 5450000 191306 213 4 1 107560004 642677 213 6 1 249217770 655065 213 6 1 1172671 136733 213 6 1 1197858 136857 213 6 1 1219682 136964 213 2 3 4000000 83508 213 6 1 1394389 137825 213 2 3 8929852 117576 213 6 1 1890036 140288 213 4 3 22071672 180607 213 6 1 2882629 145298 213 2 3 3043078 77653 213 6 1 3350924 147699 213 2 3 7012906 103548 213 6 1 4679205 154638 213 4 5 6634965 77910 213 6 1 7189047 168266 213 4 5 31292523 180782 213 6 1 22329791 263435 213 4 5 12397812 96464 213 8 1 1100000 170739 213 4 5 2000000 65000 213 8 1 1210300 171251 213 2 5 6000000 88202 213 8 1 1295000 171644 213 4 5 1460000 63602 213 8 1 2822338 178790 213 4 5 10276666 89292 213 8 1 5060804 189435 213 2 5 2700000 75703 213 2 3 1938884 71196 213 2 5 3470000 78506 213 2 3 2000000 71545 213 2 5 2798600 76058 213 2 3 3186742 78516 213 2 5 5421129 85917 213 2 3 3805883 82301 213 4 5 17276584 114591 213 2 3 4000000 83508 213 4 5 3342114 68566 213 2 3 5197159 91178 213 4 5 1591340 63940 213 2 3 6075993 97054 213 4 5 1264550 63102

PAGE 178

178 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 2 3 7423000 106466 213 2 5 2807739 76091 213 2 3 8861138 117056 213 2 5 1371870 71025 213 2 3 9118903 119013 213 2 5 2867000 76305 213 2 3 12728717 148244 213 2 5 29482249 217955 213 2 3 15605648 173843 213 2 5 5879000 87721 213 2 3 18981143 206121 213 6 5 32989965 172320 213 2 3 19593720 212198 213 2 5 1801000 72515 213 4 3 1301978 52861 218 4 1 1651718 61774 213 4 3 2217939 56766 218 6 1 4100000 38462 213 4 3 3917194 64412 218 4 1 30708210 300344 213 4 3 4011524 64852 218 4 1 2000000 63563 213 4 3 8343814 86879 218 4 1 39962233 387707 213 4 3 24640000 201780 218 4 1 3418477 71150 213 6 3 1365500 37516 218 4 1 1254161 59766 213 6 3 2944473 42128 218 4 1 1776062 62410 213 6 3 3447554 43651 218 4 1 1720000 62123 213 6 3 4804195 47893 218 2 3 6349622 50270 213 6 3 23527259 128649 218 4 3 24186687 93632 213 6 3 26958663 148231 218 2 3 49251652 231680 213 1 5 2022003 81940 218 2 3 1250000 36648 213 2 5 1007367 69776 218 4 5 30522524 150367 213 2 5 1067184 69980 218 4 5 9000000 50878 213 2 5 1197962 70427 218 4 5 52019051 287972 213 2 5 1286567 70732 218 4 5 10000000 54318 213 2 5 1345065 70933 218 4 5 1136429 27675 213 2 5 1382505 71062 218 4 5 5238917 38950 213 2 5 1423370 71203 12 3 3 3894753 21530 213 2 5 1467914 71357 12 4 3 13035225 60988 213 2 5 1581871 71752 12 4 3 3611991 25590 213 2 5 1856688 72710 12 4 3 14152837 66295 213 2 5 1870704 72759 12 4 3 17235573 82301 213 2 5 1930200 72968 12 6 3 20581900 107785 213 2 5 2514154 75036 12 2 3 13808116 49973 213 2 5 2846411 76231 12 2 3 10708107 37925 213 2 5 2964262 76657 12 4 3 10062782 48092 213 2 5 2989999 76750 12 2 3 2762067 14378 213 2 5 3749812 79541 12 2 3 13451516 48494 213 2 5 3948064 80280 12 4 3 2727098 23023 213 2 5 4000815 80478 12 4 3 10773729 51020 213 2 5 4382030 81914 12 4 3 18241179 87971 213 2 5 4428194 82089 12 4 3 12255163 57435

PAGE 179

179 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 2 5 5093984 84643 12 4 3 15362918 72335 213 2 5 5514468 86282 12 3 3 3168899 19544 213 2 5 5771377 87295 12 4 5 10045743 22810 213 2 5 6835002 91571 12 2 5 4658578 46103 213 2 5 8642282 99157 12 2 5 12160562 70098 213 2 5 9018000 100786 12 3 5 6976833 32123 213 2 5 9110435 101189 12 2 5 5049679 47198 213 2 5 9771581 104107 12 8 5 3438059 -1458 213 2 5 15034729 129379 12 4 5 9948188 22606 213 2 5 17170414 140699 12 4 5 4632350 12803 213 2 5 20314536 158503 12 2 5 9541401 60973 213 2 5 21444057 165230 12 2 5 8635874 58011 213 2 5 22927312 174326 12 2 5 1326196 37394 213 2 5 28107048 208358 12 2 5 1541191 37924 213 2 5 29549112 218427 12 2 5 6238811 50627 213 2 5 34124268 251885 12 2 5 2170201 39498 213 3 5 18154370 129402 12 2 5 9875095 62089 213 4 5 1008236 62450 12 2 5 2443824 40194 213 4 5 1031645 62509 12 2 5 3343393 42535 213 4 5 1041888 62535 12 3 5 3699005 24874 213 4 5 1050001 62556 12 2 5 2565163 40506 213 4 5 1107309 62701 12 2 5 3221396 42213 213 4 5 1150000 62810 12 2 5 7553487 54597 213 4 5 1326789 63261 12 2 5 8090665 56275 213 4 5 1353628 63330 12 4 5 12192709 27524 213 4 5 1411139 63477 12 2 5 8566395 57788 213 4 5 1458760 63599 12 6 5 4882596 39 213 4 5 1464803 63615 12 4 5 8580317 19850 213 4 5 1585790 63926 12 4 5 7224194 17284 213 4 5 1637464 64059 12 2 5 9590353 61136 213 4 5 1729804 64298 12 4 5 2190546 9039 213 4 5 1935954 64833 12 2 5 1650825 38195 213 4 5 1993452 64982 12 4 5 3407383 10862 213 4 5 1998870 64997 12 4 5 3858722 11565 213 4 5 2047680 65124 12 2 5 3798992 43752 213 4 5 2069499 65181 12 5 5 10881497 13525 213 4 5 2095207 65248 12 2 5 7778322 55295 213 4 5 2268244 65702 12 6 5 2550001 -2458 213 4 5 2272080 65712 12 4 5 7110781 17077 213 4 5 2700000 66843 12 5 5 4214954 3573 213 4 5 3462362 68892 12 3 5 17557184 63542

PAGE 180

180 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 213 4 5 3934828 70185 12 4 5 9301238 21281 213 4 5 4418140 71525 12 3 5 7991840 34578 213 4 5 5569697 74792 12 2 5 3359771 42579 213 4 5 6339884 77037 12 5 5 4948107 4517 213 4 5 6684233 78056 12 5 5 7680992 8344 213 4 5 7266605 79803 12 4 5 10308880 23363 213 4 5 7403865 80219 12 6 5 1472917 -3519 213 4 5 9239147 85932 12 4 5 3469388 10958 213 4 5 10049427 88547 12 4 5 1463715 7997 213 4 5 10560260 90227 12 4 5 5631627 14468 213 4 5 12350000 96297 12 2 5 2238403 39671 213 4 5 14034501 102285 12 3 5 2212977 21902 213 4 5 14512760 104035 12 4 5 3807129 11484 213 4 5 14663196 104590 12 4 5 1047006 7415 213 4 5 14847933 105275 12 2 5 2351420 39958 213 4 5 15520418 107795 12 4 5 1805057 8482 213 4 5 19820798 125000 12 2 5 3525944 43020 213 4 5 22572609 137035 12 6 5 2418583 -2590 213 4 5 27901688 162738 12 6 5 2007775 -2999 213 4 5 34099959 196723 12 5 5 2049415 970 213 4 5 36831121 213088 12 2 5 1567142 37988 213 4 5 154568188 652200 12 4 5 1914598 8639 213 6 5 3243778 77096 12 4 5 8562635 19816 218 4 1 1247256 59732 12 3 5 10310008 40589 218 4 1 1324179 60117 12 3 5 1461621 20469 218 4 1 1492789 60967 12 5 5 32146974 79570 218 4 1 1770818 62383 12 2 5 3445881 42807 218 4 1 2114386 64157 12 2 5 8359161 57126 218 4 1 2821967 67901 12 4 5 2436333 9399 218 4 1 3274450 70358 12 4 5 11895284 26842 218 4 1 3919380 73946 12 4 5 8488886 19672 218 4 1 35395422 345885 12 2 5 3577976 43159 218 4 1 113915158 628209 12 2 5 1143486 36948 218 2 3 14659298 76297 12 2 5 1415341 37613 218 2 3 23023341 107254 12 2 5 9309746 60206 218 4 3 1362051 21214 12 3 5 5519522 28777 218 4 3 1687559 21948 12 2 5 1041410 36699 218 4 3 17869800 69321 12 2 5 8474952 57495 218 2 5 3589000 63354 12 3 5 1270504 20112 218 2 5 5720000 71328 12 1 5 12709323 99536 218 2 5 7483865 78276 12 8 5 1438210 -3508

PAGE 181

181 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 218 2 5 7933333 80097 12 2 5 3205411 42171 218 2 5 11091826 93453 12 2 5 2301684 39832 218 2 5 63137848 391013 12 2 5 1201000 37088 218 4 5 1013299 27364 213 4 1 13100142 182127 218 4 5 2178065 30374 213 4 1 4659746 128103 218 4 5 2500000 31230 213 2 1 6800000 144010 218 4 5 3223705 33193 213 6 1 1655000 139117 218 4 5 3615000 34277 213 6 1 1900000 140337 218 4 5 3825000 34866 213 8 1 1658000 173334 218 4 5 4815000 37703 213 8 1 1610000 173110 218 4 5 5010000 38274 213 8 1 5725000 192631 218 4 5 5352000 39285 213 3 1 107722494 644256 218 4 5 5752350 40486 213 4 1 25671605 283062 218 4 5 5857787 40805 213 2 3 4000000 83508 218 4 5 8271598 48444 213 2 3 1205000 67080 218 4 5 8683500 49813 213 2 3 8227929 112324 218 4 5 22880098 109099 213 4 3 1593100 54086 218 4 5 36115970 183937 213 4 3 16207313 136240 218 5 5 4500000 26086 213 2 3 1500000 68718 12 1 1 11700000 120665 213 2 3 1500000 68718 12 4 1 9971149 62521 213 2 3 54238235 524980 12 4 1 2121593 37318 213 2 3 77981699 609760 12 4 1 4155416 43206 213 2 3 6550000 100310 12 4 1 7876752 55124 213 2 3 2208874 72746 12 6 1 5729692 26140 213 2 3 1366415 67974 12 6 1 6892822 29580 213 2 3 5143562 90826 12 6 1 16611070 66562 213 6 3 7159000 55731 12 6 1 5323553 24983 213 4 3 5278613 70923 12 8 1 6723957 16006 213 4 3 2760000 59148 12 2 3 6851870 25298 213 2 5 31792782 234563 12 2 3 5799141 22261 213 4 5 15834468 108987 12 2 3 3055488 15085 213 2 5 17259040 141182 12 4 3 4288545 27631 213 4 5 3415392 68765 12 4 3 2969782 23716 213 4 5 6435287 77319 12 4 3 10730063 50838 213 2 5 9000000 100707 12 4 3 15776518 74470 213 4 5 34886044 201348 12 4 3 10158057 48479 213 2 5 54451658 413033 12 4 3 5302399 30822 213 4 5 17176264 114194 12 4 3 15106943 71031 213 4 5 81580986 522798 12 4 3 1933328 20816 213 4 5 4666666 72221 12 4 3 3121365 24153 213 4 5 3443000 68840

PAGE 182

182 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 4 3 6899607 36182 213 4 5 10000000 88386 12 4 3 7561050 38526 213 2 5 30384990 224373 12 4 3 18425428 89034 213 2 5 21307263 164406 12 5 3 11614275 59429 213 2 5 3080000 77078 12 1 5 1081000 57124 213 4 5 8118107 82408 12 2 5 8922328 58938 213 2 5 49201128 371634 12 2 5 6801311 52302 213 4 5 2561964 66476 12 2 5 5388453 48160 213 4 5 35536766 205229 12 2 5 3633106 43306 213 2 5 1371870 71025 12 2 5 4070526 44486 213 4 5 3300000 68452 12 2 5 5014437 47099 213 4 5 9269484 86029 12 2 5 5669009 48965 213 4 5 4000000 70365 12 2 5 3429408 42763 213 2 5 7036116 92395 12 2 5 2206832 39591 218 4 1 116245584 629666 12 2 5 2613927 40631 218 4 1 8029224 99232 12 2 5 1992999 39051 218 4 1 10195255 114261 12 2 5 8080552 56243 218 4 1 1656480 61798 12 2 5 3408976 42709 218 2 3 5973235 49204 12 2 5 2427113 40152 218 4 3 58011014 262813 12 2 5 6100975 50222 218 2 5 5000000 68583 12 2 5 6285013 50763 218 4 5 9000000 50878 12 2 5 7775748 55287 218 2 5 1237030 55088 12 2 5 6106969 50239 218 4 5 9599233 52925 12 2 5 6576341 51628 218 2 5 3000000 61231 12 2 5 2012039 39099 12 2 5 6197850 50506 12 2 5 18154133 94372 12 4 5 4705619 12922 12 2 5 2325152 39891 213 2 1 6800000 144010 12 2 5 7306531 53837 213 3 1 4000000 121258 12 2 5 2689540 40826 213 4 1 29847551 319870 12 2 5 7077000 53136 213 6 1 10100000 184904 12 2 5 4658421 46103 213 6 1 200200313 654473 12 2 5 1852829 38699 213 6 1 1550000 138596 12 2 5 19031224 98348 213 6 1 10900000 189628 12 2 5 7186507 53469 213 6 1 4000000 151066 12 2 5 16362915 86597 213 8 1 94937616 588618 12 2 5 5848411 49485 213 8 1 14575435 236647 12 2 5 2874078 41304 213 8 1 6400000 195897 12 2 5 1364005 37487 213 8 1 31429350 325220 12 2 5 2733693 40940 213 2 3 4000000 83508 12 2 5 5303453 47917 213 2 3 42589140 442355 12 2 5 3031048 41714 213 2 3 3927175 83054

PAGE 183

183 Table B-9. Continued WMC Lanes TS Cost Pred. qty. WMCLanes TS Cost Pred. qty. 12 2 5 7028045 52987 213 2 3 56687098 538312 12 2 5 2121546 39375 213 2 3 4895000 89206 12 2 5 4811940 46531 213 2 3 4647500 87609 12 2 5 2029521 39143 213 2 3 2600000 75024 12 2 5 6813118 52338 213 2 3 6050000 96877 12 2 5 7024236 52976 213 2 3 40045877 420061 12 2 5 1217713 37129 213 4 3 5000000 69562 12 2 5 1044000 36706 213 6 3 2000000 39339 12 2 5 1108000 36861 213 6 3 15000500 86489 12 2 5 1447000 37691 213 2 5 20000000 156661 12 2 5 1101000 36844 213 2 5 8000000 96414 12 3 5 2481464 22426 213 2 5 1900000 72862 12 3 5 7271988 32826 213 2 5 27353088 203193 12 3 5 15207399 55348 213 2 5 1371870 71025 12 3 5 23734302 89059 213 2 5 37620116 278758 12 3 5 4381233 26302 213 3 5 12004509 102445 12 3 5 2366012 22200 213 3 5 8000000 87190 12 3 5 4803267 27206 213 4 5 1994727 64986 12 3 5 1121820 19835 213 4 5 3825238 69884 12 3 5 2277784 22028 213 4 5 2000000 65000 12 3 5 2304994 22081 213 4 5 3518000 69044 12 3 5 3785185 25052 213 4 5 30800000 178079 12 3 5 1742625 21000 213 4 5 10000000 88386 12 3 5 4585727 26738 213 4 5 42598665 250272 12 4 5 4655524 12840 213 4 5 8000000 82043 12 4 5 3034535 10293 213 4 5 12715195 97573 12 4 5 3513724 11027 213 4 5 5330000 74103 12 4 5 2318643 9226 213 6 5 25000000 139897 12 4 5 2703891 9796 218 4 1 3917121 73933 12 4 5 1631462 8234 218 4 1 4956852 79932 12 4 5 4488534 12569 218 4 1 2231848 64771 12 4 5 6141414 15347 218 4 1 1500000 61003 12 4 5 1263699 7716 218 2 5 5000000 68583 12 4 5 4182438 12077 218 2 5 1280326 55235 12 4 5 2125444 8944 218 2 5 5000000 68583 12 4 5 1640535 8247 218 4 5 9000000 50878 12 4 5 5159874 13672 218 4 5 2000000 29905 12 4 5 1419851 7935 218 5 5 5568318 28724 12 4 5 3596110 11154 12 4 5 6393357 15789 12 4 5 4485447 12564 12 4 5 3016096 10265 12 4 5 1515518 8070 12 4 5 2662205 9734

PAGE 184

184 LIST OF REFERENCES Adeli, H. and Hung, S. L. (1995) An adaptive conjugate gradient l earning algorithm for efficient training of neural networks. Applied Mathematics and Computations, 62 (1), 81-102. Adeli, H., and Karim, A. (2001) Construction Scheduling, Cost Optimization, and Management, Spon Press, New York, NY. Agapiou, A., Clausen, L. E., Flanagan R., Norm an, G., and Notman, D. (1998) the role of logistics in the materials flow control process. Construction Management and Economics, 16, 131-137. Agdas, D. (2006) An XML based solution proposal to the data management issues within the construction industry. Masters thesis, Un iversity of Florid a, Gainesville, FL. Akintoye, A. (1995) Just-in-time applicati on and implementation for building material management. Construction Management and Economics, 13, 105-113. Ambrogi, S. (2007) Sea freight co st for dry raw materials near record, Reuters, retrieved from http://uk.reuters.com/articlePrint?articleId=UKL242024720070424 at 04/01/2008. Amm ar, A. M., and Mohieldin, Y. A. (2002) Resource constrained project scheduling using simulation. Construction Management and Economics, 20, 323-330. Baye, M. R. (2006) Managerial Economics and Business Strategy, McGraw-Hill, New York, NY. Bennett, F. L. (2003) The Management of Construction, Butterworth Heinemann, Burlington, MA. Boussabaine, A. H. (1996) The use of artificial neural networks in construction management: a review. Construction Management and Economics, 14, 427-436. Dunphy, T. (2006) Evening the playing fiel d,Aggregate Manager, retrieved from http://www.aggman.com/articles/aug06b.htm 07/22/2008. Ellis, R. D., Hoit, M., an d McVay, M. (2008) Integration of XML schemas for the exchange of FDOT construction project data, research report prepared for F DOT, University of Florida, Gainesville, FL. FDOT Specifications and Estimates Office (2007) Presentation: Rising c onstruction costs, the Florida story. Proceedings of National Highway Construction Cost Workshop, St. Louis, MO. Flood, I. (1991) A Gaussian-based feedforwar d network architecture and complementary training algorithm. Int. Joint Conf. on Neural Networks, IEEE, Washington, D.C., Vol. I, 171176.

PAGE 185

185 Flood, I., and Kartam, N. (1994) Neural networks in civil engineering. I: Principles and understanding. Journal of Computing in Civil Engineering, 8 (2), 131-148. Gagarin, N., Flood, I., and Albrech t, P. (1993) Computing truck attributes with artificial neural networks. Journal of Computing in Civil Engineering, 8 (2), 179-200. Garza, J. M., and Rouhana, K. G. (1995) Ne ural networks versus parameter-based applications in cost engineering. Cost Engineering, 37 (2), 14-18. Goldberg, D. E. (1989) Genetic Algorithms in Search, Optimization, and Machine Learning, Addison-Wesley, Reading, MA. Hecht-Nielsen, R. (1989) Theory of the backpropogation neural network Proceedings of international Joint Conference on Neural Netw orks, IEEE, Washington, D.C., Vol. I, 593-605. Hegazy, T. (1999) Optimization of resource allocation and leveling using genetic algorithms. Journal of Construction Engineering and Management, 125 (3), 167-175. Hua, G. B. (1996) Residential construction demand forecasting using economic indicators: a comparative study of artificia l neural networks and multiple regression. C onstruction Management and Economics, 14, 25-34. Hua, G. B. (2000)valuating the performance of combining neural networks and genetic algorithms to forecast construction demand: the case of Singapore residential sector. Construction Management and Economics, 18, 209-217. Jang, H., Russell, J. S., and Yi, J. S. (2003) A project managers level of satisfaction in construction logistics, Canadian Journal of Civil Engineering, 30, 1133-1142. Jiang, S., and Shi, J. (2005) Exact algorith m for solving project scheduling problems with multiple resource constraints. Journal of Construction Engineering and Management, 131 (9), 986-992. Kamarthi, S. G., Sanvido, V. E., and Kumara, S. R. T. (1992) Neuroform-Neural network system for vertical formwork selection. Journal of Computing in Civil Engineering, 6 (2), 178-199. Kandil, A., and Rayes, K. E. (2006) Parallel genetic algorithms for optimizing resource utilization in large-scale construction projects. Journal of Construction Engineering and Management, 132 (5), 491-498. Kim, G. H., Seo, D. S., and Kang, K. I. (2005) Hybrid models of neural networks and genetic algorithms for predicting preliminary cost estimates. Journal of Computing in Civil Engineering, 19 (2), 208-211.

PAGE 186

186 Lampl Herbert Consultants (2007) Strategic a ggregates study: Sources constraints, and economic value of limestone and sand in Florida, research report prepared for FDOT, Tallahassee, FL. Lapedes, A., and Farber, R. (1988) How neural networks work, Neural Information Processing Systems, American Institute of Physics, 442-456. Leu, S. S., and Yang, C.H. (1999) A geneticalgorithm-based resource-constrained construction scheduling system. Construction Management and Economics, 17, 767-776. Li, H., and Love, E. D. (1998) Site-level facilities layout using genetic algorithms. Journal of Computing in Civil Engineering, 12 (4), 227-231 London, K. A., and Kenley R. (2001) An i ndustrial organization economic supply chain approach for the construction industry: a review. Construction Management and Economics, 19, 777-788 Madan, A. (2006) General appro ach for training back-propaga tion neural networks in vibration control of multi-degree-of-freedom structures. Journal of Computing in Civil Engineering, 20 (4), 247-260 Marzouk, M., and Moselhi, O. (2002), simu lation optimization for earth moving operations using genetic algorithms. Construction Management and Economics, 20, 535-543 Matthews, R. G. (2007) Ship shortage pushes up pr ices of raw material, Th e Wall Street Journal, retrieved from http://online.wsj.com/artic le/SB119301028152866449.htm l?mod=googlenews_wsj 04/03/2008 Mentzer, J. T. (2001) Supply Chain Management, Sage Publications, Thousand Oaks, CA. Michalewicz, Z. (1994) Genetic Algorithms + Data Structures = Evolution Programs, Springer-Verlag, Heidelberg, NY. Naim, M., and Barlow, J. (2003) An innovative supply chain strategy for customized housing. Construction Management and Economics, 21, 593-602. Nicolini, D.,Holti, R., and Smalley, M. (2001) Inte grating the project activities: the theory and practice of managing the s upply chain through clusters. Construction Management and Economics, 19, 37-47 Rumelhart, D. E. (1986) Parallel Distributed Proce ssing, Volume 1: Foundations, the MIT Press, Cambridge, MA.

PAGE 187

187 Senouci, A. B., and Adeli, H. (2001) Res ource scheduling using neural dynamics model of Adeli and Park. Journal of Construction Engineering and Management, 127 (1), 2834 The Balmoral Group (2007), Strategic res ource evaluation study: Highway construction materials, research report prepared for FDOT, Maitland, FL. Thomas, H. R., Horman, M. J. De Souza, U. E. L., and Zavrski, I. (2002 ) Reducing variability to improve performance as a lean construction principle. Journal of Constr uction engineering and Engineering Management, 128-2, 144-154 Tommelein, I. D., and Li, A. E. Y. (1999) Just-i n-time delivery: mapping alte rnatives for vertical supply chain integration. Proceedings of the Se venth Annual Conference of the International Group for Lean Construction, Berkeley, CA. Tsai, D. M., and Chiu, H. N. (1996) Two heuris tics for scheduling multiple projects with resource constraints. Construction Management and Economics, 14, 325-340 Vonk, E., Jain, L. C., and Johnson, R. P. (1997) Automatic Generation of Neural Network Architecture Using Evolutionary computation, World Scientific, River Edge, NJ Vrijhoef, R., and Koskela, L. (1999) Roles of supply chain management in construction. Proceedings of the Seventh Annual Conference of the Inte rnational Group for Lean Construction, Berkeley, CA.

PAGE 188

188 BIOGRAPHICAL SKETCH Duzgun Agdas was born on in 1983, in Tunceli, Turkey. After com pleting his high school education, he studied at the civi l engineering department of Mi ddle East Technical University (METU) in Ankara, Turkey. Upon getting his Bachel or of Science degree in civil engineering in June 2005, he was admitted to the University of Florida for graduate studies. After getting his Master of Engineering degree in December 2006, he remained at UF in the same department. While continuing his PhD studies, he was admitted to the University of Floridas MBA program from which he graduated in May 2008. He comple ted his graduate education by getting his PhD in civil engineering and plans to put his academ ic knowledge into use as a professional and a scholar.