Service Life of Concrete and Metal Culverts Located in Ohio Department of. Transportation Districts 9 and 10. A thesis presented to.

Transcription

1 Service Life of Concrete and Metal Culverts Located in Ohio Department of Transportation Districts 9 and 10 A thesis presented to the faculty of the Russ College of Engineering and Technology of Ohio University In partial fulfillment of the requirements for the degree Master of Science Gabriel J. Colorado Urrea December Gabriel J. Colorado Urrea. All Rights Reserved.

2 2 This thesis titled Service Life of Concrete and Metal Culverts Located in Ohio Department of Transportation Districts 9 and 10 by GABRIEL J. COLORADO URREA has been approved for the Department of Civil Engineering and the Russ College of Engineering and Technology by Shad M. Sargand Russ Professor of Civil Engineering Dennis Irwin Dean, Russ College of Engineering and Technology

3 3 ABSTRACT COLORADO URREA, GABRIEL J., M.S., December 2014, Civil Engineering Service Life of Concrete and Metal Culverts Located in Ohio Department of Transportation Districts 9 and 10 Directors of Thesis: Shad M. Sargand and Teruhisa Masada In this study, in-service conditions were evaluated to estimate the service life of concrete and metal culverts. The Ohio Research Institute for Transportation and the Environment (ORITE) and a private consulting company proposed new inspection methods and rating procedures for concrete, metal and thermoplastic pipes; concrete and metal culverts are addressed in this study. The inspection activities were developed in culverts located in Ohio Department of Transportation (ODOT) Districts 9 and 10 since the aggressive environmental conditions found in these portions of the state of Ohio. Before each field trip, culverts were selected to meet requirements of location, material, and dimensions. For dimensions, screening criteria were than 42 inch span and rise dimensions, and a maximum length of 150 feet. From each culvert, basic information was gathered from the inventory data provided by ODOT and in the field. The data gathered from the inventory and the field work was statistically analyzed to identify significant factors that contribute to material deterioration. The rating scales proposed by the ORITE and ODOT were employed in the statistical regressions as outcome variables, to measure the effectiveness and accuracy in predicting the remaining service life. Multivariable linear and nonlinear regression

4 4 models were proposed to estimate the remaining service life of existing metal and concrete structures with similar conditions in the state of Ohio. Results for concrete culverts show that the multivariable linear regression results showed that ph and resistivity of water were significant for the ODOT and ORITE rating scale but the linear model is not since the age is not included in the regression. While, the multivariable non-linear regression results indicated that ph of water, age and span were significant based on the ODOT rating scale. For metal culverts, the multivariable linear regression results showed that rise, span, age, level of abrasion, thickness of the plate, slope, velocity, and depth of the flow were significant based on the ODOT rating scale. And, age, soil cover, level of abrasion, ph of water, thickness of the plate, slope, flow velocity, and depth of flow were all significant for the ORITE rating scale, both models are not practical in estimating the deterioration of metal pipes. Non-linear regression did not generate more reliable results in predicting the service life of metal culverts.

5 5 DEDICATION To my mother and my uncle, their guidance and support made all of this possible.

6 6 ACKNOWLEDGMENTS I would like to thank to Dr. Shad Sargand for giving me the opportunity and the financial support to pursuit my master s degree in Ohio University. Also, I want to thank to the committee members Dr. Deborah McAvoy for her contribution and advice in the statistical analysis and Dr. Martin Mohlenkamp for her valuable contribution to the in this thesis. I want to express especial gratitude to Dr. Teruhisa Masada, John Hurd and Issam Khoury for his guidance and contribution during the field work and the process of this research. Finally, I want to express my deeply gratitude to my family and friends to their support during these two years.

7 7 TABLE OF CONTENTS Page Dedication... 5 Acknowledgments... 6 List of Tables List of Figures Chapter 1: Introduction Background Objectives Thesis Outline Chapter 2: Literature Review NCHRP Syntheses Rating Systems Field Performance Studies Durability Studies Concepts of Multivariable Linear Regression Variable Selection Method Assumptions of the Linear Regression Analysis Non-Linear Regression Chapter 3: Culvert Data Collection Form Inventory Data Concrete Pipe Metal Culvert Field Data Concrete Pipes Metal Pipes Concrete and Metal Pipes Culvert Rating Systems ODOT General Scale ORITE General Scale... 61

8 ODOT Joint Scale Settlement Rating Scale Abrasion Levels Chapter 4: Inventory and Field Inspection Data Selection and Localization of Culverts Inspected Inventory Data Characteristics Length Pipe Shape Span Age Type of Protection Soil Cover Height Corrugation Profile Field Data Characteristics Pipe Slope Characteristics Depth and Velocity of Flow Depth of Sediment Characteristics of Water Properties of Backfill Soil Rating Scale Classifications Basic Analysis of Field Data Basic Analysis of Concrete Culverts Basic Analysis of Metal Culverts Chapter 5: Statistical Analysis of Field Data Linear Regression Analysis Linear Regression of Concrete Culverts Linear Regression of Metal Culverts Non-Linear Regression Analysis Non-Linear Regression of Concrete Culverts Non-Linear Regression of Metal Culverts Chapter 6: Summary and Conclusions

9 9 6.1 Concrete Culverts Metal Culverts Recommendations for Future Work References Appendix A: Data Collection Forms Appendix B: Inventory and Field Collected Data

10 10 LIST OF TABLES Page Table 1.1. Culvert Materials, Types and Shapes Table 3.1. County Number and Districts (ODOT, 2003) Table 3.2. ODOT Concrete Pipe Rating Scale (ODOT, 2003) Table 3.3. ODOT Metal Rating Scale (ODOT, 2003) Table 3.4. ORITE Concrete Pipe Rating Scale Table 3.5. ORITE Metal Pipe Rating Scale Table 3.6. ODOT Joint Rating Scale (ODOT, 2003) Table 3.7. ORITE Settlement Rating Scale Table 3.8. Abrasion Levels from Highway Design Manual CHP850 (DeCou and Davis, 2007) Table 4.1. Concrete Culverts Inspected Table 4.2. Metal Culverts Inspected Table 4.3. Length Classification per Material Table 4.4. Shape Classification per Material Table 4.5. Span Classification per Material Table 4.6. Age Classification per Material Table 4.7. Type of Protection Classification Table 4.8. Soil Cover classification per Material Table 4.9. Slope Classification per Material Table Depth of Flow Classification per Material Table Flow Velocity Classification per Material... 78

11 11 Table Depth of Sediment per Material Table Water ph Classification per Material Table Resistivity of water Classification per Material Table Chloride and Sulfate Classification Table ph and Resistivity of Soil Classification Table General Condition Ratings of Concrete Culverts Table General Condition Ratings of Metal Culverts Table Inlet and Outlet Condition Values for Concrete Culverts Table Inlet and Outlet Condition Values for Metal Culverts Table Joint Ratings of Culverts Table Settlement Condition per Material Table Abrasion Level Classification per Material Table Basic Statistics of Concrete Culverts Table Descriptive Statistics of Metal Culverts Table 5.1. Variables for Concrete Culverts Table 5.2. Concrete Linear Model Coefficients (ODOT Rating Scale) Table 5.3. ANOVA Concrete (ODOT Rating Scale) Table 5.4. Concrete Linear Model Statistic Values (ODOT Rating Scale) Table 5.5. Concrete Linear Model Coefficients (ORITE Rating Scale) Table 5.6. ANOVA Concrete (ORITE Rating Scale) Table 5.7. Concrete Linear Model Statistic Values (ORITE Rating Scale) Table 5.8. Variables of Metal Culverts

12 12 Table 5.9. Service Life Add-on for Supplemental Pavings and Coatings (National Corrugated Steel Pipe Association, 2008) Table FHWA Abrasion Levels (Retrieved from: National Corrugated Steel Pipe Association (NCSPA), 2008) Table Relationship between Levels of Abrasion Table Age of Culverts with Supplemental Coating Table Metal Linear Model Coefficients (ODOT Rating Scale) Table ANOVA Metal (ODOT Rating Scale) Table Metal Linear Model Statistic Values (ODOT Rating Scale) Table Metal Linear Model Coefficients (ORITE Rating Scale) Table ANOVA Metal (ORITE Rating Scale) Table Metal Linear Model Statistic Values (ORITE Rating Scale) Table Concrete Non-Linear Model Coefficients (ODOT Rating Scale) Table Concrete Non-Linear Model Statistic Values (ODOT Rating Scale) Table Concrete Non-Linear Model Coefficients (ORITE Rating Scale) Table Concrete Non-Linear Model Statistic Values (ORITE Rating Scale) Table Metal Non-Linear Model Coefficients (ODOT Rating Scale) Table Metal Non-Linear Model Statistic Values (ODOT Rating Scale) Table Metal Non-Linear Model Coefficients (ORITE Rating Scale) Table Concrete Non-Linear Model Statistic Values (ORITE Rating Scale) Table B.1. Concrete Pipe Inventory and Field Data Table B.2. Concrete Pipe Rating Scores

13 Table B.3. Metal Pipe Inventory and Field Data Table B.4. Metal Pipe Rating Scores

14 14 LIST OF FIGURES Page Figure 1.1. Culvert Shape, Diameter, and Total Span Illustrated (ODOT, 2003) Figure 2.1. Maintenance Urgency Index (NCHRP, 1982) Figure 3.1. HD150 Distance Meter Figure 3.2. PC 300 for ph and resistivity measurements (Eutech Instruments Pte Ltd/ Oakton Instruments, 2008) Figure 3.3. Dionex IC25 Ion Chromatograph Figure DL Plus Ultrasonic Thickness Gage. ( 52 Figure 3.5. ph Measurement of the Soil with the PC Figure 3.6. Miller 400D Soil Box (M.C. Miller Co., 2010) Figure 3.7. Resistivity measurement with the Miller 400D Soil Box Figure Magnetic Digital Laser Level ( 56 Figure 3.9. SK202 Telescopic Fiberglass Measuring Pole ( 57 Figure Separated joint in a concrete pipe Figure 4.1. Water ph based on ph Values of Individual Culverts (ODOT, 1982) Figure 4.2. Percent of Abrasive Flow per County (ODOT, 1982) Figure 4.3. Geographical Location of Concrete culverts Inspected Figure 4.4. Geographical Location of Metal culverts Inspected Figure 4.5. Bed Load of a Concrete Pipe... 79

15 15 Figure 4.6. Loss of Mortar and Aggregate in a Concrete Culvert Figure 4.7. Metal culvert with extensive corrosion and pitting Figure 4.8. Separated Joint in a Concrete Culvert Figure 4.9. Metal Pipe with a Dropped off End Section Figure Metal Pipe with Abrasion Level Figure Influence of Age on ORITE Rating Score (Concrete Culverts) Figure Influence of Age on ORITE Rating Score (Metal Culverts) Figure 5.1. Standardized Predicted against Residual (ODOT Concrete Rating Scale) Figure 5.2. Histogram of Frequencies (ODOT Concrete Rating Scale) Figure 5.3. Normal Probability Plot (ODOT Concrete Rating Scale) Figure 5.4. Standard Predicted against Residual (ORITE Concrete Rating scale) Figure 5.5. Histogram of Frequencies (ORITE Concrete Rating Scale) Figure 5.6. Normal Probability Plot (ORITE Concrete Rating Scale) Figure 5.7. Standardized Predicted against Residual (ODOT Metal Rating Scale) Figure 5.8. Histogram of Frequencies (ODOT Metal Rating Scale) Figure 5.9. Normal Probability Plot (ODOT Metal Rating Scale) Figure Standardized Predicted against Residual (ORITE Metal Rating Scale) Figure Histogram of Frequencies (ORITE Metal Rating Scale) Figure Normal Probability Plot (ORITE Metal Rating Scale)

16 16 CHAPTER 1: INTRODUCTION 1.1 Background A culvert can be defined as a channel that maintains the continuity of a stream when it encounters an artificial obstacle such as an embankment, roadway, or levee according to the National Corrugated Steel Pipe Association (NCSPA, 2008). The shapes utilized in culvert construction are: Round pipe, pipe arch, elliptical pipe, and box. More specifically, the Ohio Department of Transportation (ODOT, 2003), defines a culvert as a structure with a pipe diameter, box, elliptical or arch span, or multi-cell having a total span less than 10 feet as shown in Figure 1. In recent years, concrete, metal, and plastic are the typical manufacturing materials for culverts. Table 1 shows the different types of culverts considered for each of those materials. Figure 1.1. Culvert Shape, Diameter, and Total Span Illustrated (ODOT, 2003)

17 17 Table 1.1. Culvert Materials, Types and Shapes Culvert Material Type Shape Concrete Cast-in-Place Slab on top Box Precast Elliptical Circular Pipe Metal Corrugated Metal Plate Circular Pipe Structural Plate Pipe Arch Thermoplastic PVC HDPE Circular Pipe Concrete and metal culverts corrode due to the environment they are exposed in the field. The corrosion of drainage structures is produced by acidity, alkalinity, dissolved salts, and other chemical factors presented in soil and water, these factors may be carried by groundwater, runoff waters, rain, and marine environments, affecting the service life of metal and concrete structures. The ODOT developed The Culvert Management Manual (2003) to maintain regular control of the culvert field performance and identify possible maintenance, repair, or rehabilitation activities on the culvert structures. This manual provides a guideline for inventory, inspection, and maintenance management of culverts with less than 10-foot span. ODOT (2003) requires that culverts should be inspected every five years unless a known deficiency has been detected. Additionally, ODOT established a comprehensive rating system for metal, concrete, plastic, and masonry drainage structures. Mitchell et al. (2005) performed a study to evaluate the rating system adopted by the ODOT (2003). A total of 60 culverts were inspected in Ohio: 25 concrete, 25 metal, and 10 thermoplastic. The authors proposed a new rating system for concrete, metal, and plastic culvert structures and executed a multivariable linear and nonlinear regression.

18 18 The results showed that the new rating systems proposed by Mitchell et al. (2005) reached a greater adjusted R 2 value for concrete and metal culverts compared to the systems proposed by the ODOT (2003). No statistical analyses were performed for thermoplastic pipes due to insufficient data. 1.2 Objectives The present study is part of a project proposed by the Ohio Research Institute for Transportation and Environment (ORITE) and a private consulting company to ODOT. The current thesis study is concerned with highway culverts in ODOT Districts 9 and 10 located in southeastern region of the state of Ohio. Concrete and metal culverts were inspected to: Ensure that all the influential factors identified by other researchers will be measured during the field inspections for concrete and metal pipes; and Develop multivariable regression models to estimate the service life of concrete and metal pipes based on the rating systems proposed by the ODOT and the ORITE. 1.3 Thesis Outline Chapter 1 presents the introduction, which includes a background about the ODOT Culvert Management Manual (2003) and the last research performed by Mitchell et al. (2005) to update that manual. The objectives of the present study and thesis outline are stated in this chapter. Chapter 2 provides a summary of a literature review on the past durability and field performance studies that were conducted for concrete and metal pipes.

19 19 Chapter 3 describes in detail the inspection forms for concrete and metal culverts utilize on the field work. Chapter 4 describes the locations chosen to develop the inspections activities in the state of Ohio and the characteristics that existed in the field inspection data collected in the study. Chapter 5 presents the descriptive analysis and statistical results of the multivariable linear and nonlinear regression analyses to estimate the service life of concrete and metal culverts. Chapter 6 provides the summary and conclusions of the research study.

20 20 CHAPTER 2: LITERATURE REVIEW Drainage culverts represent an essential infrastructure system which should not be ignored. Ensuring the stability of these structures over time provides safer traveling conditions for public commuters. However, the natural process of corrosion, abrasion and erosion affect the durability of drainage culverts. Studies about rating systems, field performance, durability, service life, and risk assessment of concrete and metal pipes have been conducted by different authors and transportation agencies. This chapter presents some of these studies. 2.1 NCHRP Syntheses The NCHRP (1978) developed an eight chapter synthesis to explain factors, materials, coatings, and procedures that affect the durability of culverts. The definition of corrosion process and the main environmental factors in soil and water were explained. The pipe materials analyzed in the study were identified as steel (including galvanized steel and aluminum) and concrete. Specifications for implementing those materials were stated in the study. The NCHRP (1978) synthesis 50 described coating techniques and materials used for concrete and metal culverts. Extra thickness, bituminous coating, bituminous-paved invert, and metallic coatings were characterized by their beneficial and disadvantageous properties for corrosion protection of drainage culverts. A procedure for inspecting culvert structures was also introduced. The rating systems developed by some transportation agencies were illustrated for concrete, steel, and aluminum pipes to determine in a qualitative or quantitative mode the deterioration of the structure. Of these

21 21 methods, the California method was described, which is used to estimate the service life of galvanized steel culverts. The influence of ph and resistivity of water or soil, factors utilized by the California method, are studied by some agencies and explained in this report. Some of the factors found a satisfactory agreement to estimate the culvert service life, but others did not find significant degree of association seen in the California method with corrosion rate. Guidelines of durability were established for inspection and maintenance of existing structures and design for new culvert pipes (National Cooperative Highway Research Program (NCHRP) Synthesis 50, 1978). To establish criteria for maintenance, the NCHRP (1982) defined an urgency index to prioritize the maintenance of bridge substructure elements below the waterline. The proposed guideline was developed as an adjunction to the system produced by the Federal Highway Administration in the Structure and Inventory Appraisal (SI&A). Figure 2.1 shows the system described by the NCHRP (1982) for maintenance of structures, which varies from 9 for No repair needed to 0 for facility is closed for repairs.

22 22 Figure 2.1. Maintenance Urgency Index (NCHRP, 1982) Gabriel and Moran (1998) investigated the updates of materials, coatings and practices to estimate the service life of drainage pipes in the Synthesis 254, 19 years after NCHRP released Synthesis of Highway Practice 50. Synthesis 254 was comprised of five chapters. The mechanisms that produce corrosion were again explained. Soil and water chemical factors along with additional conditions such as erosion, abrasion, precipitation, and flow velocity were described as influential factors that induce corrosion.

23 23 Pipe materials were characterized for durability purposes. Gabriel and Moran (1998) stated that concrete pipes must be designed to account for the influence of water and soil chemistry, especially sulfate content; further considerations for concrete culvert design include the water/cement ratio (to reduce porosity), protective coating for steel reinforcement (to prevent corrosion), and the cement type (since types II and V which are sulfate resistant). Other useful design techniques include increasing the cover over rebars when abrasive flows are present and possibly adding harder aggregates in the concrete mixture than bed load aggregates. As stated by Gabriel and Moran (1998), the state of Ohio developed a statistical regression to estimate the service life of concrete culverts for ph greater than 7. Corrosion of metal culverts was related to minerals, ph, and resistivity of water. Minerals in the water may accelerate or protect the metal layer from corrosion. Scale formation may perform protective coating to prevent corrosion produced by the excess of calcium carbonate that isolates the metal from water. Most states linked durability with the ph and resistivity of soil and water. Coatings listed by Gabriel and Moran (1998) for metal culverts were zinc, aluminum, asphalt, and polymers. Ph ranging between 6.0 and 9.5 was accepted in order to utilize galvanized steel. Most agencies adopted the California method based on ph and resistivity of water or soil to estimate the service life of galvanized culverts, but some agencies did not find a correlation between soil ph and culvert service life. For abrasion, bituminous coating provided a slight additional life for galvanized steel culverts. Additional service life years were determined based on the modified California chart for different coating such as bituminous, bituminous and paved

24 24 invert, concrete lined. The Aluminized type 2 might provide up to twice the service life predicted for galvanized steel pipe, but invert protection or extra thickness is required for moderate to severe abrasive conditions. FHWA recommended inspecting culverts every 2 years, except when significant problems were detected, in which case inspections were recommended annually or every six months. Also, FHWA defined a rating system from 0 (worst) to 9 (best) for installed culverts (Gabriel & Moran, 1998). NCHRP (2002) synthesis 303, performed a survey of different agencies and departments of transportation to evaluate the methods to assess the pipe condition and alternatives of pipe repair and rehabilitation. A total of 155 questionnaires were sent to different transportations agencies, departments of transportation (DOTs), and a number of localities. The report showed different approaches developed by different agencies to rate the pipe conditions. The main factors identified for determining the service life, besides the corrosion, were joint failure, deflection, and cracking. This study also reported repair activities executed by different agencies according to the deficiencies found in the inspection. Those activities might be patching, cracking sealing, invert paving, lining, or joint work (National Cooperative Highway Research Program (NCHRP) Synthesis 303, 2002). Most agencies chose lining as rehabilitation process. Additional rehabilitation methods utilized by agencies included cut-and-cover and invert paving. NCHRP concluded that there was no consistent methodology for inspection and evaluation of culvert pipes between the agencies.

25 Rating Systems Kurdziel (1988) compared the existing culvert rating systems, which were established by the different state highway agencies, for metal and concrete culvert pipes. The main objective was to unify the rating scales and have two systems, metal and another for concrete, which can be directly compared. Other objectives were to eliminate the subjective interpretation and improve the inspection and data analysis procedures. The new rating system is based on methods presented by the Federal Highway Administration (FHWA) in the Culvert Inspection Manual released in 1986 as a supplement to the Bridge Inspector s Training Manual 70. The proposed rating system includes a zero to nine scale, to conform the national bridge rating system. Most of the state rating systems are significant inconsistent in the descriptions of fair to poor condition for metal and concrete pipes. For metal culverts, there is no consistency in the exact point of failure. Some studies considered the first perforation (Kansas, Maine and Michigan) and others considered the deterioration of the entire invert (Florida, California LA County, Louisiana, Minnesota, and Mississippi). Features included in the new metal rating system are galvanizing, level of rust, depth of pitting, metal thinning, and degree of perforations. Besides these factors, a unique point of failure is proposed. For concrete, Ohio has provided most of the studies on concrete culverts. The study by J.O Hurd in 1985 provided a comprehensive rating system for concrete pipes. The new scale for concrete pipes included a gradual increment of deterioration for mortar, aggregate, softening of concrete, and reinforcement condition. The major change was the inclusion

26 26 of an intermediate condition in more than one step between the first exposure of reinforcement and the total deterioration of the invert (Kurdziel, 1988). 2.3 Field Performance Studies A statistical analysis of the field data was performed by Degler, Cowherd, and Hurd (1988). The objective was to determine modes of structural failure and interrelationships between the independent variables measured in the field, such as age, geographic location, depth of cover, and gauge of multiplate. A total of 890 metal pipearch structures were evaluated throughout the state of Ohio. The culverts assessed were installed from 1833 to 1986, although the majority of structures were from 1951 to The authors concluded that the durability of the culverts had a linear relation with the age when the structure is up to 35 years old. The culverts installed in the southeastern area of Ohio exhibit a faster rate of deterioration when compared to other areas in the state. For high rating scores, there is a linear relationship for shape and cracking problems; however, there was insufficient data to provide a relationship for low scores. On the contrary, no relations were established for distortion or cracking and durability, gauge of the multiplate, and depth of cover. It was found that the most dominant problem that could cause failure on the metal culverts was the occurrence of corrosion and pitting of the multiplate culvert, with seepage and corrosion of the bolted joints. Potter (1990) conducted a quantitative study to evaluate the field performance of the Federal Highway Administration (FHWA) culverts related to the site conditions. The author compared the FHWA installation with adjacent galvanized and aluminum-zinccoated corrugated steel pipe (CSP). Water, soil, and metal samples were gathered at each

27 27 site. The ph and resistivity parameters for soil and water were measured for every culvert inspected, from which the culvert service life was computed according by the California method. According to Potter (1990), the California approach is an accurate model to predict the service life of galvanized steel pipes. Although the conclusions are limited due to small size of the sample (total sample of 14 pipes), the author concluded that the aluminum-coated pipes provided an average of over six times more protection than the service life predicted by the California method for galvanized steel pipes (Potter, 1990). Sagüés et al. (2001) evaluated the concrete pipe performance under aggressive/corrosive environments. The research was divided into four areas: (1) laboratory assessment, (2) test yard assessment, (3) evaluation from the field, and (4) revision of durability prediction guidelines. For (1), two different concrete pipes, one consisting of Type II cement and the other comprised of cement with fly ash, were assessed under continuous seawater and cyclic saltwater. For (2), four real size pipes, two pipes of Type II cement and another two made with cement with fly ash were installed in a test yard location and partially filled with water. A salt solution was then added to determine the corrosion performance. Eight concrete culverts were inspected for (3), some cores where extracted to analyze chloride penetration. In (4), the service life estimating equation provided in the FDOT report 93-94A was reviewed for possible changes. The authors concluded that results from typical concrete utilized for pipe production did not indicate a special resistance against chloride penetration. From the yard exposure, the data supported the idea that active corrosion was present after few years, but no external evidence was shown. The field structures inspected showed that for

28 28 low chloride, sulfate, and moderate ph, no significant corrosion was found. Culverts at locations that were exposed to seawater exhibited some rust staining to extensive corrosion. Additionally, for culverts in service where the flow contains chloride in concentration greater than 2000 ppm and part of the section above the high tide the service life can be estimated by the current equation by FDOT, but the chloride content should be set at 25,000 ppm to include the corrosive effect produced by the cyclic water. From the ODOT Risk Assessment and Update of Inspection Procedures for Culverts report, two articles were generated concerning the service life of metal and concrete culverts. For metal culverts, a total of 25 culverts were inspected in the state of Ohio by Masada et al. (2006). The purpose was to evaluate the recent culvert management program developed by the Ohio Department of Transportation (ODOT). Additionally, the Ohio Research Institute for Transportation and the Environment (ORITE) introduced a complementary procedure to evaluate the culvert performance. Two levels of statistical analysis were carried out. The Level 1 analysis consisted of a simple linear regression between the culvert condition and the age. In the Level 2 analysis, multivariable linear and nonlinear regressions were computed using the SPSS computing software. From the ODOT rating system, age, rise, and plate type were statistically significant for linear regression. However, only age was statistically significant for nonlinear regression. The improved approach proposed by ORITE indicated that the variables such as plate type, ph, abrasiveness, flow velocity, age, and rise were statistically significant for linear regression, while age, ph, and plate thickness were statistically significant for nonlinear regression. The method proposed by ORITE

29 29 improved the coefficient of determination, compared to the ODOT approach. A risk assessment method was proposed based on the average of the metal plate, culvert shape, seams and joints rating, and alignment rating scores multiplied by the relation of soil cover height and culvert rise. The authors concluded that the ODOT method is accepted, but could be improved by incorporating the suggested ORITE approach. In the second paper, Masada et al. (2007) stated the same objectives but this time for concrete culverts. Twenty five concrete culverts were chosen from different ODOT districts to provide a wide variety of conditions, including age, soil cover height, geographical locations, and environmental conditions. Three different analysis levels were performed: the first level was an analysis of the data presented, the second utilized Excel to develop a two-variable linear regression between two correlated variables, and the third level established a multivariable linear and nonlinear regression using the statistical software SPSS. The multivariable linear and nonlinear analyses generated two models to predict the material rating score. The significant variables for the linear model were culvert age, ph level of drainage water, and drainage flow abrasiveness; however, only age of culvert was significant for the nonlinear model. Again, the risk assessment approach proposed by ORITE was utilized, but this time for circular and elliptical concrete pipes. The original average rating score (OAR) can be calculated for circular and elliptical concrete pipes as the average of the general material rating score (GR), culvert alignment score (CA), and seams/joints rating score (SJ). The authors concluded that culvert age, drainage water ph, and flow abrasiveness affect the service life of

30 30 concrete culverts. The other parameters evaluated were not statistically significant (Masada, Sargand, Tarawneh, Mitchell, & Gruver, 2007). DeCou and Davis (2007) conducted a research to determine the influence of abrasive flow on the thickness loss of different pipe materials. The authors placed some samples with a section of 12x12 inches composed of different materials such as concrete, galvanized CMP, ribbed PVC, and corrugated HDPE, in an extremely abrasive flow located in Nevada County, north of California. The samples were exposed to the flow five years, and readings of sample thickness were made every year to establish the material loss. Each sample included a total of nine data points for measuring thickness. After five years, 18 of the samples were completed destroyed. The authors attempted to constitute the occurrence of the first perforation, based on it, they calculated the thickness loss rate for each material. From the results, DeCou and Davis concluded that the abrasion wear to pipes does not progress linearly with time. Most of coated the steel pipes exhibited better performance than the non-coated ones. Smoother pipe profiles showed less abrasion wear than corrugated profiles. Concrete pipes showed a significantly higher abrasion wear rate compared with all other pipe materials. Additionally, the authors proposed a new abrasion flow classification. The proposed classification is composed of six levels based on the flow velocity and the size of the bed load material. 2.4 Durability Studies From the symposium on Durability of Culverts and Storms Drains organized by the Transportation Research Board, two relevant studies about durability of culverts were presented and published in the Journal of the Transportation Research Board. First,

31 31 Jacobs (1984) evaluated culverts of different materials in Maine. The main objective of the study was to determine the service life of different types of culverts materials. The service life estimation equations were set for corrugated metal pipe (CMP), reinforced concrete pipe (RCP), and bituminous coated with paved invert corrugated metal pipe (BCCMP). An evaluation scale from 0 (best condition) to 5 (worst condition) was utilized to classify the conditions of metal and concrete culverts. The author concluded that for BCCMP culverts, 103 structures were evaluated with two up to 44 years of service, 40 years of service life was estimated for a BCCMP with 14 gauge plate thickness. From the multiple linear regression analysis, the independent variables of age and water ph were significant at 1% level, and the stream flow was significant at 5% level. Resistivity of water was not significant on the culvert condition. Ninety-nine CMP culverts were analyzed, with age from nine to 45 years old. The predicted service life for a 16 gauge pipe was 28 years. From the multiple regression analysis for BCCMP, the ph, age, and stream flow were significant whereas the resistivity was not significant on the culvert condition. For RCP, a total of 64 pipes were rated. A service life of 65 to 70 years was estimated. The variables ph, age, and stream flow had a significant effect on the pipe condition. The second influential study was performed by Bellair and Ewing (1984). The authors evaluated the metal-loss of galvanized and aluminum culverts. Samples of 1 in. diameter were extracted to analyze the corrosion produced. The study was divided into two stages. First, from an obtained steel pipe of 20 years old, 66 cores were extracted to determine the best method to calculate the average metal loss. The weight-loss method

32 32 presented the best results, but the pin-micrometer provided a positive approach to predict the average metal-loss. In the second stage, a total of 30 field locations were chosen for the study. From each culvert, a total of eight samples from the invert were gathered for laboratory purposes. A total of 190 galvanized and 35 aluminum culverts were tested. It was concluded that it is not relevant to provide protection for outside or above the flow in the pipe; however, it is highly important to provide extra thickness for the invert below the flow line in order to consider the metal lost due to the flow conditions. Two studies were performed to evaluate the ODOT rating system. On one hand, Hadipriono, Larew and Lee (1988) evaluated the service life of concrete pipes. The authors explained the main causes of concrete pipe deterioration. In addition to design or construction deficiencies, sulfate attacks from the soil could trigger structural failures in concrete pipes. In this article, the ODOT rating system was used for the pipes inspected. According to the authors, the variables that affected the service life of pipes can be classified into three categories, specifically pipe geometry, load, and resistance. Those categories included pipe age, pipe size, depth of flow, flow velocity, presence of abrasive materials, presence of sediments, protecting rating, slope of pipe, and tests of ph in water. From the gathered data, a regression analysis was carried out to estimate the service life of concrete culverts. The data was provided by ODOT, and a total of 21 concrete culverts were used in the analysis. A stepwise regression was performed throughout two types of analyses: additive and multiplicative. The additive model used a standard linear form, and the multiplicative applied a logarithmic transformation. A total of 12 cases were performed, with six additive and six multiplicative models. The authors

33 33 concluded that an additive model had the best fit to predict the service life of concrete culverts with similar conditions to state of Ohio. On the other hand, Hurd (1988) compared the model proposed by ODOT with another linear model created by a research agency using data collected by ODOT. This comparison was made as a result of the large difference in high values of ph among the two models in predicting the service life of concrete pipes. A total of 196 culverts were rated, which were installed from before 1940 to Based on previous studies, the independent variables included were age of culvert, diameter, flow depth, flow velocity, sediment depth, largest bed load particle size, ph, slope, and the dependent variable was culvert pipe rating. The author concluded that both models were acceptable for predicting the service life of concrete pipes with acidic flow (ph lower than 4.5). While the linear model is not adequate to predict the service life for flows with ph above 4.5, the ODOT model exhibited an accurate estimation for existing older concrete installations. Bednar (1989) proposed a modification to the California Chart to estimate the service life of galvanized steel culverts. The modification that was established can be applied to any weather conditions. The California method is utilized to predict the service life of galvanized culverts; nevertheless, this approach is suitable for dry environments where the soil chemistry controls the corrosion of the structure. For wet environments, where runoff is present most of the time, the California method tends to be conservative for scaling waters and liberal for non- scaling waters. A modification for estimating the service life is introduced to evaluate the effect of protection provided by the hardness and alkalinity of waters as well as the corrosive effects of free CO 2. The author presented an

34 34 equation that includes the relationship between alkalinity, ph, and free CO 2 to estimate the galvanized steel pipes service life. An additional parameter is included to evaluate the effects of the temperature in the corrosion of the culverts. Also in 2000, Ault and Ellor studied a total of 32 steel pipes that are 10 to 17 years old. Among these, 24 were aluminized type 2, three galvalume, three bituminouscoated galvanized, one was aluminum alloy, and one was galvanized. For each culvert, pipe characteristics (e.g. location, corrugation type, length, diameter, age, and slope) and environmental factors (e.g. soil and water temperature, ph, resistivity, and abrasiveness of flow) were examined. Samples were gathered from the invert and crown to determine the remaining coating thickness. The data collected from the field was analyzed to demonstrate the possible advantages of aluminum type 2 over galvanized coating. The actual perforation was compared to the California method to predict the first perforation for galvanized metal pipes. From this study, although the data is limited to determine the advantages of aluminized type 2 over galvanized steel pipes, it was concluded that the service life of aluminized type 2 is up to eight times longer than predicted for galvanized pipes by the California method. The data suggested that the aluminum type 2 coating performance was equal or better than galvanized on steel culverts. For aluminum type 2, it was found that there were two pitting rates, and the highest pitting rate is mainly influenced by the presence of severe bed load. A new system to qualitatively and quantitatively evaluate culverts installed in the state of Utah was developed by Beaver and McGrath (2005). A total of 270 culverts made up of 193 metal, 40 concrete, and 37 thermoplastic pipes were inspected. Two

35 35 independent culvert conditions were rated for assessment purposes, namely waterway and structure barrel performances. Waterway assessment was rated from 9 for good to zero for closed to traffic; washed out by flood action. In this rating system the alignment, scour and obstructs/roadway/structure variables were evaluated. For the barrel structure, a rating system was developed for metal, concrete, and thermoplastic pipes. The three rating systems again varied from 9 a new culvert or a culvert with no evident deficiency to zero for complete failure and roadway closed. After the waterway and barrel structure rating was obtained, the authors proposed an important factor to define the maintenance actions. The system consisted of multiplying the minimum waterway or structure barrel rating score by a factor based on the culvert size, roadway importance, and waterway purpose. This modification factor allowed for determining whether culverts require immediately attention or delaying maintenance. The authors recommended inspecting culverts structures every five years or when activities on the road are executed. Halmen, Trejo, and Folliard (2008) analyzed galvanized culverts embedded in controlled low-strength materials (CLSM) under different environmental and soil configurations. Galvanized coupons were introduced into 13 different CLSM configurations in order to evaluate the percentage mass loss. Two different flow water environments were included: distilled water and a sodium chloride solution. Statistical analyses were performed, including multivariable regression and analysis of variance. The corrosion rate was used as the dependent variable and environment, fine aggregate type, fly ash type, resistivity, ph, metal type (ductile iron and galvanized steel), w/cm,

36 36 percent chloride ion content, and cement content were used as the explanatory variables (Halmen, Trejo, & Folliard, 2008). Although ductile iron material is included, the authors only evaluated galvanized steel. A weighted regression analysis was executed with the environment and metal type variables to obtain the largest effect on the logarithm of corrosion rate (LCR). The proposed equation determines the service life of galvanized culverts embedded in CLSM materials. It might help engineers to specify the characteristics of the CLSM materials to control the variables that contribute to the culvert corrosion. This study also compared the predicted service life with the equation proposed by California method. The authors concluded that the California method might not be appropriate to estimate the service life of galvanized culverts when CLSM materials are utilized in backfills. Also, Halmen et al. (2008) concluded that LCR decreased when ph values increased. Chloride solutions had a significant negative effect on corrosion compared to samples analyzed with distilled water. The authors concluded that logarithm of resistivity and water cementitious material are significant variables in the model. The National Corrugated Steel Pipe Association (NCSPA, 2008) developed a manual to design corrugated steel pipe (CSP). In this manual, the section on durability provides information on the factors that influence the steel pipe service life. The soil, water and abrasiveness of the flow were evaluated in terms of their influence on the corrosion and loss of thickness of CSP culverts. Aluminized type 2 (ALT2) and polymer coated were defined in this chapter. ALT2 provides a barrier layer to decelerate the corrosion when soft waters are present and it works in wider ph ranges than galvanized.

37 37 Based on the California method, the NCSPA provided a durability guide for galvanized culverts. Additionally, three supplemental pavings or coatings were described, which are asphalt coated, asphalt coated and paving, and concrete paved. The additional life provided by each coating was stated under the four levels of abrasion established by the Federal Highway Administration (FHWA). Molinas and Mommandi (2009) conducted a study to evaluate the performance of galvanized coating for the Colorado Department of Transportation by utilizing the California method. In this study, the authors inspected a total of 21 culverts to gather soil and water samples to measure ph, resistivity in soil and water, and soluble sulfate and chloride measurements in runoff water. The objective was to determine possible modification of the California method to predict the service life of galvanized culvert pipes. Environmental factors were also analyzed to establish the significance in concrete culverts corrosion rate. The abrasion flow was also analyzed to establish its influence in the service life of culvert structures. An adjustment chart was introduced for different levels of abrasion proposed by CALTRANS for concrete, steel, and aluminum, culverts. The authors concluded that service life of galvanized steel culverts is related to the ph of the soil and water, resistivity of the soil, and plate thickness. A modification for the California method was proposed, which included calculating the thickness effect in the estimated service. Molina and Mommandi (2009) proposed a power relationship when the thickness of galvanized pipe is increased. For concrete culverts, the salt contents were surpassed in magnitude to the reference values without compromising the structure stability.

38 38 Recently, Salem, Salam, and Najafi (2012) developed a preliminary model to determine the deterioration of metal culverts. They reviewed several studies where the service life is evaluated for metal and concrete culverts. Those studies included statistical analyses to assess the significance of variables such as water and soil ph, water resistivity, abrasion flow, flow velocity and age of the structure, in the durability of the pipes. Additionally, to evaluate the culvert asset management practices of the departments of transportation in the United States and Canada, a survey was sent to 50 transportation agencies and 10 provinces. Twenty six consistent responses were analyzed, and some important findings from the survey were presented. The results indicated that 42% of the respondents did not have an inventory guideline for culverts, 8% of the respondents had a model to predict the service life of culverts, 69% of respondents did not have a model and three agencies are working on developing a prediction model, only three of 26 indicated that they have a system for repair or renewal of culverts. Finally, the authors proposed a binary logistic model to determine the probability of when a metal culvert will require a repair procedure. The independent variables included in the model were age of the structure, span, slope, and protection type. The dependent variable was the current condition of the pipe based on ODOT rating system (9 for best condition through 0 for the worst). According to ODOT, culverts rated worse than six needed a repair procedure. A forward stepwise statistical method the 0.05 level of significance was utilized to select the significant variables in the model. The results showed that age and span were significant; therefore, these variables were included in the model. From the model it can be concluded that the span has a negative relation with the deterioration of

39 39 the metal pipe. This may be explained by the high quality standards of large diameter pipes. 2.5 Concepts of Multivariable Linear Regression The linear regression analysis is a method to predict an outcome variable from one or multiple independent or predictor variables by a straight line. For a multivariable linear regression the outcome variables are predicted by a combination of different predictor variables with their coefficients as shown in Eq. 5.1 (2.1) Where: Y: Outcome variable X 1,2...,n : Independent or predictor variables b 0,1,..,n : Coefficient of independent variables. The linear regression analysis employs the method of the least squares to find the line that best fits to the data. The method of least of squares applies the squared difference from the line to the data. When the squared difference is large, the line does not significantly represent the data; however, if the value is small, the line is representative of the data. The line with the minimum squared difference is taken by the method of the least squares (Field, 2009). The line that best fits the model is evaluated to determine how well the line represents the data. The R 2 term represent how well the model explains the data compared to the mean of the outcome variable. The R 2 can be calculated as shown in Eq. 2.2

40 40 (2.2) Where: SS T = The total sum of the squared difference between the observed data and the mean of the outcome variable. This value show whether the mean is satisfactory for the data as a model SS M = The difference between the SS T and the residual sum of the squared difference (SS R ) computed with the method of least squares (Field, 2009) The adjusted R 2 represents the shrinkage of the model. The R 2 represents the amount of variance for the outcome variable that is explained by the model for the sample. Meanwhile, the adjusted R 2, represents the amount of variance in the outcome variable that is represented if the model is applied to the population rather than the sample. To calculate the adjusted R 2 the Eq. 2.3 is applied (Field, 2009): Where: 1 1 (2.3) Adjusted R 2 = Adjusted R 2 value R 2 = Unadjusted R 2 value n= The number of participants k= number of predictors in the model. The F-ratio represents the ratio of improvement of the model compared with level of the inaccuracy of it. The F-ratio is calculated as the mean of the sum of squares of the model (MS M ) divided by the residual mean squares (MS R ). MS M can be obtained by dividing the SS M by the number of variables in the model. MS R is obtained by dividing

41 41 the SS R by the number of observation minus the number of parameters estimated, including the constant. The F-ratio is shown in Eq. 2.4: (2.4) Where: F: F-ratio value MS M : The mean of the sum of squares of the model MS R : The residual mean of squares. The significance of the independent variables is tested to establish the contribution to the outcome. The t-statistic tests each predictor with the null hypothesis that the coefficient is zero (i.e. b=0). If the test is significant, at the level of significance of 0.05, the null hypothesis is rejected; therefore, the b-coefficient is different than zero. This means that the independent variable that was assessed significantly contributes to the outcome variable Variable Selection Method The stepwise method was chosen to obtain the best model in the linear regression analysis. The stepwise is divided in three types of selection: Forward, Stepwise, and Backward. The Forward Selection starts predicting the outcome only with the constant. Then, it adds the variable with the highest correlation with the outcome. If the predictor significantly improves the model, the variable is retained and the method tries with a second variable. In SPSS, the Stepwise Selection (SS) is similar to the FS, except that every time a variable is added to the equation, a removal test is performed for the predictor with the higher significance in the t-statistic test. The Backward Selection (BS)

42 42 is contrary to the FS since the BS places all the variables at the first time. The BS compares the significance of the variable with highest significance with the removable criteria. If the significance is higher than the removal criteria, the variable is removed and a new iteration starts with the remaining predictors Assumptions of the Linear Regression Analysis The linear regression analysis requires that several assumptions are met. If the assumptions are true, conclusions about the population can be made based on the linear regression results. The linear regression assumptions are (Field, 2009): 1. Non-zero variance: the predictor or independent variables must have variation in value. 2. No perfect multicollinearity: Two or more predictors must have no perfect linear relationship or be highly correlated. 3. Homoscedasticity: The residuals of each predictor must have the same variance. 4. Independence of errors: the residual terms of two observations must be uncorrelated. This assumption is tested by the Durbin-Watson test. Values close to two may indicate that the residuals are uncorrelated. 5. Normally distributed errors: This assumption requires that the differences between the model and the observed data must be zero or close to zero Non-Linear Regression The non-linear regression is employed to explain non-linear relationships between the outcome and the predictors. For culverts, the rate of deterioration may be slow during the first years after the installation and increase with time (Mitchell et al., 2005). To

43 43 establish a non-linear relationship between the outcome and the predictor, Mitchell et al. (2005) proposed the following non-linear model: (2.5) The Eq. (2.5) can be transformed into a linear equation through applying logarithms as follows (2.6) The linear Eq. (2.6) can be analyzed as a linear regression model in SPSS. The logarithm should be applied to the concrete and metal data to allow the analysis. The characteristics, tests, and assumptions of the linear regression analysis described in Section 2.5 are applied to solve for the b-coefficients of the Eq. (2.6).

44 44 CHAPTER 3: CULVERT DATA COLLECTION FORM The Ohio Department of Transportation (ODOT, 2003) released a comprehensive manual to determine the main characteristics of existing culverts for inventory and inspection. The Culvert Management Manual presents a complete rating guideline to quantitatively describe the current conditions of metal, concrete, thermoplastic or masonry structures inspected. In 2005, new inspection and rating systems were developed for concrete, metal, and thermoplastic culverts by Mitchell et al. (2005), who evaluated the ODOT (2003) policies for inspection and rating approach and modified it based on what were observed in the field. Their results showed that using the new form and rating system, the service life of concrete and metal culverts can be predicted more accurate than by the Culvert Management Manual. In this chapter, the new inspections and inventory forms are introduced. The data collection form and rating scales are based on the findings of the study developed by Mitchell et al. (2005). In addition to the rating scales for concrete and metal culverts, this thesis includes the new method to classify the abrasion flow level developed by DeCou and Davies (2007) for CALTRANS. Appendix A includes concrete and metal pipe data collection forms. 3.1 Inventory Data The ORITE developed a new form for inventory based data provided by ODOT. The inventory data is acquired to determine the initial conditions of each structure when it was installed. Culvert File Number (CFN): This was introduced by the ODOT in their Culvert Management Manual (2003) to classify each culvert structure. The

45 45 components utilized to assign the CFN number are county number, route number, and culvert number. County number is composed of 2 digits as shown in Table 3.1. The route number has 3 digits and points to the main route where the culver is located. The culvert number has 4 digits randomly assigned to each culvert. Table 3.1. County Number and Districts (ODOT, 2003) Year or date installed: The year when the culvert was installed at the selected location.

46 46 Length of the pipe: The distance from the inlet to outlet of the pipe. Also, the length is checked in the field by utilizing the HD150 Handle distance Laser Meter provided by Trimble (Figure 3.1). The HD150 has an accuracy of <1/8 in up to 100 feet. Figure 3.1. HD150 Distance Meter The depth of the flow: This is obtained by subtracting from the total interior height of the pipe, the distance from the crown to the top of the water flow measured by a standard tape measure. Nominal size of pipe: Inside diameter (only for circular pipes). Vertical height (rise) or diameter: the diameter is stated for circular pipes and the maximum vertical dimension for arch and elliptical pipes. Horizontal width (span) or diameter: the diameter for circular pipes and the maximum horizontal dimension for arch and elliptical pipes.

47 47 Number of Cells: The current number of pipes present at each culvert location. Original wall thickness: The pipe wall thickness when the structure was installed. Cover over pipe (height): Total height of cover measured from the top of the culvert pipe to the top of embankment or pavement. Backfill type (and compaction if possible): Type of soil used to backfill and the degree of compaction Concrete Pipe Specific aspects for concrete pipes are stated during installation. The concrete culvert data collection form includes the following parameters: Pipe shape: The cross section of shape (circular or elliptical). Type of protection: Type of lining or coating (none, epoxy coated, vitrified plate lined). Whether a coating or lining is present in the pipe, the percentage of protection remaining is estimated Metal Culvert Corrugation Profile: Corrugated Metal Pipe (CMP) or Structural Plate Pipe (SPP). Pipe shape: The cross-sectional shape (circular or pipe arch). Metallic coating (galvanized or aluminized). Type of protection: Half or full asphalt coated, 1/2 or full asphalt coated with paved invert, concrete field paved, polymeric coated.

48 48 Headwall Type: Type of headwall at both ends (none, half-height stone, full height stone, half-height concrete, full-height concrete). 3.2 Field Data The culverts in the field are affected by soil and drainage water conditions. The most influential factors reported by different authors are included in the form employed in the present study Concrete Pipes Surface condition The condition of the concrete surface is detailed inside the pipe. The visual observations are mainly focused below the flow line. The following aspects are evaluated: Cracking: The maximum width and an approximated percentage of the total area where the cracks are present. Delamination and spalling: The amount of area in percentage where the concrete shows pits or delaminated surface. Scaling: Percentage of area flaked on the pipe surface after freezing and thawing process. Aggregate and mortar loss: Whether mortar or aggregate loss is present and the degree of the loss. Exposed reinforcing steel: Whether reinforcing is exposed and the conditions of the reinforcement exposed.

49 Water Chemistry The water chemistry is divided into two phases. Phase 1 is composed of the measurements taken in the field; phase 2 consists of the measurements taken in the laboratory. For phase 1, the ph and resistivity (ρ) of the water are measured by the PC 300, provided by Oakton (Figure 3.2). The PC300 provides measurements of ph and conductivity (σ) directly from the stream in minutes. The resistivity is calculated as the inverse of the conductivity. The PC 300 provides an automatic temperature compensation for ph and conductivity when the temperature of water is different from 25 C (Eutech Instruments Pte Ltd/ Oakton Instruments, 2008). The PC 300 sensors (See Figure 3.2) were introduced into the stream to determine the ph and conductivity of water. The PC 300 was calibrated every two weeks to ensure the accuracy of readings. The calibration of the equipment consisted of two calibrating solution points for the ph and conductivity. As mentioned before, this equipment provides temperature compensation for reading of the ph and conductivity. A couple minutes were required to obtain a stabilized measurement for both parameters.

50 50 Figure 3.2. PC 300 for ph and resistivity measurements (Eutech Instruments Pte Ltd/ Oakton Instruments, 2008) Phase 2 is composed of the chloride and sulfate contents in water. Samples of water are extracted from the stream flow and analyzed in the laboratory, using the Dionex IC25 Ion Chromatograph (see Figure 3.3). The test follows the EPA standard proposed by Pfaff (1993). The Dionex IC25 Ion Chromatograph allows computing the fluoride, bromide, nitrate, nitrite, phosphate, chloride, and sulfate content of a water sample. For this study, only the sulfate and chloride content are determined. The Dionex IC25 (Figure 3.3) analyzed a total of 18 water samples taken from concrete culvert locations.

51 51 Figure 3.3. Dionex IC25 Ion Chromatograph Hardness of Concrete A prospector s pick or geologist hammer is utilized to determine the degree of hardness of the concrete inside the pipe. The procedure consists of hitting the interior concrete wall with the hammer to determine if the concrete hardness has been affected by the exposure to the environment. The examinations are obtained close to the flow line and invert where the concrete is wet most of the time Metal Pipes Current Wall Thickness During the inspection, the current wall thickness is measured at undeteriorated location. For that purpose the 38DL Plus Ultrasonic Thickness Gage provided by Olympus is employed, see Figure 3.4. The 38DL Plus is a nondestructive equipment to measure the thickness of a variety of materials. The transducer is capable to measure thickness from 0.04 to 20 inches with a resolution of inches (Olympus NDT, Inc., 2011). The 38DL Plus was calibrated in the field for each field work.

52 52 Figure DL Plus Ultrasonic Thickness Gage. ( Surface Condition within Main Barrel Condition of metal coating or corrosion: The condition of the remaining metallic coating and corrosion present inside the pipe. This measurement is expressed in percentage of the total interior area of the pipe. Rust: Degree and presence of flakes or scaling in the interior of the pipe. The presence of rust is expressed in percentage of the total interior surface area of the pipe. Perforation: When perforations are present within the pipe, the percentage of area covered and location are specified. Pitting: Percentage of area and depth when pits are found Water Chemistry For metal culverts, ph and resistivity properties are determined. As stated previously in section , ph and conductivity are measured on field by utilizing the PC 300 (Figure 3.2). Once the conductivity is known, the resistivity can be calculated.

53 Soil Chemistry Resistivity and ph are analyzed for the backfill soil sample at each metal culvert site. Samples from the backfill are extracted to be tested in laboratory. The ASTM D standard is followed to determine the ph of the soil sample. In this standard, the soil sample is air dried and the coarse soil fraction is removed by using a No. 10 sieve. Once the soil sample is dried, 10 g mass is taken and placed into a plastic container with 10 ml of distilled water; the water and soil are mixed and left standing for one hour (American Society of Testing Material, 2009). The ph is established using the PC 300 (Figure 3.2). Figure 3.5 presents the measurement of ph for one of the soil samples taken from the field. Figure 3.5. ph Measurement of the Soil with the PC300 The soil resistivity is determined through the ASTM G (2009) protocol. This standard described how to calculate the resistivity of soil by utilizing the soil box method. The procedure consists of placing the soil sample by layers in the soil box; each

54 54 layer should be compacted by hand. Between layers, slight amount of distilled or deionized water is added to the soil sample to saturate it (American Society of Testing Material, 2009). When the soil box is completely full, the resistance measurement can be obtained, using the Miller 400D supplied by MC Miller Co., Inc (2010). Once the resistance value is known, it is converted to resistivity (see Figure 3.6). Soil samples were saturated with deionized water and placed by layer in the soil box. Figure 3.7 shows the measurement with the Miller 400D soil box of one of the soil samples taken from the field. Once the value of the resistance is met, the resistivity of a soil sample can be calculated using the following equation: Resistivity (Ω.cm) = Measured resistance (Ω) x A/L (3.1) Where: A: Area of soil box (cm 2 ) L: Distance between electrode pins (cm) The Miller 400D soil box establishes a relationship between the area (A) and distance between electrodes (L) equal to 1 (e.g. A/L=1). Substituting A/L=1 in equation (3.1), the resistivity of the sample can be expressed as: Resistivity (Ω.cm) = Measured resistance (Ω) x 1 (cm) (3.2)

55 55 Figure 3.6. Miller 400D Soil Box (M.C. Miller Co., 2010) Figure 3.7. Resistivity measurement with the Miller 400D Soil Box Metal Wall Thickness The metal wall thickness is established at the invert or at the flow line in three representative locations. This measurement may indicate the level of metal loss or level of corrosion as a result of abrasive flow or water acidity Invert or Interface Perforated The percentage of the area perforated on the invert or near the flow line by the action of abrasive flow and/or corrosion.

56 Prospector s Pick Observation A prospector s pick is utilized to determine the severity of corrosion when rust layers are found. The procedure consists of hitting the interior metal wall with the hammer close to flow line or invert. The objective is to remove the rust surface to expose the remaining metal layer and determine if it is possible to perforate the metal wall with the hammer Concrete and Metal Pipes Pipe Slope The slope of the pipe is measured by using the 12 Magnetic Digital Laser Level (model: ) provided by Johnson (shown in Figure 3.8). The instrument has an accuracy of ±0.2 for 1 to 89 (Johnson Level & Tool, 2012). Figure Magnetic Digital Laser Level ( Inside Diameter Measurements Several measurements of the interior diameter are taken when the pipe presents a circular shape, using the SK202 telescope measuring pole (shown in Figure 3.9), supplied by SK Senshin. The measuring pole was utilized for pipes with diameters greater than 6 feet. For smaller diameter pipes, the team used a measuring tape.

57 57 Figure 3.9. SK202 Telescopic Fiberglass Measuring Pole ( Description of Joints The joint sections are evaluated to determine if they are aligned properly, tight, or separated. When a joint is observed to be separated, signs of water exfiltration, backfill soil infiltration, and void formations in the backfill are noted. Figure 3.10 shows a separated joint found inside a concrete culvert.

58 58 Figure Separated joint in a concrete pipe Description, Depth, and Particle Size of Sediment within the Pipe The stream load is the material carried by a stream as a result of erosion process. The type and quantity of material transported depends on the stream velocity, amount of water, and culvert slope. The material accumulated over the pipe invert can be composed of sand, gravel, mud or clay. The depth of sediment is measured directly on the field at the inlet, outlet, and interior of the main barrel. Additionally, the approximate size of the maximum particle is recorded while inspecting the structure Depth of Drainage Flow The depth of the drainage flow is measured at the inlet and outlet of the pipe. When sediment is present the measurement of the flow depth is specified over the sediment layer Drainage Flow Velocity The velocity of the flow is calculated by recording the time that a floating element takes to travel the entire length or a segment of culvert structure.

59 Culvert Rating Systems The durability of concrete and metal culverts has been studied by different authors. The Culvert Inspection Manual, proposed by FHWA, established a rating guideline for metal and concrete culverts from 9 best condition to 0 failure of culvert (Kurdziel, 1988). The ODOT (2003) and Mitchell et al. (2005) based their rating approach also on a zero to nine scale as established by Kurdziel (1988). The following sections present details of the ODOT and ORITE rating systems for concrete and metal culverts ODOT General Scale The ODOT (2003) developed a rating system to establish the deterioration produced by environmental factors. With it, metal and concrete culverts barrels can be quantitative evaluated. The rating scale can reflect a continuous and progressive deterioration of the structures Concrete Pipe Scale The concrete rating scale can range from nine New condition, superficial and isolated damage from construction to zero The culvert is collapsed. The intermediate values in the scale describe the emergence of cracking, efflorescence, exposing of reinforcement, and loss of thickness. Table 3.2 shows the ODOT rating scale for concrete pipes.

60 60 Table 3.2. ODOT Concrete Pipe Rating Scale (ODOT, 2003) Rating Description value 9 Excellent - New Condition, Superficial and isolated damage from construction. 8 Very Good - Hairline cracking without rust staining or delaminations; surface in good condition. Isolated damage from construction. 7 Good - Hairline cracking. No single crack greater than 1/16 inch without rust staining parallel to the direction of traffic. Light scaling on less than 10% of exposed area less than 1/8 inch deep. Delaminated/Spalled area less than 1% of surface area. Note: cast-in-place box culverts may have a single large crack (less than 3/16 inch) on each surface parallel to the direction of traffic 6 Satisfactory - Hairline map cracking combined with molted areas; cracks less than 1/8 inch parallel to traffic with minor efflorescence or minor amounts of leakage. Scaling on less than 20% of exposed area less than 1/4 inch deep. Spalled areas with exposed reinforcing less than 5%. Additional delaminated/spalled areas less than 5% of surface area. 5 Fair - Map cracking. Cracks less than 1/8 inch parallel to traffic, less than 1/16 inch transverse to traffic with efflorescence and/or rust stain, leakage and molted areas. Scaling on less than 30% of exposed area less than 3/16 inch deep. Spalled areas with exposed reinforcing less than 10%. Total delaminated/ spalled areas less than 15% of surface area. 4 Poor - Transverse cracks open greater than 1/8 inch with efflorescence and rust staining. Spalling at numerous locations; extensive surface scaling on invert greater than 1/2 inch. Extensive cracking with cracks open more than 1/8 inch with efflorescence; spalling has caused exposure of heavily corroded reinforcing steel on bottom or top slab; extensive surface scaling on invert greater than 3/4 inch. (approximately 50% of culvert is affected) 3 Serious - Extensive cracking with spalling, delaminations, and slight differential movement; scaling has exposed all surfaces of the reinforcing steel in bottom to top slab or invert (approximately all exposed surfaces are 50% loss of wall thickness at invert; concrete very soft. 2 Critical - Full depth holes. Extensive cracking greater than 1/2 inch. Spalled areas with exposed reinforcing greater than 25%. Total delaminated, spalled, and punky concrete areas are greater than 50% of surface area. Reinforcing steel bars have extensive section loss and perimeter of bar is completely exposed. (Several bars in a row) 1 Imminent Failure - Culvert partially collapsed or collapse is imminent. 0 Failed - The culvert is collapsed Metal Pipe Scale The metal rating scale is also set from nine New condition; galvanizing intact; no corrosion to zero Total failure of pipe. The scale shows the progressive deterioration of pipe including factors such as: loss of galvanizing, pinholes, layers of rust on the invert, pitting, and perforations. Table 3.3 presents the rating scale for metal pipes.

61 61 Table 3.3. ODOT Metal Rating Scale (ODOT, 2003) Rating Description Value 9 Excellent - New condition; galvanizing intact; no corrosion. 8 Very Good - Discoloration of surface; galvanizing partially gone along invert but no layers of rust. 7 Good - Discoloration of surface, Galvanizing gone along invert but no layers of rust. Minor pinholes (with an area less than 3 square inches per square foot) in pipe material located at ends of pipe (length not to exceed 4 feet and not located beneath roadway). 6 Satisfactory - Galvanizing gone along invert with layers of rust. Sporadic pitting of invert. Minor pinholes (with an area less than 6 square inches per square foot, 4%) in pipe material located at ends of pipe (length not to exceed 4 feet and not located beneath roadway). 5 Fair - Heavy rust and scale. Pinholes (with an area less than 15 square inches per square foot, 10%) throughout pipe material. Section loss and perforations at ends. Holes in metal at end in invert and not located under roadway 4 Poor - Extensive heavy rust; thick and scaling rust throughout pipe; deep pitting; perforations throughout invert with an area less than 30 square inches per square foot, 20%. Overall thin metal, which allows for an easy puncture with chipping hammer. 3 Serious - Extensive heavy rust; thick and scaling rust throughout pipe; deep pitting. Perforations throughout invert with an area less than 36 square inches per square foot, 25%. Overall thin metal, which allows for an easy puncture with chipping hammer. End section corroded away 2 Critical - Perforations throughout invert with an area greater than 36 square inches per square foot, 25%. 1 Imminent Failure - Pipe partially collapsed. 0 Failed - Total failure of pipe ORITE General Scale ORITE developed their own the rating systems after evaluating the ODOT rating systems ODOT (2003). Two published papers, Masada et al. (2006) and Masada et al. (2007), and the report by Mitchell et al. (2005) show the improvement of the rating scale proposed by ORITE for concrete and metal pipes over the ODOT approach. The ORITE rating scale is composed of zero to nine values with emphasis on the worst condition detected.

62 Concrete Pipe Scale The ORITE scale also spans from nine New condition to zero Failure. Intermediate values shows the existence of cracks, spalling, loss of mortar and aggregates, and softening of concrete. The exposure of reinforcement is classified in a lower scale value compared to ODOT rating system. Table 3.4 introduces the rating scale proposed by ORITE for concrete pipes. The general rating scale, applied in the main barrel condition, will be utilized to evaluate the inlet and outlet conditions. Table 3.4. ORITE Concrete Pipe Rating Scale Rating Description Value 9 Excellent New condition 8 Very Good Discoloration of concrete; No cracking, spalling, scaling or softening of concrete; Surface in good condition 7 Good Minor hairline cracking at isolated locations; slight spalling; Light scaling (< 1/8 inch deep) on invert; Slight loss of mortar (aggregate not exposed); No softening of concrete 6 Fair Extensive hairline cracks, some with minor delaminations or spalling; Moderate loss of mortar around aggregate; Invert scaling 1/8 to 1/4 inch deep 5 Fair to Marginal Cracking open greater than 1/8 inch with moderate delamination and moderate spalling exposing reinforcing at isolated locations; Large areas of invert with spalls greater than 1/2 inch deep; Significant loss of mortar and slight loss of aggregates due to surface scaling (1/4 to 1/2 inch depth) 4 Marginal Cracks open more than 1/8 inch with effluence and spalling at numerous locations; Spalls have exposed rebars that are heavily corroded: Heavy invert surface scaling greater than 1/2 inch; Moderate aggregate loss; Concrete softening 3 Poor Extensive cracking, spalling, and minor slabbing: invert scaling has exposed reinforcing steel at isolated locations; moderate amount of concrete softening 2 Very Poor Severe slabbing has occurred in culvert wall; invert scaling below first layer of reinforcing; 50% loss of wall thickness at invert; concrete very soft 1 Critical Holes through concrete at isolated locations; 75% loss of wall thickness at invert; reinforcing exposed throughout invert 0 Failure Invert completely deteriorated; reinforcing steel gone; collapse of the culvert is imminent

63 Metal Culvert Scale ORITE introduced different metallic coating in the rating scale additional to galvanized coating stated by ODOT (2003).The metal rating scale is set from nine New condition; metallic coating intact; no corrosion to zero Inver completely deteriorated; collapse of the culvert is imminent. The ORITE scale presents the first perforations at level two, while ODOT (2003) considers perforations at level five. Table 3.5 presents the rating scale for metal pipes. For inlet and outlet conditions the general rating scale will be applied. Table 3.5. ORITE Metal Pipe Rating Scale Rating Description Value 9 Excellent New condition; metallic coating intact; No corrosion 8 Very Good Discoloration of surface; Metallic coating partially gone 7 Good Superficial or pinpoint rust spots; No pitting 6 Fair Moderate rust, rust flakes tight; shallow pitting of surface; Metallic coating gone 5 Fair to Marginal Heavy rust and scale; Moderate pitting and slight thinning of core metal 4 Marginal Extensive heavy rust, thick and scaling rust coatings; Deep pitting and significant metal loss (approximately 25%) 3 Poor Rust and pitting halfway through core metal (some deflection or penetration when struck with a prospector s pick or geology hammer) 2 Very Poor Extreme deterioration and pitting; Three quarters of core metal gone; First perforations 1 Critical Extensive or large perforations 0 Failure Invert completely deteriorated; Collapse of the culvert is imminent ODOT Joint Scale The rating scale for joints and seams is stated for key factors such as, soil infiltration and water exfiltration. The ODOT (2003) includes joint defects such as open joints, seepage at the joints and surface sinkholes over the culvert in the joint scale for concrete and metal pipes. The joint rating scale is set from nine for straight line between

64 64 sections to zero for total failure of pipe. The intermediate values describe the emergence of joints defects described above. Table 3.6 shows the joint rating scale proposed by ODOT that may be used for concrete and metal culvert. Table 3.6. ODOT Joint Rating Scale (ODOT, 2003). Rating Description Value 9 Excellent - Straight line between sections. 8 Very Good - No settlement or misalignment; tight with no defects apparent. 7 Good - Minor misalignment at joints, offsets less than 1/2 inch; Possible minor infiltration of backfill; no settlement. Minor distress to pipe material adjacent to joint. Shallow mortar deterioration at isolated locations. 6 Satisfactory - Minor backfill infiltration due to slight opening at joints; Minor cracking or spalling at joints allowing exfiltration; Dislocated end section. Extensive areas of shallow deterioration; missing mortar at isolated locations; minor cracking. 5 Fair - Joint open and allowing backfill to infiltrate; significant cracking, spalling, or buckling of pipe material; joint offset less than 3 inches; End sections dislocated and about to drop off from main portion of the structure; Mortar generally deteriorated; loose or missing mortar at isolated locations. Infiltration staining apparent. 4 Poor - Differential movement and separation of joints; significant infiltration or exfiltration at joints. Joint offset less than 4 inches; voids seen in fill through offset joints. End sections dropped off at inlet. Mortar severely deteriorated and/or significant loss of mortar; significant infiltration or exfiltration between masonry units. 3 Serious - Significant openings; dislocated joints in several locations exposing fill material with joint offsets greater than 4 inches. Infiltration or exfiltration causing misalignment of pipe and settlement or depressions in roadway. Large voids seen in fill through offset joints. Extensive areas of missing mortar for masonry structures. 2 Critical - Culvert not functioning due to alignment problems throughout. Large voids seen in fill through offset joints. 1 Imminent Failure - Pipe partially collapsed or collapse is imminent. 0 Failed - Total failure of pipe Settlement Rating Scale ORITE proposed slight modifications to the rating scale proposed by ODOT (2003) for assessing slope and settlement issues that concrete and metal culverts are experiencing (Mitchell et al., 2005). The proposed rating scale includes a continuous increment of settlement exemplified in changes of slope along the length of the pipe and

65 65 the increment of ponding depth inside the culvert. Table 3.7 introduces the settlement rating scale. Table 3.7. ORITE Settlement Rating Scale Rating Description Value 9 Excellent Uniform slope; no settlement. 8 Very Good Minor settlement at one location. 7 Good Minor settlement at a few isolated locations; Ponding of water less than 1 deep. 6 Fair Minor settlement at numerous locations; Ponding of water less than 3 deep. 5 Fair to Marginal Moderate settlement at one location; Ponding of water up to 5. 4 Marginal Moderate settlement at one location; Ponding of water as deep as 6. 3 Poor Severe settlement in one area; Ponding of water more than 6 deep; Pipe end sections are dropping off. 2 Very Poor Pipe is not functioning due to sever settlement problem; Upstream end is not visible from downstream end; Water ponding more than 50% of pipe length. 1 Critical Pipe has partially collapsed. 0 Failed Pipe has collapsed Abrasion Levels The classification of the abrasion level represents a major addition to the approach developed by Mitchell et al. (2005). The California Department of Transportation (CALTRANS) performed a study to evaluate the pipe material resistance to abrasion over a period of 5 years. New levels of abrasion, wear rates, and recommendations for culvert and lining materials in abrasive flows were proposed (DeCou & Davies, 2007). They introduced six levels of abrasion according to the performance of pipes materials assessed. The six levels were classified based on the flow velocity and the type of bed load as shown in Table 3.8.

66 Table 3.8. Abrasion Levels from Highway Design Manual CHP850 (DeCou and Davis, 2007) Level of Description Abrasion Level 1 Virtually no bed load with velocities less than 5 ft./s (1.5 m/s); Where there are increased velocities with no bed load (e.g. urban storm drain systems or culverts < 30" (0.76 m) dia.); significantly higher velocities may be applicable to Level 1. Level 2 Bed loads of sand, silts, or clays regardless of volume. Velocities > 3 ft./s (0.9 m/s) and < 8 ft./s (2.4 m/s). Where there are increased velocities with minor bed load volumes (e.g. urban storm drain systems or culverts < 30" (0.76 m) dia.); Significantly higher velocities may be applicable to Level 2. Level 3 Moderate bed load volumes of sands and gravels (1.5" (38 mm) max); Velocities > 5 ft./s (1.5 m/s) and < 8 ft./s (2.4 m/s); Where there are increased velocities with minor bed load volumes < 1.5" (38 mm) (e.g. storm drain systems or culverts < 30" (0.76 m) dia.), higher velocities may be applicable to Level 3. Level 4 Small to moderate bed load volumes of sands, gravels, and/or small cobbles/rocks with maximum stone sizes up to about 6 in (150 mm); Velocities > 8 ft./s (2.4 m/s) and < 12 ft./s (3.7 m/s). Level 5 Moderate bed load volumes of sands, gravels, and/or small cobbles with maximum stone sizes up to about 6 in (150 mm); For larger stone sizes within this velocity range, see Level 6. Velocities > 12 ft./s (3.7 m/s) and < 15 ft./s (4.6 m/s). Level 6 Heavy bed load volumes of sands, gravel and rocks, with stone sizes 6 in (150 mm) or larger. Velocities > 12 ft./s (3.7 m/s) and < 20 ft./s (6.1 m/s); Or Heavy bed load volumes of sands, gravel and small cobbles, with stone sizes up to about 6 in (150 mm); Velocities > 15 ft./s (4.6 m/s) and < 20 ft./s (6.1 m/s); Very limited data on abrasion resistance for velocities > 20 ft./s (6.1 m/s) contact District Hydraulics Branch. 66

67 67 CHAPTER 4: INVENTORY AND FIELD INSPECTION DATA The inspection activities took place in the Ohio Department of Transportation (ODOT) Districts 9 and 10, located in southeastern portion of the state of Ohio in the months of April and May during the spring of The inspections were performed for concrete and metal pipes. Inspections were performed for culverts with span and rise dimensions greater than 42 inches and a maximum length of 150 feet. The objectives of the field work are: to classify the current conditions of each pipe inspected; to validate the new data collection form proposed by ORITE for concrete and metal pipes; and to evaluate the rating systems proposed by ORITE and ODOT for quantifying the current culvert conditions and predicting the remaining service life. 4.1 Selection and Localization of Culverts Inspected The rise, span, and length dimensions were used as screening factors to select the culverts for field inspection work out of the ODOT inventory data. The rise and span were set at a minimum of 42 inches so that the culvert is large enough to allow direct man entry. The length was set at a maximum of 150 feet, since longer than this length would be higher time consuming to inspect. This stage of the project was focused on the ODOT Districts 9 and 10. Districts 9 and 10 are known for the aggressive environmental conditions for culverts according to previous studies by ODOT (1972) and Mitchell et al. (2005). Indeed, ODOT (1972) provided the distribution maps of ph and abrasive conditions around the state of Ohio. Figures 4.1 and 4.2 show the acidic and abrasive properties of the drainage flow that commonly found in ODOT districts 9 and 10.

68 68 Figure 4.1. Water ph based on ph Values of Individual Culverts (ODOT, 1982) Figure 4.2. Percent of Abrasive Flow per County (ODOT, 1982)

69 69 A total of 39 (18 concrete and 21 metal) culverts were selected for inspection and analysis to assess the current conditions and predict the remaining service life. Tables 4.1 and 4.2 list the culverts inspected. Table 4.1 provides shape, rise, span, length, and age of each concrete culvert. Table 4.2 gives similar information for each of the metal culverts. Appendix B provides the remaining data obtained in the field and laboratory work out for each concrete and metal culvert. Table 4.1. Concrete Culverts Inspected Culvert ID Shape Rise (in) Span (in) Length (ft) Age (years) ATH Elliptical NA HOC Circular NOB Elliptical NOB Circular VIN Circular VIN Circular ROS Circular ROS Circular NA JAC Elliptical JAC Elliptical JAC Elliptical JAC Circular JAC Elliptical ROS Circular HIG Circular NA ROS Circular ROS Circular JAC Circular NA= Not Available

70 70 Table 4.2. Metal Culverts Inspected Culvert ID Shape Corrugated Profile Rise (in) Span (in.) Length (ft) Age (years) ATH Circular CMP ATH Circular CMP HOC Circular SPP HOC Pipe Arch SPP HOC Circular CMP NOB Circular CMP MOR Circular CMP ROS Circular CMP ATH Circular CMP VIN Circular CMP VIN Circular SPP NA ROS Circular CMP NA ROS Circular CMP ROS Pipe Arch SPP JAC Circular CMP JAC Circular CMP JAC Circular CMP JAC Circular SPP JAC Circular CMP JAC Pipe Arch CMP JAC Circular SPP NA= Not Available CMP= Corrugated Metal Pipe; SPP= Structural Plate Pipe Figures 4.3 and 4.4 present the geographical location areas of the concrete and metal culvert pipes (each dot may represent more than one culvert). The pipes inspected belong to Highland, Jackson, Ross and Vinton counties in District 9, and Athens, Hocking, Morgan, and Noble counties in District 10.

71 71 Figure 4.3. Geographical Location of Concrete culverts Inspected Figure 4.4. Geographical Location of Metal culverts Inspected

72 Inventory Data Characteristics After the inspection activities were completed, the ORITE team analyzed the data collected to characterize the drainage structures and their environment settings. For concrete and metal culverts, length, span or diameter, shape, corrugation profile, type of protection, age, and cover were evaluated Length The actual lengths of the culverts inspected have been presented in Tables 4.1 and 4.2. Table 4.3 shows the length according to the type of material. For concrete culverts, the average length was 70 feet. Four concrete pipes were equal or larger than 100 feet: Ross , Ross , HIG , and ROS These pipes measured 111,100,108, and 142 feet in length, respectively. For metal culverts, the average length was 55 feet. Two metal pipes had a length slightly above 100 feet: ROS and VIN These pipes reached 112 and 108 feet in length, respectively. Table 4.3. Length Classification per Material Material Length of Pipe [L] (ft) L 25 25<L 50 50<L 100 L>100 Concrete Metal Pipe Shape As described in Table 1.1, concrete and metal culverts are made up of different cross-sectional shapes. Table 4.4presents the shape classification for each material type. Of the total pipes inspected, 76% of concrete and metal pipes were circular in shape, 33%

73 of the concrete pipes had an elliptical shape, and only 14% of the metal pipes had a pipeach shape. 73 Table 4.4. Shape Classification per Material Material Pipe Shape Circular Elliptical Pipe Arch Concrete Metal Span Table 4.5 shows the span/diameter classification for metal and concrete culverts. The structures inspected have an average of 65 and 67 inches for concrete and metal culverts, respectively. The NOB culvert reached 106 inches, which is the maximum span for concrete culverts. For metal, 168 inches was the maximum span, which was found in the VIN culvert. Table 4.5. Span Classification per Material Material Span Dimension [S] (in) 42 S 60 60<S 84 S>84 Concrete Metal Age Table 4.6 presents the age range for the culverts inspected. The age classification is specified by the type of material. Concrete culverts did not show a wide variety of age. The average age was 54 years, with the majority of them falling between 54 and 66 years.

74 On the contrary, metal pipes had an average age of 47 years and, a wider range 22 to 74 years. 74 Table 4.6. Age Classification per Material Material Age (years) Age 25 25<Age 50 50<Age 60 Age>60 Unknown Concrete Metal Type of Protection The type of coating or lining presented in the inventory data and confirmed in the field is presented in Table 4.7. Only metal pipes had any type of protective coating. In the field, bituminous coating (BC) and bituminous coating with invert paving (BCIP) were the only type of protection found in some of the metal culverts inspected. Of the metal culverts evaluated, 43% (9 structures) exhibited a BC protection while 9.5% (2 structures) presented BCIP. A total of 10 metal pipes (47.5%) did not have any type of coating. Table 4.7. Type of Protection Classification Type of Protection BC BCIP None BC= Bituminous Coating BCIP= Bituminous Coating Invert Paving

75 Soil Cover Height Table 4.8 shows the soil cover classification for concrete and metal pipes. In general, the metal and concrete pipes inspected presented low soil covers. The average soil cover height was 7.0 feet for concrete and 4.5 for metal. The maximum cover for concrete culverts was 17 feet, which was reached by two structures. Also for metal culverts, the maximum soil cover height was 17 feet, which existed at one structure. Table 4.8. Soil Cover classification per Material Soil Cover over Pipe [H] (ft) Material 0 H 5 5<H 10 10<H 15 15<H 20 Concrete Metal Corrugation Profile The corrugation profile is only relevant to metal culverts. The profiles are classified as CMP and SSP. In Table 4.2, it can be seen that 15 metal culverts (71%) possessed a CMP profile, while 6 culverts (29%) presented a SPP profile. 4.3 Field Data Characteristics The inspections were performed during the months of May and June of 2014 by the ORITE team. The pipes should satisfy the requirements for span or rise greater than 42 inches. The selected pipes were located by using the coordinates provided in the inventory data through the software Google Earth. The pipes were grouped in order to accomplish as many inspections as possible by each trip. The inspection activities were set to assess the current condition of concrete and metal structures. During each field inspection work, the following activities were

76 76 executed 1) confirming the location, material type, and shape; 2) measuring the rise/span dimensions, length, and slope 3) measuring the soil cover height; 4) assessing the culvert conditions 5) assessing the flow conditions (depth and velocity); and assessing the sediment conditions (depth and particle size range). Conditions of the concrete culverts were evaluated in terms of the degree of scaling/spalling/delamination, the degree of cracking, the softness/hardness of concrete material, the area and he conditions of the rebars exposed, the degree of misalignment, and the tightness of joints. Conditions of the metal culverts were evaluated in terms of the span/rise dimensions, metal plate thickness, the degree of rust/scale, the area perforated, the degree of cracking, the softness/hardness of metal late, the degree of settlement, and the tightness of seams and joints The stream water was tested for ph and conductivity at the inlet end. Samples of the backfill soil and stream water were obtained to determine in the laboratory sulfate and chloride concentrations of the water and resistivity and ph of the soil. The inspected structures were rated utilizing the rating scales described in Section 3.3. This section presents the values obtained for concrete and metal pipes in the rating scales applied to establish the current conditions of the barrel, inlet, outlet, settlement, joints of the pipes, protective coating condition, and abrasive flow level. This section also contains the results of the inspection activities and tests executed in the laboratory Pipe Slope Characteristics Table 4.9 presents the range of slope values found in the field during the inspections. Of the concrete culverts, half of them presented a slope close to 1%. The

77 77 average slope for concrete culverts was 2.1%, with the maximum slope of 9.2%, found at the ROS culvert. For metal culverts, the average slope was 2.7%, and the HOC culvert presented the steepest slope of 5.9. Table 4.9. Slope Classification per Material Material Slope [m] (%) m 1 1<m 3 m>3 Concrete Metal Depth and Velocity of Flow Table 4.10 classifies the depth of flow conditions for concrete and metal pipes. The average depth of flow for concrete and metal pipes was 4.7 and 2.2 inches respectively. In general, concrete pipes presented a deeper water flow than metal pipes. The maximum depth of flow in concrete pipes was 25.5 inches for the elliptical pipe ATH Metal pipes exhibited low flow depths in all the structures evaluated. The maximum flow depth was found in the ATH and VIN pipes with a value of 7 inches. Table Depth of Flow Classification per Material Material Depth of Flow [d] (in) No flow d 1 1<d 5 d>5 Concrete Metal According to Table 4.11 most of the pipes had normal flow velocities slower than 1ft/s. Concrete pipes presented an average velocity of 1.02 feet per second. The HOC-56-

78 culvert reached a maximum velocity of 1.85 feet per second. Metal culverts exhibited an average velocity of 0.8 feet per second. The ATH reached a maximum velocity of 2.45 feet per second. Table Flow Velocity Classification per Material Material Flow Velocity [v] (ft/s) No Flow v 1 v>1 Concrete Metal Depth of Sediment The depth of the bed load was measured at every pipe location. Figure 4.5 shows the bed load seen inside a concrete pipe. During the inspection activities, the structures presented bed loads composed mainly of sand, gravel, and silt with different maximum particle sizes. Table 4.12 introduces the sediment depth condition inside concrete and metal pipes. Concrete pipes exhibited higher sediment accumulations, which was around 50% for two structures.

79 79 Figure 4.5. Bed Load of a Concrete Pipe The average depth of sediment for concrete pipes was 6.5 inches. The NOB culvert presented a bed load of 29.4 inches. A total of eight concrete pipes (44%) did not have any bed load. For metal pipes, the average of sediment was 2.7 inches, while the maximum of 8.7 inches was found in the MOR culvert. A total of 10 metal pipes (47%) did not present bed load along the length. Table Depth of Sediment per Material Material Depth of sediment [ds] (in) ds=0 0<ds 10 ds>10 Concrete Metal Characteristics of Water Chemical characteristics of water are determined in the field and laboratory. As explained in Section , the ph and resistivity can be measured directly in the stream, while the sulfate and chloride concentration levels must be detected in the

80 80 laboratory. In this section, the results of the water chemistry data mentioned above are summarized for the pipes analyzed in this study ph and Resistivity of Water Table 4.13 classifies the ph level of water for concrete and metal pipes. 61.5% of the concrete and metal pipes evaluated were exposed to acidic water (ph<7). Extreme acidic conditions were detected at four concrete pipe sites VIN , VIN , and JAC , where the ph obtained was 3.39, 3.55, and 2.54, respectively. The average ph level of water for concrete pipes was Metal pipes were mostly situated at milder acidic conditions compared to concrete pipes. The minimum ph values were found in HOC and NOB with values of 5.10 and 3.97, respectively. The average ph level of water for metal pipes was Table Water ph Classification per Material Material ph of Water ph 6 6<pH<7 ph 7 NA Concrete Metal NA= Not available The resistivity of water was computed once the conductivity measurements were obtained directly from the stream at each location. Table 4.14 presents the classification of the water resistivity values for concrete and metal culverts expressed in Ω.cm. Low water resistivity values may lead to increased corrosion rates in metal and concrete pipes. Most of the pipes (74%) were subjected to water resistivity values lower than 5,000 Ω.cm. The average value was 6,146 and 3,535 Ω.cm for concrete and metal pipes,

81 81 respectively. The minimum water resistivity value for concrete pipes was 602 Ω.cm at the JAC culvert site. Metal culverts were overall exposed lower values compared to concrete pipes, with the minimum resistivity values of 418 and 485 Ω.cm found in the JAC and NOB culvert sites. Table Resistivity of water Classification per Material Material Resistivity of Water [ρ] (Ω.cm) ρ <ρ ρ>10000 NA Concrete Metal NA= Not Available Chloride and Sulfate Content Table 4.15 classifies chloride and sulfate concentration levels found in the water samples. Chloride levels were relatively low, with an average of 6.2 mg/l and a maximum of 30.7 mg/l found at the ATH culvert site. Sulfate level was relatively high (above 160 mg/l) at six culvert sites. The VIN site presented the highest value of mg/l. Table Chloride and Sulfate Classification Chloride [Cl] and Sulfate [SO 4 ] Content (mg/l) Cl 10 Cl>10 SO <SO SO 4 >

82 Properties of Backfill Soil Samples of backfill soil were taken at each metal pipe location. The soil samples were analyzed in the laboratory to determine ph level and resistivity. Table 4.16 presents the ph and resistivity values of the soil samples. At every location, the ph of the soil was basic (ph>7), with an average ph of Most of the locations had soil resistivity values of 2000 to 5000 Ω.cm (71.4%), and the average was 4017 Ω.cm. Table ph and Resistivity of Soil Classification ph of Soil Resistivity of Soil [ρ s ](Ω.cm) 7 ph<8 ph 8 ρ s < <ρ s 5000 ρ s > Rating Scale Classifications The pipes were quantitatively evaluated to establish their current state and produce a model to estimate their remaining service life. The inspected concrete and metal pipes exhibited conditions which varied widely from like-new conditions to deteriorated conditions such as extensive cracking and reinforcement exposed for concrete pipes and rust and pitting through core metal for metal pipes. This section summarizes the ratings of general condition (ODOT and ORITE rating scale), inlet and outlet, joints, settlement, and abrasion conditions that the pipes inspected had received General Condition Ratings of concrete Culverts The condition of the concrete pipes was rated by utilizing the ODOT and ORITE general condition rating scales as described in Sections and 3.3.2, respectively. The general conditions for concrete pipes varied from like-new condition to presence of

83 83 extensive cracking, and loss of mortar on invert. Two concrete pipes had reinforcing bars exposed near the normal flow line. The results showed that culverts in district 9 exhibited better conditions. Most of the concrete culverts in District 9 were rated at 8 in the ORITE and ODOT general rating scales. The worst condition was found in the VIN and VIN culverts located in District 10. These culverts presented significant loss of mortar, softening of concrete, and exposed reinforcement bars. Table 4.17 presents the classification of the general condition ratings of the concrete culverts. Table General Condition Ratings of Concrete Culverts Culvert ID ORITE ODOT ATH to 8 7 HOC NOB to 6 NOB VIN to 4 VIN to 4 ROS ROS JAC JAC JAC JAC JAC to 6 ROS HIG ROS ROS JAC Figure 4.6 shows the concrete culvert, JAC , rated as five (5) in the ORITE general rating scale. This pipe presents a significant loss of mortar and slight loss

84 of aggregates over its invert due to surface scaling. The deterioration may be caused by the acidic conditions found at this location since the water ph Figure 4.6. Loss of Mortar and Aggregate in a Concrete Culvert General Condition Ratings of Metal Culverts The metal pipes were inspected in the ODOT Districts 9 (11 pipes), and 10 (10 pipes).the conditions of the metal pipes were also rated by utilizing the ODOT and ORITE general condition rating scale described in Sections and The general condition for metal pipes varied from like-new condition to heavily deterioration (pitting, rust, and invert deteriorated). For both districts, the metal culverts inspected evidenced favorable and poor conditions according to ORITE and ODOT general rating scale classification. Table 4.18 presents the classification of the general condition ratings of the metal culverts.

85 85 Table General Condition Ratings of Metal Culverts Culvert ID ORITE ODOT ATH to 3 4 to 5 ATH HOC HOC to 3 4 to 5 HOC NOB MRG ROS ATH to 8 7 to 8 VIN VIN to 7 ROS ROS to 4 4 to 5 ROS to 8 7 to 8 JAC JAC to8 7to8 JAC JAC to 6 5 to 6 JAC to 4 4 to 5 JAC to 6 JAC Figure 4.7 shows a metal pipe, HOC , rated at one (1) by the ORITE general rating scale. This structure had perforations throughout its invert with a possibility of collapse. The condition of this pipe might have been provoked by the acidic condition of the water flow (water ph of 5.1) during its 66 years of service life.

86 86 Figure 4.7. Metal culvert with extensive corrosion and pitting Inlet and Outlet Condition Ratings The conditions of the inlet and outlet sections can vary from those of the main barrel due to exposures to sunlight, acid rain, deicing operations, or deterioration produced by the stream load. The inlet and outlet conditions of each culvert were rated by utilizing the same general condition scales proposed by ODOT and ORITE. In this study, the inlet and outlet condition only considers the material condition, structural issues are not included. Table 4.19 presents the ratings that the inlet and outlet sections of the concrete received. For these pipes, the inlet and outlet conditions were always identical rating values compared to the main barrel classification for the ORITE and ODOT rating scale, respectively.

87 87 Table Inlet and Outlet Condition Values for Concrete Culverts Inlet Outlet Inlet Outlet Culvert ID (ODOT) (ODOT) (ORITE) (ORITE) ATH to 8 7 to 8 HOC NOB to 6 5 to NOB VIN to 4 3 to VIN to 4 3 to ROS ROS JAC JAC JAC JAC JAC to 6 5 to ROS HIG ROS ROS JAC Table 4.20 lists the ratings that the inlet and outlet ends of the metal culverts received. JAC and JAC showed different values compared to the main barrel rating values for ORITE and ODOT rating scale. This difference can be explained by the loss of protection coating at inlet and outlet sections, what triggered the corrosion on the invert at both ends.

88 88 Table Inlet and Outlet Condition Values for Metal Culverts Culvert ID Inlet (ODOT) Outlet (ODOT) Inlet (ORITE) Outlet (ORITE) ATH to 3 4 to 5 2 to 3 4 to 5 ATH HOC HOC to 4 NA 2 NA HOC NOB MRG ROS ATH to 8 7 to 8 7 to 8 7 to 8 VIN VIN to 7 6 to ROS ROS to 5 4 to 5 3 to 4 3 to 4 ROS to 8 7 to 8 7 to 8 7 to 8 JAC to 4 3 to 4 3 to 4 JAC to 4 3 to 4 3 to 4 3 to 4 JAC JAC to 6 5 to 6 5 to 6 5 to 6 JAC to 5 4 to JAC to 6 5 to 6 5 to 6 5 to 6 JAC Joint Condition In the inspection forms, the joint sections of the concrete and metal culverts were also quantitatively rated. The joint evaluation was made based on the rating scale proposed only by ODOT (2003). Table 4.21 presents the joint ratings that the concrete and metal culverts received. The concrete culverts had four pipes (16.7%) with poor to fair (4 to 5) rating, and most of the remaining concrete pipes exhibited an acceptable joint condition. Also, the metal culverts presented an adequate joint condition. Metal culverts

89 89 were all rated from satisfactory to excellent (6 to 9). The joints of one pipe were not classified due to the fact that the pipe had no joints within it. Table Joint Ratings of Culverts Joint Condition Material NA Concrete Metal Figure 4.8 shows the joint sections inside of the JAC concrete culvert. This structure exhibited a large separation at some of the joints with signs of backfill soil infiltration. The joint of this pipe was rated at four (4) by the ODOT scale. Figure 4.8. Separated Joint in a Concrete Culvert

90 Settlement Condition Ratings The concrete and metal pipes inspected presented a variety of settlement conditions. Table 4.22 presents their settlement ratings. In general, settlement was not an issue for most of the culverts evaluated. Of the total pipes inspected, 87% received a very good to excellent (8 to 9) settlement rating. Concrete pipes exhibited better settlement performances compared to the metal culverts. The lowest settlement value of six (6) was detected at for the NOB culvert. Two metal pipes HOC and JAC showed marginal to poor conditions. Figure 4.9 shows the JAC metal culvert, where the end section is dropped off. Table Settlement Condition per Material Settlement Condition Material Concrete Metal Figure 4.9. Metal Pipe with a Dropped off End Section

91 Abrasion Level Classifications The abrasion is one of the main factors evaluated in this study. The abrasion levels proposed by DeCou and Davis (2007), as described in Section 3.3.5, were applied during the field inspection work. The concrete and metal pipes exhibited different levels of abrasion according to the bed load material and a potential velocity along the culvert. Figure 4.10 shows the metal pipe HOC with an abrasive flow level of 5. The level 5 was assigned to this pipe due to the size of the bed load with stones up to 12 inches. Figure Metal Pipe with Abrasion Level 5 Table 4.23 presents the abrasion levels found at the culvert sites. A total of nine structures (23%) were exposed to high levels of abrasive flow. No concrete structures were found at the level 6 of abrasion, while four metal culverts existed at the highest level.

92 92 Table Abrasion Level Classification per Material Abrasion Level Material Concrete Metal Basic Analysis of Field Data A preliminary analysis was conducted with the data gathered from the ODOT inventory and the field work. This analysis consisted of statistical descriptions of the data. A simple linear regression analysis was performed for the concrete and metal culverts. The objective was to establish whether environmental factors or just the age affect the service life of the concrete and metal culverts in ODOT Districts 9 and Basic Analysis of Concrete Culverts Table 4.24 introduces basic statistics for concrete culverts. Conditions of concrete pipes varied from pipes with low scores close to failure to those that presented excellent conditions. The average general ODOT and ORITE rating scores were similar, which means, that the rating scales did not differ for this set of pipes.

93 93 Table Basic Statistics of Concrete Culverts N Minimum Maximum Mean Std. Deviation Rise (in) Span (in) Length (ft) Shape Age (years) Soil cover (ft) Abrasion (3) 1.7 (2) ph of water Resistivity of water (Ohm.cm) Chloride[Cl] ( mg/l) Sulfate[SO 4 ] (mg/l) Slope (%) Velocity (ft./s) Depth of flow [d] (in) Depth of sediment [d s ] (in) ODOT (General) (7) 1.75 (2) ORITE (General) (7) 1.88 (2) Joint (1) Settlement (8) 0.81 (1) Inlet (ODOT) (7) 1.75 (2) Inlet (ORITE) (7) 1.75 (2) Outlet (ODOT) (7) 1.75 (2) Outlet (ORITE) (7) 1.75 (2) Shape: 1= Circular; 2= Elliptical The data gathered was used to examine the influence of the age on the deterioration of the concrete pipes. Figure 4.11 presents the simple linear regression correlating the ORITE general rating scale to the age of pipe. According to the plot, the age plays no roles in predicting the deterioration of the concrete pipes. This suggests that other factors must affect the deterioration of concrete pipes. A highly similar plot can be

94 envisioned between the ODOT general rating and the culvert age, since the ORITE and ODOT ratings were basically identical for these concrete pipes. 94 Figure Influence of Age on ORITE Rating Score (Concrete Culverts) Basic Analysis of Metal Culverts Table 4.25 Table 4.25 presents basic statistics obtained for the metal culverts inspected. Conditions of the metal pipes varied from pipes with a score of one (close to the failure) to those with a score of eight (almost new). The averages of the general ODOT and ORITE rating scores were similar; therefore, the rating scales did not differ for this set of pipes. The difference in the number of samples for joint, settlement, and outlet can be explained by the fact that some of the pipes either presented just one section or the outlet was connected to another structure which hindered the evaluation of the section.

95 95 Table Descriptive Statistics of Metal Culverts N Minimum Maximum Mean Std. Deviation Corrugated Profile Shape Rise (in) Span (in) Length (ft) Age (years) Soil Cover (ft) Abrasion (3) 1.7 (2) Soil ph Soil Resistivity (Ohm.cm) Water ph Water Resistivity (Ohm.cm) Plate Thickness (in) Slope (%) Velocity (ft/s) Depth of flow (in) Depth of sediment (in) ODOT (5) 2.16 (2) ORITE (4) 2.42 (2) Joint (8) 0.96 (1) Settlement (8) 1.66 (2) Inlet (ODOT) (5) 2.09 (2) Inlet (ORITE) (4) (2) Outlet (ODOT) (5) (2) Outlet (ORITE) (4) (2) Corrugated Profile: 1= CMP; 2= SPP Shape: 1= Circular; 2= Pipe-Arch Figure 4.12 presents the simple linear regression that relates the ORITE general rating scale to the age of pipe. According to the plot, the age again plays almost no roles predicting the deterioration of the metal pipes. This implies that other factors may affect the deterioration.

96 Figure Influence of Age on ORITE Rating Score (Metal Culverts) 96

97 97 CHAPTER 5: STATISTICAL ANALYSIS OF FIELD DATA The Ohio Department of Transportation (ODOT) and the Ohio Research Institute for Transportation and Environment (ORITE) have been working on developing more accurate models for durability of culverts in order to prevent catastrophic events in the state of Ohio. Given this, several studies such as Degler, Cowherd, and Hurd (1988), Hurd (1988), and Masada et al. (2006, 2007) have developed models to estimate the service life of concrete and metal culverts and establish the urgency when maintenance activities are required. This study updates the service life models by introducing new factors in the analysis related to durability of drainage culverts in Ohio. This chapter analyzes statistically the data gathered during the inspection activities. Linear and non-linear analyses are performed using Statistical Package for the Social Sciences (SPSS) to determine the significant variables that contribute to the deterioration of concrete and metal culverts. The ODOT and the ORITE rating scales are evaluated to point out their effectiveness in estimating the service life of the culverts inspected. Appendix B presents the data utilized for the regression analysis of the concrete and metal culverts. 5.1 Linear Regression Analysis The linear regression analysis was performed for the concrete and metal pipes. The data gathered during the inspection activities provided the information required for the analysis. The objective of the regression analysis was to establish independent equations for metal and concrete culverts with significant factors that affect their durability and explain the rate of deterioration.

98 Linear Regression of Concrete Culverts The linear regression analysis was first performed for the concrete culverts. Two linear equations were generated based on the ODOT and ORITE rating scales. The proposed equations may be applicable to culverts with similar conditions as those presented in the ODOT Districts 9 and 10 of the state of Ohio. The variables introduced in SPSS for the linear regression of concrete culverts are shown in Table 5.1. Table 5.1. Variables for Concrete Culverts Variable Unit Rage ODOT Rating Scale ORITE Rating Scale Shape = Circular 2= Elliptical Rise in Span in Length ft Age years 6-68 Soil Cover ft ph of Water Resistivity of Water Ω.cm Chloride Concentration mg/l Sulfate Concentration mg/l Slope % Flow Velocity ft/s 0-3 Depth of Flow in Depth of Sediment in Level of Abrasion Linear Regression based on ODOT Rating Scale The linear regression analysis was performed for the concrete culvert with the ODOT rating scale as the outcome variable. The analysis was performed in two stages.

99 99 The first stage included two blocks in the SPSS interface. Block 1 contained the significant variables (age, ph w, and abrasion) in the study conducted by Mitchell et al. (2005). Block 2 included all independent variables listed in Table 5.1. The second stage encompassed each of the predictor variables from Table 5.1. Two stages were performed with the Stepwise method (FS, BS, and SS) as explained in Section Table 5.2 presents the b-coefficients of the linear regression. The independent variables ph and Resistivity of water were significant at the level of significance of The significance for ph and resistivity of water are 4.00E-5 and 0.028, respectively. Also, Table 5.2 shows the variance inflation factor (VIF). In accordance with Field (2009), values greater than 10 cause concern and the average VIF values greater than one may be biasing the model. VIF values are close to one; hence, multicollinearity is not a concern for the regression. Table 5.2. Concrete Linear Model Coefficients (ODOT Rating Scale) Unstandardized Standardized Collinearity Model Coefficients Coefficients Statistics t Sig. Std. B Error Beta Tolerance VIF (Constant) ph w E Resistivity (Ohm.cm) E E Dependent Variable: CODOT From Table 5.2, the equation of the linear regression based on the ODOT rating scale can be expressed as:

100 Ω (5.1) Where: ρ= Resistivity of water and ph w = ph of water The Analysis of Variance (ANOVA) is shown in Table 5.3. The F-test is significant at the level of significance of The model presents significantly better prediction compared to the mean of the outcome. According to the ANOVA results, the ability to predict the rating score of a concrete pipe was significantly improved by the model compared to the level of inaccuracy. Table 5.3. ANOVA Concrete (ODOT Rating Scale) Sum of Squares df Mean Square F Sig. Regression E-4 a Residual Total a. Predictors: (Constant), ph, Resistivity (Ohm.cm). Dependent Variable: CODOT Table 5.4 shows the effect of the model on the R 2 and the Durbin-Watson test. In predicting the rating score, the model accounts for the 78% of the variance. The model is still representative when the data is from the population instead of the sample (adjusted R 2 =0.676). Additionally, the Durbin-Watson shows that the assumption of independence of errors is met.

101 101 Table 5.4. Concrete Linear Model Statistic Values (ODOT Rating Scale) R R 2 Adjusted R 2 Std. Error of the Estimate Durbin- Watson a a. Predictors: (Constant), ph, Resistivity (Ohm.cm). Dependent Variable: CODOT According to Figure 5.1, the residuals of the model present a random distribution around zero. This distribution confirms that the assumptions of linearity and homoscedasticity are met. As stated by Field (2009), a standardized residual greater than two may cause concern at 95% of confidence. From the data, it is expected that one case (5%) has a standardized residual higher than the limit as can be seen in Figure 5.1. Figure 5.1. Standardized Predicted against Residual (ODOT Concrete Rating Scale) The normality of residual assumption can be tested with the histogram and normal probability plot. In accordance with Figure 5.2 and 5.3 the histogram is similar to the normal distribution with deviations of non-normality; hence, the assumption of normal distribution of residual is met.

102 102 Figure 5.2. Histogram of Frequencies (ODOT Concrete Rating Scale) Figure 5.3. Normal Probability Plot (ODOT Concrete Rating Scale) Linear Regression Based on ORITE Rating Scale The differences between the ORITE and ODOT rating scales were stated in Section In the field, minor differences existed occasionally between them for the conditions of the pipes inspected. Table 5.5 presents the significant (level of significance of 0.05) independent variables in the linear model based on the ORITE rating system. The significant variables for the model were ph of water and resistivity, with a

103 significance of 2.69E-5 and 0.038, respectively. Multicollinearity is not a concern, since VIF is close to one. 103 Table 5.5. Concrete Linear Model Coefficients (ORITE Rating Scale) Unstandardized Standardized Collinearity Coefficients Coefficients Statistics t Sig. Std. Model B Error Beta Tolerance VIF (Constant) ph w E Resistivity (Ohm.cm) E E Dependent Variable: CORITE From Table 5.5, the equation of the linear regression based on the ORITE rating scale can be expressed as: Ω (5.2) Where: ρ= Resistivity o water; ph w = ph of water The results of ANOVA are presented in Table 5.6. ANOVA shows that the F-test was significant at level significance of The results of ANOVA are slightly better than those presented in Table 5.3 for the ODOT rating scale.

104 104 Table 5.6. ANOVA Concrete (ORITE Rating Scale) Sum of Squares df Mean Square F Sig. Regression E-5 a Residual Total a. Predictors: (Constant), ph, Resistivity (Ohm.cm). Dependent Variable: CORITE The model accounts for 79% of the variance in predicting the rating score. The adjusted R 2 =0.693 shows that the model is still representative when the data is from the population instead of the sample. The Durbin-Watson test was close to two hence the assumption of independence of error is true (See Table 5.7). Comparing to the model based on the ODOT rating scale, the ORITE model slightly improved the accuracy of the regression. Table 5.7. Concrete Linear Model Statistic Values (ORITE Rating Scale) R R 2 Adjusted R 2 Std. Error of the Estimate Durbin-Watson a a. Predictors: (Constant), ph, Resistivity (Ohm.cm) Dependent Variable: CORITE Figure 5.4 confirms that the assumption of linearity and homoscedasticity are true due to the random distribution. As expected, one case (5%) has a standardized residual higher than two (for a 95% level of confidence). The distribution is similar to Figure 5.1 based on the ODOT rating scale.

105 105 Figure 5.4. Standard Predicted against Residual (ORITE Concrete Rating scale) Figure 5.5 and Figure 5.6 show that the assumption of normality is met. The histogram of frequencies is similar to the normal shape distribution. The normal probability plot tends to be normal. Figure 5.5. Histogram of Frequencies (ORITE Concrete Rating Scale)

106 106 Figure 5.6. Normal Probability Plot (ORITE Concrete Rating Scale) Linear Regression of Metal Culverts The data gathered from the inventory database and in the field was utilized to perform the liner regression analysis for the metal culverts. The linear regression equations were obtained for the ODOT and ORITE rating scales. The variables introduced in SPSS are listed in Table 5.8.

107 107 Table 5.8. Variables of Metal Culverts Variable Unit Rage ODOT Rating Scale ORITE Rating Scale Corrugated Profile = CMP 2= SPP Shape = Circular 2= Pipe-Arch Rise in Span in Length ft Age years Soil Cover ft ph of Water Resistivity of Water Ω.cm ph of Soil Resistivity of Soil Ω.cm Thickness in Slope % Flow Velocity ft/s Depth of Flow in 0-7 Depth of Sediment in 0-12 Level of Abrasion Analysis of Coating Protection for Metal Culverts As stated in Table 4.7, a total of 10 metal culverts (48%) possessed a protective coating in addition to the galvanized layer. The presence of bituminous coating (BC) and bituminous coating with invert paving (BCIP) is known to provide additional service life to a metal culvert (National Corrugated Steel Pipe Association or NCSPA, 2008). The NCSPA, in its Corrugated Steel Pipe Design Manual, provided the additional service life, in years, when BC or BCIP is installed on metal pipes at different levels of abrasion (See Table 5.9).

108 108 Table 5.9. Service Life Add-on for Supplemental Pavings and Coatings (National Corrugated Steel Pipe Association, 2008) FHWA Abrasion Level Coating Material Add-on Service Life (years) Asphalt coated N/R N/R Asphalt coated and Paved N/R N/R= Not Recommended The NCSPA (2008) referred to the abrasion levels proposed by the Federal Highway Administration (FHWA). The abrasion is classified into four levels based on the bed load and flow velocity. Table 5.10 describes the abrasion levels proposed by FHWA. Table FHWA Abrasion Levels (Retrieved from: National Corrugated Steel Pipe Association (NCSPA), 2008) No bed load regardless of velocity; or storm sewer Level 1 (Non-Abrasive) applications Minor bed loads of sand and gravel and velocities of 5 ft/s or Level 2 (Low abrasion) less Bed loads of sand and small stone or gravel with velocities Level 3 (Moderate Abrasion) between 5 and 15 ft/s Heavy bed loads of gravel and rock with velocities Level 4 (Severe Abrasion) exceeding 15 ft/s In this study, abrasion is classified into six levels per Davis and Decou (2007). The additional service life provided by the supplemental coating (Table 5.9) can be applied if the six levels of abrasion are related to the four levels proposed by the FHWA. Table 5.11 presents the relationship between the abrasion levels of abrasion proposed by

109 DeCou and Davies (2007) and FHWA, in NCSPA (2008). This relationship was devised in the current study. 109 Table Relationship between Levels of Abrasion FHWA Abrasion Level Abrasion Level (DeCou and Davis 2007) ,3 3 4,5 4 6 The analysis of the supplemental coating requires some years to be subtracted from the original age of the metal pipe. To estimate the years to subtract from each pipe inspected, Table 5.9 and Table 5.11 are applied in agreement with the type of coating and the level of abrasion. In essence, a number of years are subtracted from the original age of the metal pipe if the culvert has a BC or BCIP. Table 5.12 presents the change of age of the metal culverts which had supplemental coatings. The new age after considering the additional service life provided by the supplemental coating is applied just for the statistical regression analysis purpose.

110 110 Table Age of Culverts with Supplemental Coating Culvert ID Type of Coating Original Age (years) Age with Coating (years) ATH BCIP ATH BC HOC BC ATH BCIP ROS BC ROS BC JAC BC JAC BC JAC BC JAC BC Linear Regression Based on ODOT Rating Scale The regression analysis for the metal culverts based on the ODOT rating scale was performed in two stages similar to the linear analysis on the concrete pipes. The first stage was composed of two blocks. Block 1 included age, rise, and corrugated profile (Mitchell et al., 2005). Block 2 encompassed all the variables listed in Table 5.8. The regression analysis was performed in SPSS utilizing the FS, BS, and SS variable selection methods (Section 5.2.1). Table 5.13 introduces the b-coefficients that resulted from the Block 2 linear regression. The independent variables of Span, Age, Abrasion, Thickness, Slope, Flow Velocity, and Depth of Flow were significant at significance level of Rise attained a significance slightly higher than 0.05, but it is included in the model. VIF values are not close to one, but they do not represent a violation of the assumption of multicollinearity.

111 111 Table Metal Linear Model Coefficients (ODOT Rating Scale) Model Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics B Std. Error Beta Tolerance VIF (Constant) Rise (in) Span (in) Age (years) Abrasion Thickness (in) Slope (%) Velocity (ft./s) Depth of flow (in) Dependent Variable: MODOT From Table 5.13, the equation of the metal linear regression based on the ODOT rating scale is as follows: Where: % (5.3) R= Rise; S= Span; t= Thickness; m= Slope; v= Flow Velocity; and d= Depth of Flow Table 5.14 presents the ANOVA for the statistical regression. The F-test is significant at the level of significance of The model represents a significantly better prediction compared to the mean of the outcome.

112 112 Table ANOVA Metal (ODOT Rating Scale) Sum of Squares df Mean Square F Sig. Regression a Residual Total a. Predictors: (Constant), Depth of flow (in), Slope (%), Velocity (ft./s), Thickness (in), Abrasion, Rise (in), Age (years), Span (in) Dependent Variable: MODOT In Table 5.15, the values of R 2, adjusted R 2, and Durbin-Watson test are reported. The model accounted for 81% of the variance in predicting the ODOT rating score of metal culverts. The Durbin-Watson value is considered close to two; therefore, the assumption of independence of errors is true. The model needs more data to be validated, since the adjusted R 2 value was only due to the high number of predictors (eight) and the small sample (18 samples analyzed). Table Metal Linear Model Statistic Values (ODOT Rating Scale) R R 2 Adjusted R 2 Std. Error of the Estimate Durbin- Watson a a. Predictors: (Constant), Depth of flow (in), Slope (%), Velocity (ft./s), Thickness (in), Abrasion, Rise (in), Age (years), Span (in) Dependent Variable: MODOT Figure 5.7 presents a random distribution around zero for standardized residual versus predicted. The random distribution confirms that the assumption of linearity and homoscedasticity are met. No cases presented a standardized residual greater than two (concern for a 95% of confidence).

113 113 Figure 5.7. Standardized Predicted against Residual (ODOT Metal Rating Scale) Figure 5.8 and 5.11 show the histogram of frequencies and the normal probability plot. This histogram possesses a shape comparable with the normal distribution. The normal probability plot tends to be normal. Figure 5.8. Histogram of Frequencies (ODOT Metal Rating Scale)

114 114 Figure 5.9. Normal Probability Plot (ODOT Metal Rating Scale) Linear Regression Based on ORITE Rating Scale In the field, the ODOT and ORITE rating scales exhibited some differences for the metal pipes at low rating scores. The occurrence of the first perforation and the amount of area corroded triggered the difference between the two rating scales. Just as the analysis made for the ODOT rating scale, the linear regression based on ORITE rating scales consisted of two stages. The first stage was comprised of Block 1, which incorporated corrugated profile, ph, abrasion, flow velocity, age, and rise. These variables were significant at the level of significance of 0.05 in the previous study by Mitchell et al. (2005). Block 2 included all the variables listed in Table 5.8. Table 5.16 introduces the b-coefficients that resulted from the Block 2 linear regression. The independent variables Age, Soil Cover, Abrasion, ph of Water, Thickness, Slope, Flow Velocity, and Depth of Flow were all significant at the significance level of Resistivity of Water, ph of Soil attained significances slightly higher than 0.05; however, only the ph of soil is included in the model, since the b-

115 coefficient of resistivity of water is 1.69E-4, close to zero. VIF values are lower than 10, and they do not represent a violation of the assumption of multicollinearity. 115 Table Metal Linear Model Coefficients (ORITE Rating Scale) Model Unstandardized Standardized Collinearity Statistics Coefficients Coefficients t Sig. Std. Beta Tolerance VIF B Error (Constant) Age (years) Soil Cover (ft) Abrasion ph soil ph water Resistivity water 1.69E E (Ohm.cm) Thickness (in) Slope (%) Velocity (ft./s) Depth of flow (in) Dependent Variable: MORITE described as: Where: Based on Table 5.16, the ORITE rating scale for the metal culverts can be % (5.4) H= Soil cover; ph s = ph of soil; ph w = ph of water; t= Thickness; m= Slope; v= Flow Velocity; and d= Depth of Flow

116 116 Table 5.17 presents the ANOVA for the statistical regression. The F-test is significant at the level of significance of The model presents a significantly better prediction compared to the mean of the outcome. Table ANOVA Metal (ORITE Rating Scale) Sum of Squares df Mean Square F Sig. Regression a Residual Total a. Predictors: (Constant), Depth of flow (in), Slope (%), Soil Cover (ft), ph soil, Velocity (ft./s), ph water, Abrasion, Thickness (in), Age (years) Dependent Variable: ORITE Table 5.18 reports the values of R 2, adjusted R 2, and Durbin-Watson test. The model accounted for the 87% of the variance in predicting the rating score of metal culverts. The Durbin-Watson value is close to two, and the assumption of independence of errors is met. The model needs more data to be validated, since the adjusted R 2 value was due to the high number of predictors (10) and the small sample (18 samples analyzed). Table Metal Linear Model Statistic Values (ORITE Rating Scale) Std. Adjusted Error of R R 2 Durbin-Watson R 2 the Estimate a a. Predictors: (Constant), Depth of flow (in), Slope (%), Soil Cover (ft), ph soil, Resistivity water (Ohm.cm), Velocity (ft./s), ph water, Abrasion, Thickness (in), Age (years) Dependent Variable: MORITE

117 117 Figure 5.10 introduces a random distribution around zero for standardized residual versus predicted. This distribution confirms that the assumption of linearity and homoscedasticity are met, similar to Figure 5.7 prepared for the ODOT rating scale. No cases presented standardized residual greater than two (concern for a 95% of confidence). Figure Standardized Predicted against Residual (ORITE Metal Rating Scale) This histogram possesses a shape comparable to the normal distribution (see Figure 5.11). Comparing to the ODOT regression model, the ORITE model presents a better approach to the normal distribution. The normal probability plot tends to be normal (see Figure 5.12).

118 118 Figure Histogram of Frequencies (ORITE Metal Rating Scale) Figure Normal Probability Plot (ORITE Metal Rating Scale) 5.2 Non-Linear Regression Analysis The non-linear regression analysis was performed for the concrete and metal pipes inspected. The non-linear regression employed transformed data to allow the analysis. The transformations consisted of applying a logarithm to each variable. Results from the non-linear regression were compared to those from the linear regression to establish the best model that describes the culvert durability.

119 Non-Linear Regression of Concrete Culverts The non-linear regression analysis generated two equations for concrete culverts based on the ODOT and the ORITE rating scales. The proposed equations may be applied to culverts with similar conditions as those presented in the ODOT Districts 9 and 10. The variables included in the analysis are listed in Table 5.1. These variables are identical to those used in the linear regression analysis Non-Linear Regression Based on ODOT Rating Scale The analysis was performed in two stages again. The first stage analysis included two blocks. Block 1 included Age as stated by Mitchell et al. (2005) for non-linear regression. The second stage was performed in one block with all the independent predictors. Table 5.19 presents the b-coefficients that resulted from the non-linear regression. The independent variables ph of water, Age, and Span were significant at the level of significance of The significance for ph of water, Age, and Span are 0.000, 0.002, and 0.045, respectively. Table Concrete Non-Linear Model Coefficients (ODOT Rating Scale) Model Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics B Std. Error Beta Tolerance VIF (Constant) Age (years) ph w E Span (in) Dependent Variable: CODOT

120 120 Based on Table 5.19, the non-linear regression equation based on the ODOT rating scale can be expressed as: (5.5) Where: ph w = ph of water, and S= Span (in) Table 5.20 shows the effect of the model in the R 2 and the Durbin-Watson test. The model accounts for 82% of the variance in predicting the rating score. The model is still representative when the data is from the population instead of the sample (adjusted R 2 =0.688). Table Concrete Non-Linear Model Statistic Values (ODOT Rating Scale) R R 2 Adjusted R 2 Std. Error of the Estimate Durbin-Watson a a. Predictors: (Constant), Age (years), ph, Span (in) Dependent Variable: CODOT The nonlinear model results present better R 2 value compared to the linear model obtained for the ODOT rating scale. Results suggest that the ODOT rating scale is nonlinear among the different deterioration levels Non-Linear Regression Based on ORITE Rating Scale The non-linear analysis was performed in two stages as usual with the ORITE rating score. The first stage analysis included age as the only significant variable in

121 121 Mitchell et al. (2005). The second stage was performed in one block with all the independent predictors. Table 5.21 presents the b-coefficients that resulted from the linear regression. Two independent, variables ph of water and Age, were significant at the level of 0.05 of significance. The significance for ph of water and Age are and 0.005, respectively. Table Concrete Non-Linear Model Coefficients (ORITE Rating Scale) Unstandardized Standardized Collinearity Coefficients Coefficients Statistics Model t Sig. Std. B Beta Tolerance VIF Error (Constant) ph w E Age (years) Dependent Variable: CORITE The non-linear regression equation based on the ORITE rating scale can be expressed as: (5.6) Where: ph w = ph of water Table 5.22 shows the effect of the model in the R 2 and the Durbin-Watson test. In predicting the rating score, 77% of the variance is accounted for by the model. The model is still representative when the data is from the population instead of the sample (adjusted R 2 =0.659).

122 122 Table Concrete Non-Linear Model Statistic Values (ORITE Rating Scale) R R 2 Adjusted R 2 Std. Error of the Durbin-Watson Estimate a a. Predictors: (Constant), Age (years), ph w Dependent Variable: CORITE The non-linear model did not improve the linear model results previously obtained for ORITE rating scale. This suggests that the ORITE rating scale presents a linear deterioration rate among the levels. For concrete culverts, the best model that describes the pipe deterioration was the non-linear model based on the ODOT rating scale Non-Linear Regression of Metal Culverts The non-linear regression analysis was also performed for the metal culverts in two stages with respect to the ODOT and ORITE rating scales. The first stage was based on the results presented by Mitchell et al. (2005) for non-linear regression. The second stage incorporated all the independent variables described in Table 5.8. The modified age proposed in Table 5.12 was applied to non-linear regression analysis for pipes with supplemental coating. The non-linear analysis was performed after applying a logarithm transformation to the data gathered in the field Non-Linear Regression Based on ODOT Rating Scale The analysis for the ODOT rating score encompassed the Age for the first stage, based on the result reported by Mitchell (2005). Table 5.23 presents the b-coefficients that resulted from the non-linear regression analysis. The independent variable of

123 123 Abrasion was the only significant variable at the level of significance of The significance for Abrasion is Table Metal Non-Linear Model Coefficients (ODOT Rating Scale) Unstandardized Standardized Collinearity Coefficients Coefficients Statistics Model t Sig. Std. B Beta Tolerance VIF Error (Constant) E-8 Abrasion Dependent Variable: MODOT Thus, for the metal pipes the non-linear regression equation based on the ODOT rating scale can be expressed as: (5.7) Table 5.24 introduces the effect of the model in the R 2 and the Durbin-Watson test. Only 37% of the variance is accounted for by the model in predicting the rating score. The non-linear model does not represent the variance of the data. The model presents an adjusted R 2 = Table Metal Non-Linear Model Statistic Values (ODOT Rating Scale) R R 2 Adjusted R 2 Std. Error of the Durbin-Watson Estimate a a. Predictors: (Constant), Abrasion Dependent Variable: MODOT

124 124 The non-linear model for the metal pipes did not improve the linear model previously established for the ODOT rating scale. This suggests that the ODOT metal pipe rating scale presents a linear distribution among the deterioration levels Non-Linear Regression Based on ORITE Rating Scale The analysis for the ORITE rating score incorporated the Age, ph, and thickness for the first stage, based on the results obtained by Mitchell (2005). Table 5.23 presents the b-coefficients that resulted from the non-linear regression analysis. No independent variables were significant variable at the level of significance of The significance for ph w was 0.058, slightly higher than Table Metal Non-Linear Model Coefficients (ORITE Rating Scale) Model Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics B Std. Error Beta Tolerance VIF (Constant) ph w Dependent Variable: MORITE The non-linear regression equation based on the ORITE rating scale can be expressed as: (5.8) Where: ph w = ph of water

125 125 Table 5.26 presents the effect of the model in the R 2 and the Durbin-Watson test. Only 21% of the variance is accounted for by the model in predicting the rating score. The non-linear model does not represent the variance of the data since the model presents an adjusted R 2 = Table Concrete Non-Linear Model Statistic Values (ORITE Rating Scale) Std. Adjusted Error of R R 2 R 2 the Durbin-Watson Estimate a a. Predictors: (Constant), ph w Dependent Variable: MORITE The non-linear regression analysis did not yield satisfying results for the ORITE rating scale for the metal culverts. This suggests that the ORITE metal culvert rating scale presents a linear distribution among the levels.

126 126 CHAPTER 6: SUMMARY AND CONCLUSIONS Drainage culverts represent an essential infrastructure system which should not be ignored. Ensuring the stability of these structures over time provides safer traveling conditions for public commuters. However, the natural process of corrosion, abrasion and erosion affect the durability of drainage culverts The current study was conducted with following objectives (1) ensure that all the influential factors identified by other researchers will be measured during the field inspections for concrete and metal pipes; and (2) develop multivariable regression models to estimate the service life of concrete and metal pipes based on the rating systems proposed by the ODOT and the ORITE. During the literature review stage of the study, it was learned that different factors affect the durability of concrete and metal culverts. Other researchers have found that chloride concentration, age, depth of flow, flow velocity, abrasive flow, slope, water ph were significant in the deterioration of concrete culverts. For metal pipes, the California method, including some modifications of it, has provided reliable results in predicting the service life of galvanized metal culverts in some states. A different analysis has been developed for Ohio, where factors such as plate type, water ph, abrasiveness, flow velocity, and age were significant in the deterioration of metal pipes. In this study, in-service conditions were evaluated to estimate the service life of concrete and metal culverts. The Ohio Research Institute for Transportation and the Environment (ORITE) and a private consulting company applied the inspection methods and rating procedures established by ODOT and ORITE for concrete and metal pipes.

127 127 The inspection activities took place at highway culvert sites located in Ohio Department of Transportation (ODOT) districts 9 and 10, since quite aggressive flow conditions exist in these portions of Ohio. Before each field trip, culverts were selected to meet requirements of location, material, and dimensions. For dimensions, screening criteria were a minimum span/rise of 42 inches and a maximum length of 150 feet. For each culvert, basic information was gathered from the inventory data provided by ODOT. Then, more detailed and site-specific data was collected in the field. The data gathered from the inventory and in the field were statistically analyzed to identify significant factors that contribute to culvert material deterioration. The rating scales proposed by the ODOT and ORITE were employed in the statistical regressions as outcome variables, to measure their effectiveness in predicting the culverts remaining service life. Multivariable linear and nonlinear regression models were employed to estimate the remaining service life of existing metal and concrete structures. Based on the limited data and the findings made in the field and the statistical analysis, the following conclusions can be stated: 6.1 Concrete Culverts Concrete culverts exhibited acceptable conditions except at 2 locations (VIN and VIN ) where the structure exhibited severely deteriorated conditions characterized by mortar and aggregate loss and exposing of reinforcement.

128 128 The conditions of joints and settlement of concrete sections were in general satisfactory. None of the concrete pipe had settlement related issues. Only 2 (VIN and VIN ) pipes possessed poor joint conditions with signs of backfill infiltration. Inlet and outlet rating scores were identical to the main barrel scores. This implies that concrete culverts in ODOT Districts 9 and 10 tend to experience a uniform rate of deterioration throughout their lengths. The ODOT and ORITE general rating scales showed slight differences in describing the current condition of the concrete pipes. The means were 6.67 and 6.61 for the ODOT and ORITE rating scale, respectively, among the concrete pipes inspected. The multivariable linear regression results showed that ph and resistivity of water were significant at the level of significance of 0.05 for the ODOT and ORITE rating scale. The model, based on the ORITE rating scale, accounted for the 79% of the variance in predicting the rating score (R 2 =0.791). Both linear models do not provide a practical estimation of the service life, since the age was not a significant factor in the deterioration of concrete pipes. The multivariable non-linear regression results indicated that ph of water, age and span were significant at the level of significance of 0.05, based on the ODOT rating scale (R 2 =0.823). While, ph of water and age were significant with respect to the ORITE rating scale (R 2 =0.768).

129 129 The model with the best results was the non-linear model based on the ODOT rating scale, which includes ph of water, age, and span as predictor variables with R 2 = According to literature review, water ph and age were found significant by other researchers. Other factors were not significant for this data. Results suggest that the ODOT rating scale is non-linear among the different deterioration levels, while the ORITE rating scale presents a linear deterioration rate. 6.2 Metal Culverts Metal culverts presented a variety of conditions in the main barrel. Some culverts possessed acceptable conditions with minor deteriorations. Other pipes exhibited extreme deteriorations characterized with heavy rust, pitting, and perforation along the main barrel. Joint and settlement condition were favorable in metal pipes. All metal culverts had satisfactory conditions at the joints. JAC culvert exhibited poor conditions due to settlement issues. Inlet and outlet ends at JAC and JAC metal pipes obtained different ratings compared to their main barrels. The difference may be explained since these two pipes had bituminous coatings, which were completely deteriorated at both ends. Beyond these two culvert locations, metal pipes in ODOT Districts 9 and 10 tend to experience a uniform rate of deterioration throughout their lengths.

130 130 The ORITE and ODOT general rating scales presented considerable differences in describing the current conditions of the metal pipes. The mean was 5.11 for the ODOT rating scale, while the mean was 4.28 for the ORITE scale. The multivariable linear regression results showed that rise, span, age, level of abrasion, thickness of the plate, slope, velocity, and depth of the flow were significant at the level of significance of 0.05, based on the ODOT rating scale. The model accounted for the 81% of the variance in predicting the rating score (R 2 =0.814). For the linear regression performed for the ORITE rating scale, age, soil cover, level of abrasion, ph of water, thickness of the plate, slope, flow velocity, and depth of flow were all significant at the level of significance of And, ph of soil obtained a significance of 0.068, slightly higher than The model accounted for the 87% of the variance in predicting the rating score of metal culverts (R 2 =0.874). The linear models do not provide a practical estimation of the service life, since the coefficient of the age is positive. This means that the ratings will increase when the age increases; however, the correct correlation must be that the ratings will decrease when the age increases. The adjusted R 2 for the linear metal culverts did not produce acceptable values, since the high number of predictors included in the model (eight and nine, for ODOT and ORITE, respectively) and the limited number of data (18 cases

131 analyzed). The linear model for metal pipes must be validated using additional data. 131 Non-linear regression did not generate more reliable results in predicting the service life of metal culverts. The non-linear model based on the ODOT rating scale provided the level of abrasion as the only significant variable with a R 2 = For the ORITE scale, no variables were significant at the level of significance of Only ph of water acquired a slightly significance above the 0.05 value, but the model does not represent the variance of the data (R 2 =0.207). The model with the best results was the linear model based on the ORITE rating scale, which included age, soil cover, level of abrasion, water ph, soil ph, thickness of the plate, slope, flow velocity as predictor variables with R 2 = According to literature review, level of abrasion, water ph, soil ph, flow velocity, and age were found significant in the deterioration of metal pipes by other researchers. Other factors were not significant for this data. Results suggest that the ODOT and ORITE rating scales are linear among the different deterioration levels. 6.3 Recommendations for Future Work Based on the results of the statistical models for concrete and metal pipes, the following recommendations can be made for the future work for this project: Additional data is required to improve the statistical models, since the linear regressions did not provide practical results in estimating the service life of concrete and metal pipes.

132 132 It is suggested to obtain information from weather stations to estimate the impact of the rainfall in the deterioration of the culverts. This condition may provide new information on the flow conditions present in the culvert most of the time. To evaluate additional parameters in the streaming water chemistry to establish a correlation between the water ph and the resistivity. The water ph map of Ohio needs to be updated based on the field measurements being taken. The ODOT s previous culvert durability models (ODOT, 1982) should be tested in light of the new data that are collected in the new study. Coupon samples should be taken out of selected metal structures to quantify the loss of protective layer and the loss of metal plate thickness.

133 133 REFERENCES American Society of Testing Material. (2009). ASTM D , Standard Test Methd for ph of Soils. In ASTM, Annual Book of ASTM Standards (pp ). West Conshohocken, PA: American Society of Testing Material. American Society of Testing Material. (2009). ASTM G187-05, Standard Test Method for Measurement of Soil Resistivity Using the Two-electrode Soil Box Method. In ASTM, Annual Book of ASTM Standards (pp ). West Conshohocken, PA: American Society of Testing Material. Ault, J. P., & Ellor, J. A. (2000). Durability Analysis of aluminized Type 2 Corrugated Metal Pipe (No. FHWA-RD ). Office of Infrastructure Research and Development, Federal Highway Administration. Beaver, J. L., & McGrath, T. J. (2005). Management of Utah Highway Culverts. Journal of Transportation Research Board, 1904, Bednar, L. (1989). Plain Galvanized Steel Drainage Pipe Durability Estimation with a Modified California Chart. Journal of Research Transportation Board, 1231, Bellair, P. J., & Ewing, J. P. (1984). Metal-Loss Rates of Uncoated Steel and Aluminum Culverts in New York. Journal of Transportation Research Board, 1001, DeCou, G., & Davies, P. (2007). Evaluation of Abrasion Resistance of Pipe and Pipe Lining Materials (FHWA/CA/TL-CA ). Sacramento, CA: California Department of Transportation.

134 134 Degler, G. H., Cowherd, D. C., & Hurd, J. O. (1988). An Analysis of Visual Field Inspection Data of 900 Pipe-Arch Structures. Journal of the Transportation Research Board, 1191, Eutech Instruments Pte Ltd/ Oakton Instruments. (2008). Instruction Manual PC 300. Field, A. (2009). Discovering Statistics Using SPSS (Third Edition ed.). Washington D.C: SAGE Publications Ltd. Gabriel, L. H., & Moran, E. T. (1998). NCHRP Synthesis 254: Service Life of Drainage Pipe. Washington, D.C: Transportations Research Board, National Research Council. Hadipriono, F. C., Larew, R. E., & Lee, O.-Y. (1988, March). Service Life Assessment of Concrete Pipe. Journal of Transportation Engineering, 114(2), Halmen, C., Trejo, D., & Folliard, K. (2008). Service Life of Corroding Galvanized Culverts Embedded in Controlled Low-Strength Materials. Journal of Materials in Civil Engineering, 20(5), Hurd, J. O. (1988). Service Life Model Verification for Concrete Pipe Culverts in Ohio. Journal of the Transportation Research Board, 1191, Jacobs, K. M. (1984). Durability of Drainage Structures. Journal of Transportation Research Board, 1001, Johnson Level & Tool. (2012). Magnetic Digital Laser Level Model No Retrieved from

135 135 Kurdziel, J. M. (1988). Culvert Durability Rating Systems. Journal of Transportation Research Board, 1191, M.C. Miller Co., Inc. (2010). MILLER 400D Digital Resistance Meter User's Manual. Sebastian, FL. Retrieved from Resistance%20Meter%20User's%20Manual.pdf Masada, T., Sargand, S. M., Tarawneh, B., Mitchell, G. F., & Gruver, D. (2006). New Inspection and Risk Assessment Methods for Metal Highway culverts in Ohio. Journal of the Transportation Research Board, 1976, Masada, T., Sargand, S. M., Tarawneh, B., Mitchell, G. F., & Gruver, D. (2007). Inspection and Risk Assessment of concrete Culverts under Ohio's Highways. Journal of Performance of constructed Facilities, 21(3), Mitchell, G. F., Masada, T., Sargand, S. M., & Jobes Henderson & Associates, Inc. (2005). Risk Assessment and Update of Inspection Procedures for Culverts (FHWA/OH ). Columbus, Ohio: Ohio Department of Transportation. Molinas, A., & Mommandi, A. (2009). Development of New Corrosion/Abrasion Guidelines for Selection of Culvert Pipe Materials (No. CDOT ). Colorado Department of Transportation. Research Brand. National Cooperative Highway Research Program (NCHRP) Synthesis 251. (1982). Assessment of Deficiencies and Preservation of Bridge Substructures Below the Waterline. Washington, D.C: Transportation Research Board, National Research Council.

136 136 National Cooperative Highway Research Program (NCHRP) Synthesis 303. (2002). Assessment and Rehabilitation of Existing Culverts. Washington D.C: Transportation Research Board, National Research Council. National Cooperative Highway Research Program (NCHRP) Synthesis 50. (1978). Durability of Drainage Pipe. Washington, D.C: Transportation Research Board, National Research Council. National Corrugated Steel Pipe Association (NCSPA). (2008). In Corrugated Steel Pipe Design Manual (pp ). Dallas, TX: National Corrugated Steel Pipe Association. Ohio Department of Transportation. (1982). Culvert Durability Study. Columbus, Ohio. Ohio Department of Transportation. (2003). Culvert Management Manual. Columbus, Ohio. Olympus NDT, Inc. (2011). 38DL PLUS Ultrasonic Thickness Gage Basic Operation Manual. Peter, J. A., & Ellor, J. A. (2000). Durability analysis of aluminized type 2 corrugated metal pipe (No. FHWA-RD ). Pfaff, J. D. (1993). Method Determination of Inorganic Anions by Ion Chronomatography. Cincinnati, OH: USEPA Environmental Monitoring Systems Laboratory. Potter, J. C. (1990, March/April). Aluminum-Coated Corrugated Steel-Pipe Field Performance. Journal of Transportation Engineering, 116(2),

137 137 Sagüés, A. A., Peña, J., Cotrim, C., Peach-Canul, M., & Urdaneta, I. (2001). Corrosion Resistance and Service Life of Drainage Culverts. University of South Florida, Department of Civil and Environmental Engineering. Florida Department of Transportation. Salem, O., Salman, B., & Najafi, M. (2012). Culvert Asset Management Practices and Deterioration Modeling. Journal of the Transportation Research Board, 2285, 1-7. Trimble Navigation Limited. (2003). HD150 Distance Meter. Retrieved from

138 138 Concrete Pipe Data Collection Form APPENDIX A: DATA COLLECTION FORMS Location Latitude: District: Longitude: County Route Section Number: Inventory Data: ODOT Culvert File Number (CFN): Year or Date Installed: Length of Pipe (ft): Nominal Size of Pipe (inside diameter, in): Vertical Height (rise) or Diameter (in): Horizontal Width (span) or Diameter: Number of Cells: Pipe Shape (circular or elliptical): Original Wall Thickness (in): Class of Pipe (C76 Class II, III, IV, or V): Type of Protection (none, epoxy coated, vitrified plate lined): Condition of coating: Cover over Pipe (height, ft): Backfill type (and compaction if possible):

139 139 Field Data: Surface Condition: Cracking (max width, % of area): Delamination and Spalling (% of area, depth): Scaling (% of area, depth): Aggregate and mortar loss?: Exposed reinforcing steel? (describe state and amount): Inlet Condition (ORITE & ODOT concrete ratings): Outlet Condition (ORITE & ODOT concrete ratings): Pipe Slope (%): Settlement observations (ORITE settlement rating): Headwall type: Description of Joints (tight, separated, groundwater infiltration, backfill infiltration, voids in fill, alignment, opening, cracking/spaling): Depth of Sediment Inside Pipe (in): Depth of Drainage Flow (in): Drainage Flow Velocity (ft/s): Water Chemistry: ph: Resistivity (Ω.cm) : Chlorides (lab mg/l): Sulfates (lab mg/l): Size Range of Streambed Material (if applicable, in): % of Protection Remaining (if applicable): Description of Invert or Interface (if sedimentation): Softening of Concrete?: Prospector s Pick Observation:

140 140 Ratings: ORITE Pipe Scale: Prominent criteria for rating above: ODOT Pipe Scale: Prominent criteria for rating above: ODOT Joint Scale: Prominent criteria for rating above: Abrasion Level: Is this a system with low bed load and higher velocity? Metal Pipe Data Collection Form Location Latitude: Longitude: District: County Route Section Number: Inventory Data: ODOT Culvert File Number (CFN): Year or Date Installed: Length of Pipe: Nominal Size of Pipe (inside diameter): Vertical Height (rise) or Diameter: Horizontal Width (span) or Diameter: Number of Cells: Corrugation Profile (CMP or SPP): Pipe Shape (circular or pipe arch): Original Wall Thickness: Metallic Coating (galvanized or aluminized): Type of Protection (1/2 or full asphalt coated, 1/2 or full asphalt coated with paved invert, concrete field paved, polymeric coated): Headwall Type: Cover over Pipe (height): Backfill type (and compaction if possible):

141 Field Data: Current Wall Thickness (at undeteriorated location): Surface Condition within Main Barrel: Condition of coating or corrosion (% of area): Rust (degree, flakes, scaling): Perforation (% area covered or area ratio, location): Pitting (% of area, depth): Inlet Condition (ORITE & ODOT metal pipe ratings): Outlet Condition (ORITE & ODOT metal pipe ratings): Pipe Slope: Settlement observations (ORITE settlement rating): Headwall type: Description of Joints (tight, separated, groundwater infiltration, backfill infiltration, voids in fill, alignment, opening, cracking): Depth of Sediment Inside Pipe: Depth of Drainage Flow: Drainage Flow Velocity: Water Chemistry: ph: Resistivity: Soil Chemistry: ph: Resistivity: Size Range of Streambed Material (if applicable): % of Protection Remaining (if applicable): Description of Invert or Interface (if sedimentation): Prospector s Pick Observation: Metal Thickness (at least 3 representative locations at Invert or Interface): % of Invert or Interface Perforated: Inside Diameter Measurements (if applicable): 141

142 Ratings: ORITE Pipe Scale: Prominent criteria for rating above: ODOT Pipe Scale: Prominent criteria for rating above: ODOT Joint Scale: Prominent criteria for rating above: Abrasion Level: Is this a system with low bed load and higher velocity? 142

143 143 APPENDIX B: INVENTORY AND FIELD COLLECTED DATA Table B.1. Concrete Pipe Inventory and Field Data Culvert ID Shape Rise (in) Span (in) Length (ft) Age (years) Soil cover (ft) Slope (%) Vel. (ft./s) Depth of flow (in) Depth of sediment (in) ph Resistivity (Ω.cm) Chloride [Cl] (mg/l) Sulfate [SO 4 ] (mg/l) ATH NA HOC NOB NOB VIN VIN ROS ROS NA JAC JAC JAC JAC JAC ROS HIG NA ROS ROS JAC Shape: 1= Circular; 2= Elliptical. Vel.= Flow Velocity

144 144 Table B.2. Concrete Pipe Rating Scores Culvert ID Abrasion ORITE ODOT Joint Settlement Inlet (ORITE) Inlet (ODOT) Outlet (ORITE) Outlet (ODOT) ATH HOC NOB NOB VIN VIN ROS ROS JAC JAC JAC JAC JAC ROS HIG ROS ROS JAC

145 145 Table B.3. Metal Pipe Inventory and Field Data Culvert ID Shape CP Rise (in) Span (in) L (ft) Age (yr.) Soil Cover (ft) ph s Resistivity (Ω.cm) (soil) ph w Resistivity (Ω.cm) (water) Thickness (in) ATH ATH HOC HOC HOC NOB MOR ROS ATH VIN VIN NA ROS NA ROS ROS JAC JAC JAC JAC JAC JAC JAC Shape: 1= Circular; 2= Pipe Arch. CP: Corrugated Profile. 1= CMP; 2=SPP. ph w = water ph. ph s = soil ph. Slope (%) Vel. (ft./s) Depth of flow (in) Depth of sedime nt (in)

146 146 Table B.4. Metal Pipe Rating Scores Culvert ID Abrasion ORITE ODOT Joint Settlement Inlet (ORITE) Inlet (ODOT) Outlet (ORITE) Outlet (ODOT) ATH ATH HOC HOC NA NA HOC NA NOB MOR ROS ATH VIN VIN ROS ROS ROS NA NA JAC JAC JAC JAC JAC JAC NA JAC NA= Not available

147 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Thesis and Dissertation Services!