Transcription

TRC1203Data Preparation for ImplementingPavement-ME Design (DARWin-ME / MEPDG)Kevin D. Hall, Kelvin C. P. Wang, Joshua Q. LiFinal Report2015

FINAL REPORTTRC-1203Data Preparation for Implementing Pavement-ME Design (DARWin-ME / MEPDG)ByKevin D. Hall, Ph.D., P.E.University of ArkansasKelvin C. P. Wang, Ph.D., P.E.Joshua Q. Li, Ph.D.Oklahoma State UniversityConducted byDepartment of Civil EngineeringUniversity of ArkansasDepartment of Civil and Environmental EngineeringOklahoma State UniversityIn Cooperation withArkansas State Highway and Transportation DepartmentU.S. Department of TransportationFederal Highway AdminstrationUniversity of ArkansasFayetteville, AR 72701May 2015

TABLE OF CONTENTSCHAPTER 1INTRODUCTION . 81.1Background. 81.2Objectives and Tasks . 91.3Report Outline . 10CHAPTER 2OVERVIEW OF PREP-ME 3.0 SOFTWARE . 122.1General Overview . 122.2Traffic Data Import. 132.3Traffic Data Check . 142.4Traffic Data Export. 142.5Climate Module . 152.6Material Module . 152.7Preliminary FWD Module . 162.8Prep-ME Tools . 16CHAPTER 3TRAFFIC DATA IMPORT . 173.1Traffic Data Formats and Naming Convention . 173.2Travel Monitoring Analysis System (TMAS 2.0) Data Check . 173.3Prep-ME Software Interface . 18CHAPTER 4TRAFFIC DATA CHECK . 204.1Weight Data Check. 20ii

4.1.1Automatic TMG Data Check Algorithms . 204.1.2Data Sampling and Replacement . 224.2Traffic Classification Data Check . 24CHAPTER 5TRAFFIC DATA EXPORT . 265.1Traffic Data Export Levels . 265.2Output Level 1- Site-Specific . 285.3Output Level 2 -Michigan DOT Clustering . 295.3.1Methodology. 295.3.2Prep-ME Interface - Setup Clusters . 305.3.3Prep-ME Interface - Run Discriminant Analysis . 345.4Output Level 2 -NCDOT Clustering . 355.4.1Methodology. 355.4.2Prep-ME Interface . 375.5Output Level 2 - KYTC Method . 425.6Output Level 2 -TTC Clustering . 455.6.1Methodology. 455.6.2Software Interface . 465.7Output Level 2 -Simplified TTC Clustering . 485.7.1Methodology. 485.7.2Software Interface . 505.8Output Level 2 - Flexible Clustering . 51iii

5.9Output Level 3 . 525.10Mixed Output Levels and Output Data Review . 52CHAPTER 6CLIMATE MODULE . 546.1Climate Data Import . 546.2Export Climate Data . 55CHAPTER 7MATERIAL MODULE . 577.1Introduction . 577.2Dynamic Modulus (E*) for HMA . 577.3Coefficient of Thermal Expansion (CTE) for PCC . 587.4Soil Map for Subgrade. 59CHAPTER 8Preliminary FWD Module . 638.1Introduction . 638.2FWD Capabilities . 63CHAPTER 9PREP-ME TOOLS . 669.1File Name Change . 669.2AADTT Calculation Based on Short Term Traffic Counts. 67CHAPTER 10CONCLUSIONS . 69CHAPTER 11REFERENCES . 71iv

LIST OF FIGURESFigure 2.1 Prep-ME 3.0 Main Interface . 13Figure 3.1 Interface of data import . 19Figure 3.2 Google Map 3.0 Utility . 19Figure 4.1 Detailed Traffic Information by Lane . 21Figure 4.2 Daily Check and Sampling . 23Figure 4.3 Classification Data Check by Direction and by Lane . 25Figure 4.4 Classification Daily Data Check . 25Figure 5.1 Three-Level Outputs . 28Figure 5.2 State Average for Number Axles/Truck . 28Figure 5.3 Output Level 2 – Michigan DOT Method . 31Figure 5.4 Set Up Michigan DOT Clusters . 32Figure 5.5 Identifying Traffic Pattern . 35Figure 5.6 NCDOT Method . 39Figure 5.7 Traffic Output by Class Comparison . 40Figure 5.8 Traffic Output by Load . 40Figure 5.9 Traffic Output by Station Information . 41Figure 5.10 KYTC Method . 44Figure 5.11 DARWin-ME TTC Values . 45Figure 5.12 TTC Clustering Method . 47Figure 5.13 Review TTC Clusters . 47v

Figure 5.14 Check TTC Plots . 48Figure 5.15 Simplified TTC Approach (Li et al, 2012) . 49Figure 5.16 Simplified TTC Clustering Method . 50Figure 5.17 Flexible Clustering Method . 51Figure 5.18 Displaying Output Data . 53Figure 6.1 Importing Climate Files . 54Figure 6.2 Google Map 3.0 Utility for Climate Data . 55Figure 6.3 Interpolating Climate Files. 56Figure 6.4 Selected Climate Stations on Google Map . 56Figure 7.1 Retrieving Dynamic Modulus (E*) Data . 58Figure 7.2 Retrieving CTE Data . 59Figure 7.3 Soil Map Module in Prep-ME . 61Figure 7.4 Retrieved Soil Properties . 61Figure 7.5 Generated Soil Property File for Pavement-ME Design . 62Figure 8.1 Import FWD Data . 64Figure 8.2 Generate Report for FWD Back-Calculation . 64Figure 8.3 Output FWD XML File for Pavement-ME Design . 65Figure 9.1 Change File Name Interface . 66Figure 9.2 AADTT Prediction Based on Short Term Traffic Count . 68vi

LIST OF TABLESTable 5.1 Traffic Input Level for Rigid Pavements (Haider et al 2011) . 31Table 5.2 Algorithm for Recommending an ALDF Group for NCDOT . 41Table 5.3 Aggregation Class of roadway in Kentucky. 44vii

CHAPTER 1 INTRODUCTION1.1BackgroundPavement-ME Design (previously DARWin-ME; also known as the MechanisticEmpirical Pavement Design Guide [MEPDG]) is a significant advancement inpavement design technology. In this report, the terms Pavement-ME Design ,DARWin-ME, and MEPDG are used interchangeably. AASHTO, FHWA, NCHRP,and many state highway agencies in the US have spent well over 50 million in the pastdecade on developing, refining, and calibrating the MEPDG procedure. Arkansas StateHighway and Transportation Department (AHTD) as a leader in MEPDG studies startedsupporting MEPDG research early on.As the next-generation pavement designprocedure, Pavement-ME Design is embraced by many state highway agencies. As itrequires a magnitude more data inputs, some of which are not familiar to pavementdesigners and not systematically stored and archived, it is imperative to have a processin place for AHTD to collect, analyze, prepare, and use the input data sets forPavement-ME Design . Equally important, Pavement-ME Design will be also usedas an analysis tool for pavement engineering due to its inclusion of many engineeringprinciples, including prediction models, materials analysis, construction and as-builtdatabase, environment, and qualification of traffic data. This research project relies onknow-how and experience from past AHTD sponsored projects on MEPDGdevelopment and establish a workflow in implementing Pavement-ME Design atAHTD with the long-term goal of establishing a supporting infrastructure for pavement8

engineering at AHTD using Pavement-ME Design as the core pavement designengine.1.2Objectives and TasksThe primary objective of the proposed study is to establish a workflow for AHTD tostart implementing DARWin-ME for production and develop relevant technologies sothat positive impacts of DARWin-ME will be fully exploited in pavement design,management, materials, construction, and traffic data collection. The objectives of thisproject include: Develop a DARWin-ME Implementation Plan for AHTD. Develop necessary software tools and processes for integrating numerousAHTD data sets for multiple purposes such as design, management,construction activities etc. Develop statewide database of traffic and materials for the initialimplementation of DARWin-ME. Develop new pavement design manuals for the implementation of DARWinME in the state of Arkansas. Conduct training and workshops for AHTD designers and industryrepresentatives to use DARWin-ME.More specifically, there are five tasks for this project: Task 1: Review of Literature and State of Practice Task 2: Software Development to Integrate Data from Different Divisions Task 3: Development of Statewide Database Task 4: Specifications and Design Document9

Task 5: Education, Training, and Final ReportThe University of Arkansas functioned as the contractor on the project;however, a large portion of the work was performed under a subcontract to OklahomaState University. It is noted that AHTD chose not to pursue Task 3 as originallyproposed. Task 2 – software development – is the primary work, and comprises theprimary deliverable of the project.1.3Report OutlineThis report documents the research, mainly focused on the development of the newversion of the Prep-ME software. The capabilities of the updated Pre-ME version 3.0software are introduced in the following chapters.Chapter 2 provides an overview of the Prep-ME 3.0 software.Chapter 3 illustrates the traffic data import functionalities in Prep-ME;Chapter 4 provides a detailed documentation of traffic data checks for bothweight and classification data. Automated data check algorithms in accordance withFHWA Traffic Monitoring Guide (TMG), but also various data operations such asmanually process, daily sampling and monthly sampling are available in Prep-ME 3.0for users to perform comprehensive WIM traffic data checks.Chapter 5 emphasizes on how to export traffic data for Pavement-ME Design software for specific pavement design based on available WIM data. Several clusteringmethods are implemented in the software.Chapter 6 demonstrates the capabilities of Prep-ME for climate module.Chapter 7 demonstrates the material module in Prep-ME. Dynamic modulus(E*) for HMA materials and coefficient of thermal expansion (CTE) for PCC materials10

can be retrieved from the statewide material library. In addition, Prep-ME is able toretrieve soil maps and related soil property data describing the soil-water characteristiccurves (SWCC) from the pedologic soil family national database developed by theNCHRP 9-23A project.A preliminary Falling Weight Deflectometer (FWD) module and Prep-ME toolsare included in Chapters 8 and 9, respectively.11

CHAPTER 2 OVERVIEW OF PREP-ME 3.0 SOFTWARE2.1General OverviewIn Prep-ME 3.0, the database platform has been changed from Microsoft Access to SQLServer. As a result, the data storage capability has been increased from 2GB to 10 GB(for Express version of SQL Server) or 16 TB (for Standard version of SQL Server).The computation efficiency has been improved dramatically in the new Prep-ME byimplementing several new programming algorithms.As shown in Figure 2.1, Prep-ME 3.0 software includes four menus: Traffic,Climate, Materials, and Tools. For traffic module, Prep-ME contains five main submodules: Import Traffic Data, Check Station Data, Check Weight Data, CheckClassification Data, and Export Traffic Data. For climate module, Prep-ME can importraw traffic data (Import Climate Data) and interpolate virtual climate files (ExportClimate Data) for the Pavement-ME Design software. In Material Module, dynamicmodulus (E*) for HMA (HMA E*), Coefficient of Thermal Expansion (CTE) for PCC(PCC CTE), soil map data (Soil Map), and FWD data (FWD) can be imported in PrepME and output data for Pavement-ME Design . Prep-ME also provides tools to aidstate DOTs in using the software.12

Figure 2.1 Prep-ME 3.0 Main Interface2.2Traffic Data ImportThe Import Traffic Data sub-menu is able to: Import raw traffic data provided by state highway agencies. Regardless oftraffic data collection techniques (such as Weigh-In-Motion, AutomaticVehicle Classification) and time coverage (such as permanent long term,short term counts), the traffic data cannot be imported ONLY if the data filesare saved strictly following the formats defined in the FHWA's TrafficMonitoring Guide (TMG), namely S-Card, C-Card, and W-Card. Conduct Travel Monitoring Analysis System (TMAS 2.0) data check foreach line of raw data, and report errors into an error log file for eachimported file. The TMAS 2.0 data check is documented in the 2013 versionof Traffic Monitoring Guide, and provided in Appendix A. The data withcritical errors are not imported into the Prep-ME database.13

Process the raw data which have passed the TMAS data check and savethem in the Prep-ME database tables.2.3Traffic Data CheckThe Traffic Data Check sub-menu is able to: Conduct QC check for both classification and weight data by direction andlane of traffic using data check algorithms defined in the TMG. Provide interfaces to review monthly, weekly and daily traffic data. Provide various manual, replacement, and sampling operations to analyzeand utilize incomplete or failed data.2.4Traffic Data ExportThe Export Traffic Data for traffic data is able to: Provide three levels of traffic outputs: Level 1 Site Specific, Level 2Clustering Average, and Level 3 State Average. The Level 1 traffic inputscan be generated based on a WIM station or one direction of traffic. Thereare in total five clustering methods for Level 2 traffic inputs, includingNCDOT method, Michigan DOT method, KYTC Method, Truck TrafficClassification (TTC) method, simplified TTC method, TPF-5(004) Method,Flexible Clustering. State average values or Pavement-ME Design defaults can be used for Level 3 inputs. Prep-ME allows each type of trafficdata to select its own output level. For example, Level 1 is selected forVehicle Class Distribution (VCD) data, while Level 3 data may be used forhourly adjustment factors.14

Implement independent C codes of Ward-based HierarchicalAgglomerative clustering algorithm, which is used in both NCDOT andMDOT clustering analysis, is implemented in Prep-ME. This algorithm willallow users to evaluate existing clusters and define new clusters if necessary. Generate 11 traffic input files in text file format for MEPDG and two XMLtraffic files for Pavement-ME Design software.2.5Climate ModuleThe climate module in Prep-ME 3.0 is able to: Import Hourly Climate Data (HCD) files, including those from thePavement-ME Design software and new data sources provided by stateDOTs, into Prep-ME database. Conduct preliminary data checks to the raw climate data. Interpolate ICM file and XML file that can be directly imported to forMEPDG and the Pavement-ME Design software.2.6Material ModuleThe Material Module in Prep-ME 3.0 is able to: Import raw FWD F25 data into Prep-ME database, output a summary reportfor back-calculation software, generate FWD XML file for Pavement-MEDesign . Retrieve dynamic modulus (E*) data for HMA materials from statewidematerial library for Pavement-ME Design . Retrieve Coefficient of Thermal Expansion (CTE) data for PCC materialsfrom statewide material library for Pavement-ME Design .15

Retrieve NCHRP 9-23A subgrade soil map data for Pavement-MEDesign .2.7Preliminary FWD ModuleThe FWD module in Prep-ME 3.0 is able to: Import raw FWD F25 data and pavement structure data into Prep-MEdatabase; Output a summary report including pavement structure data along with thedeflection data for use in back-calculation process; 2.8Generate FWD XML file for Pavement-ME Design .Prep-ME ToolsCurrently, Prep-ME 3.0 provides two tools to: (1) change traffic file names that don'tcomply with the Traffic Monitoring Guide name conventions; (2) calculate AnnualAverage Daily Truck Traffic (AADTT) and Vehicle Class Distribution (VCD) factorsbased on 24-hour or 48-hour short term traffic count data.16

CHAPTER 3 TRAFFIC DATA IMPORT3.1Traffic Data Formats and Naming ConventionThe Prep-ME 3.0 software can only import traffic data that comply with the dataformats recommended in the FHWA Traffic Monitoring Guide (TMG). Collected trafficdata are classified into four types in TMG: station description data, traffic volume data,vehicle classification data, and truck weight data. Specific coding instructions andrecord layouts can be found in Chapter 6 in the 2001 Traffic Monitoring Guide. Therecommended file naming conventions are "ssyy.STA", "ssyy.CLA", and "ssyy.WGT"for station, classification and weight data sets, where ss is state postal abbreviation andyy is the last two digits of the year. In case that state DOTs don't follow therecommended name conventions to store traffic data, Prep-ME provides a tool tochange the file names in a batch mode so that the data can be imported to the Prep-MEdatabase.The 2013 version of TMG guide also provides record layouts with minorchanges. In addition to the four files above, the 2013 TMG guide requires collectingtwo more data files (speed data and the per vehicle data referred to as PVF). Each typeof data has its own individualized record format.3.2Travel Monitoring Analysis System (TMAS 2.0) Data CheckTMAS stands for Travel Monitoring Analysis System. TMAS provides online datasubmitting capabilities to State traffic offices to submit data to FHWA. Access toTMAS is obtained through the FHWA Division office in the individual State. TMAS17

2.0 provides a set of traffic data checks, as provided in Appendix A. All the TMASchecks are implemented in Prep-ME 3.0 during traffic data import.3.3Prep-ME Software InterfaceAfter selecting a file folder and clicking “OK” button, all classification, stationdescription and weight files in this file folder and its sub-folder will be imported to thePrep-ME database. Figure 3.1 shows a screen shot of data import processing. Current/Total Files: The index of current processing file verse total number offile selected to be imported; Imported (Rows): number of rows of data imported1into the database; Failed TMAS (Rows): the number of records (rows) that failed the TMAS21check; 3 records to the total number of dataFailed Rate: the percentage of failed TMAS4imported;11 Duplicate: number of rows (records) that are duplicate in the raw data sets; Currently Import File: The path of current raw data file under importprocessing; Total Processing Time: the processing time of data importing in seconds; Stop Importing: user can stop importing the data being processed.A detailed TMAS checking error report file will be generated for each importedfile and located in the same directory as the raw files that have been processed. Datalines with critical errors will not be imported by Prep-ME.18

Figure 3.1 Interface of data importAfter data importing, the geo-referenced Google Map 3.0 is activated to showthe geographical relationships among the design project, WIM stations, and thesurrounding area. This mapping utility has all major functions of Google Map 3.0, suchas displaying satellite imagery. Users can click the traffic station legend for moredetailed information (Figure 3.2).Figure 3.2 Google Map 3.0 Utility19

CHAPTER 4 TRAFFIC DATA CHECK4.14.1.1Weight Data CheckAutomatic TMG Data Check AlgorithmsThe algorithm used in the 2001 3rd Edition of TMG for weight is adopted for weightdata check. There are two basic steps to evaluate recorded vehicle weight data. Firstly,to check the front axle and drive tandem axle weights of Class 9 trucks. The front axleweight should be between 8,000 and 12,000 lb (10,000 2,000 lb). The drive tandemsof a fully loaded Class 9 truck should be between 30,000 and 36,000 lb (33,000 3,000lb). Secondly, to check the gross vehicle weights of Class 9 trucks. The histogram plotshould have two peaks for most sites. One represents unloaded Class 9 trucks andshould be between 28,000 and 36,000 lb (32,000 4,000 lb). The second peakrepresents the most common loaded vehicle condition with a weigh between 72,000 and80,000 lb (76,000 4,000 lb).Figure 4.1 demonstrates the interface for weight data check. Default TMG QCCriteria are built into Prep-ME and the stations are automatically classified as"Accepted" and "Unaccepted". Because a minimum of 12-month data within a year(from January to December) are required to prepare the loading spectra data inputs forthe Pavement-ME Design software, the Prep-ME software will classify stations as"unaccepted" if they don't have a minimum of 12-month data that pass the QC. PrepME also allows users to adjust those parameters. In addition, users can opt not to applyone or all the QC criterion for weight data check by unselecting them.20

For each station, the detailed traffic information can be reviewed by users. Thecorresponding histograms for each data check criterion can be checked by switching theradio buttons. The monthly QC check results can be viewed by WIM station, bydirection of a station, and by direction & lane of a station.For WIM stations don't have a minimum of 12-month data, Prep-ME providesfunctionalities on how to use those incomplete traffic data sets for the Pavement-MEDesign software through various operations, such as manual, sampling andreplacement operations.3Figure 4.1 Detailed Traffic Information by Lane21

4.1.2Data Sampling and ReplacementFour sampling and repair options are provided in the Prep-ME: Manual Operation(Accept and Reject), Replacement (Copy and Paste), and Sampling Operation(Daily Sampling and Monthly Sampling). Prep-ME uses five different backgroundcolors to differentiate various QC checking status as shown in Figure 4.1.Manual Operation (Accept/Reject) allows users to review and double check theautomated QC results. If users confirm that the software has misclassified the datacheck status, users could manually accept or reject this month’s data.Daily check and sampling operation is useful in three situations: It can be used as a diagnostic tool to investigate the reason(s) for bad datathat cannot pass automatic data check. When WIM sensors malfunction in the middle of a month, samplingoperation can be used to prune failed daily data. When multiple days’ data is missed in a month, sampled weekly data can beused as a substitute for that month.Occasionally, multiple days' of data are missing within a month for some WIMstations. In this case, users may want to sample the available data to represent thismonth. In addition, users may be interested in investigating the data trend for a specificDay of Week (for example, all 5 Mondays as shown in Figure 4.2). Therefore, the PrepME software has designed the function that allows user to select multiple days of dataand show the results in the QC Plots and Daily Data Summary figures.22

Figure 4.2 demonstrates the comparisons of the Gross Vehicle Weight data forall the five Mondays in the selected month. It is anticipated that the data be consistentamong the five Mondays. However, it is seen that the data for the first Monday showsdifferent trend from those for the other Mondays. Users may investigate the data anddecide whether the data is reasonable.Figure 4.2 Daily Check and SamplingWhen one month data is missing or fail to pass the data check algorithms, userscan apply "Copy" and "Paste" operation by checking the similarity of the data inadjacent months, opposite direction, or different lane, same month but different year,and then identify a suitable month which can be used as the “source month” tosubstitute the failed or missing month (the “target month”).23

Since WIM sites can collect many years of data, users may only be interested inusing twelve consecutive months’ data right after a WIM system calibration or 12selected months' data based on engineering judgment for pavement design. Prep-MEprovides users with monthly sampling either by direction or by lane.4.2Traffic Classification Data CheckClassification data check follows the four-step algorithms defined in the TMG guide:(1) to compare the manual classification counts and the hourly vehicle classificationdata. The absolute difference should be less than five percent for each of the primaryvehicle categories. (2) To check the number of Class 1 (motorcycles). The evaluationprocedure recommended that the number of Class 1 should be less than five percent1unless their presence is noted. (3) To check the reported number of unclassifiedvehicles. The number of unclassified vehicles should be less than five percent of thevehicles recorded. (4) To compare the current truck percentages by class with thecorresponding historical percentages. No significant changes in the vehicle mix areanticipated. The first step is not processed since no manually data are available. Thesecond and third step can be checked with the imported vehicle classification data. Inthe fourth step, the TMAS2.0 consistency check is applied. By default, MADT fromsame month previous year should be within 30%.The Prep-ME software provides similar

In Prep-ME 3.0, the database platform has been changed from Microsoft Access to SQL Server. As a result, the data storage capability has been increased from 2GB to 10 GB (for Express version of SQL Server) or 16 TB (for Standard version of SQL Server). The computation effici