Evaluation and Analysis of LTPP Pavement Layer Thickness Data

This chapter summarizes the results of the data availability and completeness assessment for tables related to pavement layer structure. First, the LTPP data source used for this study is presented. Then, LTPP data availability and quality control (QC) are discussed, which explains the QC process of the LTPP data and why some data collected are deemed "unreleasable" to the public. After that, layer structure and thickness data are assessed for their quality level and completeness.

LTPP Data Source Used in This Study

LTPP data release 11.5 version NT3.0, obtained on June 8, 2001, was used for this study. LTPP tables with layer material type and thickness data for individual layers at the section level are evaluated for data availability and completeness for the relevant sections. Tables TST_AC01 and TST_L05 were not included in this study.

Table TST_AC01 was not evaluated in this study because it contains measured core thickness, which may represent thickness from multiple layers. For example, a single AC core identified in the field as AC material and with measured thickness in the TST_AC01 table may contain hot-mix asphalt concrete (HMAC) wearing, binder, and base layers.

Table TST_L05 was not used because it contains information only for SPS projects at the project level. Many SPS projects contain multiple sections at the same site (e.g., SPS-1 and SPS-2). This table is useful for researchers who would like to link material type information from multiple sections at the section level together for a given SPS project.

The following LTPP tables were assessed for data availability and completeness:

• TST_AC01_LAYER-Core examination and thickness information. Contains field layer and real layer number.
• TST_PC06-Core examination and thickness.
• SPS*_LAYER-Layer descriptions (Sheet 4).
• SPS*_LAYER_THICKNESS-Layer thickness measurements (Sheet 12).
• INV_LAYER-Layer descriptions (Data Sheet: Inventory 3).
• RHB_LAYER-Layer descriptions (Data Sheet: Rehab 2).
• TST_L05A-Table containing layer descriptions for all constructions.
• TST_L05B-Table containing layer descriptions for all constructions.

LTPP Data Availability and Quality Control Checks

The quality of the data is the most important factor in any type of pavement performance analysis. From the onset of the LTPP program, data quality has been considered of paramount importance. Procedures for collecting and processing data were defined (and are modified as necessary) to ensure consistency across various reporting contractors, laboratories, and equipment operators. Although these procedures formed the foundation of quality control/quality assurance (QC/QA) and data integrity, many more components of a QC/QA plan were necessary to ensure that the data sent to researchers were as error-free as possible.

LTPP has developed and implemented an extensive QC program that classifies each of the data elements into categories, depending upon the location of the data in this QC process. Several components or steps comprise the overall QC/QA plan used on LTPP data, as discussed in the following paragraphs [20].

1. Collect Data: Procedures for collecting data are documented for each module in the LTPP database. These procedures are intended to ensure that data are collected in similar formats, amounts, conditions, and so on. Documentation references include the Data Collection Guide and various module-specific guides.
   
   
2. Review Data: Regional engineers review essentially all data input into the regional LTPP databases to check for possible errors related to keystroke input, field operations, procedures, equipment operations, and so on. The regional review is intended to catch obvious data collection errors. In addition, some data are preprocessed before they are entered into the LTPP database. For example, PROFCAL software is used on the profilers to provide a system check by comparing measurements taken at different speeds. PROFSCAN is a field QA tool that allows an operator to identify invalid data while still in the field, thus saving costly revisits to the site.
   
   
3. Load Data in LTPP Database: Some checks are programmed into the LTPP database to identify errors as data are entered. The LTPP database contains mandatory logic, range, data verification, and other miscellaneous checks that are invoked during input.
   
   
4. QC/QA: Once data are input into the LTPP database and reviewed by regional engineers, formal QC/QA software programs are run on the data.
   
   
   • Level A - Starting point. When records are first input into the IMS they are assigned a status of A. Records failing the level B or level C checks will have a status of A. At present, data for SPS supplemental test sections, which by policy are not subjected to QC checks, are left at level A in most tables.
     
     
   • Level B - An old check that is being replaced in some modules. Originally, level B was a dependency check on the availability of certain critical data contained in other tables. In some modules, this check has been phased out and replaced with level E checks and changes to the structure of the EXPERIMENT_SECTION table. There are cases where records with RECORD_STATUS=B exist due to restrictions imposed by the software used to perform manual upgrades.
     
     
   • Level C - Availability of critical data fields in a record. These are checks to see if certain data fields have non-null values. As an example, test section coordinates are required for all entries in INV_ID and SPS_ID. Some of the level C checks are conditional checks on several fields. Another example, in MON_DEFL_DROP_DATA, of the 7 to 9 possible deflection values, at least 5 must be non-null. These checks are not performed on key fields and fields defined as non-null, since these fields must be populated in order to create a record.
     
     
   • Level D - Range checks on the values contained in single fields. While these are called expanded range checks, they are refined range checks on the reasonableness of the magnitude of a number or code value. When data is entered, its range must match the field format logic, for example, a value of 999 can not be entered in a field defined as NUMBER(2,0). These checks are more stringent than logical range values, but in some instances are set to a rather large range of values to encompass typical conditions. For example, the range of air temperature must accommodate conditions spanning from Arizona to Alaska. In other instances, the range limits are based on traditional practice in order to flag outliers and suspect values. For example, the percent longitudinal reinforcement in PCC pavements is limited to 1% since it is very rare that pavements are built with even this very high level of steel reinforcement.
     
     
   • Level E - Relational checks between data elements in the same record and data elements contained in other records. Although previously described as intra-modular checks, these checks have been expanded to include record level inter-field and inter-modular checks. Some of the types of level E checks include:
     
     
     • Logical relationship between related values. For example, a minimum value must be less than or equal to the average, which must be less than or equal to the maximum.
       
       
     • Parent-child integrity checks. For example, every record in MON_DEFL_LOC_INFO must have a matching record in MON_DEFL_MASTER.
       
       
     • Range checks between related values. For example, the difference between the daily maximum and minimum air temperature must be less than 50º C.
       
       
     • Referential cascading parent-child level E relationships. For example, for records in MON_T_PROF_MASTER to reach level E, all matching records in MON_T_PROF_PROFILE must be at level E.
       
       
     • Compliance with LTPP rules and test protocols. Many level E-QC checks are based upon LTPP rules for pavement-structure-material layer types, sequence and LTPP test protocols. For example, the surface layer of a GPS-3 test section should consist of portland cement concrete.
       
       
     • Computed parameter referential level E checks on records in source tables. For example, for records that contain results of FWD backcalculation computations to reach level E, matching data from the FWD deflection tables must also be at E.

Once the QC/QA programs are completed, the regional engineers review the output and resolve any data errors that they can. Often, the data entered are accurate and legitimate but do not pass a QC/QA check. When this occurs, the regional engineer can document that the data have been confirmed using a Comments table in the database and manually upgrade the record to Level E.

There are many reasons that some important data may not be available from the publicly released LTPP database at the time of analysis. The following are some possible examples:

• Data are yet to be collected or the laboratory tests have not been performed on samples that have been taken.
• Data are under regional office review.
• Data have failed one of the quality checks and are being reviewed.
• Data have failed one of the quality checks and were identified as anomalies.
• Data need to be quality checked.
• The development of the SPS-8 requirements took place over time, and some of the earlier projects may have had different requirements.
• The monitoring requirements for some sites may have changed over time.

As such, the unavailable data identified in this section do not necessarily mean the data were not collected or submitted by the States. There are several instances where data may have gotten held up and did not reach Level E. The LTPP program is continuing on a system-wide effort to resolve all unavailable data so they will be available to future researchers.

Assessment of the LTPP Layer Thickness Data Availability and Completeness

An overview of the available LTPP data, both at all QC levels and at Level E for regular LTPP sections (non-supplemental sections), is provided in table 1.

This overview is presented at three levels to provide a complete picture:

• Record level - Number of records in each of the layer material and thickness tables.
• Section level - Number of sections having data in each of these tables.
• Pavement layer structure level - A unique combination of STATE_CODE, SHRP_ID, and CONSTRUCTION_NO comprises a pavement structure.

Generally, the proportion of records at Level E is good, ranging from 92 to 100 percent. The percentage of records at Level E is especially good for the SPS*_LAYER and SPS*_LAYER_THICKNESS tables, ranging from 96 to 100 percent, with many at 100 percent.

A summary of the data availability assessment for LTPP supplemental sections is presented in table 2. It is the policy of the Federal Highway Administration (FHWA) that records for the supplemental sections should not be at Level E. Therefore, no Level E data availability assessment is given in table 2.

Pavement structures that do not have any records in either table TST_L05A or table TST_L05B are listed in table 3. There are 16 regular pavement structures and 1 supplemental pavement structure that currently do not have any data in these tables.

For the Level E data to be used in the subsequent evaluations of the layer thickness data, a more detailed assessment was performed to find out how many pavement layer structures have data in these layer thickness related tables for each LTPP experiment. The results are presented in table 4 for the pavement structure records at Level E in table EXPERIMENT_SECTION. As shown, the experiments contain data in different layer structure related tables, ranging from one table to seven tables, with most experiments having Level E data in four tables.

Summary

The layer thickness data availability is very good in tables TST_L05B and TST_L05A, which contain the representative layer structure and thickness information for section-level analysis. Only 16 pavement structures from LTPP regular sections and 1 pavement structure from a supplemental section do not have any layer structure information in either TST_L05B or TST_L05A.

Out of 3,457 pavement layer structures at QC Level E in table EXPERIMENT_SECTION, 3,240 layers (93.7 percent) have records in table TST_L05B and 3,229 layers (93.4 percent) have records in table TST_L05A. There are a significant number of records in all the layer structure tables.

A total of 217 pavement layer structures do not contain Level E data in table TST_L05B. Other thickness-related tables contain data for selected experiment or layer types. A more detailed summary of the SPS and General Pavement Studies (GPS) pavement structures that do not contain Level E information in the TST_L05B table is provided below:

• GPS-1 – 8 pavement layer structures.
• GPS-4 – 1 pavement layer structures.
• GPS-6B – 3 pavement layer structures.
• GPS-6S – 1 pavement layer structures.
• SPS-1 – 6 pavement layer structures.
• SPS-3 – 4 pavement layer structures.
• SPS-5 – 56 pavement layer structures.
• SPS-6 – 76 pavement layer structures.
• SPS-7 – 7 pavement layer structures.
• SPS-8 – 18 pavement layer structures.
• SPS-9 – 37 pavement layer structures.