Evaluation and Analysis of LTPP Pavement Layer Thickness Data

Data Evaluation Overview

One of the project objectives was to identify and explain anomalous observations and provide recommendations for layer thickness characterization for each LTPP section. The following potential issues related to layer thickness data were identified during the preliminary data review:

• Unusually high or low thickness values for certain layers.
• Lack of consistency among different data sources.
• Erroneous layer types in materials testing tables.
• Excessive variation in layer thickness or material types among different locations within a layer.

Data Sources

To fulfill this task's objective, the layer thickness data in the following LTPP tables were evaluated for reasonableness and consistency (using cross-table comparison):

• TST_L05B.
• TST_L05A.
• TST_AC01_LAYER.
• TST_ PC06.
• INV_LAYER.
• RHB_LAYER.
• SPS*_LAYER.

Table TST_AC01 and table TST_SAMPLE_LOG in the LTPP database also contain thickness related information. Table TST_AC01 contains AC core thickness measurements from the field. Table TST_SAMPLE_LOG stores information about the samples taken from holes, pits, and probes, and is a good raw data source for unbound layers. However, records in these two tables are not keyed to the layer numbers as stored in TST_L05B and other above listed layer thickness related tables (field LAYER_NO). Therefore, the thickness measurements from these two tables can only be manually matched to the layers established in the TST_L05B table. Furthermore, some measurements span more than one layer, and thus cannot be used for any layer thickness comparison at all. As a result, tables TST_AC01 and TST_SAMPLE_LOG are not included in this evaluation. Nevertheless, these two tables can be used as raw layer thickness related data sources and be consulted for layer thickness measurements on a case-by-case basis.

The main data elements related to pavement layering structure from each of these tables are illustrated in figure 1. Double sided arrows between the table TST_L05B and tables TST_L05A, TST_AC01_LAYER, TST_ PC06, INV_LAYER, RHB_LAYER, and SPS*_LAYER schematically show that the data elements in the later tables were compared against similar data in TST_L05B table.

Essential Fields for Data Analysis

Based on the analysis of the fields in the above tables related to pavement layering structure, the following data elements were selected for detailed pavement layering data examination:

1. Layer functional description (e.g., surface, overlay, base, subgrade).
2. Material type description.
3. Representative layer thickness.
4. Layer thickness variability (discussed in the next chapter).

These four essential pavement layering characteristics (schematically identified in figure 2 as question marks and circled numbers 1 through 4) serve as key inputs for many types of pavement analyses. The selected data elements were examined and compared between different data sources (LTPP tables). The comparisons were done individually for each layer and each LTPP section. Additionally, layer thickness variability indicators were examined, as discussed in chapter 4.

Analysis Steps

The data review activities carried out in this task included the following:

1. Selection of pavement layering data from different LTPP data sources.
2. Development of a master data analysis table with the layering information from different sources included for each pavement layer.
3. Evaluation of consistency in layer functional description.
4. Evaluation of reasonableness and consistency in material type description.
5. Evaluation of reasonableness of layer thickness data and layer thickness consistency between different sources.
6. Evaluation of layer thickness variability indicators from different data sources (chapter 4).
7. Summarize evaluation outcomes and identify reasons for data inconsistencies.
8. Preparation of feedback reports to help ensure the data issues are resolved.

The flowchart identifying different data analysis and data evaluation activities is shown in figure 3.

In steps 1 and 2, all the data elements from different sources were prepared for the layer-by-layer review for each section. Steps 3 through 5 were used to evaluate information for major layer structure data components available in the LTPP database. Results of step 6 are presented separately in chapter 5. Under steps 7 and 8, the anomalies or suspect data in the LTPP layering information were identified, examined, and reported back to the FHWA. These activities are discussed in more detail in the following sections.

Step 2 - Analysis Data Set

Master Table for the Pavement Layering Data Evaluation

To analyze pavement layering information from different sources, a master list of all pavement layers available in the LTPP database was created. The master list contains the maximum number of unique records obtained for each LTPP section, layer number, and construction event. These records were obtained from the INV_LAYER, RHB_LAYER, TST_L05B, TST_L05A, TST_AC01_LAYER, TST_PC06, and SPS*_LAYER tables.

Reference Table Selection

The initial data review indicated that table TST_L05B contains the most recent and most complete LTPP section layering information for each layer. The main attributes of the TST_L05B table are:

• Thickness data based mostly on core measurements
• Representative and most accurate layer thickness
• Most complete layer and material type description
• Highest number of layer records

A total of 96.7 percent of the unique GPS layer records (5,938 records) and 83.8 percent of all SPS layer records (9,360 records) were included in the TST_L05B table at the time of the study. As such, TST_L05B was selected as the target or reference table for the selection of analysis components and cross-table comparison of pavement layering data. Layers not included in the TST_L05B table were not used in the cross-table pavement layering data analysis. These records were examined individually for data reasonableness and identification of anomalous data.

Correspondence in Layer Numbering System between Different Sources

The review of the layer numbering scheme used in different tables indicated that layer numbering is consistent among all the tables except INV_LAYER. Thus, before the layer-related information between different tables could be compared, layers from the INV_LAYER table were aligned with the layers from the other tables.

To align the INV_LAYER records, the TST_L05B table was used as the reference. The TST_L05B table contains two fields (INV_LAYER_NO and INV_LAYER_NO2) that provide information about the corresponding inventory layers. Based on the values in these fields, several different scenarios are possible regarding layer correspondence between the INV_LAYER and TST_L05B tables. The INV_LAYER layer correspondence scenarios and consequent actions are summarized in table 5 below.

Table 5. Evaluation of layer numbering correspondence between the INV_LAYER and TST_L05B tables.

Description	Number of Records (GPS and SPS)	Action
Layer numbers are the same	2803 (72%)	Analyze
Layer numbers are different	488 (12%)	Align and analyze
2 INV_LAYER layers correspond to 1 TST_L05B layer	90 (2%)	Analyze combined thickness
Only part of INV record corresponds to TST layer	69(2%)	Exclude from cross-table analysis
INV_LAYER records exist but not referenced in TST_L05B	468 (12%)	Exclude from cross-table analysis
Total number of records in INV_LAYER	3918 (100%), with 3381 (86%) analyzed

Using the scenarios outlined in table 5, 3,381 records (86 percent) with layer-related information from the INV_LAYER table were aligned with the rest of the data sources.

Data Availability for Consistency Evaluation

Based on the number of data sources available for the analysis of each pavement layer, different data availability codes were assigned to each layer:

• Code 1-layer-related data are available from the TST_LO5B table and one or more other tables.
• Code 2-layer-related data are not available in the TST_LO5B table but are available in one or more of the following tables: TST_L05A, TST_AC01_Layer, TST_ PC06, INV_LAYER, RHB_LAYER, or SPS*_LAYER.

Because the TST_L05B table was selected as a reference table, only records with analysis data availability code 1 were used in the cross-table pavement layering data analysis. Records that did not have a corresponding entry in TST_L05B were reviewed individually for data reasonableness. Table 6 summarizes the number of records used in the analysis for each LTPP experiment.

Table 6. Summary of the number of records used in the cross-table pavement layering analysis.

Experiment		Number of Pavement Layers Analyzed
Type	No.	TST_L05B	TST_L05A	TST_AC01_LAYER	TST_ PC06_LAYER	INV_ LAYER	RHB_ LAYER	SPS*_ LAYER
G	1	1460	1452	526	-	961	-	-
G	2	972	971	366	-	648	29	-
G	3	516	510	13	126	455	-	-
G	4	247	247	1	62	223	-	-
G	5	342	342	22	84	292	4	-
G	6	1763	1725	636	-	327	877	-
G	7	555	553	111	44	171	249	-
G	9	146	145	21	48	115	12	-
S	1	1214	1162	420	-	-	32	1102
S	2	693	656	-	176	-	-	655
S	3	3664	3648	1065	-	-	313	-
S	4	496	496	-	-	-	-	-
S	5	1682	1664	553	-	165	665	1612
S	6	779	746	48	55	24	159	654
S	7	282	282	-	59	-	105	208
S	8	112	104	30	2	-	-	91
S	9	416	401	56	34	-	12	409
Total		15276	15041	3856	690	3381	2391	4731
Notes: G = GPS experiment. S = SPS experiment.

Step 3 - Layer Functional Description Evaluation

The pavement layer functional description provides information about the functionality of a given pavement layer, such as overlay, surface, base, or subgrade. LTPP uses a list of codes to describe layer functional description, as shown in table 7.

Table 7. LTPP layer function description codes.

Code	Description
1	Overlay
2	Seal Coat
3	Original Surface Layer
4	AC Layer Below Surface (Binder Course)
5	Base Layer
6	Subbase Layer
7	Subgrade
8	Interlayer
9	Friction Course
10	Surface Treatment
11	Embankment Layer
12	Recycled Layer

In this study, the values from the layer functional description field were compared among the following tables: TST_LO5B, TST_LO5A, INV_LAYER, RHB_LAYER, TST_AC01_ LAYER, and SPS*_LAYER. The description field in the TST_L05B table served as a reference for the functional layer description information, and the description fields from the other tables were compared against it.

The procedure for layer functional description consistency evaluation is shown schematically in figure 4.

The results of the layer functional description consistency evaluation are summarized in table 8 and are shown in figure 5 separately for the GPS and SPS sections.

Records with a functional layer description field that is inconsistent between different data sources were reported to the LTPP data managers in feedback reports.

Table 8. Summary of the layer functional description consistency evaluation.

Experiment	Percentage of Records with Matching Layer Functional Description
Type	No.	TST_L05A	TST_AC01_ LAYER	INV_ LAYER	RHB_ LAYER	SPS*_LAYER
G	1	100.0%	92.8%	91.9%	-	-
G	2	99.9%	95.1%	92.3%	93.1%	-
G	3	100.0%	100.0%	95.8%	-	-
G	4	100.0%	100.0%	97.3%	-	-
G	5	100.0%	100.0%	97.9%	100.0%	-
G	6	99.8%	93.3%	91.7%	88.7%	-
G	7	100.0%	98.2%	94.7%	83.5%	-
G	9	100.0%	90.5%	90.4%	100.0%	-
S	1	100.0%	86.7%	-	-	68.8%
S	2	100.0%	-	-	-	-
S	3	100.0%	87.0%	-	-	79.2%
S	4	100.0%	-	-	-	-
S	5	100.0%	96.2%	-	90.9%	80.3%
S	6	100.0%	89.6%	-	100.0%	74.8%
S	7	100.0%	-	-	-	68.6%
S	8	100.0%	96.7%	-	-	-
S	9	100.0%	75.0%	-	-	16.7%
Notes: G = GPS experiment. S = SPS experiment.

Note that in figure 5, the chart slice labeled "Inconsistent" represents the layers that had at least one of the evaluated tables with data (functional description) inconsistent with the data in the TST_L05B table. Similar statement applies to all other pie charts presented in Chapter 3.

Step 4 - Material Type Reasonableness and Consistency

The material type description is very important pavement layering information. Material type description data are found in tables TST_LO5B, TST_LO5A, INV_LAYER, RHB_LAYER, and SPS*_LAYER. These data were examined to determine:

• Reasonableness or validity of the material type codes in each table.
• Consistency of the material type description from other tables with that in the TST_L05B table.
• Consistency of the material type description available in the TST_L05A table for different locations along the section.

Material Type Reasonableness

The purpose of the reasonableness check was to evaluate whether the material description code for the layer is consistent with the layer functional description. For example, soil material descriptions are not adequate for the paved surface layers. Table 9, based on the SPS Pavement Layering Methodology, Operational Guide [5], was used as a primary reference for evaluating material type reasonableness.

Table 9. Criteria for evaluation of material code validity.

Layer Description Code	Description	Valid Material Code
1	Overlay	01-08, 13, 16-20, 901
2	Seal Coat	71-73, 74-852
3	Original Surface Layer	01-08, 17-20
4	AC Layer Below Surface (Binder Course)	01, 03, 13, 20
5	Base Layer	302-310, 319-350, 21-492
6	Subbase Layer	302-310, 319-350
7	Subgrade	100-178, 200-294, 51-652
8	Interlayer	71-80, 85, 81-842
9	Friction Course	02, 20
10	Surface Treatment	11, 12, 20, 823
11	Embankment Layer	100-178, 200-294, 51-652
Notes: 1 For SPS-7 only. 2 Based on Appendix A of LTPP Data Collection Guide. [2] 3 Based on reference. [27]

While most of the records had valid material codes, some records in the evaluated tables had material codes different from those specified in table 9. Table 10 provides a summary of the records with identified erroneous material codes. Additionally, some records were missing material codes. The identified records were reported to the FHWA in the data analysis/operations feedback report.

Table 10. Summary of the records with erroneous material codes.

Table Name	Number of Erroneous Records	Total Number of Records	Percentage of Records with Erroneous Codes
TST_L05B	53	15,298	0.35%
TST_L05A	49	15,189	0.32%
RHB_LAYER	99	2,841	3.48%
INV_LAYER	368	3,918	9.39%
SPS1_LAYER	1	1,021	0.10%
SPS2_LAYER	0	621	0.00%
SPS5_LAYER	18	1,056	1.70%
SPS6_LAYER	13	402	3.23%
SPS7_LAYER	8	135	5.93%
SPS8_LAYER	2	155	1.29%
SPS9_LAYER	31	475	6.53%
Total	642	41,111	1.56%

Material Type Consistency among Different Tables

To evaluate consistency between material types reported in different tables, LTPP material code lists were reviewed first. Two sets of material codes are used in the LTPP database to describe material types in the testing tables (TST_L05A and TST_L05B tables) and in inventory-type tables (including INV_LAYER, RHB_LAYER, and SPS*_LAYER tables) in the LTPP database. As a result, for some layers, material type descriptions in tables TST_L05B and TST_L05A do not have exact corresponding material type descriptions in tables INV_LAYER, RHB_LAYER, and SPS*_LAYER. For these layers, manual reviews of individual layer descriptions and engineering judgment are necessary to identify whether the material descriptions from different tables are consistent (or similar enough).

Correlated material codes need to be formulated to evaluate the consistency in material data from all LTPP tables containing material types. For the material type codes that do not have the exact same descriptions, "similar" material groupings were developed to correlate material codes in the inventory tables and material codes in the testing tables. The reasoning for the assignment of different material categories is summarized below for different material types.

Similar Material Type Grouping for Base and Subgrade Materials

The AASHTO classification system [28] was considered the best way to group "similar" soil or granular materials. For example, clayey materials were grouped as "clayey soils," as per the AASHTO group classification A-6 and A-7. The same criteria were applied to other typical soil types, such as gravels (A-1, A-2), silty soils (A-4, A-5), sand (A-1, A-2), clayey sand (A-2), silty gravel (A-1, A-2), and silty sand (A-2). In addition, the following criteria were applied:

• Stone and rock materials were assigned in two different categories to differentiate between rock that is entrapped and stone or cobbles that are loose unbound aggregate particles no longer intact in their original formation. [29]
• Limerock and caliche were grouped into an individual category because of their specific characteristics (i.e., used only in very specific parts of the country, such as Florida) that differentiate them from typical embedded rock.
• Soils that are treated in some manner were grouped as "stabilized subgrade soil," and the same criteria were applied to create a group of "stabilized base materials," which includes soil cement and aggregate mixtures. [30]
• Textiles and geo-grid products cannot be defined as materials in the common sense, but they are part of the pavement system. These materials were grouped as "geomaterials."
• Processed aggregates such as crushed aggregates and stone should not be grouped with natural-occurring gravelly subgrade soils; therefore, a new group called "processed granular base materials" was defined.
• The "fine soil" and "unbound base/subbase" groups were combined in a new similar group denoted "subgrade soils" that includes fine, unbound/untreated soils. Although some fine-grained soils are grouped as "subgrade soils," little information about the material properties can be conveyed by the existing definition.

Similar Material Type Grouping for Asphalt Concrete Materials

The basis for grouping "similar" asphalt concrete materials included a decision-tree process. The materials were first aligned by mixture gradation (sand, open- or dense-graded) as a first filtering step. The method of production (hot- or cold-laid) was the second criterion used to distinguish asphalt groupings. Recycled asphalt concrete, maintenance seal coats, and special plant mixes (emulsions, cutbacks) were retained in individual groupings. [27]

The table of new correlated groupings of "similar" materials and corresponding material codes from inventory and testing tables is presented in appendix A.

Material Type Consistency Criteria

To test the consistency of material type data between different tables, the TST_L05B table was used as the reference for material type description information. The material type description data from other tables were compared against it using the criteria outlined in table 11 below.

Table 11. Material type consistency criteria.

Criteria Name	Description	Evaluation Code
Consistent	Material type descriptions are the same.	0
Similar	Material types are similar based on a broad material categories developed for geological materials using the dominant material component(s).	1
Inconsistent	Material type descriptions are different.	2
Not evaluated	Material types cannot be evaluated because no material codes are available in one of the tables that make comparison pair (or if material type is available only at one location for "along the section" consistency test).	3

Figure 6 shows schematically the testing procedure used for evaluation of consistency in the material type description between different tables.

The results of layer material type consistency evaluation between different data sources are summarized in table 12 and figure 7, separately for GPS and SPS sections.

Table 12. Summary of the layer material type consistency evaluation.

Experiment		Percentage of Layers with Layer Material Type Records Matching with Records in TST_L05B
Type	No.	TST_L05A		INV_ LAYER		RHB_LAYER		SPS*_ LAYER
		Exact	Similar	Exact	Similar	Exact	Similar	Exact	Similar
G	1	98.9	0.6	32.6	39.4	-	-	-	-
G	2	99.0	0.7	41.0	25.8	40.0	20.0	-	-
G	3	98.2	1.0	41.1	24.5	-	-	-	-
G	4	97.6	1.6	37.6	25.3	-	-	-	-
G	5	98.8	0.6	36.6	25.3	-	-	-	-
G	6	99.6	0.2	46.4	26.9	70.3	10.3	-	-
G	7	98.1	1.1	43.8	31.4	61.4	11.4	-	-
G	9	98.6	0.7	45.2	20.9	-	-	-	-
S	1	99.9	0.0	-	-	-	-	35.9	37.4
S	2	99.6	0.0	-	-	-	-	24.0	56.5
S	3	95.6	1.5	-	-	53.0	23.2	-	-
S	4	100.0	0.0	-	-	-	-	-	-
S	5	100.0	0.0	55.2	33.3	73.1	3.4	42.4	29.8
S	6	98.9	0.5	66.7	33.3	64.9	12.2	30.7	31.3
S	7	98.6	1.4	-	-	63.1	33.8	24.0	47.1
S	8	100.0	0.0	-	-	-	-	34.1	38.5
S	9	99.5	0.0	-	-	25.0	0.0	33.0	31.1
Notes: G = GPS experiment. S = SPS experiment.

Records with inconsistent material codes were identified and reported to the FHWA in the form of feedback reports.

Material Type Consistency along the Section

Table TST_LO5A contains information about layer material types evaluated at up to three locations (the beginning, the middle, and the end) along the LTPP section. In this task, the consistency of the material type along the LTPP section was evaluated using the process shown schematically in figure 8.

In the TST_L05A table, 5,795 GPS records (97 percent of all GPS records) and 2,581 SPS records (28 percent of all SPS records) had layer material type information for more than one location along the section. The evaluation results of layer material type consistency along the section are summarized for GPS and SPS sections in table 13 and figure 9.

Table 13. Summary of the layer material type consistency evaluation along the LTPP section length (TST_L05A table).

Experiment		Percentage of TST_L05A Layers with Material Types along the Section
Type	No.	Consistent	Similar	Inconsistent
G	1	87.0	4.8	8.2
G	2	89.4	3.0	7.6
G	3	88.2	5.1	6.7
G	4	84.7	5.6	9.6
G	5	87.5	5.2	7.3
G	6	89.9	3.7	6.5
G	7	91.4	4.0	4.6
G	9	88.2	3.5	8.3
S	1	99.3	0.0	0.7
S	2	96.3	0.5	3.2
S	3	99.9	0.0	0.1
S	4	-	-	-
S	5	99.9	0.1	0.0
S	6	98.1	0.9	0.9
S	7	95.6	4.4	0.0
S	8	96.2	3.8	0.0
S	9	85.5	3.4	11.2
Notes: G = GPS experiment. S = SPS experiment.

Step 5 - Reasonableness and Consistency of Layer Thickness Data

Evaluation of the layer thickness data was one of the most important activities under this project. Layer-specific thickness data are found in the following tables: TST_LO5B, TST_LO5A, TST_AC01_LAYER, TST_PC06, INV_LAYER, and RHB_LAYER, SPS*_LAYER, and SPS*_LAYER_THICKNESS.

The layer thicknesses in the SPS*_LAYER_THICKNESS tables are reported for different locations along the section; these data are grouped by layer type (surface, base, etc.) and material type (AC, PCC, aggregate) categories, rather than using the LTPP consecutive layer numbering scheme. The SPS*_LAYER tables contain the summary information from the SPS*_LAYER_THICKNESS tables.

The TST_LO5A table contains layer thickness measurements obtained at up to three locations (the beginning, the middle, and the end) along the section. These data serve as a source for representative layer thickness values reported in the TST_LO5B table.

The TST_PC06 table contains layer thickness measurements for PCC layers obtained using individual pavement core samples. The TST_AC01_LAYER table contains layer thickness measurements for AC layers obtained using individual pavement core samples.

The layer thickness data from the above tables were analyzed to determine:

• Reasonableness of the thickness data.
• Consistency of the thickness data with the representative thickness data in table TST_L05B.

Reasonableness of the Layer Thickness Data

To evaluate reasonableness of layer thickness data, representative layer thickness ranges were determined for different layer types. The criteria specified in SHRP-LTPP Interim Guide for Laboratory Materials Handling and Testing (PCC, Bituminous Materials, Aggregates and Soil), Operational Guide No. SHRP-LTPP-OG 004 [3] (SHRP-LTPP Lab Guide), were used to set reasonable layer thickness ranges based on the layer description codes, as shown in table 14.

Table 14. Thickness ranges used for reasonableness checks.

Layer Description Code	Description	Range (mm)	Range (inches)
1	Overlay	13 - 229	0.5 - 9
2	Seal Coat	3 - 38	0.1 - 1.5
3	Original Surface Layer	13 - 330	0.5 - 13
4	AC Layer Below Surface (Binder Course)	13 - 254	0.5 - 10
5	Base Layer	25 - 610	1 - 24
6	Subbase Layer	76 - 1217	3 - 47.9
7	Subgrade	N/A	N/A
8	Interlayer	3 - 152	0.1 - 6
9	Friction Course	3 - 64	0.1 - 2.5
10	Surface Treatment	3 - 38	0.1 - 1.5
11	Embankment Layer	76 - 1217	3 - 47.9
12	Recycled Layer	N/A	N/A

The SHRP-LTPP Lab Guide [3] does not provide guidance for the representative thicknesses of the prepared subgrade and recycled layers. Also, only a few records had subgrade thickness data in the LTPP database. Thus, thickness reasonableness was not evaluated for the subgrade and recycled layers. Layer description codes from each table were used as a reference to obtain reasonable thickness ranges for different layers listed in table 14. Based on the representative layer thickness ranges, minimum and maximum thickness values were determined for each layer type.

The TST_PC06 table does not contain a field with layer functional description. To evaluate reasonableness of representative layer thicknesses reported in this table, the layer functional description from the TST_L05B table was used for the corresponding records. Thicknesses for the layers from the TST_PC06 table that did not have matching layer numbers in the TST_L05B table were not evaluated for reasonableness.

The TST_L05A table could contain thickness measurements at different locations. Reasonableness of layer thicknesses at all locations was evaluated in the study. If at least one out of the possible three layer thickness measurement values was outside of the reasonable thickness range for a given layer type, the layer was flagged as one with unreasonable layer thickness.

Table 15 provides the layer thickness reasonableness evaluation results grouped by LTPP table name and experiment type.

Table 15. Summary of the layer thickness reasonableness evaluation results1.

Experiment		Percentage of Layers with Reasonable1 Layer Thickness
Type	No.	TST_L05B	TST_L05A	TST_AC01_LAYER	TST_ PC06	INV_ LAYER	RHB_ LAYER	SPS*_LAYER
G	1	98.3	97.3	98.8	-	98.9	-	-
G	2	98.2	96.2	99.7	-	99.5	100.0	-
G	3	98.9	96.8	100.0	98.4	98.6	-	-
G	4	100.0	97.6	100.0	100.0	99.4	-	-
G	5	99.6	98.8	100.0	100.0	99.1	-	-
G	6	95.3	93.9	99.1	-	99.1	98.4	100.0
G	7	98.1	97.5	100.0	100.0	100.0	99.3	100.0
G	9	89.1	85.3	100.0	77.1	91.6	-	-
S	1	99.8	99.8	100.0	-	-	100.0	99.7
S	2	99.4	99.4	-	100.0	-	-	99.2
S	3	98.4	98.4	99.3	-	-	97.8	-
S	4	100.0	100.0	-	-	100.0	-	-
S	5	93.2	93.0	98.7	-	98.2	99.0	92.0
S	6	99.1	100.0	100.0	100.0	100.0	97.1	100.0
S	7	100.0	100.0	-	88.1	92.3	100.0	85.2
S	8	97.5	97.5	100.0	100.0	-	-	97.4
S	9	96.7	96.3	-	100.0	100.0	55.6	96.2
Note: 1 Based on the criteria from the SHRP-LTPP Lab Guide. [3] G = GPS experiment. S = SPS experiment.

As a result of the layer thickness reasonableness evaluation, all thickness values outside the acceptable thickness ranges were identified and reported to the FHWA for review.

Layer Thickness Data Consistency

One of the objectives of the study was to evaluate the consistency between section-level layer thickness values available from different data sources (tables). Section-level layer thickness values could be found in the following LTPP tables: TST_LO5B, INV_LAYER, RHB_LAYER, and SPS*_LAYER.

In addition, table TST_L05A contains layer thickness values at up to three different locations along the section (beginning, middle, and end) and serves as a source of the representative layer thickness values included in the TST_L05B table. Layer thickness data from the TST_L05A table was considered consistent with the data from the TST_L05B table if at least one of the possible three thickness values in the TST_L05A table passed the consistency test. This criterion is based on the procedure for determination of the representative layer thickness, as explained in the SHRP-LTPP Lab Guide. [3]

Tables TST_AC01_LAYER and TST_PC06 contain layer thickness measurements obtained from the pavement cores taken at different locations along the section. These measurements were used to compute representative layer thicknesses for the records included in the TST_AC01_LAYER and TST_PC06 tables.

To evaluate the consistency of the layer thickness data from different sources, the criteria for allowable differences in layer thickness were developed first. The criteria were based on the layer thickness consistency values utilized in the SHRP-LTPP Lab Guide [3]. The values reported in the guide were developed for evaluating layer thickness consistency between the ends of the LTPP section (i.e., between minimum and maximum values). The comparison carried out in this study is between the representative or "average" thickness values obtained from different data tables. Based on the difference in the data statistics used in the current study compared to the analysis outlined in the operational guide ("range" versus "average" value comparison), the allowable differences used in the current study were reduced by half for the comparison of the average thickness values. The representative thickness data in table TST_L05B were used as a reference for the comparison with the representative thicknesses in the other tables.

Table 16 provides a summary of the allowable differences between representative layer thicknesses that were used in this study to evaluate layer thickness data consistency between different tables. Figure 10 schematically shows the procedure used for evaluation of consistency in layer thickness data between different tables.

Table 16. Criteria used for evaluation of layer thickness consistency between different tables.

Type of Layer Materials	Layer Type Code from TST_L05B	Layer Thickness from TST_L05B (h), mm	Allowable Difference in Layer Thickness, mm
PCC	PC	< 203 > 203	38*½ = 19 50.8*½ = 25.4
Bituminous	AC	< 51 > 51	0.5*h*½ = 0.25*h 0.3*h*½ = 0.15*h
Bound Base or Subbase	TB, TS	Any	0.3*h*½ = 0.15*h
Unbound Base or Subbase	GB, GS	Any	0.5*h*½ = 0.25*h

For thin AC layers (less than 51 mm), if the allowable difference computed using formula provided in table 16 was less than 2.5 mm (0.1 inch), the value of 2.5 mm was used as a criterion for evaluation. This decision is based on the fact that layer thickness values are recorded in the IMS database to the nearest one-tenth of an inch.

Layer thickness consistency for the subgrade or engineering fabric layers were not evaluated because no comparison criteria for these layers were established. Additionally, if layer thickness in the TST_L05B table was marked as 999.9, no comparison with the corresponding layer thicknesses from the other tables was carried out. A thickness value of "999.9" indicates that there is a considerable difference in pavement thickness values between section ends, so that no representative thickness value could be established.

Representative layer thickness values were obtained from different data tables and compared with the representative thickness data in table TST_L05B. The outcome of the thickness data consistency evaluation is summarized in table 17 and figure 11 separately for GPS and SPS sections.

Table 17. Summary of the layer thickness consistency evaluation results1.

Experiment		Percentage of Layers with Consistent Layer Thickness
Type	No.	TST_L05A	TST_AC01_LAYER	TST_ PC06_LAYER	INV_ LAYER	RHB_ LAYER	SPS*_LAYER
G	1	99.8	97.7	-	73.2	-	-
G	2	100.0	99.0	-	72.7	87.5	-
G	3	100.0	100.0	100.0	79.5	-	-
G	4	100.0	100.0	100.0	83.3	-	-
G	5	100.0	100.0	100.0	81.0	-	-
G	6	99.9	90.6	-	63.2	60.6	-
G	7	99.8	96.4	97.7	73.1	68.9	-
G	9	100.0	94.7	95.8	69.7	-	-
S	1	99.9	80.1	-	-	-	90.8
S	2	99.6	-	90.3	-	-	87.7
S	3	99.5	88.0	-	-	48.8	-
S	4	100.0	-	-	-	-	-
S	5	98.7	91.7	-	74.3	61.8	62.8
S	6	98.7	93.8	100.0	87.5	69.8	82.0
S	7	97.6	-	84.7	-	93.9	63.1
S	8	100.0	93.3	100.0	-	-	93.8
S	9	100.0	76.8	94.7	-	0.0	72.9
Note: 1 Based on the criteria from the table 16. G = GPS experiment. S = SPS experiment.

Records with layer thickness differences between the tables exceeding the values shown in table 16 were reported to FHWA.

Step 7 - Evaluation Outcome Summary and Resolution

The anomalies, suspect data, and inconsistent information found during the pavement layering data evaluation are described below, along with a discussion of possible causes of their occurrence. Corrective or remedial measures taken to address these data issues are also discussed. Identified layer thickness data issues were reported to the FHWA for data resolution in numerous LTPP Data Analysis and Operations Feedback Reports (feedback reports).

1: Inconsistent Layer Descriptions

A total of 1,067 records had layer functional descriptions different from the description provided in the TST_L05B table-304 records from GPS experiments and 763 from SPS experiments. A feedback report was generated and sent to the FHWA for the data in these records.

2: Erroneous Material Type

Data evaluation of material and layer functional description codes indicated that, in some instances, the material description codes for the layer were inconsistent with the layer functional descriptions. For example, soil material descriptions were used for the base layers. This means that either the material code or the layer functional description code is incorrect. The summary of records with invalid material codes for specified functional layer type is provided below:

• 53 layers out of 15,298 layers in the TST_L05B table.
• 49 layers out of 15,189 layers in the TST_L05A table.
• 99 layers out of 2,841 layers in the RHB_LAYER table.
• 368 out of 3,918 layers in the INV_LAYER table.
• 1 layer out of 1,021 layers in the SPS1_LAYER table.
• 0 layers out of 621 layers in the SPS2_LAYER table.
• 18 layers out of 1,056 layers in the SPS5_LAYER table.
• 13 layers out of 402 layers in the SPS6_LAYER table.
• 8 layers out of 135 layers in the SPS7_LAYER table.
• 2 layers out of 155 layers in the SPS8_LAYER table.
• 31 layers out of 475 layers in the SPS9_LAYER table.

In addition, material or functional layer description codes were missing for some records. A feedback report was generated and sent to the FHWA for the data in these records.

3: Different Material Type Coding Schemes

The review of material type data used to describe different pavement layers showed inconsistencies in the material naming conventions and material codes used in the testing tables and in inventory-type tables (including INV_LAYER, RHB_LAYER, and SPS*_LAYER). As a result, for some layers, material type descriptions in tables TST_L05B and TST_L05A do not have exact corresponding material type descriptions in tables INV_LAYER, RHB_LAYER, and SPS*_LAYER. There are no established reference criteria that could be used to determine whether material types in the above tables are similar or significantly different.

As a remedial action, a materials expert was contacted to develop a methodology for evaluation of material code compatibility. As a result, a table of correlated material codes was created to enable cross-table comparison of the material codes between inventory- and testing-type tables. The results are presented in appendix A.

4: Inconsistent Material Types

A substantial number of records from the SPS*_LAYER, INV_LAYER, and RHB_LAYER tables had material types significantly different from those specified in the TST_L05B and TST_L05A tables, as summarized below.

INV_LAYER Table:

• GPS experiments-31.5 percent (990 of the 3,147 layers with material codes) had inconsistent material types.
• SPS experiments-10 percent (19 of the 189 layers with material codes) had inconsistent material types.

RHB_LAYER Table:

• GPS experiments-22 percent (100 of the 455 layers with material codes) had inconsistent material types.
• SPS experiments-22 percent (147 of the 655 layers with material codes) had inconsistent material types.

SPS*_LAYER Tables:

• SPS-1 experiment-27 percent (294 of the 1,102 layers with material codes) had inconsistent material types.
• SPS-2 experiment-19.5 percent (128 of the 655 layers with material codes) had inconsistent material types.
• SPS-5 experiment-28 percent (449 of the 1,612 layers with material codes) had inconsistent material types.
• SPS-6 experiment-38 percent (248 of the 654 layers with material codes) had inconsistent material types.
• SPS-7 experiment-29 percent (60 of the 208 layers with material codes) had inconsistent material types.
• SPS-8 experiment-27.5 percent (25 of the 91 layers with material codes) had inconsistent material types.
• SPS-9 experiment-36 percent (147 of the 409 layers with material codes) had inconsistent material types.

Some of these inconsistencies could be explained by different material coding lists used in these tables. In some instances, it was difficult to establish material "similarity." In other cases, more than one layer with different material codes in the INV_LAYER table corresponded to a single layer in the TST_L05B table. Identified problems were reported to the FHWA in the form of feedback reports

5: Unreasonable Thickness Values (Outside the Recommended Range)

The LTTP material testing guide provides typical thickness ranges for most layer types. [3] These values were compared with entries in the TST_L05B, TST_L05A, TST_AC01_ LAYER, TST_ PC06, INV_ LAYER, RHB_ LAYER, and SPS*_LAYER tables. Records that fall outside the recommended range are summarized below for each table.

TST_L05B Table:

• GPS experiments-2.7 percent (125 of the 4,639 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS experiments-2.2 percent (164 of the 7,399 layers with thickness data) had thickness values outside the recommended thickness range.

TST_L05A Table:

• GPS experiments-4.1 percent (192 of the 4,638 layers with thickness data) had thickness values outside the recommended thickness range (least at one location along the section.)
• SPS experiments-2.5 percent (118 of the 4,777 layers with thickness data) had thickness values outside the recommended thickness range (at least one location along the section.)

Computed Representative Values based on the TST_AC01_ LAYER Table:

• GPS experiments-0.7 percent (10 of the 1,364 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS experiments-0.8 percent (12 of the 2,903 layers with thickness data) had thickness values outside the recommended thickness range.

Computed Representative Values based on the TST_ PC06 Table:

• GPS experiments-3.6 percent (13 of the 364 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS experiments-2.3 percent (7 of the 311 layers with thickness data) had thickness values outside the recommended thickness range.

INV_LAYER Table:

• GPS experiments-1.2 percent (32 of the 2,694 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS experiments-1.5 percent (5 of the 344 layers with thickness data) had thickness values outside the recommended thickness range.

RHB_LAYER Table:

• GPS experiments-1.5 percent (7 of the 470 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS experiments-2.0 percent (15 of the 732 layers with thickness data) had thickness values outside the recommended thickness range.

SPS*_LAYER Tables:

• SPS-1 experiment-0.3 percent (3 of the 928 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS-2 experiment-0.8 percent (4 of the 532 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS-5 experiment-8.0 percent (156 of the 1,953 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS-6 experiment-0 percent (0 of the 811 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS-7 experiment-14.8 percent (32 of the 216 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS-8 experiment-2.6 percent (3 of the 114 layers with thickness data) had thickness values outside the recommended thickness range.
• SPS-9 experiment-3.8 percent (24 of the 630 layers with thickness data) had thickness values outside the recommended thickness range.

No remedial action was taken for the identified records. However, comment codes were assigned in the analysis summary table to the records containing such data. A feedback report was submitted to the FHWA for further data review. If the review of data sources would indicate that the reported thickness values are "true" data, we recommend adding a comment field to the relevant layer thickness tables explaining the reason for the unusual layer thickness.

In addition, in the RHB_LAYER table, thickness values of 0.0 are used to identify:

• Thin layers (friction course, surface treatment, seal coat) with a thickness that cannot be established.
• Removed layers.

This creates some confusion because it is unclear whether the layer is removed or whether it is too thin to establish representative thickness. In the future, it is recommended to use a minimum thickness of 3 mm (0.1 in) for thin layers instead of 0.0 to differentiate between "removed" layer and existing thin layers (with thicknesses too small to determine).

6: Inconsistent Thickness Values

Based on the criteria established in table 11 in this report, layer thickness values were compared with the values in the TST_L05B table. Records that had layer thickness values significantly different from those reported in TST_L05B are summarized below.

TST_L05A Table

:

• GPS experiments-0.09 percent (4 of the 4,612 layers with thickness data) had thickness values significantly different from those in the TST_L05B table at all locations along the section.
• SPS experiments-0.7 percent (33 of the 4,721 layers with thickness data) had thickness values significantly different from those in the TST_L05B table at all locations along the section.

Computed Representative Values based on the TST_AC01_ LAYER Table:

• GPS experiments-5.2 percent (86 of the 1,670 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS experiments-12.7 percent (272 of the 2,144 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.

Computed Representative Values based on the TST_ PC06 Table:

• GPS experiments-0.8 percent (3 of the 364 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS experiments-8.7 percent (27 of the 311 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.

INV_LAYER Table:

• GPS experiments-26.0 percent (612 of the 2,355 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS experiments-24.4 percent (38 of the 156 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.

RHB_LAYER Table:

• GPS experiments-36.4 percent (147 of the 404 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS experiments-38.5 percent (196 of the 509 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.

SPS*_LAYER Tables:

• SPS-1 experiment-9.2 percent (79 of the 859 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS-2 experiment-12.3 percent (61 of the 497 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS-5 experiment-37.2 percent (493 of the 1,325 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS-6 experiment-18.0 percent (88 of the 488 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS-7 experiment-36.9 percent (58 of the 157 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS-8 experiment-6.2 percent (4 of the 65 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.
• SPS-9 experiment-27.1 percent (88 of the 325 layers with thickness data) had thickness values significantly different from those in the TST_L05B table.

No remedial action was taken for the identified records. However, comment codes were assigned in the analysis summary table to the records containing such data. A feedback report was submitted to the FHWA for further data review.

7: Multiple Records in the RHB_LAYER Table

A number of layers in the RHB_LAYER table had multiple records for the same layer and construction number. Only records with the most recent "date complete" were used in the analysis. A feedback report identifying multiple records in the RHB_LAYER table was submitted to the FHWA.

8: Missing Records in the TST_L05B Table

Analysis of the data indicated that the TST_L05B table is the most complete source of layer thickness information. However, there are still 203 (3.3 percent) GPS layers and 1,813 (16.2 percent) SPS layers available in the other tables that are not included in the TST_L05B table. Layers that are available in at least one of the following tables but not available in TST_L05B Level E release 11.5 version NT3.0 were reported to the FHWA: TST_LO5A, TST_AC01_LAYER, TST_PC06, RHB_LAYER, and SPS*_LAYER.

There are 468 (12 percent) records in the INV_LAYER table that are not referenced in the TST_L05B table. These records were reported to the FHWA for data review.

Summary of Pavement Layering Data Evaluation

The results of the pavement layering data evaluation were assessed to determine the consistency of pavement layering information between different sources. In addition, within-section layer material type consistency and material type reasonableness were evaluated using selected tables where these parameters were available.

The consistency of pavement layering data between different sources was evaluated for three data categories:

• Layer functional description
• Material type description
• Representative layer thickness

In this evaluation, data pertinent to the layer functional description, layer thickness, and layer material type were obtained from multiple LTPP data tables for each pavement layer and each LTPP section. The data were reviewed to determine consistency between multiple data sources. A layer was considered to have consistent information between different data sources if all the tables containing pertinent information had the same data for this layer. The only exception to this rule was allowed for evaluation of the layer material types. If material type records from multiple data sources had a "similar" material type, as identified in table 66 of appendix A, these records were considered "consistent." This exception was used to accommodate the comparison between the values from the tables utilizing different material classification codes (i.e., material codes for testing versus material codes for inventory tables.)

If there was inconsistency in data from one or more data sources, a layer was flagged for further review. Inconsistencies in pavement layering data were reviewed and reported to the LTPP data managers in the form of data analysis/operations feedback reports, along with recommendations for data anomaly resolution.

Table 18 contains summary results for the pavement layering data consistency evaluation for each LTPP experiment.

Additionally, reasonableness (or validity) of material type description was evaluated. The purpose of the reasonableness check was to evaluate whether the material description code for the layer is consistent with the layer functional description. While most of the records had valid material codes, 642 records out of 41,111 (1.56 percent) had erroneous material codes, and some records were missing material codes. The identified records were reported to the FHWA in the data analysis/operations feedback report.

Reasonableness of layer thickness data was evaluated using representative layer thickness ranges specified in SHRP-LTPP Lab Guide [3]. As a result of the layer thickness reasonableness evaluation, thickness values outside the representative thickness ranges were identified and reported to the FHWA for the data review.

Table 18. Summary of layering data consistency evaluation for each LTPP experiment.

Experiment		Number (percentage) of Pavement Layers Analyzed
Type	No.	Layer Functional Description		Material Type Description		Representative Layer Thickness
		Consistent	Inconsistent	Consistent	Inconsistent	Consistent	Inconsistent
G	1	1410 (96.4%)	53 (3.6%)	1180 (81.6%)	266 (18.4%)	933 (82.1%)	203 (17.9%)
G	2	927 (95.4%)	45 (4.6%)	748 (77.8%)	214 (22.2%)	622 (81.1%)	145 (18.9%)
G	3	496 (96.7%)	17 (3.3%)	354 (69%)	159 (31%)	306 (82.5%)	65 (17.5%)
G	4	243 (98.4%)	4 (1.6%)	165 (66.8%)	82 (33.2%)	143 (85.1%)	25 (14.9%)
G	5	336 (98.2%)	6 (1.8%)	231 (67.5%)	111 (32.5%)	209 (84.3%)	39 (15.7%)
G	6	1583 (92.8%)	122 (7.2%)	1539 (91.2%)	148 (8.8%)	1160 (82.1%)	253 (17.9%)
G	7	490 (91.4%)	46 (8.6%)	452 (84.5%)	83 (15.5%)	352 (82.1%)	77 (17.9%)
G	9	129 (92.1%)	11 (7.9%)	101 (72.1%)	39 (27.9%)	84 (75%)	28 (25%)
S	1	1138 (93.7%)	76 (6.3%)	872 (74.8%)	294 (25.2%)	794 (84.3%)	148 (15.7%)
S	2	633 (91.3%)	60 (8.7%)	559 (81.1%)	130 (18.9%)	457 (85.4%)	78 (14.6%)
S	3	3549 (96.9%)	115 (3.1%)	1353 (94.9%)	73 (5.1%)	1335 (87.3%)	194 (12.7%)
S	4	496 (100%)	0 (0%)	21 (100%)	0 (0%)	14 (100%)	0 (0%)
S	5	1393 (82.8%)	289 (17.2%)	1191 (71.8%)	467 (28.2%)	819 (59.8%)	550 (40.2%)
S	6	698 (89.6%)	81 (10.4%)	488 (66%)	251 (34%)	446 (80.9%)	105 (19.1%)
S	7	233 (82.6%)	49 (17.4%)	219 (78.5%)	60 (21.5%)	144 (67.9%)	68 (32.1%)
S	8	112 (100%)	0 (0%)	87 (77.7%)	25 (22.3%)	75 (92.6%)	6 (7.4%)
S	9	323 (77.6%)	93 (22.4%)	268 (64.4%)	148 (35.6%)	232 (69.9%)	100 (30.1%)
Total		14189 (93%)	1067 (7%)	9828 (79.4%)	2550 (15.6%)	6570 (79.1%)	1736 (20.9%)
Note: G = GPS experiment. S = SPS experiment.

Layer Material Type and Thickness Data Status Summary Table

Using the outcome of the data evaluation for the four major parameters related to layer structure and layer thickness (layer functional description, material type, representative thickness, and variation in thickness measurements), the quality assurance codes indicating consistency and reasonableness of pavement layering data from different data sources were assigned to each layer. A data analysis summary table containing QA codes for major layer-related parameters evaluated for each layer was submitted to the FHWA on a CD with the final report. This table includes the following information for each LTPP section on a layer-by-layer basis:

• Layer functional type and material type codes, thickness, and thickness summary statistics indicators extracted from multiple data sources.
• Indicators of functional layer data consistency between sources.
• Indicators of layer material type reasonableness from each source data table.
• Indicators of material type data consistency between sources.
• Indicators of layer thickness data reasonableness from each source data table.
• Indicators of layer thickness data consistency between different sources.
• Within-section layer variability indicators, including excessive variability flags (where available).
• Recommended representative layer thickness for each pavement layer (for layers that satisfied data reasonableness and consistency evaluation criteria).
• List of tables where layer thickness data are available for each pavement layer.