|<< Previous||Contents||Next >>|
Use of PMS Data for Performance Monitoring with Superpave as an Example
6. Concepts for Linking PMS and Materials and Construction Databases to Performance Evaluation
6.1 Current Situation
Typically, when engineers and analysts think of performing an evaluation or analysis, they think of the final data set needed to make the analysis, in other words they think of the "combined performance evaluation database". While that is the ultimate goal, it is not a feasible starting point, since such a database must be constructed from other available data sources first.
6.1.1. PMS Database
It's important to remember that the average PMS database is very large in terms of number of sections or subsections entered. In Arizona DOT, for example, there are more than 7,000 sections and in Texas there are more than 70,000 sections. On the other hand, for a particular performance evaluation module, the number of sections to be analyzed may range from a few to several hundreds at the most. Keeping this size differential in perspective should make it clear that it is impossible to put all the detailed materials data into the very large PMS database. Rather, separate databases for each individual purpose can be maintained or assembled and those linked to create the performance evaluation database or module. These facts are born out by the knowledge that early in the development of pavement management, it was thought that all project details could be assembled into a database. In each case the process bogged down. Only when states like Washington and Arizona [Kulkarni 82] started with the idea of keeping simple performance data at the network level, did pavement management move forward rapidly.
For the purposes of material evaluation and this project, reality starts where the states currently exist. Good PMS databases for network level use are simple and contain relatively few data elements because excessive detail can bog down the process. The data elements used in five states that were visited in this project are given in Section 8 and Appendix D.
In most PMS databases, information on cracking, roughness, and other performance measures, are stored in summary form such as, IRI or a cracking index of some sort. Some states use the serviceability index or a related pavement quality rating.
It is essential to define for states the elements of a minimum PMS database. However, that is not the goal of this project. The minimum PMS data expected for use in performance evaluation of Superpave, should be defined including a number of elements appropriately coming from PMS. Where that is not possible, it is suggested that they be collected as part of the auxiliary "materials database and construction database." Subsequently, it would be desirable to move key elements into the broader PMS database.
Most pavement management databases do not include details about actual pavement thickness and material type below the surface layers. This is true in part because construction records were not readily available at the time when the pavement management system was implemented. Time and resources have not permitted coring and searching permanent record files to supply this missing data.
In those states where these type of data are missing from the PMS data file, it should be provided for selected sections in the materials database or data sub-files for performance evaluation.
Similarly, reliable traffic load data are often not available in the PMS database. For the present, it is recommended that the traffic data be obtained and entered into an analysis file. However, in some state DOTs a traffic data file is available and this information can be provided into the PMS database. In most cases, it is desirable that the traffic and load data element be added to the PMS database. Originally this element would contain data only for those sections being considered for evaluation of Superpave, other new design, or materials concept. As these data are collected and processed, it may become clearer to the DOT staff that it is important to add this information to the entire PMS process. If this is not possible, then a supplementary data element called the "Detailed Data" should be developed.
Depending on the materials or conditions begin evaluated, other classes of data not available in the PMS database may be needed. This may involve detailed climatic or history, structural evaluation such as repeated FWD measurements, and in the cases of some materials more detailed distress data related to the purpose of the material. For example, it may be necessary to evaluate rutting more extensively for certain types of mix designs than for others. Therefore, rutting or detailed cracking while combined as an index in the PMS database, should remain separate for those sections being evaluated for Superpave.
It will also be valuable if auxiliary short-term performance information such as, wheel track testing, Stabilometer values or other heavy vehicle simulator or other short-term estimators of performance could be restored in a permanent database. This will depend upon the testing resources available to the individual DOT and the information shown here is by way of example only, assuming a particular class of available information.
6.1.2. Materials and Construction Databases
As discussed in Section 4, much of the required data necessary for the performance monitoring of individual design and construction concepts exist in the materials and construction databases, not in the PMS database. Two major problems exist. The shear magnitude of the design and construction information exceeds anything that can be entered into a routine PMS database. Thus, the data, if stored, must be summarized on a statistical basis keyed to individual locations, including average value, standard deviation and number of samples. Probably more important is the fact that the data exist in a separate database obtained for a different purpose. First of all, the design data are probably not in a database, but on a set of hard copy files or hand written notes that define the design parameters and the mixed design characteristics obtained for the materials and the pavement structure for the individual section. Secondly, as-constructed pavement data exist in a separate construction file. In some cases, and hopefully in the future, this will exist in a file such as Site Manager or in another suitable electronic data subsystem. These will make the process somewhat easier, but it must be remembered that data collected for one purpose, often have limitations for a second purpose. The transposition of these data must be done carefully, effectively, and in a timely manner.
Another major problem with linking data is the source of the data, which may be some other section of the DOT, the files of a contractor, a materials testing company working for the contractor, or in a field materials laboratory. Other complications occur when field changes are initiated due to unexpected conditions such as weather, varied subsoil conditions, or existing condition of old pavements where rehabilitation or reconstruction is involved. This latter condition is further compromised and complicated when the section is to be built under existing traffic. Other complicating factors occur when the section is a widening of an existing pavement where drainage patterns are interrupted.
In these subsidiary data sets, the data may be keyed to local field identification and location, which must be transferred to a permanent mile point GPS location system. This must be done carefully with cross checking if it is to be useful. Data improperly located in an electronic data format are useless and have a negative impact since purportedly it represents conditions incorrectly. It is better to have missing data then bad data.
Only with special effort made at an early date will the linking of the data take place. There are many examples where important past research is not accessible for subsequent analysis and in fact the data have been effectively lost.
6.1.3. Summary of Main Needs
It is important in summary that the data from a materials and/or construction database be appropriately and effectively linked to the PMS database at an early date within the efforts of the history of the data. No major effort should be made to retrieve all possible project level or research level data for a twenty-year period. However, in many state DOTs it would be desirable to initiate a retrospective effort to enter such data for the past year and to setup the database so that the data could be effectively entered in the future. In this way, important data would not be lost. This is especially important for the Superpave study which serves as a primary example in this report. Superpave sections have been constructed for the past several years and because of the critical focus on this element nationwide, such data should still be available in most state DOTs and as soon as possible should be entered into an electronic database keyed to the PMS database.
6.2 Concept for Linking Databases
It is proposed to use the concept of linking two central electronic databases, one for Materials/Construction data and one for PMS data. Relevant data from these two databases should be made available and transferred electronically to a third Performance Analysis database, which should contain all required information. For the performance analysis, use can be made of a commercially available spreadsheet system (e.g. Excel or Lotus) that extracts relevant information and makes overviews, graphs, and reports. A possible structure for this methodology is given in Figure 2 below.
Figure 2. Example of a structure to link materials and construction data to PMS data
The key to linking databases, as shown in Figure 2, is to have precise unambiguous location identification and date/time information. Only in this way can it be assured that the data are comparable. For example, in the case of multi-lane roadways involving bi-directional interstate highway lanes and frontage roads, the new material may be used in only one or two lanes or in one roadway direction. This identification must be unambiguous over time. When the material is first placed, it is clear in everyone's memory, which is which, but four or five years later or after personal changes, when a long-term analysis is needed, this becomes more difficult. Unambiguous locations can be provided by GPS measurements and these are relatively economical at the present time. However, they must be tied to traditional location identification information such as project number, mile point, lane, direction, date, etc.
6.3. Practical Implementation of Data Linking
While the concept of data linking is relatively obvious to all, it became clear in this project, by visiting five DOTs, that additional resources and energy will be required to actually capture the information in usable form. As long as the data elements remain in paper files, field logbooks or other auxiliary locations, there will be a very difficult transition process.
The good news is that several states have already started to put materials and construction data in electronic format. An effort should be made in other states to transfer the data from hard copy files into electronic form quickly, soon after construction. This would introduce an additional check into the system and there is a good chance that the quality of data can be improved. With a properly designed system much of the detailed design, materials and construction data, only have to be entered into the system once. But this initial entry information should be done carefully and uniformly.
The new web-based database with analysis and evaluation capabilities as developed by the University of Washington, shows a lot of promise as a system for practical implementation of linking. Data can be entered with Excel, Access or comparable files. The University of Washington system offers a range of evaluation, analysis and reporting possibilities, but it is also possible to download the raw data and perform these analyses in another way at your own location. Appendix F gives further details.
Administratively it will be necessary to assign someone for data processing and storage. In computer parlance a "Data Administrator " will be needed. If the data in each of the two or three subdata-elements remain independent of each other, there may be a gap in the compatibility of the data for future analysis. Rather than create a separate position or slot for this data administrator, it is likely that someone charged with data responsibility in materials, can start or Pavement Management should be responsible. If the Pavement Management Section is charged with the responsibility of evaluation and analysis, then the best place for data administration is PMS. However, this data administrator must be aware of the needs and workload imposed on the materials section or other sources for the detailed information. As in most other undertakings good relations and coordination between the sub-elements are required.
Ideally, the pavement management or asset management section within a DOT would recognize the multiple goals of a good pavement or asset management system: Network level, Project level and Research and Evaluation. In this case, recognition can be extended beyond Superpave analysis to other new concepts such as an evaluation of the results of using the AASHTO 2002 Guide, new concrete pavement technologies, etc.
In the specific case of Superpave, someone intensely interested in the performance of Superpave can become the data administrator and insure that the materials and auxiliary data are collected and then interfaced with the PMS database to obtain from that source the additional information needed for the ongoing analysis.
Since Superpave performance has a time history dimension to the data evaluation follow-up, the data required for analysis should be collected and evaluated as early as possible and the process should be continued on an annual basis for five to 10 years to better define the performance histories. We must remember from the AASHTO Road Test [Hudson 68b] and other long term studies, that only abnormal failures are observed in the first two or three years. True performance based on long term observations, usually take a minimum of five years and more often 10 to 15 years of observation. For true success, some administration unit should be charged with the data activities.
|<< Previous||Contents||Next >>|