U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
This magazine is an archived publication and may contain dated technical, contact, and link information.
|Publication Number: Date: September/October 2004|
Issue No: Vol. 68 No. 2
Date: September/October 2004
The new guide and software may become the national approach for creating and rehabilitating roadway surfaces.
In the early 1960s, the American Association of State Highway Officials (AASHO)—the precursor to the American Association of State Highway and Transportation Officials (AASHTO)—conducted the road tests that would become the basis for most pavement designs. More than four decades have passed since the AASHO study, and many methods and design procedures have evolved, stemming from new technologies and the increasing traffic demands on pavements.
Researchers are now incorporating the latest advances in pavement design into a new set of design procedures. Over the past 7 years, these researchers have been developing, testing, and refining the data and procedures that will become the new Mechanistic-Empirical Pavement Design Guide.
|Analysis of typical pavement rehabilitation projects is not possible with the current pavement design guide. The new guide will enable evaluation of both simple and complicated rehabilitation projects, such as this flexible pavement rehabilitation project in eastern Oregon (above) and this rigid pavement rehabilitation on I-40 in Raleigh, NC (below).|
The new guide was developed through the National Cooperative Highway Research Program (NCHRP Project 1-37A). Recently, NCHRP submitted the guide to AASHTO for adoption within the next 2 to 3 years. Currently, NCHRP is distributing the design guide and associated software as a research product at www.trb.org/mepdg for agency representatives and other users to acquaint themselves with the material.
The new guide offers several dramatic improvements over the current one, AASHTO's 1993 Guide for Design of Pavement Structures. "The most significant change is that it offers a much more sophisticated design procedure, using mechanistic-empirical analysis," says Dr. Leslie Myers, asphalt pavement engineer with the Federal Highway Administration's (FHWA) Office of Pavement Technology and a member of the FHWA Design Guide Implementation Team (DGIT). A mechanistic pavement design approach uses a model to calculate the response of the pavement to traffic loads. A mechanistic-empirical approach uses both experimental data and mathematical models to predict pavement performance. Myers continues, "These procedures should support a higher degree of predictability in pavement performance, since they are based on fundamental material properties."
Another major improvement is that the new guide will enable designers to evaluate pavement rehabilitation projects, whereas the current guide only handles new pavements. NCHRP indicates that approximately 73 percent of the Nation's pavement design dollars are spent on rehabilitation, so the ability to evaluate rehabilitated pavements is a critical feature.
The guide will include a user-friendly software package designed for flexibility, offering engineers three levels of input data from which to choose depending on the amount of available data. Below is an overview of the limitations of the current design procedures and highlights of the new guide's features.
The new design guide eventually will replace the existing one, which is based purely on empirical principles. Using an empirical approach, the pavement designer determines the relationships between design inputs (such as loads, materials, layer configurations, and the environment) and pavement failure through field observations and experiments. An empirical analysis, however, does not establish the fundamental basis for the relationships between variables and outcomes.
The empirical equations used in the 1993 AASHTO guide are largely a result of the original road tests performed by AASHO from 1958 to 1961, in what was the largest experiment on highway pavements of its time. The study evaluated the performance of both portland cement concrete (PCC) and asphalt concrete pavements using known thicknesses under traffic loads of known magnitude and frequency.
Although the empirical methods for pavement design that ultimately resulted from the study represented a giant leap forward, the road test had a number of limitations that constrain its continued usefulness as a basis for modern designs. Because the test was conducted at one specific geographic location, in Ottawa, IL, it does not capture the effects of differences in climatic conditions on pavement performance. For example, pavements in the road test experienced a significant amount of distress (such as thermal cracking) due to cold weather and the spring thaw. However, since a significant portion of the country, including most of the south and west, does not generally experience these extremes, this type of distress is not an issue.
Another limitation is that all of the variables in the test (such as vehicle characteristics and assumptions about traffic loads and pavement materials) were representative of the conditions of the late 1950s. Many of these assumptions are outdated today.
A primary difference between the new guide and previous versions is that it is based on a mechanistic-empirical approach. "Using real traffic data, local climate conditions, material behaviors, and known field performance, engineers can predict the future performance of new pavement sections more accurately," says Myers.
"One of the premises of the current empirical design procedure is that if you have an increasing amount of traffic then you need an increasing amount of thickness in the pavement," says Dr. David Newcomb, vice president for research and technology at the National Asphalt Pavement Association. "What we know in reality, and what the mechanistic-empirical approach can model, is that you can hit a point of diminishing returns. In other words, at some traffic level, increasing thickness is not going to buy you anything."
|This pavement is suffering from thermal cracking, a typical pavement distress due to freezing temperatures. Local validation and calibration of distress predictions are key to the successful implementation of mechanistic-empirical design.|
The new approach also takes into account the properties of the actual materials that will be used in the design. "Take for example an asphalt pavement design," Myers says. "The new design procedure takes into account the asphalt binder content, binder grade, and the aggregate gradation. It takes real material properties and brings them into the analysis tool, which can then be used to make decisions on what the ultimate pavement design should be. Likewise, for a PCC pavement design, the guide uses detailed material properties, such as the coefficient of thermal expansion, to project the expected performance of the pavement."
The mechanistic-empirical procedures also can evaluate the potential impacts of construction control. "If you have a pavement where a lot of segregation has taken place in the asphalt—separation between the coarse and the fine particles—that could affect the variability of the material properties," Newcomb says. "The modeling inputs can include these kinds of material properties, so you can get a sense of what the segregation would do to the performance of the pavement."
Another major departure from the current design guide is that the new guide will provide analysts with predictions for pavement performance rather than pavement thickness values. "When you run a design using the current procedures, you're given thickness values for each of the layers in your pavement design," Myers says. "The new guide is more of an analysis tool. It doesn't give you the thickness of a layer, but instead it predicts pavement performance. So for a layer of concrete, the user inputs a trial layer thickness and other design data, and the procedure tells you how much cracking, punchout, or other pavement distress you'll see for the life of your pavement."
The results of the analysis then can be used to refine the pavement design. "It's an iterative process," Myers says. "If the analysis indicates that cracking will occur, a thicker concrete pavement or a shorter joint spacing may be warranted. The users come up with a different design, rerun the analysis, and evaluate the results again. They can keep making changes, whether to the smoothness or the thickness of the pavement, or the materials they use, until they come up with an acceptable design that meets the distress criteria determined for their local conditions."
Benefits of a Mechanistic-Empirical Pavement Design Approach
What about projects to rehabilitate existing pavements? With the majority of the Nation's pavement design dollars spent on rehabilitation, another critical enhancement offered by the new guide is the ability to analyze rehabilitation designs. Specifically, the guide includes procedures for evaluating existing pavements and recommendations on rehabilitation treatments, subdrainage, and foundation improvements.
"Take, for example, a concrete pavement rehabilitation project that involves rubblizing the pavement and putting additional concrete on top—or even a flexible pavement rehabilitated with PCC whitetopping," Myers says. "Using the current design procedures you couldn't analyze that. But with the new design guide procedures you can."
|Material tests provide valuable inputs to the new design analysis tool. This concrete specimen awaits testing in a water bath, which will measure the change in length of the specimen over a given temperature range to determine the concrete's coefficient of thermal expansion (CTE). The amount of calculated curling stress is sensitive to CTE values, making it an important parameter in the software program for designing rigid pavements.|
|The new procedures enable designers to analyze the impact of construction conditions on pavement performance. Here, designer are sampling hot-mix asphalt from the back of a paving truck.|
One reason a rehabilitation design is possible now is because the new procedures enable the analyst to indicate the smoothness of the existing pavement, typically characterized by the International Roughness Index (IRI). "The analyst enters information into the model on the existing pavement, including the existing IRI, and planned rehabilitation," Myers says. "The model then predicts how the rehabilitated pavement will perform, outputting a new IRI and indicating when cracking might occur."
The new design guide also includes procedures for analyzing more pavement types than the existing guide and software. Although the existing guide includes procedures for evaluating flexible and rigid pavements, the new guide and software include procedures for these two types of pavements, as well as semirigid composite pavements.
And according to NCHRP, the procedures and software provide consistent trial design inputs for each pavement type. The rigid and flexible pavement modules have the same inputs and interface wherever possible. The consistency between the rigid and flexible modules enables the user to experiment with both rigid and flexible solutions to design problems.
|Surface cracking in asphalt pavement, like that shown here, is one of the distress forms that designers will be able to predict using the mechanistic-empirical pavement design tool.|
Perhaps one of the most valuable improvements in the new design guide is that it can accommodate data on local conditions, such as traffic and climate. According to Linda Pierce, a pavement engineer at the Washington State Department of Transportation (WSDOT), including local climate data will be helpful in her pavement designs. "Washington State has microclimates in every climate area," she says. "I can be at the bottom of a hill, and 5 miles [8 kilometers] away, on the top of the hill, it's an entirely different climate region. And yet in my current design procedure, it's all classified as the same climate. With the new procedures, I can actually go to a weather station that's within 5 miles of the project [to obtain climatic data]."
The software also can be calibrated to reflect the properties of local pavement materials. "The new design guide is based on data from the Long-Term Pavement Performance [LTPP] program and other field and lab performance data," Myers says. "But an agency can take its own material, complete a series of laboratory tests on it, and then use those as the inputs instead."
Washington State plans to do just that. "The researchers did a fantastic job of pulling together as much data as they possibly could from all over the country to help generate the models," says Pierce. "But some States don't have the quality of aggregate that is available in Washington State. So we will calibrate the model to our own material properties. And when it's adopted by AASHTO, I suspect that there will be a strong recommendation that the models be calibrated to local conditions."
|A work crew rehabilitates a flexible pavement using the whitetopping approach on I-40 near Henryetta, OK.|
The mechanistic-empirical procedures should lead to a longer life for many pavements. According to FHWA, pavements typically are designed for a 20-year period of performance, although some States use 30 to 40 years. Some pavements, however, last much longer. These may be called long-life pavements.
The NCHRP research team that developed the new guide estimates that pavement performance will improve considerably by using the new design procedures, potentially resulting in substantial cost savings. For example, NCHRP estimates that the new procedures for pavement rehabilitation alone could result in nationwide savings of $1 billion per year over the next 50 years.
A major source of potential cost savings stems from reducing premature pavement failures. The research team's analysis assumed that early pavement failures (pavement life of less than 10 years) would drop from 5 percent to 0.5 percent with the new design procedures. The analysis further assumed that using the new procedures would add an additional 5 years to the current performance life standards of 10 to 30 years.
|Mechanistic-empirical pavement design incorporates data from actual laboratory tests. (Above) A technician prepares asphalt mixture samples. (Below) A researcher runs the Simple Performance Tester to evaluate an asphalt mixture for its response to permanent deformation (rutting) and fatigue (cracking).|
These predictions may, however, represent long-term results, and States may not experience dramatic near-term cost savings. "For some States it will be cheaper, for other States it will be more expensive," says Pierce from WSDOT. "It's all in how you're currently designing pavement, and whether you're overestimating or underestimating traffic levels, material properties, or climatic effects."
The NCHRP research team designed the software to be user-friendly and flexible. As a member of the panel reviewing the new guide, Pierce also evaluated the software. "It's actually a much better software program than we ever anticipated," she says.
The software features three levels of data input from which a pavement engineer can choose, depending on the amount and type of data available. The most complex is a Level 1 design, which represents a full mechanistic-empirical procedure. "A Level 1 design means you've done all the high-level testing and evaluation of pavement materials in the laboratory," Pierce says, "and you've quantified the material properties according to the design procedures." Users can conduct the Level 1 design for any type of pavement project, assuming the engineers have the capability of gathering the traffic data and have the equipment necessary to test the pavement materials.
The Level 2 design relies on both field data and default values from the design guide. According to Myers, a Level 2 design enables users to provide some information, for example, traffic data from a weigh-in-motion station, but they do not need the equipment to do testing on soils or other components. The procedure will take any real data, or a designer can use the default inputs, which are based on national averages of long-term pavement performance data, for the data that cannot be measured. "So it is a bit of a mixed analysis," Myers says.
The final level of design is Level 3, which relies completely on default values from the models and is the closest to an empirical analysis, most similar to what is done now. Level 3 is flexible because it does not preclude someone from using the new procedures and guide if they do not have all the test data from the laboratory or the field data. They still can use the default values.
Ideally, however, pavement engineers should work toward the Level 1 analysis. Washington State plans to approach the new procedures incrementally. "I'm probably going to start my State out at Level 3, the very entry level of the data inputs," says Pierce, "and then enhance it with more data to get it to a Level 2 and Level 1 analysis."
The software is in English units, and it includes a user's manual and onscreen help. The software developers designed the program to run on Microsoft® Windows® 98, Windows 2000, Windows NT®, and Windows XP®, and the results can be output both as hardcopy and electronically in either HTML or Microsoft Excel workbook formats.
"[The software] is a great design as far as the way they have laid out the program and the way that it processes," Pierce says. "It's actually pretty slick. And it's significantly better than the existing software that goes with the 1993 guide."
Introducing… the Design Guide Implementation Team
The Federal Highway Administration (FHWA) has taken a leading role in helping implement the new design guide by creating the Design Guide Implementation Team (DGIT). The team's purpose is to inform and educate FHWA division offices, State highway agencies, industry representatives, and others about the new design guide and assist with implementation.
In January 2004, FHWA distributed a survey to each division office to assess the readiness of the States to adopt and implement the new guide. The results indicated that about 70 percent of respondents believed that their States would be receptive to adopting the mechanistic-empirical approach. A number of respondents, however, requested additional information before they would be ready to use the new guide. In response, the DGIT developed a program to help States understand and implement the new procedures.
FHWA is holding a series of workshops for Federal and State pavement designers and industry and academic partners over the next couple years. "The workshops provide an overview of what is different in the new guide, versus the current design procedures," says Dr. Leslie Myers, a member of the DGIT. "We offer details regarding what you really need to run a mechanistic-empirical design, the testing involved and the lab equipment needed, and the expected benefits."
Future DGIT Activities
As agencies use and assess the guide, the DGIT will provide technical assistance to the States. The team will assist with technical issues, such as local calibration, new materials, and unique load configurations, and arrange small working sessions to address local concerns.
FHWA's new and improved Mobile Asphalt Pavement Mixture Laboratory and the Mobile Concrete Laboratory will help employ and demonstrate the concepts covered in the new guide. "The labs will give States a preview of what they can to do to begin using the design guide in terms of Level 1 materials testing and how they can make implementation easier," Myers says.
Members of the team are Katherine Petros from FHWA's Office of Infrastructure R&D at the Turner-Fairbank Highway Research Center, Leslie Myers and Sam Tyson from FHWA's Office of Pavement Technology, and Monte Symons and Timothy Barkley from FHWA's Resource Center.
For more information, visit www.fhwa.dot.gov/pavement/dgit.htm.
Although a Level 1 design requires many more data inputs than a Level 3 design, some inputs may be more critical than others. Dr. Kevin Hall, a professor in the department of civil engineering at the University of Arkansas, is working on a sensitivity analysis of the procedures and models used in the design guide. His research involves establishing a baseline set of values for all the inputs the user is expected to provide, then varying a single input across a range of values to see how much that variable impacts the resulting answers. Other researchers and agencies also are undertaking similar efforts.
|The graph illustrates the performance of existing pavements, (solid black curve) versus desirable performance (dotted curve). Under the current procedures for pavement design, a sizable number of pavements experience premature failure, shown in the curve on the left, and require rehabilitation less than 10 years after construction. Using the mechanistic-empirical approach, however, (shown in the dotted curve on the right) researchers expect that no pavements will experience premature failures.|
Hall acknowledges that one limitation of this approach is that it does not capture interactions between inputs. "Two inputs into the procedure for concrete pavement design are the unit weight and the compressive strength," Hall says. "Our approach changes only one of these inputs at a time, whereas in the real world there may be a relationship between concrete strength and unit weight such that if one of the values varies, the other might also [vary]."
But even with the limitations of the approach, according to Hall the research should result in valuable information that will be useful to pavement designers. "We can check to see if the models and procedures contained in the design guide give reasonable answers," he says. "For many of the inputs, a pavement engineer will have an intuitive feel for how that input should affect the resulting answer. In concrete pavement design, for example, increasing the compressive strength of the concrete should improve the performance of the pavement in many areas. The other potential value of this analysis is to identify any inputs that do not seem to have a significant effect on the resulting answer."
In other words, Hall's research will help identify inputs that do not seem to have much impact on predictions about pavement performance. "If we can identify those inputs that fall into such a category, we can tell designers to try and obtain reasonable values for those inputs, but not to spend large amounts of time or money refining the input value since the system is not significantly sensitive to the value anyway." This could result in cost savings for agencies during the development of designs.
|Resilient modulus test in progress on a sample of subgrade soil.|
Although NCHRP released the new design guide and software during the summer of 2004, several years may pass before AASHTO officially approves it. While waiting for AASHTO's approval, States can begin getting ready for the improved design approach. FHWA established the Design Guide Implementation Team (DGIT) to help States prepare.
Some States are well on their way. The Montana DOT is midway through a 5-year research effort that involves collecting new data and conducting extensive testing of the most common materials used in Montana. According to Jon Watson, a pavement engineer with the Montana DOT, the State plans to use the data and analysis from the project to calibrate the new design guide model locally.
According to Pierce, Washington State also is taking steps to prepare for the release of the new guide. "I will use my existing software to improve my comfort level and help in the calibration procedure," she says. "I am actually starting the calibration process now and getting a plan sketched out on how we are going to do this."
Because Pierce has been reviewing the design guide and software, her State may be further ahead than others. However she has some advice for other States. "Take one piece at a time and try not to make the transition to the new design procedures a much larger project than it actually is," she says. "Take the new procedures in sizable chunks, evaluate and analyze one feature, and go on to the next."
"I think the new design guide is a great thing," Pierce says. "Although it does not include everything yet, it is a good move forward, and a lot of people will be happy with it."
John D'Angelo is an asphalt materials engineer in the Office of Pavement Technology at FHWA. He has been with FHWA for 27 years. For 12 years he was involved with the implementation of SuperpaveTM through the former Strategic Highway Research Program. He has published numerous papers on material testing and quality control.
Suneel Vanikar leads the concrete group in the Office of Pavement Technology at FHWA. An FHWA employee for more than 20 years, he is responsible for activities related to concrete pavements and materials, including policy, guidance, and technology transfer. He earned his M.S. degree in civil engineering from Colorado State University.
Katherine Petros is the team leader of the Pavement Design and Performance Modeling Team at the FHWA Turner-Fairbank Highway Research Center. She also chairs FHWA's Design Guide Implementation Team.
For more information, visit www.trb.org/mepdg or contact the FHWA Design Guide Implementation Team at firstname.lastname@example.org.