Giant PANDA

Software could make huge impact on asphalt industry

Asphalt Article May 04, 2015
Printer-friendly version

Since asphalt roadways began to appear in the U.S. in the mid-19th century, Americans have been seeking to enhance the quality of their roads.
 
Improved pavement performance has long been the goal of the modern asphalt industry. But even as pavement designs continue to advance, our highway infrastructure ages at a steady pace. So sustainability of asphalt-pavement performance is receiving national and international emphasis. Contractors and highway agencies are facing significant challenges in establishing warranties for the long-term performance of asphalt pavements. Knowing which materials will work best and how long a road will hold up is a key to being able to warranty pavements for a long life—sometimes for as much as 50 years.
 
More mechanistic
One of the historical challenges is that pavement design and analysis has been based largely on empiricism, in which laboratory test data are related to observed performance. The new American Association of State Highway & Transportation Officials’ (AASHTO) Mechanistic-Empirical Pavement Design Guide (MEPDG) marks a historic advancement from empiricism to a more mechanistic approach. Following the MEPDG development, it is important to continue to seek an even more solid, mechanics-based (or mechanistic) foundation for our models, one that more directly considers how the laws of physics and chemistry impact material properties and  pavement responses under traffic loads and fluctuating environmental conditions. In fact, that quest must be an ongoing one.   
 
“There has been a continuing effort to replace empirical methods of pavement design and analysis with more mechanistic approaches,” said Dallas Little, Regents Professor and Snead Chair Professor in the Zachry Department of Civil Engineering at Texas A&M University and Texas Transportation Institute (TTI) senior research fellow.  
 
The latest innovation being developed at Texas A&M University and the Texas Transportation Institute (TTI) is just such—a mechanistic method of simulating pavement behavior and predicting pavement damage that can help produce longer-lasting, distress-resistant pavements.
 
“This mechanistic method is based on fundamental aspects of pavement mechanics, as well as more fundamental characterization of the materials that comprise asphalt pavements, such as asphalt binder, mastic and aggregate,” said Little. “Empirically based models can be very misleading if the specific conditions used to create the models are not representative of a region, climate, or if materials outside of those considered in the empirical model development are used. We use mechanistic-modeling techniques based on the laws of physics and chemistry to create a much broader, more general model that can then be used across many different pavement design and analysis scenarios.”  
 
Texas A&M has researched and developed this new method as part of the Asphalt Research Consortium (ARC). The Federal Highway Administration (FHWA) is funding the consortium with $27 million over a five-year period to evaluate asphalt-infrastructure performance. The effort has now produced a state-of-the-art, three-dimensional finite-element computational model called the Pavement Analysis using Nonlinear Damage Approach (PANDA). 
 
Predicting behavior
Engineers will start with their proposed pavement design and use the software by inputting that information, along with climatic data, regional location, temperature variations, moisture conditions and traffic projections over time. While the model itself is multidimensional and sophisticated, the software interface is friendly and guides users to put in correct data and characterization of their materials.  
 
Once the data are in, PANDA will predict when and where pavement damage will occur, along with what specific type of damage, such as permanent deformation (rutting), might occur under various traffic and environmental (temperature and relative humidity) loading conditions. PANDA considers the impact of moisture intrusion, aging, healing and temperature on how the asphalt-composite mixture responds under traffic and can predict the fatigue damage (fatigue cracking and damaged regions) and rutting depths in asphalt pavements and their progression with time. 
 
Output can be viewed graphically (with color contours) so that users can visually see where the damage and deformations will occur in the pavement, rather than trying to visualize this from looking at reams of data. Users of PANDA also can see how the pavement responds under regional environmental conditions. This approach allows the user to readily compare the usefulness and effectiveness of various material combinations.
 
One unique feature of PANDA is that, rather than considering the pavement as a continuum, it also can focus on the microstructure of the asphalt mixture. This helps the pavement engineer to assess specific mechanisms that can cause failure or to define the weak link in the design.
 
“Our vision is for wide use of PANDA, both within the United States and around the world,” said Rashid Abu Al-Rub, assistant professor in the Zachry Department of Civil Engineering and TTI assistant research engineer. “The results are easy to evaluate even if the user isn’t knowledgeable about mechanics.” 
 
The research team includes Professor Eyad Masad, assistant dean of engineering at Texas A&M University–Qatar, and TTI research engineer, who is assisting with software development and performing validation in both field experiments and laboratory tests. Work continues to make the model more user-friendly for widespread use. Little is quick to praise Professor Bob Lytton, Benson chair professor and TTI research engineer, for his tireless leadership and guidance in advancing the understanding of mechanics and fundamental properties, and their application within the PANDA model. These are the basis of PANDA and help fuel ARC research in many other important areas.
 
“There is not another model like this out there,” said Abu Al-Rub. “The software represents the next generation of mechanistic-pavement models and pavement-design methods.” Abu Al-Rub’s comment prompts the question, “How is PANDA different from MEPDG?” The answer is that while MEPDG is a huge advancement in overall pavement design and analysis, PANDA provides a specific focus on the damage that occurs in the asphalt layers, with a high level of sophistication, as it considers a number of different interacting mechanisms that influence the asphalt layer’s performance. 
 
Sophisticated manner
A good analogy for understanding the benefits and cost savings that can come from utilizing mechanistic models is seen in the automobile industry’s use of computational simulations. Today, the crashworthiness of vehicles is thoroughly tested during the design process using crash-test simulations based on computational mechanics software, which is created based on the laws of physics. This saves the industry a lot of money that would otherwise be spent on multiple, expensive, full-scale vehicle crash tests. The industry can perform hundreds of simulated crash tests to refine and perfect auto designs. PANDA will give the highway community the ability to easily perform similar “what-if” scenarios for asphalt pavements, tweaking mix designs and compositions for maximum performance, while producing quality asphalt mix in compliance with required specifications.   
 
A specific example of a beneficial potential use for PANDA will come on projects where the design-build approach is being used, such as with the Texas State Highway 130 corridor toll-road project. In these types of projects, a consortium of teams combines expertise in design, construction and ownership of a project over an extended warranty period. Since such consortia need to accurately predict costs over the life of the project, they need as reliable a predictive model as possible. Such groups can benefit greatly by using PANDA. Besides predicting performance, the engineers also can use PANDA for selecting compatible highway materials to increase the life of the pavement. Highway agencies, contractors and consultants need a reliable performance-prediction tool to allow them to optimize the use of local materials. Lowering risks of early failure translates to lower cost to the highway agency and the public.
 
PANDA also is a forensic tool. For example, DOTs often use reduced-scale, accelerated loading tests to predict pavement performance. A number of these tests simulate traffic loading in the laboratory and determine the number of wheel passes the asphalt sample can withstand before a threshold level of rutting is reached. But such tests cannot recreate actual traffic or environmental conditions. The tests can only tell you if your asphalt is able to stand up to the specific test condition. Therefore, these tests do not give insight to the reasons for certain trends in behavior of the specimens. PANDA provides the tools to investigate why—to figure it out. So it is more than just a pavement-design tool; it’s an analysis tool. Once again, the thing that separates PANDA from other approaches is the level of sophistication of the analysis, which is able to consider a wide range of interaction mechanisms and properties of the constituent materials.
 
Data to support it
PANDA’s mechanistic constitutive and computational models are developed based on laws of thermodynamics and the continuum damage mechanics framework. The mechanistic constitutive models in PANDA include: 
 
Linear and nonlinear viscoelasticity; 
Viscoplasticity; 
Time-dependent damage; 
Moisture-induced damage; 
Micro-damage healing; and 
Short-term and long-term
oxidative aging. 
 
PANDA also incorporates coupled transport/diffusion models for predicting the distributions of heat, moisture and air within the asphalt-pavement structure, and their changes with time depending on current weather conditions. Therefore, the coupling between these mechanistic models allows engineers to predict the effects of moisture, temperature, aging and healing on the fatigue damage and rutting performance of asphalt pavements with a level of accuracy and precision that is not possible with the current empirical models or performance-related testing methods.  
 
The mechanistic models in PANDA have been validated against a large set of laboratory and field experimental data. PANDA can be calibrated based on standard laboratory tests with a reasonable investment of time and effort, and then used for predicting the performance of flexible and rigid pavements, rehabilitated pavements, overlaid pavements of any type and pavements made with innovative cross sections and/or materials. 
 
The mechanistic approach allows for virtual experiments where variables can be changed—giving the user the opportunity to select the variables in the physical experiment that really matter—to make the actual, physical experiment “smart.” We now have a model based on fundamental concepts and data.  
 
So people may ask—“How do you know it really predicts actual pavement performance?” To validate the model’s results, the team is using controlled field experiments from accelerated loading facilities such as those from Nottingham University’s wheel tracking facility in the UK, from the accelerated loading facility (ALF) at FHWA’s Turner-Fairbank Highway Research Center, from the WesTrack Facility in Reno, Nev., and the NCAT test track.
 
The next steps for ARC involve generating more funding for development of all the products for PANDA. Workshops and an implementation program will get people up to speed on how to use it as well as generate feedback from users on what adjustments can improve the model. In January, the ARC team updated Congressional delegations from the various entities and states represented by ARC, as well as certain Congressional committees on the progress and deliverables of ARC, including PANDA. 
 
“PANDA offers substantial improvements in modeling capabilities,” said Michael Harnsberger, principal scientist at the Western Research Institute. “It will have a great impact on the pavement design community in selecting materials and modeling performance.” AT

Overlay Init