U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

 
SUMMARY REPORT
This summary report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-13-103    Date:  December 2013
Publication Number: FHWA-HRT-13-103
Date: December 2013

 

The Exploratory Advanced Research Program

Multiscale Materials Modeling Workshop Summary Report April 23–24, 2013

Part Two: Discussions

Following the presentations, there was an opportunity for group discussion. Workshop participants and presenters discussed several topics over the 2-day workshop. Five overall themes are summarized in the following pages.

Role of Modeling

The diagram depicts the relationship between computation models and the engineering tools they support. In this diagram, empirical research both leads to and benefits from computational models such as HydratiCA, which assesses IT maturity. Computational models and architecture (model constituency, integration) lead to abstractions. Abstractions then lead to engineering tools such as MEPDG (Mechanistic Empirical Pavement Design Guide) and STADIUM (Software for Transport and Degradation in Unsaturated Materials), which reference vendor documentation. Finally, new techniques for material assessment both lead to and benefit from the various engineering tools and require periodic recalibration.

Figure 13. The relationship between computational models and engineering tools.

Workshop participants and presenters discussed the role of modeling and why these models are being developed and exercised. Participants suggested that there are essentially two purposes for multiscale modeling.

First, the models are used to support engineering tools that will allow the construction of better roads, as shown in figure 13. For example, an admixture company might want to understand the effect on a concrete mixture of adding 20-percent fly ash. A team preparing to lay pavement could use a model-based engineering tool to compute suggested parameters for the paving process. The second purpose is to develop an improved understanding of the physics of the materials processes involved. Participants noted modeling can be used for hypothesis testing when experimentation is infeasible.

 

Motivation for Materials Research

The group noted the importance for researchers to be aware of the end goal of materials research is better roads. The group noted several traditional performance goals during discussion, including friction, rideability, durability, sustainability, and reduced construction time and cost. The group also noted the potential for new material functions or smart materials. One question put forward was how do computational material models help establish or support quantitative performance targets for the materials being studied and developed?

Measurement and Modeling Feedback

The group went on to discuss the essential relationship between materials modeling and measurements. For example, modeling can assist in the interpretation of measurements by providing insights into dynamics at scales below that at which measurements were taken.

Addressing the following two specific needs could improve the relationship between materials modeling and measurements. First, for modeling to advance, new experimental data are required. Current models have been built on measurements that were designed and in some cases conducted years or decades ago. As understanding of material dynamics has evolved, so too must the experiments that validate that understanding so we can go beyond current frontiers. Second, to predict the long-term performance of materials, there must be a tighter coupling between tests and modeling.

Participants noted the point where admixture design parameters could be entered into a model-based engineering tool to reliably predict durability is far away; however, it is possible to do a much better job of predicting long-term performance with interim performance data. If a sample of material was aged and tested and then modeling conducted on the basis of those test results, it would enable researchers to look much farther into the future.

Questions put forward by workshop participants included asking what experiments will provide critical improvements to existing computational models of material performance? Additionally, what new methods of material measurement would allow for new or expanded material modeling capabilities? Another question was what aging and testing protocols would best position modelers to make long-term performance predictions?

Finally, the group discussed development and maintenance of a set of reference materials and problems as a means of comparing models and thereby advancing the state of the art.

Grand Model Versus Focused Models

Workshop participants discussed the state of the practice in modeling. They determined there may be value in developing a grand model that stitches dynamics across the length scales, although participants questioned the feasibility of developing a useful grand model.

Participants noted a grand model could possibly be constructed by marrying the models from the workshop that work at the micron level and below with the NIST effort, which works at larger length scales. This might work well as a modular system but participants noted not all analyses would necessarily be best accomplished with a grand model. Accordingly, focused computational models or tools that deal with a particular mechanism or problem would continue to have a place. Participants also noted that many questions can be answered without considering multiple scales.

Participants went on to discuss how models can be used to improve understanding of many properties. The following properties were discussed:

Several questions were raised as a result of this discussion. Participants questioned whether focused models are extendable and, if so, which ones are. Moreover, they asked what it is about a model that most influences its extendibility. Additional questions asked if it is conceivable that a grand model could work for any combination of aggregates and cement, and what elements a grand model needs to operate concurrently or sequentially. During discussion, it was noted that to understand cracking, model elements must run concurrently. Participants questioned if existing models allow for integration—they asked if there is a need for common or coordinated approaches for integration or a need to improve the ability to abstract computation models for improving engineering tools. Participants also queried how engineering tools have used computation model improvements. One final question asked if the grand model is an integration of focused models or if it requires an entirely separate approach.

 

Gaps and Opportunities for Advancing the State of the Practice

The final discussion theme examined some of the challenges ahead. Participants considered the various scales at which the models operate and concluded the most challenging length scale is the upper end of the mesoscale, near 50 nanometers (nm). Participants questioned whether anyone was focusing modeling efforts at scales near 50 nm and attempted to identify the key challenges in that modeling. Additionally, participants queried whether anyone is developing computational models for the interface between aggregates and binders.

Discussion proceeded to address the topic of applying microscale modeling to engineering problems. A critical question identified here is how and when to abstract heterogeneous materials. For example, finite element models need homogeneous cells to predict deformation. Variability was a feature identified as an essential part of asphaltic and cementitious material engineering and analysis; however, stochastic modeling was put forward as one method that could be used to manage the uncertainty associated with that variability.

Next Steps

Following discussion, workshop participants identified a couple of follow-on activities as next steps following the workshop.

These activities included: (1) characterization of existing computational material model research; and (2) additional discussion to gain feedback from people developing engineering tools.