By Jay Landers
Proper compaction is key to the performance of asphalt pavement. Improper compaction, on the other hand, can lead to cracks, ruts, potholes, and other failures that can impair pavement performance, shorten roadway life spans, and damage vehicles.
Density is the main parameter that is used to determine whether asphalt has been compacted properly. For this reason, roadway owners typically require that contractors extract a certain number of core samples from a newly paved road and have their density tested to verify that they conform to the specified density levels of a design. However, this verification method has its drawbacks, chief among them the inability to test asphalt density levels during construction.
A newly available verification technology aims to overcome this drawback by providing real-time assessments of asphalt density during the compaction process, offering contractors the means to ensure proper pavement quality during construction. In this way, the technology — which is known as RTDensity for the real-time nature of its density readings — can improve the performance and longevity of asphalt roads and lower construction and maintenance costs, according to its developers.
Using machine learning
RTDensity debuted in March at the construction trade show known as the ConExpo-Con/Agg in Las Vegas. The device consists of an integrated measurement system that includes a temperature sensor to record the temperature of the asphalt as it is laid, a GPS to record positioning, and an accelerometer. The latter is attached to the axle of a vibratory roller to detect and measure the vibrations experienced by the roller as it compacts asphalt.
RTDensity was developed by Sesh Commuri, Ph.D., a professor of electrical and biomedical engineering at the University of Nevada, Reno. Aware of the many problems and expenses associated with improperly compacted asphalt roads, Commuri decided that there “had to be a better way” than extracting and testing core samples to determine asphalt density, he says. “One of my focal areas is in building systems that are intelligent.” Enter machine learning.
At the heart of the RTDensity technology is a proprietary algorithm that estimates asphalt density based on the vibrations detected by the accelerometer. Using artificial intelligence and a form of machine learning known as clustering, RTDensity is calibrated in such a manner that it learns to associate different vibration signals with certain densities. “We train the system to learn these vibrations,” Commuri says. “The system learns how these different densities look.”
The state of Nevada holds the patent for the RTDensity system, which has been exclusively licensed to G4 Technologies Inc., a subsidiary of the construction company George Reed Inc. G4 Technologies is manufacturing the system.
Drawbacks with existing methods
To make hot mix asphalt, heated aggregates are combined with liquid asphalt cement at temperatures ranging from about 300 F to 350 F. Delivered by truck to the construction site and applied by pavers to the desired thickness, the asphalt is then compacted to the desired density by a series of vibratory rollers and ultimately by a static roller to provide the final finish. “If you don't compact (the asphalt) well, the quality of the road is shot,” Commuri says.
Temperature is a key factor in ensuring the success of compaction. Below a certain threshold, asphalt becomes too cold to be compacted properly, Commuri says. “You have a narrow window during which you have to do the compaction.” Over-compaction is another concern. Asphalt that is compacted too much will break down and start to fall apart. In either case, the asphalt does not achieve the proper density, potentially leading to poor roadway performance.
The most used method of verifying the density of asphalt involves letting the asphalt cool and then extracting three to five cores of the pavement per lane mile of roadway. The densities of the asphalt cores then are verified in a laboratory. “That's the gold standard,” says Garrett Winkelmaier, Ph.D., the technology development manager for G4 Technologies.
This approach has its drawbacks. “Three to five measurements are not going to tell you anything about how a road that is a mile long is going to function,” Commuri says. What is more, obtaining density results after construction has been completed obviates the possibility of modifying the construction process, if necessary, to obtain the desired density.
Meanwhile, the coring process itself damages the asphalt and can contribute to poor road conditions in the future. “No matter how you patch (the area from which the core has been taken), it’s the first place where you start having all kinds of issues,” Commuri says. “These things spread over time.”
Another tool for estimating asphalt density, the nuclear density gauge, also has its drawbacks, Commuri says. The handheld device has a radiation source that emits particles and uses a sensor to measure how much radiation is returned. By comparing the result to a calibrated reading, the gauge estimates the density of the asphalt. Because of the radiation, use of the device is “very tightly regulated,” rendering it expensive, Commuri says. Furthermore, each reading takes several minutes, limiting the gauge’s usefulness as a quality-control tool during construction.
Responding in real time
By contrast, density results from RTDensity are displayed on tablets during compaction, informing roller operators of their performances in real time. Following calibration, the device provides continuous measurements of asphalt density, stiffness, and surface temperature for each pass and for every point along the entire road. “We can provide (contractors) a complete profile of their construction process,” Winkelmaier says.
Armed with this data, roller operators can adjust their operations as necessary to avoid under- and over-compacting asphalt, improving productivity and reducing costs. “You can go and make a change right there while the asphalt is still hot,” Winkelmaier says. “If you are not getting high enough densities, you might need to make a change to your mix or increase the number of passes that you make over that road. Such changes are going to improve the quality of the product that you're constructing.”
If the desired density is achieved sooner than expected, roller operators can avoid making unnecessary passes over given sections of asphalt. “That will speed up production and reduce costs as far as the fuel that you're using and the time that your crews are out on the road,” Winkelmaier says.
In some cases, contractors that fail to achieve desired asphalt densities may receive reduced payments or be required to reconstruct significant portions of roadways at their own expense, Commuri says. Real-time density readings can go a long way toward helping contractors avoid such costly outcomes, Commuri says.
Contractors also can use RTDensity to provide their clients with complete documentation of compaction quality, Commuri says. “We have a back-office solution that will help contractors to upload the compaction data onto our portal, and it generates all the reports that they need on the fly. At the end of a project, they can take the entire report and give it to” the owner of the roadway, he says. Instead of knowing only the compaction data associated with a small number of core samples, the client will have such data for the “entire extent of the roadway,” Commuri says.
Independent validation
Beginning this summer, G4 Technologies will conduct additional “independent validation” of RTDensity, Winkelmaier says. “We are looking for about three to four independent contractors to test the technology for a one- to two-monthlong project,” he says.
By the end of this summer, G4 Technologies expects RTDensity systems to be available for the public. The technology is expected to have a list price of approximately $65,000.
Ultimately, RTDensity could be configured to determine the density of other compacted materials, including soil, though further research is needed, Commuri says.
This article is published by Civil Engineering Online.
