College of Engineering Blog

Day: August 2, 2017

Verification and validation—Predicting uncertainties early on

Published
By

Shabakhti Research

Mahdi Shabakhti
Mahdi Shahbakhti
Mechanical Engineering–Engineering Mechanics

The verification and validation (V&V) process for a typical automotive vehicle and powertrain electronic control unit takes approximately two years, and costs several million dollars. V&V are essential stages in the design cycle of an industrial controller, there to remove any gap between the designed and implemented controller. Computer modeling has brought about improvements over the years, but the gap remains.

Michigan Tech researcher Mahdi Shahbakhti has made significant progress to remove that gap, using system models to easily verify controller design. His solution features an adaptive sliding mode controller (SMC) that helps the controller deal with imprecisions in the implementation of the system.

The research is funded by the National Science Foundation GOALI program, or Grant Opportunities for Academic Liaison with Industry. Shahbakhti’s team and fellow researchers from the University of California, Berkeley, and Toyota USA in Ann Arbor, Michigan are nearing the end of their three-year collaborative GOALI project.

“Analog-to-digital conversion (ADC) is one of the main sources of controller implementation imprecisions, mostly due to sampling and quantization,” says Shahbakhti. “Our approach mitigates ADC imprecisions by first identifying them in the early stages of the controller design cycle. We first developed a mechanism for real-time prediction of uncertainties due to ADC and then determined how those uncertainties propagated through the controller. Finally we incorporated those predicted uncertainties into the discrete sliding mode controller (DSMC) design.”

“Analog-to-digital conversion (ADC) is one of the main sources of controller implementation imprecisions, mostly due to sampling and quantization.”

– Mahdi Shahbakhti

Shahbakhti and his team tested an actual electronic control unit at Michigan Tech in a real time processor-in-the-loop setup. Their approach significantly improved controller robustness to ADC imprecisions when compared to a baseline sliding controller. In a case study controlling the engine speed and air-fuel ratio of a spark ignition engine, the DSMC design with predicted uncertainty provided a 93 percent improvement compared to a baseline sliding controller.

Toyota works closely with the research team to integrate GOALI project results into the design cycle for its automotive controllers. The company provided team members with an initial week of training on its V&V method of industrial controllers, and also participates with Shabakhti’s team in online biweekly meetings. “The concept of this project is fundamental and generic—it can be applied to any control system, but complex systems, such as those in automotive applications, will benefit most,” notes Shahbakhti.

Continue Reading →

What’s in the air? Understanding long-range transport of atmospheric arsenic

Published
By
Coal-fired power plant on the Navajo Nation near Page, Arizona
Coal-fired power plant on the Navajo Nation near Page, Arizona

Once emitted into the atmosphere, many air pollutants are transported long distances, going through a series of chemical reactions before falling back to the Earth’s surface. This makes air pollution not just a local problem, but a regional and a global one.

Shiliang Wu
Shilliang Wu, Geological & Mining Engineering & Sciences, Civil & Environmental Engineering

“If you’d been living in London in December 1952, you’d probably remember what air pollution can do—in just a couple of weeks, a smog event killed thousands of people,” says Michigan Tech researcher Shilliang Wu.

“Today, photos of air pollution in China and India flood the Internet,” he adds. “Air pollution remains a significant challenge for the sustainability of our society, with detrimental effects on humans, animals, crops, and the ecosystem as a whole.”

An assistant professor with a dual appointment in Geological and Mining Engineering and Sciences, and Civil and Environmental Engineering, Wu examines the impacts of human activities on air quality, along with the complicated interactions between air quality, climate, land use, and land cover. Using well-established global models, he investigates a wide variety of pollutants including ozone, nitrogen oxides, volatile organic compounds, aerosols, mercury, and arsenic.

Wu’s research team recently developed the first global model to simulate the sources, transport, and deposition of atmospheric arsenic including source-receptor relationships between various regions. They were motivated by a 2012 Consumer Reports magazine study, which tested more than 200 samples of rice products in the US and found that many of them, including some organic products and infant rice cereals, contained highly toxic arsenic at worrisome levels.

“Our results indicate that reducing anthropogenic arsenic emissions in Asia and South America can significantly reduce arsenic pollution not only locally, but globally.”

– Shilliang Wu

“Our model simulates arsenic concentrations in ambient air over many sites around the world,” says Wu. “We have shown that arsenic emissions from Asia and South America are the dominant sources of atmospheric arsenic in the Northern and Southern Hemispheres, respectively. Asian emissions are found to contribute nearly 40 percent of the total arsenic deposition over the Arctic and North America. Our results indicate that reducing anthropogenic arsenic emissions in Asia and South America can significantly reduce arsenic pollution not only locally, but globally.”

Wu’s model simulation is not confined to any region or time period. “We can go back to the past or forward to the future; we can look at any place on Earth. As a matter of fact, some of my colleagues have applied the same models to Mars,” he says, adding: “In any case, the atmosphere is our lab, and we are interested in everything in the air.”

 

Continue Reading →