2019 Finalist: ASTech Awards
Inferential sensing reveals new oil sands insights
Many researchers focus on addressing the environmental consequences of the processes that make up oil sands operations. Dr. Biao Huang focuses on the systems that control these processes by using predictive modelling and inferential sensing developments to vastly improve accuracy and reliability. This technology is being used by Syncrude, Suncor, Cenovus, Imperial Oil, Husky, CNRL and Teck Metals and is demonstrating significant economic and environmental benefits.
What problem or opportunity did you identify and seek to address?
In the energy sector, the oil sands process industry is of particular interest in Alberta. My research focuses on process control, modelling, soft-sensing, data analytics and machine learning in both upstream and downstream applications.
In the oil sands process industry, there are many challenging practical problems in taking the oil sands from the mining site, transporting it to the extraction facility for separation and then to the downstream refinery to separate the oil into the different products, such as diesel and gasoline.
One of the main issues with the oil industry is in the sensing technology. The sensors are similar to our eyes, nose, or ears. If you put your head underground what will you see, smell or hear? Without input from your senses, your brain can’t interpret the environment. In our field, the brain is the automation control system. To operate the process safely, profitably and without risk, the automation system needs accurate and reliable measurements.
Part of my work focuses on developing inferential sensing technology. Rather than putting hardware sensors into the oil sands, we use process variables such as temperature and flow rate, which are relatively easier to measure. With the easier-to-measure process variables, we use mathematical equations and machine learning tools to predict the critical variable we are interested in, but unable to easily measure directly.
Inferential sensing technology is critical: to control the process you need to understand the process. By using simple or easier to measure process variables together with mathematic prediction and inferences, you are able to predict and understand the difficult to measure variables, such as steam quality or pipeline transported mixture content.
What has been the impact?
When we produce oil in Alberta, there are associated energy costs such as the energy intensive heat needed for bitumen separation. If you are able to improve the efficiency by a few percent, the savings are huge. Furthermore, accurate reading of critical variables, such as steam quality, can ensure steam generators are operating efficiently and without the risk of damaging the equipment, which can cost significant amounts of money in replacement and operation downtime.
With more accurate measurements in the process control system, the operation is more efficient in both cost and environmental impact. This ensures separation is efficient and as little oil as possible is wasted. Now not only are you producing more oil, you’re also improving the environmental performance. It’s very beneficial to have good automation systems, but it greatly depends on the sensing technology.
How has being in Alberta helped you find success?
The Alberta oil sands industry has a lot of very good engineers. They support new research and new technologies. We have lots of collaborators from Alberta industry and these companies support our students with mentors. They treat us as part of their team and we feel we are always welcome by our collaborators in both the upstream and downstream sectors. It really gives us strong motivation and encouragement because we feel our work is really useful and welcome by our industrial partners.
What are the plans for the future?
We are working with most of Alberta’s oil companies and now a lot of other companies are interested in the machine learning and intelligent control aspects of this work. I think in the long-term this technology will not be limited to the oil industry.
In the metallurgical sector, pulp and paper and even in the banking industry, inferential analysis is very important. In the broad sense we are working toward many industrial sectors to tackle sensing and inferential prediction problems through automation and machine learning.
How does it feel to be an ASTech Finalist?
It’s great that we get to go to the awards and be recognized at least as a finalist by the adjudication panel. I really thank them for recognizing our team for all this work we’ve done. Of course, we did a small piece of work in part of the whole picture, but we are proud of our research and we certainly look forward to learning from the other finalists and hearing their stories.