When Anthony Chan is looking for new employees, he's not just looking at their past work. He's looking to see if they have the skills needed for the future, including machine learning and data analytics.
Chan is the CEO and co-founder of Venturi Engineering Solutions, a boutique engineering firm specializing in energy and pipeline systems. He says emerging technologies like machine learning are having massive impacts on the industry. So when it comes to hiring new engineers, he's looking for people who can work in this new world.
"When we go into career fairs and meet students — yes, engineering is important. But we're also looking at complementary skill sets. Can you work with large amounts of data?"
Chan thinks that flexibility will be a necessity in a rapidly-changing industry. It's a value he's held for a long time, one that led to the birth of Venturi. At the time, Chan was working in a large engineering firm. While he enjoyed the experience, large organizations have a lot of inertia and can't be as agile as some clients need. He started Venturi with the goal of creating an engineering firm that could be highly responsive to individual needs — whether it was those of his clients or his employees.
Predicting proper pipeline pressure
Venturi works with oil-and-gas companies to help them operate and optimize their pipelines: everything from controlling the pressure in the pipeline, to analyzing energy use, to supporting leak detection systems and increasing pipeline safety.
Much of that work relies on working through vast amounts of data to give pipeline operators the information they need to keep things running efficiently. One example he points to is pressure prediction. A pipeline that runs a long distance — say Edmonton area to the U.S. border — is broken up with stations along the route that control the pressure of the pipeline.
If a pipeline doesn't have the optimal pressure, it runs the risk of damaging equipment or compromising safety.
"When we go into career fairs and meet students — yes, engineering is important. We're also looking at complementary skill sets. Can you work with large amounts of data?"
Determining the necessary pressure is no simple task. It can be affected by a host of different factors, including temperature, density, and the type of product running through the pipeline.
Currently, operators predict pressure requirements using traditional calculations. But Chan says it's not a perfect process. Pipelines don't always carry the same product through them. A batch of heavy crude oil might be followed by a batch of light crude or other fluids, adding uncertainty to the equation. Chan says using the pressure data from the periods of transition from one batch to another, and not just from the batches themselves, can make predictions more accurate.
Venturi originally didn't have much experience in machine learning, but Chan had a feeling that it could be a key to solving the problem.
"When we first started, we needed to understand what machine learning was, and we needed to partner with experts. That's why we approached Amii," he says.
"And that was kind of a gateway to understanding what this world is all about."
Saving time, money and energy with machine learning
Venturi worked with Amii through the Supply Chain West Program and the Reducing Emissions through Machine Intelligence (REMI) program. The experience helped them build a solid understanding of machine learning and helped the company develop strategies to incorporate it into their existing data analysis work.
"The initial result we're seeing right now is we have the potential to save these companies 10-15 per cent annually. Let's say a pipeline system costs $100 million per year to operate; we're talking about .. $10 - 15 million a year [saved]," he says.
He's quick to point out that it’s not just about cost savings, however. Using too much pressure to push product through a pipeline can mean a lot of wasted energy. Better predictions can help oil-and-gas companies produce fewer emissions and meet their obligations under environmental regulations.
Chan says it also means less stress on the workers who are operating the pressure stations. The less time that they need to spend on predictions, the more time there is to focus on more important aspects, such as pipeline safety.
Chan is convinced that data analytics and machine learning will continue to transform the way Venturi works and will be vital skills for new engineers entering the field. And his experience working with Amii had helped give Venturi the foundation it needed to keep preparing for that future.
"Amii has a really great approach in framing the problem and helping us understand it deeper. And once we are there, providing the expertise and assistance to help us develop that technology. "