In Vancouver, Canada the SkyTrain network connects the city through three lines and 53 stations. All three lines on SkyTrain run fully automated trains on the nearly 80 km of track making it the longest fully automated driverless system in the world.

In 2009 the third of the three lines, The Canada Line was opened, connecting Vancouver city centre and downtown to Vancouver International Airport and the neighbouring city of Richmond.

MAP OF SKYTRAIN NETWORK

Ever since the Canada Line was opened it has running non-stop with record breaking efficiency. The line has been running for the 13 years with 99.9% availability and 99.8% punctuality, metrics that are unmatched by any other commuter rail line in North America.

Now the line has reached a stage where the wear and tear of over a decade of use is going to require major maintenance and even replacement of important assets and that could threaten the high-performance metric of the line.

The Canada Line’s general manager Ron Powell believes there is a way to maintain the impressive performance numbers even during the increased maintenance. “We have a history of continuous improvement and innovation on the line to get the numbers that we’ve got to sustain the performance that we have. And it’s all aspects of our performance, we’re talking about environment sustainability, we’re talking about safety, we’re talking about security, we’re talking about our quality, performance, and all the metrics that go with that. To sustain that you must have a culture which is going to continuously try to improve.”

In the last couple of years Ron Powell and his team at the Canada Line have been working on building a digital twin of the entire line.

Building the foundations of a digital twin

To build a digital twin of a real-life asset particularly on as big as an entire train line requires a lot of data.

All the way back in 2014, years before creating a digital twin for the Canada Line was even considered, the team has working on developing a variety of software and sensors that could allow them to better understand the condition of assets on the track.

Ron Powell explains how they continuously developed their data collection capabilities, “What we focused on was developing things that would improve our productivity and our safety. It has morphed though into trying to get us to a point where we can actually now start to see more on the condition of our assets than we could with our naked human eye and we’re talking about using advanced sensors, lasers, we’re talking about high speed line cameras, we’re talking about ultrasonic tests, we’re talking about sensors, we’re talking about using gyroscopes to measure to a precision and a level of precision that we can’t do manually.”

Then, in 2018 the director for digital solutions at SNC Lavalin, Johan Germishuys started working on the Canada Line. His first job was to make use of the huge amounts of data that was being collected.

“At that point in time, they had huge amounts of data that they were trying to analyse. But just like anybody else, they didn’t have quite the amount of time to do it. So introduce them to some dashboarding software, Tableau and Power BI to give them some insights on some of the operational challenges.”

With major maintenance challenges coming in the next few years, the Canada Line team wanted a way to bring together all of their data and use it to better predict when the major assets on the line would fail and use that information to better optimise the maintenance and replacement of those assets. So, they decided to start building the digital twin.

To build the digital twin they first performed a LIDAR scan of the entire track. By performing a LIDAR scan they would no longer have to send out rail workers on to the line to check the track alignment.

“The highest risk activity that you can take on a rail line is putting a person on it.” Says Germishuys, “So by doing a laser scan, we were able to bring the laser scan within a real time engine where they could take fairly accurate measurements that avoided having to go out into the field to do those measurements or to see issues over time. And the benefits of that laser scan have been, you know, exponential because we’ve been able to use it to do assessments of various kinds. activities or situations that people didn’t even intend for the laser scan to help them with, but it’s really been very beneficial.”

They brought in new data collection tools to further increase their understanding of the track and its key assets. High speed cameras that can detect submillimetre defects in the track were installed, as were microphones underneath the trains, so sound recording can be analysed to see when the wheels are getting worn down.

Building the digital twin

When it came to setting up the digital twin Johan’s first job is to make sure that the data being collected was useful.

Johan explains “75% of data engineers time that they spend is cleaning data, not actually performing the analytics, the data needs to be of good quality and good labelling, to really get a good prediction out of it.”

Then comes building out the algorithms that can analyse that data, understand the conditions of the track and their assets and then in real time alert operators to any potential issues.

Johan describes digital twins as a constantly evolving system, “Digital twins are not a one off. It’s an evolving system or an evolving methodology.”

The machine learning algorithms allow the digital twin to go beyond just understanding the current state of certain assets, it can begin to make predictions about when assets will start to fail and when the optimum time to replace them is.

“It doesn’t mean you stop things and fix the defects, it means that you start monitoring them. So, then you are getting better ideas about the failure and the failure mode and predicting when that will happen. So that you’re able to plan when you must go in and intervene.” Says Ron Powell.

One of the most important assets to maintain on a train line are the switches If a switch fails it can result in huge amounts of the service being down.

For almost a decade the Canada Line’s switches have been monitored using sensors that can provide data on exactly how forcefully and quickly the switches swing. Now the digital twin project into the future when that switch is likely to fail.

Ron Powell explains how the years of data they already had. Can be used to improve the algorithms that are making the predictions. “We collaborated with a professor in the UK and he was able to take us to a different level of control using regression models, and feeding this information through so that we get much better prediction with confidence intervals around when that switch is most likely going to fail by trending it forward.”

The digital twin is also helping the Canada Line meet its sustainability goals. The aim is for all propane, natural gas, gasoline, and diesel that is used across the line to be eliminated by 2025.

“All of that initiative is also supported by the digital twin because as we push out the assets, we get a better understanding of the condition of the assets and we don’t have to replace assets that aren’t worn. We can delay some of the unnecessary elements of carbon footprint and we also can now take a very strong look at what we’re going to use as replacements and where they’re coming from.”

Improving Maintenance

The Canada Line digital twin isn’t just used to predict when maintenance is required but also improve rail workers ability to do the maintenance.

Maintenance on the Canada Line takes place during the three hours at night when the line doesn’t run, so using those few hours for repairs efficiently is very important.

Certain maintenance tasks will only be done very infrequently meaning rail workers will be far less experienced at it and this is an area where the digital twin can help improve their work through rehearsing the maintenance in a digital space.

Utilising virtual reality (VR) headsets allows repair work to be rehearsed and make sure everyone knows what they will be doing during the crucial three-hour maintenance window. Ron Powell explains the benefits of VR training “You can see exactly how you’re going to do it, you can look at your hazards, you can speak to colleagues about it, and really get it refined so that when you go out there in those three hours, you know exactly what you’re going to be doing.”

Maintenance work and training can also be helped by using augmented reality (AR). When out on the track, work instructions can be displayed using a headset and can help guide a worker through the maintenance process while they are working.

Johan Germishuys explains how the Canada Line has already begun utilising AR, “As long as your work areas are laid out in a similar way, every single time, it will guide you through the process of getting the spare, getting the right piece of equipment, removing it, placing new bolts into it. Over time, they’ve seen benefits from using the tool, because they don’t have to have supervisors necessarily overlook all of the work, they are able to convey a much clearer message and much clearer instructions on what the operator needs to do to maintain those pieces of equipment.”

Connecting Digital Twins

A more major pieces of infrastructure adopt digital twins there is the potential for them to be connected to create a city-wide digital twin, something that has already happened in Singapore.

The Vancouver International Airport already has a digital twin of its own and the Canada Line connects the city of Vancouver to the airport. This means the two digital twins could be connected to help improve the service on the Canada Line.

Although it is not something that is being planned right now Ron Powell explains how connecting the digital twins could help, “We serve that airport, and a lot of those passengers, they come out and get on to the Canada Line. So I absolutely could see, in the future, there’s no reason why there couldn’t be some co joining of the digital twins, where there’s feedforward information associated with planes that have arrived.”

Vancouver also has a large cruise ship industry that results in thousands of passengers arriving all at once. That can result in a sudden increase in need for access to the Canada Line and if they could feed forward passenger number information, they could better plan their services to accommodate the extra passengers.

Digital twins are in an early stage of development and as they become more utilised the opportunity to connect multiple forms of public transport together could improve their efficiency and the customer experience.

For now though, the aim for the Canada Line’s digital twin is to help them maintain their status as North America’s most efficient commuter rail even as they go through a period of increased maintenance requirements.

Ron Powell says that while maintaining the line reliability is key they will continue to explore ways the digital twin can help improve all aspects of the line, “From our perspective, reliability is key. So, we don’t want unplanned, unwanted shutdowns on the line, we want it to be the most reliable service that you could have between point A and point B that we serve. Then the other aspects of it, the safety, the security, the sustainability, they all have to support it and the digital twin will allow us to do that.”