When an operator commissions a digital twin for a large asset they expect to visualise their entire plant and ‘walk through it’ in 3d to orientate themselves. They want to zoom in for details of a particular piece of equipment. Increasingly, they also want to see live data getting piped in from sensors attached to the physical asset.
Designing and building these kinds of sophisticated virtual worlds is the job of the development team at Silverhorse. The technical challenges involved are pretty extraordinary, and yet very little of the work we do is really understood or explored by a wider audience.
So in this first article I’d like to shine a light on some of the cool and innovative stuff that the Silverhorse development team is working on right now. Starting with the video game techniques we’ve deployed to create a reliable and dynamic digital twin that also delivers measurable gains in efficiency and safety.
Big and complex
Let’s start with the main problem. 3d models are just as big and complex as the physical assets they represent. But they also have many additional layers of coding and complexity below the surface that users can’t see. This is the digital pipework and wiring that is needed to make the digital twin work smoothly and provide reliable information to the user.
At Silverhorse, the assets we digitalise tend to be large industrial sites such as gas plants. They often have decades of maintenance history behind them that have passed through the hands of various contractors. Inevitably the records are stored away across a multitude of different filing cabinets and databases.
Our job is to build a working 3d world that can somehow bring together and capture that vast quantity of information in digital form. We then need to organise the data, make sense of it in the form of data insights, and then pipe it via digital workflows to the people in operations who are meant to use it to boost efficiency.
Real world challenges
Naturally there are all sorts of practical issues to consider. For example, on a large asset the number of people interacting with the 3d model can easily exceed 100 at any one time, which puts pressure on the system. Connectivity is often patchy or non-existent on remote or offshore assets, so this needs to be planned for and managed as well. Quite often, staff are not exactly keen to embrace new ways of working, and are easily deterred from using digital tools by glitches or bugs. So user experience is paramount if onboarding is going to be a success.
From a technical point of view, this presents – shall we say – a number of headaches. Which nearly all arise from the huge volume of data involved. It makes downloading slow. It causes browsers to struggle. This forces operators to buy expensive machines with extra memory to handle the data. Even then, results can be mixed.
The old approach
The traditional technical approach to getting around (rather than solving) this problem of size and complexity is to slice up the 3d model of the asset into chunks, and only download the bit that is relevant to the job in hand.
This has drawbacks however. Let’s say that a 2km pipeline needs to be inspected. It is going to straddle at least two chunks of the asset. One way of resolving this is to download both chunks and then inspect them side-by-side, but the visualisations often don’t align, and it’s just not very practical on a mobile device as part of an inspection pack. What if the pipeline straddles three chunks or more, for instance?
Clearly, this approach is never going to align with Silverhorse’s guiding objective of creating an intuitive, future-proofed system for our clients that can scale at pace in a low cost, low risk way. So that’s why we decided to invest in an entirely new approach and build from the ground up, drawing on the video game industry for inspiration.
Why video gaming?
Video gaming is a hugely competitive global industry that has poured billions of dollars into creating increasingly realistic digital worlds for its players to roam around. In these worlds, players perform sophisticated tasks and interactions that need to be intuitive and smooth-running, often with limited bandwidth. The parallels with the challenges of 3d asset management are pretty obvious from a technical standpoint.
Unfortunately it was not going to be a straightforward case of switching out a fantasy world populated by zombies for a real world gas plant! But it was a useful starting point.
Our first decision was to use Unity as an engine, which is also used in gaming. This allowed us to deploy the AssetHive 3d model viewer application to both the web, mobile platforms and desktops. We then built the 3d model to replicate an intuitive video game experience, where the user gets to ‘point the camera’ and zoom in on objects they want information on. The gaming trick we used here was to download only the data that appeared in the user’s view, with an outline or ‘wireframe’ to capture the rest of the model. This makes downloads much faster than a non-gaming solution.
When bundled with our AssetHive Client app, the static assets can also be preloaded locally, meaning that there’s no need to wait for any massive downloads. This added bonus enables the user to get straight into the action and complete their task efficiently with all the relevant data at their fingertips embedded in the digital workflow. All on a standard machine.
Because less data needs to be processed using this approach, interactions with the model are faster and more agile, enabling smooth camera movements from all angles, similar to a drone.
Similar benefits apply to planners in the office. This powerful and agile model enables them to step back and ‘connect the dots’ for better strategic decision making, with fresh data flowing into the model from the field as soon as the worker regains connectivity. The digital workflows carrying this data can also be connected to sensors and AI as required.
A poor user experience can be the kiss of death to the roll-out of a digital strategy. That’s why our longer term objective is to develop a photographic level of 3d visualisation without sacrificing performance or budget, where users can digitally see rust or corrosion on a piece of equipment, for example, rather than be told that it’s present.
Just to be clear, this is not about creating shiny 3d models for their own sake. We build these worlds because they are the best proven way to deliver transformative efficiency and safety gains for our clients. By taking inspiration from other sectors, and building out our own bespoke solutions to address validated client problems, I believe we can stay at the forefront of next generation asset management for years to come.