Digital twins optimize investment
iStockIn our previous story on digital twins, we pointed out why they should be an industrial CEO’s best friend, as they deepen the instrumentation and understanding an executive can have of the business. This allows companies to both identify risks and make better judgments about what will happen in the future.
That insight is important not only to operational efficiency but also regarding how companies can best allocate the money they’re spending to improve their environment. In this piece, I again talk to GE Digital’s John Renick about the way that a CEO and a CFO can be much more confident they’re spending money in the right place during their digital transformation journey because the information from their digital twins guides them in investment allocation decisions.
The Link Between Digital Twins and Investments
Many companies use digital twins to drive improved operational efficiency in large industrial environments. However, they can inform many other sorts of processes, including providing important clues about how to allocate an investment portfolio. Here’s how such financial insight comes to life.
Generally, when starting out with digital twins, the primary mission of using these models is to create a high-resolution data-based view of the equipment and key processes inside an industrial facility. Digital twins model components inside devices or assets, the assets themselves, systems of assets, or larger processes involving many systems.
John Renick, Sr. Director, Product Management at GE Digital
GE DigitalIn the first phase, implementations are focused on optimizing operations of the equipment in a variety of ways including:
- Improving reliability
- Standardizing preventative and condition-based maintenance processes
- Understanding how to optimize performance of assets and systems of assets
- Using AI and ML in the most effective way to find events that are predictors of problems
But once a digital twin has been in place for a time, its use greatly expands, as does the way in which it adds value to the business. The end result of high-potency digital twins is the creation of a comprehensive model of activity that can be used to guide investments in new and existing equipment, technology, and instrumentation toward the assets and processes that will have the greatest ROI.
Think of the theory of constraints made popular in the Eliyahu Goldratt’s book The Goal. A high-resolution operational model can show where in the larger system constraints exist. Investment can then be targeted to remove the constraints (for example, by improving performance or expanding capacity or refreshing technology). Then a new constraint can be identified.
Such a program optimizes investments based on data, analysis of that data, and knowledge of the application and best practices, rather than instinct. Here are some examples of how digital twins lead to the ability to optimize technology investments by providing more detailed views of what’s happening in the business, potential risks, and the best places to leverage investments.
Filling Gaps in Coverage
Renick gave an example of how mapping a complex digital twin to a failure curve can greatly improve a company’s investment decisions in technology to support the performance and availability of assets and processes. “Long before a piece of equipment fails, there are generally things an operator can do to mitigate or even prevent that failure from occurring,” he said. “And then even before that, there are insights that can tell you well in advance what you should be doing to proactively maintain that equipment.” Digital twins give you insight into these signals and how often you should be looking at those signals to mitigate risk.
In the real world, an example of this is the vibration of a rotating piece of equipment. Most of these types of equipment have alarms that allow companies to know when they’re close to failure. But well before these alarms occur, the owner or operator of this equipment should be looking at data to understand the performance and risk of failure. How frequently data should be collected and analyzed can be associated with the importance of this piece of equipment relative to its overall process or operation. For a noncritical run-to-failure asset, it may be sufficient to collect data on a weekly or even monthly basis. For more critical rotating equipment, like a generator, high-resolution and high-frequency data would enable better insights and risk mitigation, making a case for investing in better sensing technology. Digital twins can identify these gaps in coverage so that companies can make the right investments based on their business needs.
Improving Systems and Processes
Another example Renick gave of the way digital twins help shape improved performance within a business is with fluids moving through motors and pumps in an industrial system. With digital twins, companies can measure flow rates over time and recognize when and how a pump is degrading and do so much earlier in the product lifecycle than they could otherwise. This can lead to entire system improvements.
“The pump is having an issue that may actually be detected by the amperage on the motor going up,” Renick said. “So the motor is working harder to compensate for the pump and therefore the motor will experience problems because of the issue associated with the pump. That’s where the system aspect comes in; you get the interactions of the equipment in this case, or multiple parts in a larger system.” You can therefore pinpoint how a single asset failure is caused by something else and can affect an entire system.
Using AI and ML More Effectively
Every company is aware of the hope and promise of AI and ML. But digital twins enable companies to move beyond the potential and actuallyapply the analytic and transformational insights of AI to improve the bottom line.
Digital twins can do this because they help to identify the problems within your business. One of the major challenges of AI is that for it to address a problem, it needs some form of data structure and context. Digital twins provide the necessary training data as well as full source data, modeled in a usable fashion, along with the knowledge to go with it. Without this, AI algorithms won’t know what to look for.
“In the industrial space, it’s not uncommon for AI investments to underwhelm and underdeliver when they have been made prior to digital twins. Studies have shown that a significant portion of an AI project’s time and cost can be spent trying to get the right data into the right data model, resulting in budget overruns or project failures. Without the right understanding of historical behavior, current behavior, and desired future behavior, it is challenging for data experts to validate any results produced by AI. This approach is labor intensive and fraught with risk. Whereas if you start to implement digital twins, and you’re collecting data and gaining insights along the way, that’s going to help guide the AI in the most effective manner and achieve benefits much more readily and with more assurance than you would with the first approach,” said Renick.
No Longer Living in the Dark
Companies that have implemented digital twins typically hit an “aha!” moment. They understand what digital twins can do for the business and realize that, for so long, decisions have been made in the dark. Until now, companies have had to make decisions based on far less information, using far less sophisticated models than they can now.
“With digital twins, we’re actually making informed decisions – not just based on data but also on knowledge – instead of making ‘educated’ guesses, which is the in-the-dark approach,” said Renick.
Additionally, Renick pointed out that digital twins make the entire staff more efficient and better able to perform their jobs.
“In a large manufacturing organization, there are always teams charged with making sure that we have reliable operations. Today those people still spend much of their time gathering data from multiple sources, scrubbing that data, cleaning that data, and trying to figure out what the data is saying versus making reliability decisions that help their organization. With the information provided by digital twins, that’s no longer the case. People can do the job that you think they’re doing today but they’re not.”
Ultimately, the end result with digital twins is that the percentage of work on value-added analysis goes up and the percentage of work done on data engineering and gathering data that a digital twin could do a lot better goes down. This helps companies to reduce costs across the board.
