December 2018

Process Control and Instrumentation

It’s not double trouble—it’s digital twingenuity

Much ink has been spilled over the concept known as the “digital twin.”

Henn, J., MAVERICK Technologies

Much ink has been spilled over the concept known as the “digital twin.” However, very little has been done to implement this technology in the oil and gas industry (especially in the midstream and downstream sectors). The industry still relies heavily on using traditional data analytic tools and techniques while continuing to run on legacy computing power. These tools and legacy systems lack the processing capability needed to decipher and enable the data that is generated by sensors and transmitters.

To stay competitive, refiners can leverage the large quantity of historized and sensor-generated data to create and implement an accurate digital twin model. By utilizing new algorithms, tools and techniques, companies can efficiently distill crude data into refined, usable information. An operator can then realize the benefits of implementing digital twin technology. A digital twin can help increase manufacturing equipment performance, reduce operations and process variability, and improve engineering design and change execution. However, as with implementing any new technology, there are operational challenges and resource bandwidth issues to overcome.

A mirror image

First, what exactly is a “digital twin?” A 2002 University of Michigan report1 originally introduced the digital twin concept as a digitalized mirror of a physical asset and/or process. For years, implementing a digital twin into industry was restricted by prohibitively expensive costs and technological limitations. However, recent advances and improved cost effectiveness in machine learning, data analytics, artificial intelligence (AI), Internet of Things (IoT) technologies and cloud connectivity have allowed for the digital twin to become a reality in modern oil and gas facilities.

A digital twin provides unique capabilities distinct from traditional computer-aided design (CAD) and process simulators. It can act as an extremely powerful tool to complement operational and business decisions on both a large and small scale. CAD is completely encapsulated in a computer-simulated environment and has demonstrated moderate success. However, the digital twin’s strength lies in its ability to provide a near-real-time comprehensive linkage between the physical and digital worlds in modeling complex environments. For refiners, leveraging this strength and implementing the technology in a usable manner presents some distinct challenges that they must overcome.

Digital twin challenges

One primary challenge of the digital twin concept is the interoperability between disparate vendor software systems. Due to the longevity of various control system and equipment life cycles, most facilities have a patchwork of different systems handling data. Getting these different platforms and vendors to play nicely together to aggregate the information in the digital twin can be a daunting task. Typically, this requires third-party expertise on the various systems to ensure data integrity. Vendors typically only have expertise in their specific platform or products. They may not be up to speed on the capabilities and unique features of other systems. This is where engaging platform-independent automation solution partners is key, as they have consulting expertise in multiple platforms and services.

Another challenge is determining the fidelity of the information needed for the digital twin. Each case will vary in the amount of data required, and most applications will be unique. For example, a digital twin used to monitor a motor may only require real-time, sensor-generated information. However, a digital twin used for predictive maintenance on the same motor may require much more information to create the models for heat transfer and mechanical torque. Too much detail, however, may create a model where the overwhelming complexity may make it much more difficult to understand. Conversely, too little detail may cause the value of the digital twin to drop precipitously. Determining the optimal level of detail is a major step in deriving value from the digital twin.

As with anything in the technology field, cybersecurity is always a concern. A high priority must be placed on ensuring that the data being conveyed between the control systems and field devices to the digital twin environment within cloud or edge devices is both secure and protected. While any security lapse can threaten worker safety and the business in general, it can also destroy any value you gain implementing a digital twin.

The human element is an abstract concern that may be a challenge. Many people worry that AI will replace them, costing them their jobs. It should be emphasized that implementing digital twin technology is complementary and not meant to compete. Its purpose is to make life easier and increase productivity. Its benefits can be seen where employee attrition or loss of tribal knowledge is an issue.

Closing the knowledge gap

A major concern in the oil and gas industry is how to close the workforce knowledge gap. Many experienced engineers, operators and maintenance professionals are most likely retiring within the next 5 yr–10 yr. Without a way to efficiently capture and transfer this knowledge to new employees and the next generation, this undocumented tribal knowledge may be lost—possibly leading to significant operational challenges and significant costs. What if you could use a digital twin to capture some of this knowledge? For example, a digital twin could be used to capture experienced operator actions and responses. These responses can be analyzed, and experienced employees could train the digital twin model and then use it to train new operators.

With these challenges and concerns, why should a company implement digital twin technology at all? The following are some justifiable cases for digital twin deployment.

The digital twin in action

With a digital twin, a person can create a digitalized model of refinery assets or processes. This model allows for both predictive maintenance and analytics, and then recommends (and potentially executes) prescriptive actions. To put its capabilities in perspective, think about a modern-day search engine. When you initially start typing a word (or letter) into the search bar, it will attempt to autofill the result with predicted searches. By utilizing billions of different searches and data, the search engine’s AI algorithms are trained to predict the full search, becoming increasingly more accurate over time. This same concept is applied to the digital twin. However, when used in an oil and gas facility, instead of analyzing user searches, the digital twin looks at transmitter and instrument data, bills of materials (BOMs), equipment and physical properties, or even operator actions to develop its model.

By factoring in the sustained wear and tear that occur naturally over the equipment lifecycle of an oil and gas facility, the digital twin can accurately reflect all manufacturing defects and perform continuous updates based on real-time data. Because this model is dynamic and responsive, it has a significantly enhanced ability to detect irregularities in equipment and processes compared to visual inspections or traditional modeling and simulation methods—essentially functioning like a high-powered “check-engine light.”

Engineers and technicians can analyze this digital twin model to look for both equipment and operational inconsistencies. They can also monitor an idealized digital twin in parallel with the physical twin (the actual process or asset) to gain valuable online diagnostic insight. By uncovering unacceptable performance trends compared to the ideal range of tolerable performance, the digital twin can help predict when and where a problem might arise as the physical twin’s response drifts from the idealized model. This digital twin capability also helps determine the ideal maintenance schedule to effectively reduce extremely costly downtime.

Working in conjunction with process simulators, a digital twin can also recommend actions and predict outcomes to a much higher degree of accuracy than a traditional CAD-based simulation. By using AI and algorithms to train the model with past and current data, it is possible for engineers to validate and verify process improvement designs much more accurately than was previously possible. Keying off specified constraints and key performance indicators (KPIs), the digital twin can determine optimal running conditions for the process. For example, it can minimize energy consumption while maintaining product specifications, maximize production of more valuable products, or potentially reduce emissions to comply with environmental permitting standards.

Initial steps to get started

What are the steps for deploying digital twin technology, and how can a company start reaping its substantial benefits?

Define a goal. Deploying a digital twin requires intelligent planning and having a well-defined goal, which are both paramount for success. Before any model is developed, a refinery should determine a shortlist of scenarios that would benefit from a digital twin. A company must consider whether these scenarios are valuable enough for the organization to invest in building a digital twin for a significant return on investment (ROI). Each of these hypothesized scenarios should be assessed to identify pieces that provide quick wins—in particular, at the early stages of digital twin deployment. Another major driver of value is the repeatability and scalability of the model, so the scenario should be deployable widely across the facility.

Design a model. After a scenario has been selected, the next step is to identify the model’s configuration and design. This step requires insight from experts of the physical twin, the software systems and vendors involved, and the digital twin technology itself. Without cross-functional expertise and communication between each of these areas, the digital twin’s impact will be severely diminished, and the effort required for implementation will increase, along with the expenditure. This is the stage in which the fidelity of information, the variables and data sources involved, the digital twin software, and the goal of the digital twin’s behavior are fully fleshed out. Engaging a platform-independent consultant who can take a holistic look at the existing system and truly determine the best solution going forward is extremely important. If your manufacturing facility uses a mix of different technologies, this broad experience is especially beneficial compared to the knowledge base of an original equipment manufacturer (OEM), which typically has expertise in just its own products.

Develop a pilot. Using the identified configuration, the implementation team should start developing a digital twin prototype. The team should be adaptable and agile to maximize the ROI. New technologies and data sources (e.g., new transmitters or equipment) should be adopted and integrated throughout the development. The team should embrace using iterative cycles to accelerate learning and decrease the amount of wasted time and effort. One of the digital twin’s powerful capabilities is to create various “instances” (or copies), and to be able to test a variety of configurations throughout different points during the digital twin’s lifecycle.

Deploy the pilot. After the digital twin pilot is deployed and is successfully delivering value in the field, it should be continuously monitored to measure value and optimize performance. At this point, the team should look at opportunities to scale up the digital twin (such as modeling similar assets or processes). Changes to the digital twin should be made iteratively over time, and the results should be analyzed to determine the best possible configuration for future use. The lessons learned, along with the tools, best practices and techniques from the development should be aggregated for future use across the enterprise at large.

The pros outweigh the cons. Digital twin technology—like most digitalization efforts—is not something to approach tepidly. Accomplishing this daunting task requires a very organized approach and a wide array of cross-functional third-party expertise to help overcome potential challenges. However, the benefits and value gained from the digital twin’s powerful analytics far outweigh any issues or challenges. To stay ahead of the competition, it is not a question of if an organization should utilize this technology or not. To maximize the benefits of this cutting-edge digital technology, it is a matter of when and how soon to implement it. HP

The Author

Related Articles

From the Archive



{{ error }}
{{ comment.comment.Name }} • {{ comment.timeAgo }}
{{ comment.comment.Text }}