Digital Twins — the scalable value chain optimization building blocks

Like business assets, process manufacturers are not isolated; they exist within the context of supply chains and their surrounding business environment. When a process manufacturer and supply chain respond in unison to market signals and disturbances, value chain optimization is possible.

A key starting point for value chain optimization in the era of digital transformation (DX) is a digital twin.​As virtual or digital copies of humans, devices, systems, or processes, digital twins vary significantly, with different types of digital twins serving different purposes. But generally, digital twins consume data from connected sensors to help plant operators understand all aspects of an asset throughout its life cycle. They strive to answer three questions about the future of an asset: What is next? What if? What is best?

Utilizing a historical view combined with extensive data, the digital twin takes a holistic approach to helping plant operators make decisions on asset performance and other factors. This facilitates faster, more complete, and more consistent human decision-making. In order to best support value optimization, a digital twin should start at the molecular level and then build along the value chain. This allows it to hold a portfolio of physical assets optimal for satisfying market demand.

Molecular thermodynamics and physical property model(s) to start

The most basic components of a molecular approach are the molecules themselves, their thermodynamic and physical properties (including electrolytic properties), and the ability to model their transformation across the ​value chain.​

Examples of transformation across the value chain where this kind of modeling applies include the following:

from reservoir to topside facilities in oil and gas production

from ships, pipelines, and tankers to storage facilities

from individual refinery units to entire refineries, petrochemical and chemical complexes, and associated supply chains

Consistent use of thermodynamics and physical property calculations across the value chain ensures consistent results. Furthermore, the use of the best models achieves the most accurate results.

Digital twin modeling to provide asset context

How molecules behave and their characterization depend upon real-world conditions, and how a process manufacturer operates an asset affects these conditions. For example, the dynamics of a reservoir, the pressure of a separator, and distillation cut point temperature can all impact molecule behavior and characterization.

Because of this, digital twins use steady state and dynamic models to represent the asset. They bring together three things:

the thermodynamics and physical property model(s)

different “unit” operations, such as reservoirs, wells, pipelines, chokes, risers, separators, etc.

various technical operating processes

In doing so, the digital twins develop forecasting, prediction, and optimization scenarios. Such scenarios help answer the aforementioned three key questions—What is next? What if? What is best?—in order to improve future asset performance.

Digital twins are unique in that they function for both design and optimizing asset performance in operation. Predictions are vital in allowing systems to interact with the real world and prevent failure. The digital twins respect the laws of nature, can handle nonlinearities and complex relationships, and can infer information that it is not possible to measure directly.

When models provide deep insight into molecular transformation mechanisms and process operating conditions, they are able to predict corrosion rates and types, and they enable the development of i​ntegrity operating windows​ to prevent asset failure. For example, a digital twin could monitor vapor-liquid equilibria and net positive suction head pressure to avoid phase envelope boundaries where pump cavitation occurs.

Deep insight into molecule transformation mechanisms and process operating conditions also allow flow assurance, field and energy optimization and facilitate the prediction of the following:

Utility digital twins to power the asset(s)

Many facilities in the energy and chemical industries are energy intensive, and in many cases, a facility’s product is also its energy feedstock. As a result, its process and utility systems are closely linked and interdependent.

For example, in liquefaction plants, refrigeration system performance and production of ​boil off gas​ (BOG) interlink via the ​main cryogenic heat exchanger ​(MCHE). The refrigeration compressors often directly couple with gas turbines that impact the fuel balance.

In turn, this fuel balance needs to further balance to the BOG production rate, providing another level of interaction. Considering refrigeration helper turbines or motors adds yet another layer of interaction. These devices link into the power or steam system, with fuel-burning generating power or steam and create system wide impacts.

Similarly, hydrogen plants primarily satisfy refinery hydrogen needs, with the balance recovered from hydrogen-containing streams generated in the refinery process itself. On offshore platforms, gas turbines consume natural gas from the production process to generate power.

Therefore, incorporating the utility system into the digital twin is essential for incorporating a trade-off between power and production.

Building Energy Management System (BEMS)

In building facilities’ aspect with the objective to drive sustained savings by operating facilities at optimal efficiency, Building Energy Management System (BEMS) streamlines electricity, water, gas, solar, chiller equipment performance and environmental sensor data on a single cloud web application for a complete picture of the building performance. The building’s characteristics are learned and modelled automatically by advanced predictive analytics with machine learning capabilities.

Supply chain planning and scheduling digital twins for systems thinking

Both upstream and downstream supply chains present challenges for asset owners and operators. To manage supply chains, asset owners and operators thus need to consider both upstream and downstream factors.

On the upstream supply side, they need to consider the following:

On the downstream demand side, asset owners and operators must consider factors such as weather and customer behavior. This could entail seasonal driving behaviors or social distancing.

All of these factors can have major impacts on business continuity and production. As such, it is vital for process manufacturers to consider them and prepare accordingly. To help with this, process manufacturers can leverage planning and scheduling tools.

Refinery planning tools, for example, can answer questions such as: “Given forecast demands and economics, and the capabilities and constraints of my operating assets, which feedstocks should I buy, and which products will I make, and in what quantity?”

In the same situation, scheduling tools can answer this question: “Given my choice of feedstock and its estimated yield of products, how do I feasibly receive and blend feedstocks, arrange my storage facilities, move materials, and blend and ship the finished product?”

It is important to note, however, that optimal answers to questions such as these are dependent on the model that an organization is using. An overly simple model will result in a nonoptimal answer.

While planning and scheduling tools are typically separate now, they do not need to be. Eventually, these two types of tools will converge so that a single tool can answer both of the above questions. Going forward, these tools will also use more rigorous models. Currently, planning and scheduling tools compromise rigor for convenience.

This is due to the limitations of the hardware and software that are available today. However, ongoing advancements in compute power will enable a single rigorous product to solve multiple practical problems.

Real-time supply chain data feeds

To best respond to market changes and asset disturbances, a supply chain digital twin combines and curates supply statistics with satellite and shipping data. This enables asset owners and operators to improve feedstock selection decisions based on real-time, analysis-ready feedstock data on a global scale. Such data includes dynamic supply-demand balances, integrated cargo tracking and inventory monitoring, and real-time supply chain tracking.

In the oil industry, for example, supply chain digital twins facilitate integrated cargo tracking and inventory monitoring and real-time oil barrel tracking from oil field to refinery. They also align reservoir management activities over the life of the field with product export market dynamics, optimizing the entire production system.

In chemicals distribution, cloud-based solutions that support vendor-managed inventory are vital. Real-time access to inventories at end-user sites allows asset owners and operators to boost intimacy with their customers by offering guarantees that they will never run out of product.

APC and dynamic RTO models to move the plant in closed loop

One factor in the safe and stable operation of an asset is ​advanced process control​ (APC). Where facility operators are not able to consistently optimize independent variables, APC continuously pushes the plant to its optimum constraints.

However, some APC applications are too narrow, leading plant operators to turn them off or manipulate their variables. Additionally, objective functions sometimes fail to update along with plan adjustments, and sometimes the optimization is scope-limited—in other words, APC applications do not consider multiple-unit and nonlinear effects.

To maintain optimization, process manufacturers should consider APC together with planning, scheduling, and first principles models. The best way to do this is with dynamic real-time optimization (RTO) for processes with models that encompass multiple process units or an entire facility or plant.

Work process models to institutionalize operational discipline

Planned and unplanned upsets, slowdowns, start-ups, and shutdowns can all act as obstacles to operational discipline. This occurs both over time and across different shift teams. Human operator digital twins can address these obstacles. They capture, control, and manipulate operator actions in real time through the monitoring and control of operator work processes.

Human operator digital twins are also able to digitalize and connect process safety and operations workflow components in an integrated manner. This is possible via a historized relational database that facilitates the analysis of behaviors and application of artificial intelligence through machine learning, deep learning, and other techniques to solve actual problems.

Asset management and reliability digital twins to maintain availability

A deep understanding of reliability and degradation patterns allows process manufacturers to keep assets available in a reliable and predictable manner, providing a competitive advantage. In fact, leading operators deliver levels of integrity and availability that are 2–8% higher than their competition at a 20–35% lower cost.

Mean time before failure (MTBF) models can predict equipment failure, but to do so accurately, they require a deep understanding of the operating conditions of the equipment. Running equipment aggressively and near maximum design limits causes many failures. As a result, it is easy to assume that an asset running within design capacity will have higher reliability and lower maintenance cost. Unfortunately, in this scenario, reliability and maintenance costs often worsen.

To maintain asset operations within optimal windows, it is important for the digital twins to combine asset reliability models with integrated asset process models for added operational context. This helps balance the multiple factors involved in maximizing asset availability.

Business-wide models to drive value capture at scale

C-suite executives and teams on the front line often have very different understandings of operational models and their relevance. This can lead to poor alignment as organizations undergo DX.

To address this, digital twin technology allows the C-suite to use the same models as operations teams, but with new intuitive user interfaces in dashboards with advanced logic. These dashboards “hide” the complexity of underlying submodels without sacrificing the power and fidelity of the answers they deliver. In doing so, they enable less-expert users to still gain valuable information on asset performance and capability.

How Yokogawa can help

Organizations in the process industries are currently experiencing massive market and operational changes as they undergo DX. In order to stay competitive amid this transformation, process manufacturers should consider value chain optimization using a digital twin that starts at the molecular level.

A partner with extensive industry experience and domain knowledge can help facilitate this value chain optimization. ​Yokogawa is a leading provider of industrial automation and test and measurement solutions for the process industries. Combining superior technology with engineering services, project management, and maintenance, Yokogawa delivers field-proven operational efficiency, safety, quality, and reliability.

For more information on how Yokogawa can help process manufacturers stay competitive amid industry change, check out Yokogawa's DX web page​ ​and consider inquiring about its value-chain optimizing digital twin solutions.