The twin that’s key to decoding asset system DNA

Share this page

October 2020
Sushil Kulkarni

Head - Digital Solution Delivery
Process Automation Digital, ABB

Understanding assets and asset systems holds the key to achieving several of the business outcomes and strategic objectives that industries target; and data is a critical enabler of this understanding.

Using both historic and real-time data to determine diverse asset performance parameters and running simulations to assess potential failures, drops in performance and other key factors can help industries enable operationally robust and commercially vibrant enterprises. Digital twins have the answer as long as they are built using deep domain understanding along with technological innovation.  

The unique relationship that is a twin

There is a (purportedly true) story about twins who were put up for adoption and separated at birth in the 1940s. When they met for the first time at the age of thirty-nine, it turned out that both men had married twice – first to women named Linda, and then to women named Betty. Both had identically named childhood dogs. Their sons’ names were James Allen and James Alan. Both worked as sheriff’s deputies and had the same preferences for various products, right down to the car they drove.


That is how harmonized twins can be, down to the basic DNA level – true of human beings themselves and of their creation – digital twins. Asset systems, their constituent assets and components within assets are always evolving – changes driven by how the system and asset constituents react and respond to various triggers, whether age or operating condition related. Such changes have a direct bearing on process effectivity, and, therefore, an impact on overall operational and commercial performance of plants.


Many of these changes happen in a black box – invisible to plant owners and operatives, who continue to act on intuition or use traditional methods for asset maintenance and manufacturing operations management. Digital twins help demystify this black box by providing an essential replica of the physical asset, system or process in digital form, enabling interventions by people at the plant before problem occurs. Digital twins help understand and assess impact of any intervention in digital form first, before making that intervention in the physical world; thus helping run process plants most optimally and achieve overall performance objectives.

Structuring digital twins for impact

Orienting digitalization to achieve key industrial business outcomes

The core aim of digital twins is to create digital replicas of a physical entity. With this intent, they must cover key aspects of the physical entity – including a digital model, its look and feel, forward simulation capabilities, relational interactions with system of assets or processes, connects with real time data to mirror reality and 360o information.

To achieve this, digital twins work bottom up - starting with digitally replicating components of assets to come together for asset twins; and then consolidating various assets into a system of assets to create system twins. If done comprehensively, the opportunity is also to create a complete process twin, aligned to having a transformative impact on outcomes.

The two important parameters of a digital twin are information twin and model twin. The former captures, in detail, diverse static and dynamic / variable data about the physical entity, and the latter captures behavior of the asset or physical entity, hence replicating it completely. A digital twin is able to transform productivity and throughput, quality, operational efficiency, asset integrity, safety, supply chain optimization and sustainability performance across assets, system of assets and processes.

 

Data coverage – the foundation of digital twin success

The ideal digital twin model captures a plethora of information – including engineering design, relationships, documentation, operating parameters, spatial information, asset history including failures, work order history, component replacement history and other related transactional data. At its very core level, the digital twin acts as a highly effective integrated source of information on all aspects relating to an asset.

In addition to this, a wide variety of variable metrics, primarily sensor-driven real-time data, is an important aspect of the digital twin. In all of this, it is important that the digital twin leverages as much contextualized data from engineering, spatial, operational and IT systems as possible. 

Ingesting, contextualizing and analyzing cross-functional data is key to a successful digital twin

Data capture is just the starting point

Information by itself does not indicate behaviour. The ideal digital model is designed to capture behaviour and apply physics-based or AI / ML models to replicate the behaviour of the physical entity in the digital world.

This is used primarily for three intents – what-if analysis, optimization and simulation. Outcomes from digital model inferences can be used to replicate a wide range of actual scenarios. This provides the flexibility to run plant use cases in the digital world for generating right model inferences and outcomes without having an impact on the running of the physical entity or process.

Digital twins leveraging trained models can aid immensely in performing data-driven descriptive, diagnostic and predictive analytics. Predictive analytics can provide accurate target parameter predictions in near real-time, thus allowing for process interventions faster and without having to intervene based on actual outcomes, which may imply lag. For example, if the intervention is based on quality results from laboratories, the action would be delayed significantly. In such scenarios, accurate model inferences can provide near real-time predictions enabling timely interventions.

What-if scenario and simulation capabilities using trained model inferences can be very useful in gauging impact of certain variable disturbances on the overall functioning of the asset or process; or to optimize the asset / process operation responding to such changes.

Optimization analysis helps understand best decisions for an asset to get optimum performance. Take, for example, condition monitoring, which is key to asset performance optimization and enabling predictive maintenance. Digital twins can help understand the condition of an asset at any given point of time using sensor and instrumentation data; and deriving model inferences based on this dataset. The digital twin can also facilitate unsupervised anomaly detection by tracking variables based on thresholds and Integrity Operating Windows. Data descriptive analysis, time series analysis and prediction based on model inferences can accurately predict asset performance in the immediate future.

Digital twins allow for what-if analysis, optimization and simulation using digital methods, without interrupting the operation of the physical entity 

Digital twins that work best simulate reality most effectively

Strong visualization and intuitive interfaces are key to utilization of digital twins to their maximum potential. The desired characteristic of effective visualization is a seamless, hierarchy-driven drilldown of asset engineering, IT and OT information; and effortless access to all relevant associated information, including drawings & documentation, based on asset object relations. Well-designed visualization also effectively utilizes graphic visualization such as 2D and 3D models. An effective visualization is important for unlocking the full potential of digital twins.

Finally, it is critical that the digital twin structure be based on a true distinction between the physical and digital world; and how they connect. The connect between the digital twin and the physical asset plays a role in autonomous operations, where outcomes based on model inferences are relayed back to the physical world to close the loop.

Visualization and intuitiveness of interfaces play a key role in maximizing the value of digital twins

Wide-ranging impact across industries

The importance of application of digital twins is recognized across industries. Digital twins of assets play an important role in condition monitoring and condition-based maintenance. Process digital twins can accurately predict target process variables, thus facilitating timely interventions ensuring safety, reliability, sustainability, improved throughput and on-spec quality.

 

Few examples of application of digital twins:

  • In the cement industry, near real-time Blaine prediction can improve cement quality, throughput, energy savings by optimizing the mill process
  • In mining, for ore processing, the froth floatation processes can be made most productive on a real-time basis to optimize the feed-to-throughput cycle
  • Heat exchanger fouling and its usable life can be very accurately predicted thus allowing timely and right interventions to be made


Digital twins designed using industry knowledge can act as a very powerful tool to optimize manufacturing operations, assets and processes, thus ensuring process integrity, operational excellence, asset integrity, safety and health of personnel, supply chain excellence and sustainable operations. 

Digital twins have powerful impact across functions, processes and industries


About the author

Sushil Kulkarni is Head – Digital Solution Delivery for ABB’s Process Automation Digital business. In this role, he spearheads the development of repeatable deployable industry business value applications using the ABB Ability™  Genix Industrial Analytics & AI Suite; and digital twin platform components such as System Twin Integrity Hub, Augmented 3D, Model Fabric for auto ML and calculation engine for physics-based modelling for the Suite. He also leads digital solution engineering, digital services & solution delivery, pre-sales & proposal development. 

Starting his career as an instrumentation engineer, he has worked for 32 years across industries – nuclear power, petrochemicals, chemicals and IT. He has in-depth understanding of process control, multi-discipline engineering, plant construction & commissioning, plant asset information lifecycle, system integration, software development and analytics; and has participated in several large digital transformation projects for global owner / operator corporations. These projects have encompassed digitization, asset lifecycle information management, solution engineering, solution blueprinting, software development, system integration and deployment. He has a bachelor’s degree in electrical engineering and a post-graduate diploma in business administration.   

Learn more

  • Contact us

    Submit your inquiry and we will contact you

    Contact us
Select region / language