The 5 Components of the Digital Twin for Water
By Colby Manwaring, P.E., CEO — Innovyze
What is a Digital Twin?
The concept of the Digital Twin started as a methodology for virtually representing physical components of any given closed system. This application became prominent in the manufacturing sector over time.
To summarize, a Digital Twin is a virtual replica of physical asset that is updated in real-time via a two-way data connection and, as such, representative of its live characteristics.
So, how has this been applied to the water industry? The first applications of Digital Twin for water were implemented on the ‘plant side’ of water and wastewater systems. Because these assets are contained, the design of their virtual replica was a natural first step in the digitalization of the entire water or wastewater system.
However, a deeply impactful application of the Digital Twin is underway for linear networked water and wastewater assets and the systems that control them.
In this regard, Innovyze has been creating Digital Twins for 30 years. Our hydraulic models are purposefully built to digitally represent water distribution, sanitary sewer, and stormwater piping networks – with their dynamic physical attributes conveyed. We are now collectively taking the next step.
The increasing installation of field telemetry sensors, SCADA, and advanced metering infrastructure is connecting live-data streams to hydraulic models and water infrastructure analytics solutions. For these reasons, the Digital Twin at Innovyze is defined as such:
“A Water Digital Twin is an integrated multiphysics, multiscale, probabilistic simulation of the assets of a water, wastewater, stormwater, or river system that uses the best available physical models, real-time sensor updates, historical performance data, machine learning/AI, etc., to replicate the life of its corresponding real-world twin.”
So, what does this look like in practice? What are the scalable ‘pieces’ that make up a useful Digital Twin? In this article, I will break down the five main components of the Digital Twin for Water.
Component 1: Geospatial and physical data about the assets in the system
The first question one can ask when developing a Digital Twin is – what makes up the system we are replicating, where is it, and what condition is it in?
For water and wastewater utilities, this usually starts with a GIS. Mapping vertical assets like pumps and tanks, along with horizontal linear infrastructure, like pipes, give system managers an understanding of what they have, and where it is.
Now, these are networked systems and – in the world of water – they are hydraulically connected. So, when considering the physical data about these assets – a Digital Twin will show the relative positioning and how one asset affects another.
This can and should go beyond a simple list of what assets make up the system. A comprehensive asset registry will contain current and historical data on not just location, but also the asset’s material, its age or installation date, size, and other unique characteristics that may impact its condition (corrosivity of the soil, proximity to sensitive infrastructure like railways or highways, etc..).
Then, as water and wastewater system managers have this physical foundation of data for their networked assets, they can begin to integrate observed and live data from the field.
Component 2: Direct observation or sensor data about the affecting environment
When the physical characteristics of the networks are defined, we can integrate data about the environment that affects the asset – these are the inputs that affect asset and relational behavior.
Considering the spectrum of maturity that exists for water and wastewater utilities, observed data could take the form of recorded inspection data, CCTV records, or similar types of manually collected field data.
However, as advancing utilities pursue industrial internet of things (IIoT) applications to their networks – in the form of field telemetry sensors, SCADA, and smart metering infrastructures – the opportunity to enhance the volume of real-time data grows in tandem.
Both observed and sensor data give system managers a glimpse into what is occurring in the field and what has occurred historically.
As a simple example, a single sensor may indicate a specific rainwater tank level and answer basic questions such as: how much water is in this tank now? What about yesterday at the same time? Last month? How has rainfall impacted the rate at which it fills?
A useful Digital Twin exceeds these observatory questions and can tell you what will happen to an asset based on environmental factors and gives a diagnostic foundation as to why it occurred this way.
Further to this point, it gives a prescriptive idea of – if it rains X amount, then the tank will fill at X rate. This then becomes the foundation of predictive analytics and can be scaled to very large and complex systems.
Component 3: Performance data
The more historical performance data that exists, the better we will be able to replicate the asset’s behavior.
These can be stored as static data – a single data point at a single time – but as Digital Twin technology continues to evolve, there is a compelling push for dynamic data.
For example, this may take the form of a continuous record of time vs. depth for a given tank in the system. If these variable data points are historically available, say over the course of 10 years, then managers are empowered to confidently predict asset performance.
In this regard, there is a substantial and concerted effort for utilities to “sensor up”. However, this presents a challenge as many water and wastewater systems are expansive and complex. The number of sensors needed to represent these networks in real-time would be extreme.
Securing funds, sourcing and installing instrumentation, reliability complexity – these are the limiting factors, and they are very real for many utilities.
It is appropriate here to reiterate, that for the useful Digital Twin, purpose comes first – the volume and methods of data capture need to align with specific and obtainable goals for stakeholders.
Component 4: The analytics
The analytics are the engine that powers the Digital Twin to replicate real-life.
At Innovyze, we specialize in a special kind of hydraulic and water quality modeling. We develop numerical engines that can perform advanced hydraulic, hydrological, and water quality analysis.
Our engines also analyze asset performance, asset failure – considering both the likelihood and consequence of potential failures – as well as real-time operational analytics to define KPIs for critical event response, and water loss control.
These are all analytical engines that take the data points outlined in components 1-3 of this article and activate them to replicate the life of an asset to a degree of confidence that becomes predictive.
A multi-physics, multi-scale model is a critical part of a Digital Twin. There may never enough data to deduce the behavior of everything out there in the system – even if we had billions of dollars – you cannot put enough sensors everywhere.
But with models based on physics – also known as First Principles models – we can use the laws of physics to predict behavior without needing sensors everywhere.
As the opportunity to pair up Machine Learning with physics-based models continues to grow, it allows us to have more power and control over where we can manage our data input to solve real-world problems.
Component 5: Digitalization
The final component takes the analytics and visualizes it so that the clearest catalysts for action are clearly understood.
For a Digital Twin to be useful, it needs to visualize what’s out there. If complex analytics are underway, translating the results of predictive modeling is essential. Raw data is the foundation, but the main drivers for action become apparent as their analysis is digitalized and representative. The useful Digital Twin facilitates communication. It will visualize and represent outcomes in a deeply analytical, but accessible way.
A useful Digital Twin is prescriptive, provides insights, and most importantly – provides answers to the problem that you need to solve. That means there is not a “standard” Digital Twin that all water utilities must create, but there can be a standard framework, common elements, and digital tools that allow application of the Digital Twin concept in an incremental, needs-driven way.
A Digital Twin is what you need it to be, what you need it to represent, and it provides a pathway for you to improve the performance of your water or wastewater system where it is most needed to better serve the public.
Colby Manwaring will be presenting an ‘Innovation in Action’ Case Study on Innovation and Acceleration: Empowering the Digital Water Utility alongside Carol Haddock, Director, Houston Public Works at the World Water-Tech North America Summit.
Innovyze empowers water professionals around the world to create, manage, and maintain water services. We are the global leader in water infrastructure data analytics software, providing
enduring support for customer success.