Operational Digital Twin or Avatar

Operational maintenance optimisation with Digital Twins/AI/ML

Advise on performance improvement using AI/ML/DT for a major UK Water & Wastewater company

In autumn 2022 we were asked to support and advise a major UK Water & Wastewater company. Their operational issues stemmed from a complex system of assets, many diferent skillsets and time critical maintenance and incidents.

Workshops held with the client team exposed the areas that were causing problems. They were typical for an operational / time critical environment - too many things to fix in too little time, firefighting rather than getting ahead of issues, resourcing & skills. Additional significant issues were emerging due to weather impacts. Extreme events such as the 2022 drought which was quickly followed by huge storms. These events had exposed that the maintenance strategy was not improving resilience of the overall asset network.

Taking all the factors into account we created; a high level process map, suggested a technology roadmap and set of 'Operational Digital Twins' or 'Digital Avatars' to support the operations team, linked the operational goals to strategic goals to improve performance and operational resilience.

Why bother to invest in an Operational Digital Twin (ODT)?

The ODT objective will assist operational decision making towards achievement of the benefits below:

Asset Monitoring and Proactive Maintenance:

Reduction in number of asset and service failures
Reduction in non-value add fixed time maintenance
Ability to reduce operational expenditure by running more assets to replace on failure

Works, assets and networks:

Ability to monitor site performance to reduce unnecessary expenditure on consumables such as energy, chemicals etc
Ability to give site operatives a clear view when attending a site of what needs their attention, improving effectiveness and efficiency of site visits
Ability to view the performance of assets on more of a catchment basis, allowing better and more effective identification and analysis of emerging risk to customer and environment

Use of an ODT will improve the evidence of ‘Duty of Care’ and ‘Due Diligence ’to Ofwat and Investors, a critical communication aid when confidence in the Industry is waning.

Operations supported by Data, AI/ML & ODT

Digital Twin Simulation models represent (sections of) the Works, Assets & Networks and simulate their operation to predict likely outcomes under different environmental and loading conditions. For example, what electrical failures are likely due to strong wind, what overflows will occur with heavy rain after a dry period, or what are the tasks I need to focus on as winter arrives. This combination of AI/ML and Digital Twin Simulation models can transform the performance of a business.

It provides insight, foresight, and direction. Identifying the actions that should be taken to achieve cost, performance and risk aims. The processes and supporting tools described by a recent client demonstrate a gradual move towards a data-driven & AI/ML approach. However, most tools were dashboards or pattern recognition for sensor alarms. Whilst these are key elements, they provide minimum predictive ability and little decision support. To truly improve performance predictive tools are needed that provide multiple ‘what if’ outputs, and that provide optimum or least worst choices.

What does a ODT look like and how does it fit into other business process?

Operational processes do not exist in isolation of the rest of the business. In addressing a potential ‘end state’ Digital Twin use for the Operations processes we need to note these other significant processes, and business needs. They interface and impact operations as much as external environmental influences and asset use, duty cycles or deterioration.

Figure (above) shows three interconnected processes at a very high level, with the operations process in slightly more detail.

These three processes are interconnected. They share the same data but operate at very different granularity and over different timeframes.

Strategic – the Strategic process is focussed on periodic (5 year) and long-term (25-35 year) planning and forecasting. In water/wastewater utilities this is heavily influenced by Regulatory requirements from OFWAT & Environment agency but also to provide board a future view of risk, cost and potential profitability. The Figure shows the simplified process looping around via;

Construction / Repair / Rebuild represents the build actions. These are informed by detailed project plans and construction plans. They may even have full construction Digital Twins. New build and repair projects should improve asset data quality.
Asset Register & GIS the central data system that contains all asset descriptions, data and history. Accuracy and completeness of this live data system is paramount for all asset and network related processes. It informs all of them and receives new data from across the business continually.
Periodic Planning Analysis using the live and historic data from the asset register, GIS and all asset works this significant process will use Modelling & AI/ML to forecast the future needs of the business and assets. Whether to fulfil regulatory obligations or board information, Periodic planning will provide fully costed options for future works, describe the performance, risk and cost/profit profiles that different levels of investment will give so that the board can choose their preferred route and set budgets.
Capital Projects Management as a result of periodic planning, and business led needs, a set of capital projects to be implemented will be managed. These can range from large replacement projects, building new treatment works to large IT projects. Each should have a cost and performance benefit associated so that the impact of delays or not implementing the project can be assessed and managed within the overall business context. Modelling/Optimisation software is often used here to help prioritise the very large set of potential projects.
- Design Refinement (Tactical) projects and works within the Capital Project Management process will be at different stages of refinement and definition. It is typical that outline projects are initially promoted into the overall plan. These are then further refined, with alternate implementation options considered and costed/performance forecast. This is often considered/performed by a team looking in a more tactical timeframe.

Tactical – the Tactical process looks in the shorter term (1-5 year) timeframe to understand the behaviour of the assets & network in order to identify issues early and improve reliability.

Asset Register & GIS
Condition, Risk & Reliability Based Maintenance is a set of analytic techniques used to understand the behaviour over time of the assets and network so that actions can be identified to improve Reliability (see Condition Based Maintenance (Tactical Process)). This process uses data from the Asset Register & GIS, hence the need for accuracy in that data system. It produces new Work Orders and potentially candidate capital projects.

Operational – the Operational process ensures the good working order of the assets and network from day to day, with a 7-14 day planning horizon.

Key to this process is A) the Works Order Management system and process, with links to or embedded within the Asset Register system. B) Asset Monitoring & Proactive Maintenance system(s) that provide alarms and warning of issues arising on the network.

However, there are key artefacts which are shared and should be aligned across the entire business

The component of the ODT, namely

Artificial Intelligence
Statistical Modelling
Machine Learning
Data analysis

These will be explained in our next Blog but as a taster below is a schematic of where the end state of ODT components sit.

...and finally do the Boards of Water Companies understand the benefits of an ODT approach? Put brutally NO. The use of analytics, modelling and optimisatio is understood for regulatory strategic business plan submission but not running the business.