Case Studies

Creating a ‘Digital Twin’ of the UK’s National Infrastructure

"To ensure the £8m investment delivers its promised value, diverse dataset and models must all work together to provide a trusted and transparent platform that allows repeatable research."

Data allows the UK to make evidence-based infrastructure decisions, but making that data useable is a big job.

 

STFC DAFNI Promo full-1

 

Increasing data capture, and the ability to build predictive models with it, brings considerable benefits to infrastructure planning. Smart meter data helps match energy use to supply; sensor data optimises train schedules and fuel efficiency; climate and environmental data tells us where to build flood defences.

Individually these models provide tangible insights, but most are developed in isolation. Big infrastructure decisions need to bring many diverse considerations together: for example, a new rail line must consider demographic trends, the changing environment, future power supply and myriad other factors. Doing so through data means combining lots of different models and executing them through simulation scenarios to arrive at a robust, scrutinisable decision.

To support the UK’s future infrastructure planning, the government has embarked upon an ambitious project to bring together relevant infrastructure data and models - including, population demographics, energy supply and demand, transportation utilities, and environmental - in one place. The resulting Data & Analytics Facility for National Infrastructure, or DAFNI, implemented and hosted by STFC on behalf of the EPSRC-funded UKCRIC, will become a cloud-based ‘digital twin’ of the UK’s national infrastructure.

Through DAFNI, researchers and planners will be able to access diverse datasets and models – such as those on population growth, rail network demand and flood risk - and run analytics, modelling and simulations. They can study the interplay of factors within complex infrastructure systems and model ambitious ideas in silico in low cost, low-risk ways, helping develop infrastructure plans – such as where to build railways, or expand energy capacity - fit for a rapidly changing world.

This is not the first time a platform has brought together related data sets to improve research capabilities. Rothamsted Research do similar things with agricultural data and NERC with earth observation and environmental data. However, this is one of the most ambitious in its goals to bring together such vast amounts of diverse data from so many different areas.

Pull Away from the Pack with Our Guide to Maximising AI’s Value

How Tessella is helping create the facility

To ensure the £8m investment delivers its promised value, diverse dataset and models must all work together to provide a trusted and transparent platform that allows repeatable research.

Due to Tessella’s unique combination of technical understanding of data and modelling, and experience delivering infrastructure data projects, we were invited to support two significant aspects of the project.

Stage 1: Defining requirements for a UK Infrastructure Digital Twin

Tessella leads a joint academic-industry engagement with potential end users – identifying the key capabilities the platform must deliver to realise the aspirations of the diverse community it supports and ensure the final system their demands now and for many years to come.

We gathered input from many highly respected academics in infrastructure and environmental modelling including Jim Hall, Professor of Climate and Environmental Risk at Oxford University; Mike Batty of UCL - a leading authority on city modelling; and Sir Alan Wilson, Director of Special Projects at the Alan Turing Centre.

This process drew in a diverse range of people, with different ways of working, all of which has to be considered and accommodated. For example, different generations of researchers favoured different technologies and approaches and often had different attitudes to data sharing.

Interoperability was identified as one of the biggest challenges. More established models have evolved over 10-20 years and were created according to the best judgement of their creators at the time. Many used different programming languages and input/output formats. For the system to work, these diverse models need to be able to talk to each other in a consistent language.

Security of data was another major concern. Similar scientific research platforms go big on open data to make access easier. But infrastructure data can be highly sensitive and poses a significant risk to the UK in the wrong hands. Ensuring security is vital to getting people to share data, whilst balancing this with ease of access was vital to encourage use.

The output of the consultancy was a requirements specification and implementation roadmap, and a set of technical, engagement and utilisation KPIs, to assess how well intended value was being generated as the project progressed.

Stage 2: Getting different models to talk the same language

DAFNI users will need to select models they need for their research and have confidence that their specified combination of models will work together to produce a trustable output.

With implementation underway, Tessella, in collaboration with STFC, has been developing the software architecture to link the different models together, and hook them into bigger modelling and simulation networks.

This is done through containerisation. Each model is packaged in a self-contained ‘black box’ container. Within each container, interfaces take in data in the DAFNI common format, and convert it to formats required by the model. The model then executes in a self-contained way and produces its own output, which is reconverted back into DAFNI’s common format before leaving the container.

The work also develops intelligent systems to scale in the cloud responsively. Projects often require many stages using different models. Some are simple and require minimal processing power; others may take weeks and require substantial processing resources. Tessella aim to support STFC as they develop a predictive framework elastically increasing and decreasing the allocation of new resources based on anticipated demand of a submitted modelling or simulation activity.

Get Data Analytics Right – 5 Rules for Delivering Value from Your Data

When can I use DAFNI?

The first publicly available release of DAFNI will launch in June 2019, providing functionality enabling an early adopter community to begin utilising its services. There are still two years of implementation ahead which will improve interfaces, data availability and searchability. Over time, DAFNI will open up beyond researchers to help companies such as Network Rail, TFL, National Grid, utility and telecoms suppliers and civil engineers to plan and implement major infrastructure initiatives from HS2 and HS3 to 5G network coverage, flood alleviation schemes and major urban developments. Early adopters from June will be the first to benefit from the system, as well as having the chance to shape its ongoing development.

Samuel Chorlton, DAFNI Project Lead at STFC said: “DAFNI is an ambitious project. In order to deliver the required level of insight, there was a need to bring together a huge number of existing models and ensure they all integrate through a single interface which works for a wide range of users. Since the underlying models and data were not designed to work together, this was a mammoth and complex task, which required deep understanding of data, modelling, software and IT.”