Digital Twins rely on capturing and processing accurate real-world data by combining it with a virtual model. The three pillars for developing a successful, real-world data-driven, functional digital twin which delivers results are:
- Sensor deployment
- Data processing and
- Accurate model building
As experts in both the real and virtual worlds, the team at Xi combine advanced computing with our long-established knowledge of specialised measurement techniques to ensure optimal value for our clients. This is our step-by-step guide on how to create a real-world digital twin.
Step 1: Define the purpose of your Digital Twin
Who needs to get benefit from the Digital Twin? Who is going to use it? The answer to these questions often is multiple groups within an organisation or different users. Digital Twins can be created and used in several ways. Here, let’s take a look at some of the ways they can be used. Often, Digital Twins are used to create a virtual version of a product to accelerate its development and improve its capabilities, or to replicate a process that needs to be monitored and optimised. A Digital Twin can also be used as a tool to increase the efficiency of a complex system with many variables and to improve multiple parts or just specific components. It can often be a mix of these elements. To get the most out of your Digital Twin, the most crucial part at this stage is to consider:
- What your digital twin needs to do and
- What you would like to achieve.
Taking the time to thoroughly consider these questions set up the design stage. It allows us to create the foundations for an efficient Digital Twin which can meet your requirements. We can then begin work on the design and build it in such a way that allows your system to deliver your intended end goals without costly future rework. Once the requirements have been laid out, we will need to look at what data the Digital Twin will need to deliver value.
Step 2: Data Collection – What should you measure?
A digital twin should respond and act in the same way its real-world counterpart does. Collecting data or using existing data from your process or product is a crucial step in creating an accurate Digital Twin of appropriate complexity. The question of what needs to be measured generates an honest debate about what data is useful and which specific part will be used for building an effective Digital Twin that serves your goals.
Many businesses are building data lakes aimlessly because they know that they should collect data. However, they are unsure how to analyse it or simply access it and use it effectively. Regardless of which way your organisation falls, our experience tells us that there is a relatively high probability that some required data is missing. Capturing this missing data can have a material impact on your product, system, or process. This is probably the most crucial part of creating a successful digital twin.
Working with the team at Xi we help you answer the question ‘what else do you need to measure?’ Xi regularly deals with huge data sets which we analyse and process with advanced computing tools and good old mathematical brain power, helping you get the true value from the data you have invested in. We can provide independent analysis and put in place the programming necessary to continually process these data streams.
Now that you have identified the purpose of your Digital Twin, what data you already have, and what missing data you will need to collect, you will need to consider what your Digital Twin might look like.
Step 3: Creating a visual representation of your Model: Initial Model Design
For those with experience in simulation steps, 2 and 3 often run in parallel as they can both influence each other. Models can tell you where best to place sensors and data lets us constrain a model in order to keep it computationally efficient, while still providing useful and accurate information.
To build a model we need its geometry and material properties and to understand how it is constrained. Receiving a client’s geometry, usually in the form of CAD drawings, is the first step to visual representation. When constructing a model, particular attention will be given to the likely parameters (or variables) that you are most likely to optimise or change over time. The model will be built in software tailored to your needs and users; to provide the outputs agreed at the design stage outline above.
Products and processes, and their various requirements in a digital space can differ greatly. There may be multiple users for a Digital Twin, those concerned with product development and those around operational efficiencies. Our team often builds the model in parallel to the development work of the measurement and data-gathering campaign to compare the real-world data against the initial model. Initial models can be used to help optimise the sensors that are placed on real-world objects. This early parallel work allows the digital twin to be built efficiently and ensures maximum value for our clients.
Step 4: Implementation, Operation, and access to your twin
You have decided what data and models you require to deliver your digital goals. How exactly this data is gathered, stored, processed, represented, and collated with your twin is the challenge of this stage. Your integration, data preparation and quality of data are crucial to this stage. Access to vast computing resources is a double-edged sword, we can have as much as we can handle but can rapidly build over complexity into systems. Structuring how your devices send data and receive data from your data store is paramount. As with any computing system, access rights and levels of detail allowed by users are needed to securely set this stage up. Multiple vendors can house your data – a large river based well-known company makes most of its revenue through its web services.
Xi’s combined experience of complex measurement campaigns which produce large Data sets alongside our specialist simulation services will help you to streamline the design of your Digital Twin. We are experts at maximising the impact and usefulness of a model while adeptly managing which data to use and which can be set aside. This helps to reduce complexity while providing deep insight into your product. We can design models that can indicate when a more detailed approach would be necessary, again optimising computational efficiency while providing critical operational decision points and information.
Finally, you will want to consider how your Digital Twin can grow with you as your product (or process) matures.
Step 5 Augmentation and optimisation based on learnings
Most Digital Twins start small and expand over time, for example monitoring the performance of a single part within an asset and expanding to the whole asset. There is a balance to be struck by ensuring that you will be able to deliver the bigger picture whilst getting some easy wins enroute.
Xi’s discerning approach to computationally efficient yet highly insightful models will help you to navigate how to continue to build your Digital Twin as your product evolves, either through ad hoc support as an R&D partner, as and when you need us, or as a Journey client, who will work right alongside you until we have helped you and your team to internalise the Digital Twin within your organisation.
Digital Twins can be extremely useful and can help you to keep at the leading edge of your sector. By working with Xi, you can get the most out of creating a real-world data informed, useful, efficient, and well-designed Digital Twin.