Renewables are growing in both scale and complexity as an increasing number of industries search for an eco-alternative to meet their energy needs.
The global lockdown has partially driven this trend, as emissions dropped drastically over the last year, with government targets also pushing an increasing number of renewable energy initiatives. The UK, for example, plans to move to net-zero by 2050 and China is aiming for carbon neutrality by 2060.
This is where simulation can help both nations and organisations meet their ambitious targets, as Jonathan Bailey, director of energy and materials, infrastructure and cities, EuroNorth at Dassault Systèmes, explained: ‘To achieve the UK’s aim by 2050, engineers need to work more collaboratively by integrating data, people and processes to optimise decision-making and to make “right-first-time” the norm.'
‘Primarily, we are seeing this done through the digitisation of businesses and them leveraging virtual twin technology, which has a key role to play in simulation,’ Bailey added. ‘In the energy industry, virtual twins are being deployed to create anything from a model of a wind turbine to nuclear reactors or even the orchestration of construction workers tasked with creating an energy storage facility. This concept reduces the need for physical prototypes and increases the accuracy of the data used to assess the viability of an innovation.’
But, from fuel cells to wind turbines and solar panels, the growing diversity of the renewables industry is introducing challenges for simulation and modelling. This is because each renewable option brings with it a unique set of challenges, many of which span multiple modelling scales and scientific disciplines, while requiring a high degree of specialist knowledge.
Understanding electrochemistry and semiconductor physics, for example, is one challenge for engineers, according to Ed Fontes, CTO of Comsol, who added: ‘When you have such an expansive field, there are simply not enough scientists and engineers that are experts in the theory and mathematical modelling of electrochemical and photovoltaic cells. So, many of our customers have to develop and learn in parallel. The same issue may be valid regarding electromagnetic motors and generators.’
To address this lack of specialist knowledge, ‘it is very important that we provide accurate and ready-made descriptions of the involved physics phenomena, reliable material properties, as well as thermodynamic and kinetics models,’ Fontes added. ‘A non-expert should be able to define the input data required for high-fidelity multiphysics models.’
Non-experts should also be able to work across different scales as Fontes explained: ‘This implies that you can stay with one tool all the way from the microscale to the macroscale. You can also link the models at the different scales to be accurate over a wide range of operating conditions.’
Wider changes must also be made to the design process at the organisational level. Dr Uwe Schramm, CTO at Altair, said: ‘Traditionally, the entire design process and, namely, simulation has been applied by means of standalone models in departments that used to work in silos.’
‘To address the challenges, companies need to break those silos, eliminate standalone simulation models, and start thinking from a multiphysics, data-integrated point of view. Model-based development is a key factor as it’s one of the best ways to increase accuracy and reduce development times,’ Schramm added.
We also need to break down these siloed environments across different organisations, as Bailey explained: ‘Engineers will need greater transparency and data management systems in place to provide analysis of energy consumption at existing energy plants, which will inform how renewable energy is implemented into society.’
To achieve this, the sector must consider ‘adopting platform-based technology for collaboration and innovation that provides a centralised view of all the relevant information,’ Bailey added. ‘When design and simulation applications are integrated on the same platform, business can access the most up-to-date data. This data can also be easily and transparently used in compliance and [when] reporting to stakeholders.’
Old and new
Some renewable energies are more mature than others. Schramm explained: ‘Wind energy and solar PV are by far the most established renewable energy technologies worldwide, especially due to the decreasing production costs across the last decade. These areas are where we see a larger adoption of simulation.’
While growth in wind and solar is expected to continue over the next few years, there is still much more work to be done to improve these technologies from both a manufacturing and operational cost perspective. ‘Continuing to decrease production costs is one of the hot topics in wind and solar,’ Schramm added.
Established renewables are also part of a much larger system, which introduces challenges. Fontes explained: ‘Such a system could be a power plant connected to the electrical power grid of a town or province, and maybe even a country. It is important that we can supply reliable lumped models (reduced order models and models based on machine learning trained using multiphysics models).
‘These are usually models that have been verified using fully-fledged multiphysics models but that are very light-weight in terms of solution time and computer memory. They can be incorporated into large system models without demanding a large increase in computing resources. Such models may also be updated in real-time to account for the measured conditions,’ Fontes added.
Innovations are also continuing within the established wind and solar sectors. For example, Altair is currently working with Vortex Bladeless, which is using the company’s simulation, IoT and data analytics tools across its design process.
Vortex Bladeless designs bladeless wind turbines where a cylinder is fixed vertically with an elastic rod to harness the power of the wind. David J Yáñez Villarreal, co-founder at Vortex Bladeless, said that simulation ‘is especially effective in introducing the variability that is difficult to reproduce in a laboratory environment and greatly speeds up the understanding and dynamics of each problem’.
However, challenges in simulation remain for Vortex Bladeless, as Yáñez Villarreal explained: ‘The greatest limitation is the hardware and its computational capacity. If things are done right, simulation tools allow us to analyse these big phenomena almost with unlimited precision. Paradoxically, and in addition to the computational time, it is the high energy consumption that is the biggest challenge and causes the highest costs.’
Vortex Bladeless uses Altair’s tools to conduct its simulations, where the Altair nanoFluidX particle-based fluid dynamics tool addresses this problem. ‘This increases the options that best fit the available tool hardware. Specifically, the appropriate use of GPUs greatly enhances the potential of the package,’ Yáñez Villarreal added.
Many novel renewable energy sources are also benefiting from simulation, including marine energies. Bailey explained: ‘Marine energies are emerging as a credible means for electricity generation from non-carbon sources. With its expansive coastlines France has very high marine energy production potential, and thus has the opportunity to race ahead in this sector.’
Tidal energy company EEL Energy is currently using Dassault Systèmes’ 3DExperience platform to develop an undulating membrane – simulating the motion made by fish swimming to generate electricity from marine or river currents.
This is biomimicry at its best and ‘several prototypes have already been successfully tested in a North Sea flume tank,’ according to Bailey. ‘With a collaborative cloud environment implemented into their design process, EEL energy has been able to evaluate the materials used to develop the undulating membrane, as well as the effectiveness of the product, before trialling it in a real-world environment.’
Batteries and electrolysers are another growing area. Both of these energy storage systems provide a better match between the supply and demand of renewable energy on the grid. ‘They can, for example, be used to avoid curtailment and to deliver electric energy when there is a demand and capacity in the grid,’ Fontes explained.
However, there are challenges ‘related to the wide range of operating conditions and intermittent use these devices have to operate at,’ according to Fontes.
‘If we look at batteries, they are available in many different chemistries, designs and for many different applications. For example, a large-flow battery system may be used for grid energy storage at a central facility, which is basically a small chemical plant. The Tesla Megapack project uses Tesla batteries for large-scale storage.’
In contrast, electrolyzers for hydrogen generation are available more or less off the shelf for one use only - water electrolysis for generation of hydrogen at steady state. But these cutting-edge systems also demand a multiphysics approach. Fontes added: ‘Such applications require new development of the electrochemical cells and the whole system with electrolyzer and fuel cell.’
Fontes said: ‘We are constantly developing tools for modelling of batteries, fuel cells, electrolyzers and solar cells. Also, the functionality for modelling of electric generators is improved for every release. The strategy is to make modelling more easily accessible for non-experts in mathematical modelling.’
In addition, Comsol is expanding its functionality for generating light-weight models. ‘This is important for including accurate models in systems models and grid models,’ Fontes added. ‘The Application Builder is another important tool that we continuously develop. This allows our customers to develop software for their customers. In this respect, we are allowing many of our customers to become software developers, who can deliver multiphysics models to a larger community of scientists and engineers than we would be able to.’
Machine learning and other data-driven technologies are expected to play an increasingly important role in the simulation for both established and new types of renewables.
Xiaobing Hu, head of design and engineering applied solutions at Hexagon’s manufacturing intelligence division, said: ‘We are just scratching the surface of what we can do with AI and machine learning in the CAE industry. By applying these technologies with the rigorous engineering and CAE experience the industry has accumulated, it’s clear we can achieve greater productivity and more sustainable and innovative renewable energy technology development and operations.’
Schramm added: ‘We see the convergence of simulation and AI, where data-driven in addition to model-based simulation will become a critical application that will be used daily by every forward-thinking energy company.’
AI-aided simulation has the potential to help across every stage of the simulation process.
Hu explained: ‘In R&D, engineers can explore the design space thoroughly and optimise systems and even the choice of materials and manufacturing processes. This front-loading of development can reduce design cycles and allow for innovation. During validation and test phases, these AI approaches can help focus effort on meaningful and relevant tests and fill gaps between physical validation from sensors and metrology with reliable virtual data points, enabling robust design.
‘Finally, creating a smart digital reality where virtual and physical data can be used interchangeably to make decisions requires high-quality data – virtual simulations or effective use of sensors and metrology – but AI and machine learning, combined with cloud computing, is key to making it efficient and scalable, stitching virtual and real together and analysing cumulative data patterns to predict outcomes or prescribe action.’
Machine learning can also help when dealing with multiphysics simulations, as Fontes explained: ‘The ability to package high-fidelity models into light-weight models using machine learning and other methods that can be trained with multiphysics models is kind of an emerging technology for off-the-shelf software. This could be a great help in predicting performance and operation of processes in the renewable energy sector.’
Going forward, collaboration will be the main barrier within the renewables sector, as Hu concluded: ‘Much of the data and physics-based simulation required to achieve these goals exists today, as do the data management tools and processes for simulation, materials and IoT. Our approach as part of Hexagon is to build scalable and open platforms that apply cloud and machine learning technologies with simulation and reality capture at each part of the asset lifecycle, with the goal of automating these processes.
‘Simply sending a data scientist into a lab doesn’t achieve this – it requires a deep understanding of the physics types, the manufacturing and measurement processes to connect these data and apply these techniques effectively.’