Ocean renewables and blade design: An analysis
23 September 2019
Located in the channel between Orkney Mainland and the island of Shapinsay, EMEC’s Shapinsay Sound scale tidal test site offers two non-grid connected testing berths for tidal energy convertors. Image: MarRINET2.eu
As EU countries increase the supply of renewables to meet the Paris climate agreement’s ambitious targets, it is of little surprise that significant investment (both public and private) has been made in ocean renewables, writes Edward Fagan.
Rapidly developing technology
Tidal stream turbines are a relatively new and rapidly developing technology for extracting power from the marine environment and converting it to electrical energy for our use.
These turbines operate in a similar fashion to wind turbines, using a multi-bladed rotor driven by tidal currents instead of the wind, to generate power.
Nevertheless, the tidal energy sector’s development is not without risks, since installing a mechanical device in the ocean for up to 25 years faces a litany of engineering challenges, a shortlist of which includes: rusted components, impact damage from detritus in the flow, extreme tidal current and wave interactions during storm conditions, fatigue and wear to components in turbulent seas and (maybe most significantly) balancing an intermittent if predictable energy source with variable demand.
The blades of tidal stream turbines are one of the key components in the machine, both from an operational and from a cost perspective, and their structural design and optimisation was the focus of my PhD project.
“Installing a mechanical device in the ocean for up to 25 years faces a litany of engineering challenges.”
To reduce the risk of damage to the devices, component testing in a lab is normally performed prior to commercialisation.
Wind turbine blade testing
However, full-scale component testing is an expensive and time-consuming process. In wind turbine blade testing, the blade is secured at its root, then loads are applied at a number of points along its length.
Clamps or pressure pads are used to transfer the large forces into the blade without damaging the surrounding structure.
Due to the higher density of water compared to air, tidal turbine blades are smaller and stiffer than wind turbine blades (for turbines of equivalent power).
Blades for tidal turbines may weigh upwards of several tonnes and experience extremely high loading that is challenging to simulate in a lab setting.
This means much larger forces must be transferred into a smaller structure. This difference, more than anything else, distinguishes between wind and tidal turbine blade design and testing.
As well as taking considerable time and resources, structural testing may produce results that are difficult to interpret, relative to the actual operation of the device at least.
To reduce the costs and provide us with a better picture of how the structure behaves, we turn to computational modelling.
Computational modelling efficient and flexible tool
Computational modelling provides an efficient and flexible tool, not to replace full-scale testing, but to enhance our understanding of the blade’s response during the test.
Working with models that have been validated against physical results, we can also expand the scope of our design to look in detail at how changes to the structure can impact the performance of the blades and the overall turbine.
This also opens up the option of applying powerful optimisation strategies, many based on concepts in evolutionary biology, to improving the design of the blades.
“[Composites] can be applied in countless different combinations tailored to meet the needs of a specific application.”
Before delving into how design optimisation is performed, there is the pressing question of what exactly to optimise.
Tidal turbine blades are made from fibre-reinforced polymer composites, such as glass-fibre or carbon-fibre materials, combined with epoxy resins.
Composites are used for several reasons: they have a high strength/stiffness to weight ratio, they don’t corrode like metals in a saltwater environment and, significant for optimisation purposes, they can be applied in countless different combinations tailored to meet the needs of a specific application.
To construct a blade, first a mold of the curved outer shape is made. The materials are then layered one on top of another into the mold, with the stiff fibres in each layer orientated in a specific direction.
This directionality of the layers allows the overall stiffness of the structure to be tailored to meet the design requirements.
Blade needs to be extremely stiff in direction of current
For example, a blade needs to be extremely stiff in the direction of the current to avoid hitting the tower as it bends, but there’s not as much concern about the deflection at 90 degrees (in the direction the blades are spinning).
This flexibility in design is where computational modelling also shows its strength. By automating the process of generating, analysing and post-processing computational models of blades, we can find the optimum configuration of materials for lighter, stronger and longer-lasting designs.
To achieve these design goals we set out the following principles in our design methodology: automate where possible, incorporate physical constraints, use well-founded design criteria and validate against experimental evidence.
The entire computational methodology was automated using the Python programming language, resulting in the development of an in-house software called BladeComp.
The software generates parametric blade models from a set of inputs, such as the geometry of the blade, the composite material properties and the hydrodynamic loads.
BladeComp then uses the commercial finite element software Abaqus to analyse the blade and process the results of the stresses, strains, deflections, natural frequencies and any other structural responses needed to assess the design.
“By automating the process of generating, analysing and post-processing computational models of blades, we can find the optimum configuration of materials for lighter, stronger and longer-lasting designs.”
Once the modelling process was automated, we needed a method of optimising the blade designs, finally settling on genetic algorithms.
Genetic algorithms are a type of search algorithm that fall under the category of ‘metaheuristics’ in the field of computational search and optimisation.
Broadly defined, metaheuristics are algorithms that do two things, (i) build upon the accumulated search experience of past solutions and (ii) locally explore options around the best solutions.
These algorithms simulate some of the concepts of evolution, such as pair reproduction, genetic mutation and improved population fitness over time.
In the context of tidal turbine blades, this means generating a population of random structural designs, assessing each of them and assigning the best performing designs a better chance of passing their ‘genes’ onto the next generation, that is, better fitness.
The intended result is that, over the course of the optimisation, the average fitness improves until the best possible configuration of variables is found.
The ‘fitness’ is defined as a mathematical formula for the objective of the design. Typical examples might be the stiffness of the blade to avoid the tip hitting the tower or the overall blade mass to limit the manufacturing costs.
Using a multi-objective version of the algorithm we can assess several objectives at once and find a set of designs that meet many assessment criteria at once.
Key component in optimisation strategy
The parametric blade models are a key component in the optimisation strategy, but without including relevant physical constraints or manufacturing limitations, the algorithm was quite likely to return a design that couldn’t be easily manufactured.
Learning from the manufacturing and testing experience of the project’s industrial partner, ÉireComposites, appropriate constraints on the designs were incorporated into the software.
This meant carefully defining the optimisation variables. For example, we had to ensure there was continuity as the different layers transition along the blade, or we limited the possible orientations of the fibres in each layer to a set that is easily manufactured with existing processes.
In addition to these constraints, the computational models were analysed using advanced material failure criteria.
These criteria highlight locations in the blades at risk of failure under extreme static or fatigue loads, which can be targeted at the detailed design stage. The failure criteria help to reduce the overall risk to the device during operation.
“A modular steel test frame and three hydraulically driven actuators can be set up to test large wind and tidal turbine blades and blade components.”
A vital step when using computational modelling is to validate the predictions against laboratory tests, to ensure they are physically representative and to limit the biases of the modeller.
Throughout the development of the BladeComp software, we have compared its predictions against material testing results and against several full-scale structural tests of wind turbine blades from 5m to 13m in length.
Natural frequencies of vibration
In these studies, the mass of the blade, the deflections, the material strains and the natural frequencies of vibration were compared to determine the level of accuracy of the models.
In the Large Structures Test Cell at NUI Galway, a modular steel test frame and three hydraulically driven actuators can be set up to test large wind and tidal turbine blades and blade components under static and fatigue conditions.
A suite of non-contact instrumentation devices are also available in the laboratory for structural testing, including a stereo camera Digital Image Correlation (DIC) device for monitoring surface strains and displacements, a 3D laser scanner for obtaining surface profiles for 3D CAD modelling and a scanning laser vibrometer for measuring velocity from 0.005 microns/s to 12 m/s, across 14 ranges.
With the optimisation process completed and the blade manufactured, full-scale structural testing is the final step performed to ensure the actual performance of the finished product meets the design requirements.
Design by its nature is an iterative process and modern engineering design transports much of the iteration into the digital space where the turnaround is comparatively quick and inexpensive.
By augmenting the design methodology with extensive automation, physically based assessments and advanced genetic-based search criteria, we have been able to explore a wide range of design options and reduce blade costs with speed, accuracy and confidence.
These achievements should have a knock-on effect on the levelised cost of electricity and, thereby, the competitiveness of tidal turbines in the global energy market.
The processes described currently form the backbone of the H2020 funded FloTec project with ScotRenewables Tidal Power Ltd, a leading manufacturer of tidal turbines, ÉireComposites Teo, a Galway-based composites manufacturer, and the Marine and Renewable Energy Ireland (MaREI) research centre at NUI Galway.
The project will result in the design, manufacture and testing (through the MaRINET-2 transnational access programme) of the next generation blades for a 2 MW tidal stream turbine device.
Recently completed testing project at NUI Galway
In addition, a recently completed testing project at NUI Galway was focused on accelerated life testing of a blade and section of the rotor structure for OpenHydro Naval Energies, resulting in a collaborative research paper in the ‘International Journal of Fatigue’.
A new research opportunity is also available in our group through a funded PhD position in collaboration with the University of Edinburgh on further developing the design, analysis and structural testing capabilities driving the development of tidal energy.
Author: Edward Fagan is a postdoctoral researcher in the Materials & Structures group of the MaREI Centre based at NUI Galway. Alongside the project to design and test a full-scale tidal turbine blade, he is engaged in research on wind turbine design and cost modelling, testing structures made from composite materials and is looking into building databases of structural models to train machine learning algorithms for design. Twitter: @edwardmfagan.