WaveDrive: a systems approach to optimising a linear generator for Trident Energy
WaveDrive is a collaborative project, led by Trident Energy, funded by Wave Energy Scotland (WES) to optimise Trident Energy's existing linear generator design into a generic power take-off (PTO) suitable for use with a wide range of wave energy converters (WECs).
Authors: Jeremy Carey and Alan Mackay
WaveDrive is a Wave Energy Scotland (WES) funded project to optimise the Trident Energy PowerPod linear generator into a generic power take-off for use in a broad range of wave energy converters. Trident Energy assembled a flexible, multi-disciplinary team by leveraging a number of external design and engineering partners, including 42 Technology who are leading the re-design of the generator. 42 Technology built on the existing Trident Energy design, addressing a number of challenges including design for manufacture and optimisation of the air gap between the magnet stack and coils to maximise performance per unit cost, and bearing redesign to improve reliability and minimise operating cost. The overall design was then fine-tuned with geometric optimisation. The WaveDrive project achieved all technical objectives including reduction of the air gap to 2 mm, performance improvement by >50%, reducing intervention rates to <1 per annum and a design life of 20 years. Success derived from the way Trident Energy outsourced work packages to a multi-functional team of flexible, external experts. The team is now working towards a further bid to WES to build and test the new generator design in an onshore test rig ahead of sea trials in late 2017.
WaveDrive is a collaborative project, led by Trident Energy, funded by Wave Energy Scotland (WES) to optimise Trident Energy's existing linear generator design into a generic power take-off (PTO) suitable for use with a wide range of wave energy converters (WECs). A PTO is a system which converts input wave/tidal mechanical power directly into the output electrical power. One example of a PTO uses a linear generator which achieves this conversion in one step, eliminating the complexity involved in first converting the reciprocating motion to uni-directional rotary motion such as required in typical hydraulic PTO solutions (see Fig. 1 ). In a linear generator based PTO system, the heave of the wave motion directly drives coils through a stationary permanent magnetic field, the varying magnetic field generates a current in the coils.
Fig 1: Generic PTO and linear generator [ 1 ]
The ambition is to create a modular generator which, by adjusting a few key parameters such as armature length and number of generators in parallel, can act as a single point PTO for many different applications. Trident Energy subcontracted significant work packages (WPs) to enable them to rapidly assemble the required expertise – 42 Technology were responsible for the concept design of the improved linear generator and are now working on the detailed design.
Trident Energy is a small and medium-sized enterprise (SME), formed in 2003 to research and develop linear generators for marine energy applications. It is developing its patented ‘PowerPod’ linear generator design based on low cost and readily available ferrite magnets . Their ambition is to create a generic linear generator design as an enabling technology with the goal of accelerating the commercialisation of a wide range of wave energy devices.
The company has worked steadily on the PowerPod linear generator designs, building and testing multiple generations of devices with each resulting in substantial improvements to performance, cost and manufacturability. During this period, the company built significant knowhow and expertise as well as a detailed numerical model of the linear generator which was validated by field testing both on and offshore (Fig 2 ).
Fig 2: Early Trident prototype linear generator being tested at NAREC, 2007 
As an SME Trident Energy has decided to limit its in house research and development capability, preferring to partner with/contract other companies and leading academic groups to bring flexibility and expertise into its product development team and project delivery. This has enabled them to deliver a complex project more flexibly and more rapidly than building an in house team.
The WaveDrive project set out to achieve a number of very specific and challenging objectives :
- Affordability – increase power output per kg (power density).
- Performance – enhance power output.
- Availability – improved maintainability and operability.
- Survivability – design for 20-year life and extreme conditions.
The first two objectives, affordability and performance, are both related to the capital cost per kW of installed capacity. A critical parameter for generator performance is the air gap between the central magnet stack and the outer coils (see Fig. 3 ). Trident Energy's linear generator design uses back-iron around the outside of the coils to improve the magnetic flux linking the coils. However, the remaining air gap acts as a resistor in the magnetic circuit – by reducing this gap the magnetic reluctance of the circuit is reduced, which in turn increases the flux linking the coils, enabling greater performance without increasing the capital cost. A small air gap and improved flux linkage also result in a higher power density package which means that the same linear generator design can be used in a wider range of WECs. However, if the gap is too small then the tolerances in the magnet stack need to be very tight, increasing manufacturing costs, or the device risks collisions and impacts which have an impact on reliability and availability and hence operating cost. The process and analysis undertaken to minimise the air gap is detailed in Section 5.
Fig 3: Illustration of linear generator showing the air gap 
Visiting and maintaining offshore marine generators is one of the biggest contributors to annualised operating cost so it is important to ensure that the intervention period is a year or more. Reliability and lifespan are key contributors to levelised cost of electricity estimates so it was critical that these were both sufficiently high.
There were additional specification requirements to be adhered to which further increased the challenge, for example:
- Temperatures in the range of −20 to +50°C (which affects thermal stress, thermal expansion and materials selection).
- Ability to withstand limited off-axis loads and axial rotation of the armature relative to the magnet stack (which affects load cases).
- Suitable for horizontal or vertical operation (which affects the load cases to be analysed and the design of the bearings).
- Suitable for sub-sea, splash zone and above water deployment (i.e. a fully marinised design
The project was divided into a number of WPs as shown in Fig. 4 .
Fig 4: Overall wave drive project structure 
The WPs were led by different project participants as follows:
|• WP1: Concept Optimisation||[Trident Energy]|
|• WP2: Linear Generator||[42 Technology]|
|• WP3: Electronics, Control & Instrumentation (EC&I)||[University of Warwick]|
|• WP4: Hydrodynamic Modelling||[West Coast Wave Institute]|
|• WP5: Test Rig Design||[Energy Technology Centre]|
|• WP6-8: Reporting, Dissemination and Project Management||[SgurrEnergy]|
42 Technology were responsible for WP2: electrical and mechanical design of the enhanced linear generator. This includes redesigning the magnet stack, armature and bearing interface as well as considering reliability, ease of maintenance and manufacturability. 42 Technology used a variant of its standard product development process (Fig. 5 ).
Fig 5: 42 Technology concept development process 
There were a number of specific challenges that merited technical ‘deep dives’ to resolve, for example the selection of magnetic material. The Trident Energy approach is to optimise for performance per unit cost. Although rare earth magnets do provide markedly improved magnetic field strengths, these more exotic materials are inherently costlier and more difficult to work with and ultimately do not provide enough performance gain to justify the increased cost. Later the magnetic design was modelled optimised and compared to the previous design iteration using magnetic finite element analysis (Fig. 6 ).
Fig 6: Modelled relative performance of candidate magnetic materials [ 4 ]
A second example of a technical deep dive was air gap minimisation to improve the capital cost/performance trade off. Reducing the air gap while maintaining reliability requires an improvement in the manufacturing tolerances associated with the magnet stack. The magnet stack is created from a large number of layers of ferrite assembled and encapsulated into axially magnetised poles separated by steel discs to help orientate the magnetic flux (see Fig. 7 ). Each ferrite/steel slice has dimensional manufacturing tolerances, as well as tolerances associated with how accurate these are assembled relative to each other and how ‘true’ the resultant assembled stack is.
Fig 7: Magnet stack assembly 
42 Technology established how accurately each the components and the stack could be made and conducted analysis to establish how small the air gap could be with current manufacturing technology without creating reliability issues or increasing manufacturing costs. By adopting a modular design for the magnet stack allows for the use of well toleranced steel discs, aligned using an assembly tool, and low toleranced ferrite segments. The resultant encapsulated module has a well toleranced outside diameter and enables running the generator with a smaller air gap.
A third example is linked to the magnet stack tolerances analysis was the design and selection of the bearing solution (Fig. 8 ). The bearing is the only wear point in the design so its performance dominates both availability and survivability and hence has a strong impact on levelised electricity cost. 42 Technology began by identifying a long list of candidate bearing types from conventional roller bearings to hydrodynamic and magnetic bearings. The team also examined a number of wear compensation mechanisms and candidate bearing surfaces and considered a number of ways of packaging these elements together into complete bearing solutions which could then be compared and scored against selection criteria such as cost, ease of marinsation, maintenance and efficiency. Ultimately, polymer wheels using sealed bearings on the wheel shaft and running on stainless steel rails inset on the magnet stack were selected as the preferred solution, offering the best balance between efficiency, maintenance, marinisation, technical risk and cost.
Fig 8: Illustration of magnet stack and bearings
Once design solutions had been found for the individual design challenges, the overall design was packaged and then optimised using Trident Energy's validated numerical model and a genetic algorithm optimisation routine. The genetic algorithm is a stochastic global search method that mimics natural biological evolution, repeatedly operating on a population of generator designs. At each step, or generation, designs are selected from the current generation based on their level of fitness which is evaluated using a scoring function. The selected designs are used as parents to produce the children for the next generation. Over successive generations, the population moves toward an optimal design.
This computer aided optimisation technique built on previous work completed by Trident Energy (using ‘The Genetic Algorithm Toolbox’ released by the University of Sheffield) to rapidly evaluate many different generator designs to find an optimal design based on a scoring function, the ‘fitness’ test, and a set of design constraints. The model analyses all the key design parameters of the linear generator design – wire diameter, coil height, coil width, pole height, pole spacing, magnet stack diameter and so on to find a combination of parameters that results in the overall maximum performance. As with any optimisation model it is paramount to constrain the variables to what is practical and achievable for real world components and manufacturing processes so that it only ‘selects’ designs that can actually be realised. 42 Technology built on Trident Energy's industry knowledge to constrain these so that the overall optimisation also addressed design for manufacture concerns. The model was run several times, each time analysing over 100,000 candidate configurations in around 24 h. The final optimised design predicts an improvement in performance of between 50–70% over Trident Energy's prototype generator. The new generator design is predicted to maintain a similar capital cost but offer reduced bearing friction and iron losses, providing a greater power output and more efficient design in the same package size.
The project successfully achieved all four of the initial target outcomes as shown in Table 1 .
Recommendations and next steps
A number of recommendations were made at the end of the concept design phase including:
- Magnetic pole shape optimisation to avoid magnetic saturation and so limit harmonic distortion and potentially further boost magnetic flux linkage.
- Reduction of periodic attractive ‘cogging’ forces by re-profiling the back-iron whilst also shortening the overall length of the armature.
- Weight reduction of magnet stack through removing some of the less utilised material in the poles or magnets.
Following the concept design phase, the project is now undergoing detailed design activities scheduled to complete in the coming months. One of the other WaveDrive partners have modelled the performance within a generic WEC (initially undertaken to provide design inputs on forces, speeds and accelerations). This modelling indicates a significant improvement in overall energy conversion performance may be possible through the use of advanced control strategies made possible through the real-time control capability of the linear generator .
The overall objective, and hence the next step, remains to build and test the design to provide real-world validation of the modelled performance by first testing onshore and then offshore in a marine environment. Consequently, the team is working on a follow up proposal to Wave Energy Scotland for a project to run during 2017 that will include sea trials of PowerPod II later that year or early in 2018.
The project reached three key conclusions:
- A systems approach to design engineering in which design challenges and their trade-offs are identified and solved individually before being integrated into a completed packaged design is a rapid and efficient way to develop complex technologies.
- Computer aided optimisation using a validated model constrained by real world manufacturing and process limitations and guided by intuitive understanding of the design trade-offs can be used to rapidly achieve significant performance and cost improvements even on already mature designs.
- The commercial model of using external experts and design specialists enabled Trident Energy to build an expert multi-disciplinary and flexible team for a limited duration project. The project could not have been completed to the same quality, time and budget had they opted to build a less specialised, less flexible team in house.
The authors acknowledge the other project participants and Wave Energy Scotland for their contributions to the project:
- Wave Energy Scotland (part of Highlands and Islands Enterprise)
- SgurrEnergy Limited
- Enercro Limited
- Technology from Ideas
- University of Warwick
- West Coast Wave Initiative
- Energy Technology Centre Limited
For further information, please contact the authors, Trident Energy, 42 Technology or Wave Energy Scotland:
- Crozier R. C.: ‘Optimisation and comparison of integrated models of direct-drive linear machines for wave energy conversion’. PhD thesis , The University of Edinburgh, 2014.
- Kelly H. P.: ‘Improvements to tubular electrical generators’, European Patent, EP2015430 (A1) , 2009.
- ‘Trident Energy Research & Development’: http://www.tridentenergy.co.uk/our-technology/research-development, accessed August 2016.
- 42 Technology Ltd: ‘WaveDrive D4.1 Linear Generator Concept Design’ (42 Technology , 2016), pp. 1–75.
- 42 Technology Ltd: ‘WaveDrive D7.1 Linear Generator Detailed Design’ (42 Technology, 2016) , pp. 1–65.
- Trident Energy: ‘The importance of real time control for commercial wave energy’ (Trident Energy, 2014), pp. 1–2.