The government has recently announced that it will enter into negotiations to fund the £1 billion pound investment needed to build the UK’s first tidal lagoon power scheme at Swansea Bay. The project is going to be expensive, very expensive. If it is completed on budget it will require a subsidy of £150 per megawatt hour (MWh), which is high even compared with the £98 figure agreed for the new nuclear power station at Hinkley.
The key to the pitch delivered by the Tidal Power Lagoon team rests on one element; learning. This is the basic idea that as you produce more of any given good, the cost per unit will decrease. As a general rule, you can expect a 20% saving every time the total unit quantity is doubled. Under this logic, while the first power lagoon at Swansea Bay may be astronomically expensive, future projects will be cheaper and may even achieve grid parity. Mark Shorrock, Tidal Lagoon Power’s chief executive and founder, has said that by the time the first two are operation “a third lagoon will be competitive with the support received by new nuclear”.
This reliance on “learning” is common among proponents of new renewable energy technologies, it is the mechanism by which pie in the sky projects today can be affordable in only a few decades time. It is almost guaranteed that comparisons with solar PV will be made and that the graph below will be wheeled out. The cost reductions in solar PV have increased in the last decade primarily due to economies of scale and improvements in the manufacturing process, almost entirely in China. With PV cells the learning has been successful because there has been limited commercial ambition in building new types of cells and because large scale production of standardised parts has increased very rapidly. This type of learning, incremental innovation, is not dissimilar to how consumer electronics such as televisions or mobile phones can be produced at ever cheaper prices.
The other side of learning, and one which is often conveniently ignored, has been seen with nuclear power. Here there has been very little evidence of learning, and interestingly the cost of nuclear power in France per MWh has consistently increased over time since the 1960s. This is due to two factors; scale and technology. There has been a conventional wisdom that to get nuclear power cheaper it needs to be done on a a larger scale. This has increased complexity and meant that lessons learnt from one project to another are minimal. Second, partly because of the focus on safety, there has been little continuity in the type of technologies used. Again this has made learning difficult. It also doesn’t help that the technology itself is fairly complex. This may be why the promise of nuclear power hasn’t really come true and is becoming increasingly unpopular with governments and the public. The inevitable disaster that is Hinkley point C is likely to be proof of this, almost guaranteed to be delivered late and over budget.
The question concerning Swansea Bay is wether it will be like solar PV or nuclear power. As much as it pains me to write this, it does seem more likely to follow the path of nuclear power. Learning in mega-projects is incredibly difficult, especially when there are so many local considerations to take into account in each of the different coastal locations. Plans relying on learning also require government support for a particular technology to go beyond just one contract, auction or five-year parliamentary term. Even if learning were possible it might be that future projects, where the knowledge could have been utilised, fail to manifest themselves for political reasons. One lesson to take away from looking at learning is perhaps that technologies which are constructed on a smaller scale and are more standardised may be a better bet for long term low carbon development. This fits in with a vision of a more decentralised energy system, where small scale generation and storage can revolutionise the how electricity is generated and consumed.