Commentators typically cite technology innovation — evidenced by the continual growth in turbine size — and industrialisation as key factors driving these dramatic cost reductions.
These two factors occupy some rich common ground. Innovations in manufacturing techniques have been shown to improve quality control in the serial production of turbine components, leading to cheaper, yet more reliable, products, for example.
However, one can also observe areas of tension between innovation and industrialisation. Henry Ford expressed it well at the dawn of the automotive industry when he famously enthused that you can have any colour of car you like ... so long as it's black.
The point is that standardisation is essential if you want a sustainable industry to emerge that leverages industrial economies of scale; but it necessarily introduces some degree of conflict with those other important pursuits of engineers, innovation and disruption.
After all, it is difficult for an OEM to optimise its supply chain if it is continually revising its opinion over what is an optimal turbine design. The art of the successful OEM lies in keeping this balance between industrialisation and innovation.
A good example of where such a balance is required is the turbine design process itself. Consider the standard process whereby machines are designed in accordance with a generic type-class.
A relatively small number of wind (I, II, III, IV) and turbulence (A, B, C) classes are defined, to tractably characterise the vast range of possible site conditions experienced by real turbines operating in real wind farms around the globe.
Of course, these standard classes are idealised simplifications of reality with all its messiness and uncertainty, but the method of simplification is nevertheless acceptable because it promotes minimum safety levels in product design and offers a means for turbine OEMs to pursue cost-efficient industrialisation.
But what are the implications for site assessment? Typically, this is carried out by checking that machines designed for standard classes have enough strength to withstand the actual environmental conditions experienced at a given project over its lifetime.
The standard approach to assess such suitability uses one of two methods; either by demonstrating the wind conditions at the site are equal to or less severe than the corresponding type class values; or by re-running the load calculations for the turbine but replacing all type class conditions with site-specific parameters to show that the loads are no greater than the design envelope.
But how is uncertainty, or variance, in site conditions dealt with? This is a question explicitly asked and addressed in the analogous assessment of project energy yield.
In that parallel world, stakeholders are keenly aware that uncertainty in site conditions can have a large impact on predicting their project's actual lifetime yield, and hence on their bottom line.
And yet, back in the world of turbine site suitability assessment, uncertainties are rarely addressed explicitly, the argument being that the standard design process has already taken care of them by means of conservative assumptions and partial safety factors.
To this, an astute project stakeholder may ask what price they are paying for such conservatism, or what level of uncertainty is assumed implicitly in the standard approach.
Applying today's standard approach to site suitability, it's difficult to arrive at satisfactory answers to such questions, pertinent though they remain.
Using digital twins
Parallel developments in digital twin technologies for wind farms may soon underline the need for standards and methods that better account for uncertainty and variance in the site-suitability process by moving the industry away from a static, one-off suitability assessment toward an online, real-time, dynamic one.
Why does this matter? A probabilistic digital twin of a wind farm will be able to assess the actual loading of its physical counterpart within some quantified level of certainty at any time during its operating life, and even predict it ahead of time.
Armed with this kind of insight, operators will know whether original siting assumptions and uncertainties were overly conservative or not. If they were, the asset may be run harder, delivering greater revenue. Conversely, turbines experiencing more severe loading than expected may be actively protected.
In other words, turbine site suitability to date has dealt in black or white. Real life, as we know, is really grey. The successful wind-farm operators of the future will be those that can resolve all of the various shades and paint their operational strategies accordingly.
Graeme McCann is the department head for turbine engineering support at DNV GL-Energy