Data – wind power's wasted opportunity

IN DEPTH | ‘Digitalisation saves’ is becoming a mantra in the wind industry, but is enough being done with the data available? asks Darius Snieckus

Digitalisation is reinventing the global energy system. But with the renewables sector heading into an subsidy-free era, the wind industry will have to take advantage of every data-driven cost-saving measure available to it to remain competitive with other forms of power — including ever-cheaper solar.

Because if it fails to find new efficiencies by integrating the “right digital intelligence” in the coming years by using advanced data science, machine learning and ‘hyperlocal’ weather forecasting, the sector’s future could hang in the balance.

“The price pressure in the market — we are being asked to drive around 25% of the cost out [of O&M] — is such that [digitalisation] is key,” says Mark Albenze, chief executive of Siemens Gamesa’s global service division. “Look back 10-15 years and we were able to gain efficiencies by improving the performance of our turbines to meet LCOE [levelised cost of energy] targets — [that is] no longer [the case].

“Further efficiencies, now that we are in an auction-driven market in Europe and facing post-PTC [production tax credit] challenges in the US and so on have to come from somewhere.”

Despite the undeniable importance of digitalisation, OEMs and developers are currently only making use of a scant percentage of the petabytes of data flowing daily from the world’s fleet of wind turbines.

"It's a missed opportunity for industry players to not take full advantage of the treasure trove flowing through their assets."

“Every single turbine generates 200-plus [data] tags every second,” says Balki Iyer, chief growth officer at Utopus Insights, a US-based energy analytics outfit bought last year by Vestas. “Yet if you consider the entire data spectrum that wind turbines generate today, and the potential to save on operational costs and increase revenue, the vast majority of operators take advantage of less than 10% of that data on a day-to-day basis.

“That’s a missed opportunity for industry players to not take full advantage of this treasure trove flowing through their very expensive assets.”

Global renewables developer RES, which has evolved its business model to now span early-stage project engineering straight through to operations and maintenance (O&M), sees “digitalisation and data-led decision-making” as key to maximising asset production and increasing returns.

“Because we work across the whole value stream — from project development to construction to asset management and operation — digitalisation and the data that we get from mining our projects has a value that can’t be overvalued,” says Rachel Ruffle, the company’s chief executive for Northern Europe.

“Applying machine learning and artificial intelligence [AI] can increase revenues and profit from the assets of our customers, [which] are increasingly institutional investors, pension funds that have budgeted for long-term, stable returns.

“A 1% increase in operational efficiency may seem like a small amount but [across a large wind farm or complex] will make a big difference to LCOE.

“[We can] use AI to figure out how to do everything better — from anticipating a gearbox failure and preventing it from happening, to charging your battery because our analytics tell us that in about three hours’ time there is going to be a spike on the electricity market.”

Digital clones

Digital twinning — high-fidelity computational models of turbines and their componentry that mirrors how the machine would run in real life — can address many of the fundamental issues involved in predicting how a turbine performs in the field.

According to US material science outfit Sentient Science, there can be a 68% difference between design tests and the actual operational life of wind farm — a lack of understanding of real-life performance that means almost two thirds of a project’s operational expenditure (opex) stems from unexpected turbine failures.

“For the wider industry, this is by far the largest contributor to opex costs, accounting for 58%, the bulk of which is reactive, unplanned maintenance,” explains Bruce Hall, chief executive of UK-based advanced digital monitoring company Onyx InSight. “The industry increasingly recognises that the best way to reduce O&M costs, and in turn total opex, is to switch to a predictive O&M regime incorporating the latest advances in digitalisation.”

Chinese turbine maker, Goldwind, for instance, is already seeing the cost benefits of its digital intelligence. “The data we see now... essentially allows us to better predict turbine behaviour, which ultimately feeds into grid-friendliness and an optimised O&M planning cycle… driving our predictive vs reactionary operational approach,” says David Sale, chief executive of Goldwind Americas.

'The Wind Challenge proves digitalisation delivers'

Read more

“Ultimately, this accumulation and processing of data will lead to a direct improvement in overall turbine availability and long-term performance.”

Digital twinning, or cloning, allows far more testing in the early stages of product design, before hardware tests begin, and so helps OEMs select longer-lasting subcomponents, stresses Ed Wagner, chief marketing officer at Sentient Science, adding that the same is true when choosing replacement parts for installed turbines.

“For the operator, digital data can be used to improve O&M by finding problem turbines and subcomponents earlier, giving operators more time and more options to maintain their turbines and improving utilisation factors.”

“Using digital data to improve fleet health ultimately will reduce the cost the insurance, financing terms and LTSAs [long-term service agreements]. Taken together, an operator can expect to recover 13% of an asset’s revenue through the use of digitalisation.”

A note of caution

Hall cautions against “unthinking” reliance on digitalisation, emphasising that the wind industry must not create blind spots for itself by depending on “the perceived power of machine-learning algorithms and vast statistical data sets [at the expense of] established engineering principles, analysing and predicting component failures using data sources that are targeted to the specifics of the turbine technology, such as vibration, inspection and lubrication data”.

“Using engineering-led datasets is typically more reliable and provides an in-depth view of the condition of critical components… but are expensive to gather, driving operators to focus their efforts on particular target areas,” he states.

“Statistical datasets provide a broader, but shallower overview of the condition of the asset — and have been shown to yield reasonable anomaly detection. But they are not a panacea and, if divorced from the physical reality of turbine mechanics and the constant evolution in turbine technologies, may flag anomalies that, on closer inspection, pose no danger from an engineering perspective.”

'Hundreds of millions' saved by digitalising O&M, game shows

Read more

The boil-down: algorithms may be learning things that have little value in solving engineering problems in the field. And the over-eager adoption of statistics-based solutions could undermine the progress made in predictive maintenance by generating a larger volume of “false positives” that drive up opex for wind farm operators.

“Every developer is going to be exposed to more merchant risk [from post-subsidy market pressures] — right now 75% of turbine are covered against this but by 2030 it will drop to 6% — and so everyone is having to get after huge improvements in O&M and wind power will only be a winner [compared to other energy sources] if we do,” says Hall. “But we must keep our attention on how we make ‘smart’ data out of ‘big’ data.”

The importance of digitalisation

There is no doubting the size of the prize on offer as digitalisation becomes a major revenue booster for OEMs, analytics companies and data-savvy asset owners, with a report last year from consultancy Totaro & Associates (T&A) pointing to a market worth more than $90bn by 2027 — driven by the 100GW-plus of assets globally that are now monitored with digital solutions.

The growth of “data-as-a-service” corroborates the trend. In the past few years, according to T&A, the number of companies engaged in digital transformation of wind energy, solar power and/or energy storage, has grown from less than 50 to over 200.

'"Buy on Life" can transform the wind industry'

Read more

Albenze underlines that whatever the potential pitfalls might be, digitalisation is now woven into the wind industry’s DNA: “Digitalisation gets us closer — every day — to guaranteeing an outcome, guaranteeing a revenue. We are quickly moving beyond how much a turbine costs in dollars to [power] output and revenue. And you can only do that with knowledge, which digitalisation affords us.”

For Iyer, the next stage of wind industry digitalisation will come from the industry moving away from thinking purely about turbine data to focusing on “the entire grid ecosystem”.

“We’re going from routine, time-based maintenance to a more intelligent operations set-up, where you have a holistic overview of your assets, components that are not working or are about to fail, what to do about it, and predicting power output by using precise power forecasting capabilities,” he says.

“Short-term forecasting is vital to reduce wind generators’ dispatch uncertainty.”

The importance of digitalisation cannot be overestimated, says Iyer. “Progress cannot happen without adopting digital — it is the path forward to bring more renewables into the global energy mix.”

Read Next

High five: The new generation of onshore wind giants

IN DEPTH | The five biggest Western OEMs have launched 5MW-plus models to slash wind power’s LCOE for the merchant market ahead, writes Darius Snieckus

29 Apr 06:53 GMT