Data Science in Renewables – Andrea Ronchi guest of HUAWEI Digital Power Club Global Tour

The reason why we created NUO is perfectly in line we the concept of this event.

We are all facing a challenge, which is probably the most ambitious challenge humanity has ever faced.

We need to decarbonize our society and most of all we must do it in less than 30 years.

Energy is of course the main pillar of this transition therefore we all must work together to speed up the diffusion of renewable energy power plants.

So, How can we play our role in this plot ?

First of all by keeping the profitability of renewable asset as high as possible

This task is growing in complexity

  • market parity,
  • dynamic energy markets,
  • participation to the grid-balancing schemes,
  • storage, etc..

are making the business models of asset owners more and more challenging.

Data, Internet of Things, and Artificial Intelligence, are the tools to face that growing complexity. These are powerful instruments that will empower Asset Managers in fulfilling their complex tasks.

I’ll give you now some examples:

  • Improve production forecasts. Even by keeping weather as exogenous variable, predicting plant availability  

more precisely is the key for correct energy dispatching and to participate to flexibility markets

  • Predictive maintenance: this is really a buzz word, but it is crucial on three time horizons:
    • Short term: being able to react to emergencies and possibly to avoid them
    • Mid term: re-scheduling programmed maintenance in order to avoid default
    • Long term: plan the revamping of asset by prioritizing the intervention based on given budget and ROI evaluations
  • Scheduling of hybrid plants: how much and when to store energy? when to use the stored power? How big should be the storage facility? These are heuristic problems in the field of operative research and Artificial Intelligence can help us in finding the optimal plans.
  • Evaluating the performance: is plant producing well or not? If not: why? – this is what we call advance analytics, right?

All these nice use cases are leveraging on DATA.

I think is worth to spend a minute on something that could sound obvious, but that in reality is not.

The photovoltaic industry is rich in data, but the data infrastructure is generally a mess! No standards, a lot of proprietary protocols, and long list of aggregators who are almost always “slave” (passive).

Again data science can help us in managing this data fragmentation but it would be definitely better to focus the efforts on the management of data rather than in heavy data cleaning activities.

We hope that players like Huawei will find the way to propose and support standards adoptable by all industry players. There is a multitude of data loggers, scada DBs etc which are apparently similar but that have different logics. This is slowing down the adoption of data science for final use cases.

In this context, I would like to give a recommendation to all the investors in green field plants: these are the times you should pay particular attention to the data infrastructure of your plants, the harmonization of data from the different component of the site are not less crucial than the hardware per se because is through those data that you will feed the tools that will help you in taking decision for making your investments profitable.

I would also like to add a personal comment on business models based on data. There is the myth of considering data as the oil of the future, but we must be aware that data alone are not solving the problems.

It’s how you treat them that solves use-cases. If everybody design business models that price every little piece of data originated or collected by itself, then, at the end of the story, the adoption of data science models  working of large databases could be not economically worth. I think it’s time for a real collaboration on those aspects: it would lead to a win win scenario for all the actors of the value chain

Talking about collaboration I would like to stress one thing that inspired also NUO’s business model. Out there there are a lot of amazing data science company, most of them are university spin-off or small boutiques. The may lack extensive domain know-how but they are for sure cutting edge in data modeling. In Data Science environment things evolve really really fast. The competitive advantage of who wants to develop everything in-house is fragile. That’s the reason why NUO wants to be the door to the market for all the innovators that decided to dedicate their data science focus on solving problems related to the renewable energy industry.

We developed a platform ready to communicate with 3rd parties developments both in the upstream side (thinking about SCADA provider OPC protocols, IoT) and downstream with AI models.

Also because in data science finding the “one-size fits all” solution is rather utopian, so customization is often the name of the game and sharing experiences with who truly know the domain is the key.