• Big data and predictive analytics can vastly improve renewable energy investment
• Key areas where the technology can be used include asset performance management, predictive maintenance and due diligence

Big data and predictive analytics are two developing trends that can dramatically improve how renewable energy investments are made. Despite the potential such technology represents, adoption has been slow. This has largely been because institutions may not have the capabilities to utilise them, and there is lack of available data within the sector to help investors generate strategies that would benefit from these trends. 

Big data is a term often used to describe the vast scale of new datasets being assembled by organisations. Formally, there is no consensus regarding how large a dataset must become to be classified as big data, as the notion of scale in relation to data is continuously changing. However, a useful rule of thumb for big data at present would be data which is too large to store or practically manage within an Excel spreadsheet.

Analytics is broadly defined as the systematic processing of data with the intention of generating insight. Predictive analytics, as a specific branch of analytics, attempts to develop forecasts from historical datasets, to support analysts undertaking forward-looking decisions. To construct these forecasts, probability and statistics are employed to develop models expressing the likelihood of an outcome occurring.

The growth of these themes in recent years has been supported  by increasing computer power, extensive data proliferation and improving analytical methodologies. These trends together allow for more information to be stored, processed and evaluated with tailored programs. This can provide the foundation for establishing empirically-based strategies as opposed to more traditional qualitative decision-making that relies heavily on cognitive heuristics. 

With these simultaneous progressions, the benefit of big data and predictive analytics has been seen in a range of contexts, with the broad applicability of these tools making them a compelling point of discussion across multiple industries.

saul butt

Saul Butt

Undertaking a big data initiative presents a multitude of challenging considerations for an organisation. There are, naturally, infrastructure-based considerations such as how to store data – through internal servers or leveraging third-party cloud storage – with the subsequent security and efficiency questions that this entails. More challenging, however, would be the conceptual topics of how to structure data. Poorly-structured data can lead to data duplication and inefficiency which, as data reserves grow, can become highly constraining.

Within renewable energy production significant digital exhaust is produced. This can be defined as data amassed simply from the operation of an asset. It is often challenging to capture this data and manipulate it in such a way that it can be used for analytics and refining performance. Recently, the use of sensors attached to assets has allowed for more granular weather and asset performance data to be assembled to evaluate the performance of each unit within a project. 

Predictive analytics has been shown to generate remarkably accurate forecasts in stable environments. However, if exogenous factors materially amend the underlying nature of the subject matter being measured, the accuracy of the forecast can deviate significantly from reality. 

Based on this understanding, to avoid the impact of fallacious assumptions it is key to continually maintain the dataset being used to train the analytics program and review the model outputs to capture any peculiar results. An example of this could be building a model based on historical data recorded from older technologies not leading to the same performance as updated technologies in the future. 

The availability of big data within the renewable energy sector which can be utilised for predictive analytics programs remains limited. This is largely because of the fragmented and predominantly private-ownership structure of the industry that has resulted in a lack of publicly published performance data. 

the renewable energy sector could benefit from predictive analytics

The renewable energy sector could benefit from predictive analytics
Photo: Stein Erik Gilje

Despite this, there has been significant progression in analytics relating to renewable energy assets in recent years, with large private companies working to consolidate sizeable datasets. Operations and maintenance (O&M) providers are beginning to collate their operating data using various technologies that has allowed them to provide advice on optimising asset performance. While IBM is an example of leveraging artificial intelligence, with its Watson platform. It is using vast amounts of data from a range of sources to predict weather patterns, which could materially improve the ability of renewable energy producers to forecast their revenue and O&M cost streams. 

Focusing on future developments in renewable energy investment and asset management, there are a few key emergent areas where the application of big data and predictive analytics has the potential to significantly improve on existing practices:

Asset performance management: APM entails collating past performance information from a portfolio of assets and uncovering correlations between certain variables and a target metric, such as maintenance spending relative to profit. Based on this, a forecast will be made to calculate how performance can be adjusted to optimise on the target metric.

Predictive maintenance: to manage O&M costs, typically reliability-centred management has been adopted to make a weighted cost-allocation model based on component durability and value. This approach has been enhanced through the use of predictive methodologies, such as artificial intelligence, which considers environmental variables within the evaluation of a component’s durability, allowing for the potential to flag and cost problems prior to their occurrence. 

• Deal sourcing and due diligence: combining web-scraping with quantitative analysis methods such as natural language processing, it is possible for managers to translate substantial reserves of online information into insights on potential acquisition targets. This allows managers to leverage a plethora of new information to complete a leaner and more detailed due diligence process to support in increasingly competitive auction processes.

The availability of data within the renewables space will fundamentally improve the ability to undertake more empirically-established forecasts

and decision-making. This will facilitate a more comprehensive understanding of the sources of volatility in performance and allow for better planning to mitigate these risks. Big data and predictive analytics, here as in many other industries, will be key levers that business leaders will need to consider when looking to refine their investment decision-making, as well as offering a compelling differentiating factor that adds value for their investor base. To prepare accordingly, steps to introduce more quantitative-decision making should start to be explored.

Saul Butt is alternative investment associate at Aquila Capital