How AI Enabled Asset Management is Driving Down the Cost of Renewable Energy

(Credit: SparkCognition)

by | Aug 24, 2021

This article is included in these additional categories:

(Credit: SparkCognition)

Renewables Are Dominating The Growth In Electricity Generation

The world is experiencing a revolution in how electricity is generated. As climate poses an existential threat to global ecologies and economies, the future—and increasingly the present—belongs to those companies that can reliably and efficiently generate electricity from low or no carbon sources.

Transitioning to wind, solar, and hydropower, often augmented with battery storage systems, presents both huge opportunities and huge challenges.

Defining the Challenges Facing Energy Producers Today

The challenges facing energy producers today are numerous and varied. Managing and optimizing massive data sets across divergent asset operations has never been easy, but now competition in the energy generation market and threats from cyber criminals have increasedthe pressure to find scalable and profit-driving solutions.

Managing huge amounts of data. Renewable energy assets generate tremendous amounts of data. While this data contains valuable information, efficiently managing the data is a difficult task. Manual and semi-automated methods used in the past are not scalable and are no longer cost effective.

Breaking out of data silos. Not only is the size of the data to be managed rapidly increasing, it’s also increasingly diverse. Operations teams, asset managers, and executives depend on data from multiple sources, including SCADA, event, condition monitoring system (CMS), production, budget, maintenance, curtailment, revenue, weather and production forecasts, ES&G, and market pricing tools (to name just a handful).

Managing these sources of data with traditional methods requires using multiple tools, which reduces efficiency and increases costs. The traditional systems also have these data sets in silos that don’t talk to each other and can’t harvest the information in the data. An example of this is the maintenance data that may be stored in a computerized maintenance management system (CMMS) and typically has information on part failures, spare parts, and replacement. This data is used in some cases for reliability-based maintenance and spare parts forecasting in isolation. The same data combined with SCADA, event data, and pricing data becomes much more powerful to predict and plan the maintenance activities for maximizing operational profit.

Balancing a diverse mix of assets. Renewable energy operators must efficiently manage multiple types of assets (e.g. wind, solar, storage, hydro, etc.) and multiple OEMs and equipment models within an asset class. This diversity in operational assets will continue to grow which makes fleet management more complex.

Adding new technologies like storage also introduces additional challenges that arise when different stakeholders have misaligned objectives and means that owners/operators will not be able to depend purely on OEMs. OEMs will want to reduce warranty risk and meet performance that may not be optimal for maximizing asset profitability.  

A competitive generation landscape. As more players enter the renewable energy generation market, competition is increasing. The most efficient operators will win more than their share of the market, while less efficient operators will lose market share.

More complex power purchase agreements. The days of 10 to 20-year fixed power purchase agreements being the standard for renewable energy projects are coming to an end. Variable pricing agreements such as bank hedge or proxy revenue swaps are growing more common. These variable pricing agreements require increased sophistication to properly evaluate, negotiate, and manage risk.  

Though it’s true that the cost of electricity from renewables ($/kWh) is significantly dropping thanks to the benefits of technology advancement and economies of scale, a coinciding dramatic reduction in the power purchase price (down to as low as $0.03/kWh) and phasing out of policy support has put additional pressure on operators. Data driven operational efficiency will play the lead role in helping energy producers meet the challenges of lower PPAs.

Cybersecurity threats. While the digital connectivity of renewable power assets has many upsides, the proliferation of highly sophisticated digital pirates poses a serious threat that requires the deployment of cybersecurity solutions that work across OT and IT systems.

New skill sets required. Power plant operators have historically been experts in operating and maintaining their generation equipment. However, skills such as data science, data engineering, full stack development, and cybersecurity have not been core competencies. These skill sets are in high demand, difficult to find, and costly to add internally.

Moving beyond the Alert

Over the last 5 years or so, power plant operators have adopted a more proactive approach in management of their assets, even if the level at which data is being used effectively varies from operator to operator.

Rules based models and, in some cases, machine learning (ML) are being used more often to drive maintenance actions. This is undoubtedly a step up from the reactive approach of the 2000s, but every implementation will have its strengths and weaknesses.

The limitations of rule based models. Rule based models have severe limitations. They work well if an operator has the same kind of assets in large numbers, but poorly if the assets are few or dissimilar.

Rule based models struggle to scale well and can’t be adapted easily to new assets. Also, failure of components such as gas turbines or power transformers are fairly infrequent, making it difficult to create rules from historical data.

The power of machine learning models. Sophisticated ML models are better solutions for the above challenges. These models will include a combination of supervised (requiring historical data) and unsupervised models.

Unsupervised models are very powerful and have been used successfully in areas with sparse true positive data such as climate informatics. An example is predicting an avalanche from images of snow-clad mountains.

Advanced machine learning models combined with the underlying physics of the assets are a powerful tool in developing an optimal operational strategy. Business decisions such as investing in software control or equipment upgrades can’t be made if the operators do not have visibility into the extent or the root cause of the problem.

Hence, a digital platform that provides an end-to-end solution from digitalization to prescriptive action is the key. This will allow for seamless decision making and efficiency.

How Energy Producers Can Win the Renewable Energy Revolution

To steer clear of the manifold challenges facing energy producers today, a holistic, scalable, platform approach must incorporate the following key elements.

Harnessing the power of cloud computing combined with advanced AI. The platform should continually monitor operating data, and automatically alert operators of anomalies that need attention. This powerful automation of tedious tasks frees up human operators to focus on higher value activities.

Combining multiple data streams into a single platform of record. The platform should unlock insights from all relevant project data, including SCADA, event, maintenance, production, budget, weather, CMS, and unstructured data sources (i.e. text) through the use of natural language processing (NLP). Data and insights are easily accessible in a single location by all platform users.

Managing diverse asset types from a single platform. Wind, solar, hydro, and storage assets must all be managed with a single, comprehensive platform. By reducing the number of tools required, operational efficiency and security are greatly improved.

Increasing energy production and reducing cost. Underperforming assets are quickly brought to the attention of platform users, along with a diagnosis of the most likely cause of the issue, with recommended corrective actions enabled by prescriptive analytics. Advance notification of failures before they actually occur permits better planning and reduces the impact of failures on project operation.

Forecasting and energy trading. The platform should provide accurate energy price forecasts, allowing operators to implement optimal operating strategies to maximize project revenue.

Securing zero day threat protection for industrial assets. As nation state actors and cybercriminals increasingly target critical infrastructure with next-gen malware (e.g. ransomware), energy operators should employ next-gen cybersecurity tools to protect both OT and IT assets from zero day attacks.

New classes of AI-based endpoint protection solutions can be integrated directly into the operator’s APM to provide the highest possible level of protection against cyber threats.

Finding the right technology partners. Digital transformation never ends, and as adoption of advanced technologies like ML and AI continue to increase, energy operators would be well served to pick the right technology partners to assist them throughout their transformation journey regardless of the current stage they’re in.

These partners can provide a full spectrum of digital services to customers, including assistance in cloud migration, integration of applications, platform customization, and the development and integration of future capabilities that are sure to come, as major advances in data science and AI continue to be made.

Electricity Generation is Rapidly Changing, Leaving Old Paradigms Behind

History has shown that companies willing to rise and meet the challenges of the new paradigm will continue to grow and thrive, while those that remain attached to old ways of doing things are destined for displacement by their competitors.

The rapid growth of wind, solar, and hydro generation, augmented by energy storage, is creating new opportunities for innovative operators to grow their businesses and enter new markets. AI enabled asset management represents the new paradigm that some companies will embrace, and others will resist too long.

Advanced technologies like cloud computing, machine learning, and artificial intelligence are market ready now. It will be exciting to watch which companies take advantage of early adoption strategies to increase the efficiency and production of their renewable energy assets, while reducing their maintenance costs and operational risk.

By Dr. Sandeep Gupta, VP of Renewables at SparkCognition

Additional articles you will be interested in.

Stay Informed

Get E+E Leader Articles delivered via Newsletter right to your inbox!

This field is for validation purposes and should be left unchanged.
Share This