Harnessing AI's Potential: A Practical Guide for Utilities

The promise and challenges of AI

Artificial intelligence (AI) is rapidly transforming industries around the world, but for utilities, effectively harnessing its potential presents unique opportunities and challenges.

As regulated monopolies responsible for delivering essential energy services, utilities operate under a capital-intensive business model with a strong emphasis on reliability. This environment fosters a cautious approach when evaluating emerging technologies.

However, AI is increasingly seen as crucial to addressing many pressing issues utilities face, from optimizing complex operations to improving customer service. 

As the grid expands in scale and complexity, AI will play an important role in enabling more dynamic and resilient energy systems. But with budgets tightly tied to demonstrating clear benefits, utilities must explore AI solutions carefully.

Here, we focus on practical, low-risk strategies utilities can take to start experimenting with AI. 

By emphasizing opportunities to streamline existing processes, utilities can begin to realize value from AI while preparing for future advancements. 

Regulated business models and risk aversion

Utilities operate under business models that differ significantly from other industries. Returns are tied to capital expenditures for new infrastructure approved by public utility commissions. This discourages large speculative investments in unproven technologies. 

Additionally, utilities must ensure long-term support for critical systems. As many control platforms underpinning grid operations are decades old, utilities face challenges modernizing while avoiding disruptions. 

These factors foster an understandably cautious culture when evaluating emerging technologies like AI. However, with careful piloting and a focus on incremental improvements, utilities can start to realize AI's benefits while gaining experience managing associated risks.

Identifying low-risk opportunities for AI 

While there are many opportunities to apply AI, we suggest utilities start with low-risk use cases that improve existing operational processes. Some of the following examples relate to optimizing front-line experiences while others focus on back-end processes. Utilities will want to select use cases that align with planned investments and areas of the business that are open to experimentation. 

Vegetation management 

Vegetation management to prevent wildfires consumes billions annually across US utilities. Utilities can use AI-powered image recognition and computer vision to analyze images of vegetation overgrowth and tower erosion captured by drones. This reduces the need for manual inspections and mitigates potential hazards.

Remote monitoring

By leveraging IoT sensors embedded in field devices and machine learning (ML) algorithms, utilities can remotely monitor field assets in real time. This enables condition-based maintenance strategies that reduce operational expenditures by minimizing unnecessary truck rolls. Remote monitoring also improves worker safety by reducing the exposure of field technicians.

Forecasting

AI and ML can help utilities with forecasting by analyzing large datasets to identify patterns. For example, ML models can analyze historical load data to predict future energy demand. Similarly, AI can process weather patterns, solar irradiance and wind speeds to generate renewable energy forecasts.

Streamlining regulatory processes 

The lengthy documentation required for regulatory filings and rate case submissions is a perpetual pain point for staff. Generative AI (GenAI) can help automate drafting responses, pulling relevant data, and generating sections of these complex documents based on previous filings and requirements. 

Capturing institutional knowledge

With thousands of experienced workers set to retire, GenAI can help capture valuable institutional knowledge. By analyzing notes, recordings and other documents using natural language processing, utilities can gain insights from retiring employees and pass that knowledge to the next generation. 

Building foundations for success

With vast amounts of operational data, utilities need policies around data access, storage and use. Before implementing new technologies, utilities must establish strong data governance and security practices. 

Also crucial for AI success is alignment between IT and lines of business. In our experience, a sustainable AI program hinges on IT and business units collaboratively identifying use cases that offer the highest value with minimal initial investment.

In the case of one WWT client, when IT began engaging business units, they learned that operations teams were running AI and ML models related to forecasting on standard PCs. Additionally, much of this work was being done over residential internet connections. Staff would start simulations on Friday at 5 p.m. and hope that they were complete, without crashing, by Monday. 

Operations considered buying expensive GPU laptops for their computing capabilities; however, this wouldn't solve the problem of huge files being transferred over the internet. 

Instead, we helped IT dedicate GPUs in their VDI environment to handle compute-intensive workloads related to AI and ML planning. 

The result was drastically reduced file transfer and simulation times. This approach quickly caught the attention of other business units that used compute-intensive applications like 3D modeling. 

Small proofs of concept allow IT to start experimenting with inserting the needed infrastructure upgrades to support AI safely while gaining business support for broader AI adoption.

Incremental progress through continued piloting expands ROI over the long term needed for large-scale projects. 

Preparing for future advancements   

While we're seeing AI being used to achieve new levels of operational efficiency, we're still a long way off from a fully AI-powered grid. However, utilities can start preparing for AI being applied to grid operations now. 

The US government is currently running simulations to figure out how AI could be used to support grid operations. It's important that utilities executives stay up to date with this research. 

We suggest stakeholders connect with industry groups who are leading this charge, such as the Electric Power Research Institution. By doing so, utilities make infrastructure investments that align to promising AI use cases related to grid operations. 

Given the long technology adoption cycles faced by utilities, staying abreast of research will also help inform strategic roadmaps outlining multi-year upgrade plans. Plus, building AI into these plans provides visibility for needed approvals down the line. 

Conclusion

Despite some hurdles, AI holds incredible potential to transform the utility industry. The trick is to take an incremental approach. By first focusing on applying AI to existing processes, utilities can realize practical gains while setting the stage for broader adoption across the business.

Constant engagement between utilities, technology partners and regulators will be crucial to refine strategies that harness AI responsibly. However, with clear communication and an emphasis on collaboration, utilities are well-positioned to harness AI to meet future energy demands.