On-Site Power O&M: How Predictive Microgrid Maintenance Drives Energy Resilience

Industries want resilient, cheap, and clean energy.

Hence, interest in microgrids is growing because of their ability to sustain electricity services when outages affect the broader grid and integrate cheap and clean renewable energy, while saving on electrical distribution grid upgrades. Operational microgrid capacity in the United States has been expanding at an annual growth rate of 32% and reached 8.6 gigawatts by the end of 2023, the most recent data available.

Microgrids can be challenging systems that require specialized skills to operate and maintain. They rely on advanced control and management systems to coordinate distributed energy resources, balance electrical loads, monitor asset health, and disconnect and reconnect a microgrid to the main grid.

The integration of renewable energy sources can increase the challenge because of the variability of energy production. Battery energy storage helps smooth out the fluctuations in power output, avoid costly distribution grid upgrades, and optimize energy usage.

Many industries buy the equipment, run it until failure, and then rip it out and replace it. This approach loses production uptime, materials (up to 800 kg of steel and copper per switchgear), and can cost up to 10x more than investing in scheduled maintenance. However, to keep these microgrids running smoothly and protect sophisticated assets, proactive maintenance is crucial.

Running a microgrid until failure or even having a static monitoring system and routine time-based maintenance isn’t good enough anymore. Microgrids supply backup power to hospitals, water treatment plants, data centers, and industrial facilities that require 100 percent uptime.

Proactive strategies like condition-based maintenance that can unlock predictive analytics contribute to reliable power and resiliency and enhance safety. By identifying potential equipment failures ahead of time, a proactive approach minimizes the risk of accidents and injuries. In dynamic microgrid environments, monitoring, alarm, and analytic systems have to be equally dynamic.

Condition-based monitoring and predictive analytics → Proactive maintenance

Regular maintenance is like following a recipe. You perform certain steps at specific times, regardless of whether your ingredients are about to spoil. Condition-based maintenance and predictive analysis are like checking your ingredients regularly and adjusting your plans based on their condition to prevent spoilage. Many organizations find that the most effective maintenance strategy involves a combination of both regular maintenance and proactive maintenance, tailored to the specific needs and nature of their assets.

Condition-based monitoring and predictive analytics are the end goals of proactive maintenance. It’s a combination of data-driven services that involves using sensors to measure the status of an asset over time while it is in operation. It shifts maintenance from a traditional time-based schedule to a data-driven approach.

The data collected can be analyzed to establish trends, predict equipment or process failures, and can calculate the remaining life of an asset. Subtle deviations in performance can help diagnose potential problems before they turn into full-fledged failures. 

A proactive maintenance approach has a wide range of applications across various industries. It plays a vital role in ensuring the reliable operation of crucial electrical infrastructure, which requires regular monitoring and inspection, especially at mission-critical facilities like hospitals.

One key example of condition-based monitoring and predictive analytics can be seen at a large hospital in Finland that our ABB team worked with. The hospital operates both dry and oil transformers across its massive site. Inspecting oil transformers requires a complete shutdown to take samples, a massive challenge for a busy hospital. The facility turned to remote, digital monitoring of its transformers. This limits the need to perform oil analysis and also provides the hospital with uninterrupted real-time data on the equipment, which isn’t possible with visual inspections. Data regarding the health of the electrical system is readily accessible in the cloud, enabling early fault detection and improved maintenance.

The cost of an unplanned shutdown can be significant. Another ABB customer, an oil and gas company in Saudi Arabia that converts natural gas into a liquid state, estimates that a production stoppage costs $100,000 an hour. Real-time monitoring provides intelligence that allows teams to prioritize and schedule corrective actions to avoid downtime.

Microgrid maintenance best practices

As with any other energy system, a microgrid must be operated and maintained regularly to ensure optimal performance. Effective maintenance involves addressing both external factors, like environmental conditions, temperature, and humidity, and internal factors, like age, usage cycles, and operational stress. Key components requiring regular monitoring include transformers, inverters, generators, switchgears, HVAC systems, circuit breakers, and relays. The most commonly monitored parameters include voltage and current, temperature, and vibration.

Condition-based monitoring and predictive analytics solutions are especially valuable for microgrids as they grow more complex with the addition of renewable energy sources like solar and wind and battery energy storage. These systems require constant monitoring to optimize energy generation and storage.

With proactive maintenance, operators can use advanced analytics to predict, diagnose, and forecast future issues. In addition to improving reliability, optimizing proactive maintenance can extend the life of a microgrid. Collected data can help maintain a unified view of the system and operational health.

Moving from run-to-fail and rip-and-replace approaches through scheduled and proactive maintenance to outcome-based service models saves production uptime, costs, and valuable raw materials. While each company is at a different maturity level, the move from run-to-fail to outcome-based models is an industrial version of moving from buying CDs to streaming songs online.