As pressure to decrease photovoltaic solar power plant operating expenses, solar asset owners are looking for the best ways to find savings while avoiding risks to equipment performance and longevity. One area where consistency is critical, even as maintenance strategies and technologies change, is remote monitoring and performance engineering. While some minimum physical maintenance should always be on the schedule – such as action items required to uphold equipment warranties – other physical maintenance items can often be driven by what the data tells us.
Is Data Telling the Truth?
There are a number of different remote monitoring software tools available in the marketplace, but two key common denominators when it comes to the value added by monitoring are data quality and automated alarm thresholds. Since small variances in the accuracy of environmental and operational data can lead to large margins of error on performance metrics, having well-calibrated environmental sensors and granular, reliable monitoring devices are investments with tremendous payoff in ensuring that the data is trustworthy.
Once confidence in the quality of the data coming through is achieved, it’s time to decide when automatic alarms should be triggered. The most cost-effective solar monitoring takes place when operators can be pointed in the right direction by alarms without having to manually comb through every data point looking for issues. A balance must be struck between not getting alarms when true causes for concern arise, and getting so many alarms that the operator is desensitized to real problems, we don’t want operators to suffer from a “cry wolf” syndrome.
A technician performs maintenance on a combiner box. Credit: Strata Solar
Zero-Power Alarms
One good example of this is an irradiance threshold for zero-power alarms. We certainly do not expect solar inverters to produce real power during nighttime hours, so we would not want alarms telling us every time an inverter has zero real power output. However, zero real power should be very alarming in the middle of a sunny day.
Studying the normal behavior of the inverter as DC input voltage rises throughout the morning and falls in the evening helps us decide when the system should flag zero real power as something worth human investigation. The same logic can be applied to alarms for any other solar power plant behavior with the final result being a system that tells us every time a human eye is truly needed to ensure normal operation and avoid lost production or even worse, equipment damage.
When to Dispatch
Now that we have alarms covered, the next steps involve deciding when a physical dispatch to the system location is warranted. Generally, the plant operator, in conjunction with the asset owner, uses the remote investigation process to arrive at this assessment.
Best practice dictates establishing general rules for this decision-making process when first commencing a contract, so that system availability does not end up fully dependent on the plant operator reaching individual points of contact in the heat of the moment.
Is there a kilowatt threshold where personnel should always be dispatched if the problem cannot be solved remotely? A certain dollar amount that can be authorized ahead of time? Whether an intense dollar-by-dollar calculation is taking place or not, there is always a cost-benefit decision that comes into play here. It may be as simple as an asset owner deciding that it will always pay off to troubleshoot a certain number of offline watts or an operator knowing that a truck roll is necessary under a response-time guarantee, but somewhere along the line, the decision has to be made, and what we learn via remote monitoring can play an important role here.
Inside the network operations center (NOC). Credit: Strata Solar
Engineering Expertise Required
The logic covered thus far should be at the core of any monitoring arrangement, but based on the value of a solar power plant and the energy it generates, there should be more analysis than just a binary online or offline status report. This is where service providers should have engineering knowledge readily available – for real-time events such as faults requiring complex diagnosis, as well as longer-term trends like underperformance or opportunities to further enhance good performance.
Higher-level analysis requires a mix of electrical expertise and familiarity with statistics. It’s not all about personnel with advanced graduate degrees, though: another key part of learning the most we can from operating data is simply good communication. If we develop solid relationships with grid operators and other solar project stakeholders, we can ensure that we share critical knowledge and jointly discover the root cause of events and performance anomalies. As an example, if a solar power plant trips offline, we should understand why it went offline and any lessons learned rather than simply working to restore the plant.
As a whole, remote monitoring is a key component of the intelligent O&M of a solar power plant. The decisions that asset owners make about how much to invest in a competent service provider, and the strategic decisions that service providers make, have everything to do with how much uptime should be expected and how well the power plant will be cared for.
Mike Loeser is Director of Operations at Strata Solar Services.
Lead image: 88-MW Solar Farm in Warsaw, NC. Credit: Strata Solar