In bulk power system operation, sustained oscillations can pose significant reliability risks. Power grid operators use high-speed synchrophasor measurements to monitor for oscillations across their territory. It is important that these detectors are configured to alert operators of significant oscillations while avoiding nuisance alarms from smaller oscillations. However, the available thresholding methods for the widely-used root mean square (RMS)-energy oscillation detector provide very limited information about the expected detection performance for the thresholds they produce. This paper addresses this challenge by proposing a conversational analytics agent, built with prompt engineering, to assist in the threshold design process by providing increased flexibility and insight into expected detection performance. The agent conducts a dialogue to understand the design goals, generates data analysis code, executes it, and reports results in context. Three examples based on real-world synchrophasor measurements are included to demonstrate its utility. This agent speeds up analysis, reduces manual coding efforts, and ensures that results are reproducible.
With the increase in power demand coupled with aging equipment and personnel shortages, we consistently must ask ourselves how to do more with less? The distributed relay-per-zone architecture wasn’t chosen because it was optimal—it was chosen because it was possible under electromechanical and wiring constraints. Centralized protection is what becomes possible when those constraints disappear through digitization.
This case study investigates mono-crystalline silicon modules from underperforming portions of a utility-scale photovoltaic power plant. Field-collected I-V curves and electroluminescence imaging suggested that increased series resistance was a primary factor driving module degradation. Selected modules were removed from the field for further analysis, including incremental damp heat accelerated testing, which confirmed a progression in series resistance degradation. Two distinct cell degradation behaviors became apparent during the investigation. Cross-sectional scanning electron microscopy (with elemental analysis) and scanning spreading resistance microscopy identified key differences between the two degradation mechanisms, primarily grid finger width and contact resistance. Additionally, the study highlights the reliability implications of retest requirements in International Electrotechnical Commission 61215 for material changes and how they may have mitigated the degradation observed at this site.
Solar plant operators have to schedule outages for periodic maintenance, where plants or subsets of plants are shut down. Outages can last many hours to several days and result in lost energy generation. Scheduling outages one or more days in advance may be required for staffing purposes and to provide adequate notification to grid operators. Because solar generation can vary from one day to the next, it is ideal for outage scheduling to be informed by the weather, with resulting generation that will or could occur on each day being considered for an outage. In this work, we demonstrate how forecasts made with open-source data and tools can improve maintenance outage scheduling relative to not using forecasts, reducing losses relative to perfect scheduling by more than half. Reference code to replicate this work, which is freely available, is also introduced.
