Data-driven approach for studying MRI units’ energy consumption

Understanding magnetic resonance imaging (MRI) machines' operational patterns and energy usage is critical for optimizing electrical infrastructure, improving energy efficiency, and ensuring reliable performance in healthcare facilities. Using energy consumption data, a data-driven framework was applied to analyze MRI machines' energy profile. Energy consumption data from 12 MRI machines were collected at one-minute intervals from various healthcare facilities across the United States at different periods. Inherently, these data included a few phases, such as hibernation and active scanning. MRI events were detected by a computer code based on the deviation from background loads. Energy and duration distributions of MRI events were reported. To evaluate the electrical demands of each MRI machine under realistic operating conditions, momentary load data, the peak power drawn by a machine or subsystem over a short duration, were extracted from manufacturer-provided cutsheets. A novel parameter called the probability of exceedance (PoE) for passing the momentary load was defined and calculated for the operation of one to four MRI machines. It was shown that the electrical design criteria can be reduced by about 5% for two concurrent MRI machines, and by more than 20% when three or four machines operate simultaneously. In contrast with Montecarlo method, the proposed approach is more precise and accounts for all feasible real-world scenarios. The current study's findings provide insights into machine usage patterns, support the foundation for electrical load demands prediction, and contribute to optimizing facility energy management strategies in healthcare facilities.