Evaluating the Incident Energy of Arcs in Photovoltaic DC Systems: Comparison Between Calculated and Experimental Data

William Sekulic, Peter McNutt

Research output: NRELPresentation


Solar Photovoltaic (PV) systems have permeated the energy generation world at a very high rate, some of the safety codes and standards are still lagging in accurately assessing the hazards and risks associated with PV array arcing energies. Safety professionals and maintenance workers using NFPA 70E have utilized the Doan, Stokes & Oppenlander or Enrique models, meant to determine arc energies in DC power systems using the maximum power method. These methods may lead to overestimating the energy available in PV systems. Since PV modules/arrays are non-linear, current limited DC devices, some of these calculation methods may overestimate the arcing energy available to small PV arrays. This paper will validate current arc energy models for PV systems by comparing experimental and calculated data. This data will help modify the current NFPA 70E models related to smaller solar arrays. Understanding where the real safety threshold for DC arc flash in PV systems exists will help maintenance and safety professionals better prepare for a variety of work related activities. This paper will analyze real arc data taken for PV systems <1000VDC and <60amps and compare this to the calculated models in 70E. Using these comparisons, it may be possible to reduce the safety hazard severity and thus relax the PPE requirements for installation and maintenance crews.
Original languageAmerican English
Number of pages37
StatePublished - 2019

Publication series

NamePresented at the Industry Applications Society 2019 Electrical Safety Workshop, 4-8 March 2019, Jacksonville, Florida

NREL Publication Number

  • NREL/PR-5K00-73213


  • arc flash
  • DC
  • direct current
  • electrical safety
  • incident energy
  • photovoltaic
  • PV
  • solar


Dive into the research topics of 'Evaluating the Incident Energy of Arcs in Photovoltaic DC Systems: Comparison Between Calculated and Experimental Data'. Together they form a unique fingerprint.

Cite this