Solar

Almost none of the generation we recover on a solar site is recovered at the panel. It is recovered at the communication layer — a Sungrow or Sineng inverter speaking Modbus TCP on one subnet, a plant SCADA expecting IEC 61850 on another, and a string combiner box whose fault contact was never wired anywhere. We build the gateway and register map that closes that gap, then prove it with a point-to-point check against the inverter's own portal.

Once the data is trustworthy, performance ratio becomes a number you can argue with a vendor about. We correlate string-level current against pyranometer irradiance and module temperature, so a 4% shortfall resolves into a specific cause — soiling, a failed string fuse, clipping, or a genuine irradiance day. For rooftop plants on an operating factory, that reporting matters more than the array: it is what decides whether the captive-consumption case holds up at the end of the year.