TL;DR: in my opinion if you have Sandia data for the actual PV panel you plan on using, use that, otherwise use the One-Diode model. Or phrased another way, maybe try to pick a module for which there is readily available Sandia data.
The two models are inherently different, not saying that one is good and the other is bad:
- The equivalent one-diode model implemented in EnergyPlus is sometimes referred to as the TRNSYS model since it was first integrated in TRNSYS in the early 90s. It's a four-parameter model (see caveat below...), for which you supply catalog data, the model then generates I-V curves from catalog data, and assumes that the I-V curve is slope of the IV curve is zero at the short-circuit condition, which the engineering reference guide qualifies as a "reasonable approximation for crystalline modules".
Pros: can be used with any module given appropriate catalog data.
Cons: It's an approximation: the I-V curve is generated from an equation with 4-parameter. is it really accurate other than for crystalline modules?
- The Sandia Model "consists series of empirical relationships with coefficients that are derived from actual testing"
Pros: Based on actual testing, so more accurate for a specific panel
Cons: can only be used for panels for which actual testing has been done.
Source:
Caveat about the one-diode model:
Is the one-diode model as 4P (= TRNSYS model) or a 5P (De Soto, 2004) model? Is the engineering reference guide outdated?
We've discussed this on Unmet Hours in the past when I was wondering whether you can use CEC data for the 4P model: Using the CEC Performance Database data for Solar PV in E+
