Predicted Mean Vote

edit

You are correct that most graphs of PMV range from -3 to +3, where (hopefully) most values fall close to 0 to indicate that the occupants are comfortable instead of too cold (negative) or too hot (positive. However, the PMV can be calculated outside of this range, as you see in your Results Viewer plots you shared. From the EnergyPlus Engineering Reference on calculating occupant thermal comfort outputs like PMV:

Note for all Thermal Comfort reporting: Though the published values for thermal comfort “vote” have a discrete scale (e.g. –3 to +3 or –4 to +4), the calculations in EnergyPlus are carried out on a continuous scale and, thus, reporting may be “off the scale” with specific conditions encountered in the space. This is not necessarily an error in EnergyPlus – rather a different approach that does not take the “limits” of the discrete scale values into account.

So, if your shoebox model of an office has a combination of thermal comfort parameters that would make occupants very cold (cold mean air temperature, low clothing level, high airflow, etc.), then you could theoretically achieve PMV of -4. Practically, you want PMV as close to 0 as possible to maximize occupant thermal comfort.

PMV is for moderate thermal environmet, and >+2 and >-2 are out of bound values (as per ISO 7730). You are very very far from comfort. For sure CLO value is not suitable for the temperatures that occur.