Interpreting the cooling energy consumption at night with shading context of neighboring buildings

I am modeling shoebox models of a few buildings with about 15–20 m of height for a warm and humid climate in India using Honeybee. The cooling is set ON only from 10 p.m. to 6 a.m. and OFF at all other times. There is no heating assigned to the building. However, I am surprised to see the output results. For some buildings, the cooling energy consumption at night is slightly higher when neighboring buildings are considered in the shading context in comparison to when no shading context is considered. Apparently, what I expected was to have less cooling energy consumption when the shading context is considered. However, the majority of the buildings modeled showed this trend, while a few did not. Could you please suggest some possible reasons?