There are three IDF files for the same building:

1) Dual
2) SingleCooling (older)
3) True SingleCooling (updated)

I suspect Corbin is using something similar to 3. 

Dual setpoint control is similar to that of the original IDF, where heating is enabled throughout
the year using a primary heating coil and a reheat coil. However, Corbin's paper cites the 
usage of only a chiller with the coasting motion, so it's likely single cooling control.

Using single cooling alone results in excess chilling for most part. The chilling sizing factor
is large in the 2nd IDF file (1.33 -> 3) to show this more evidently. The other factor that
affects the overchilling is the minimum air fraction flow in the fans (0.3). 

The last IDF file uses 1.33 and 0.1 for the chilling sizing factor and the minimum air 
fraction flow, respectively. The MATLAB code is not in charge of the on/off behavior of the
HVAC devices. 

Generally, the simulated days are near the summer design days -- in the beginning of August. 
At this time, there are intense morning gains in the middle section of the building (these
5 zones contain the x10 multiplier). These middle zones are sent to MATLAB for viewing. Could
these intense gains be responsible for promoting pre-chilling behavior? Perhaps this strategy
is only viable with a large building with potentially large external gains.