Temperature changes measured in June 2004 and February 2005

Sometime ago an analysis was made of daily temperature changes between centerbeams and conebars for June 2004 and February 2005, showing an inprovement of the situation after repairing insulation and restarting fans in the centerbeams.
Missing was still an analysis of the change of temperature differences for the rest of the telescope: the relatively narrow bars of the backing structure and the thicker and more massive conebars, spinebars, and coneheads. This was done for Sector 12 of the backing structure, in which all seven temperature sensors are functioning (sector 12 is at the bottom of the dish). The average temperature of the seven sensors were compared with the average of the three sensors in conebar, conehead, and spinebar in this sector, again for June 2004 and February 2005.
The plots below show the average temperature (for June 2004 5C were subtracted) as well as the difference between backing structure and the other sensors (the red line). The blue line indicates when roof and doors were open. The dotted lines are the times of sunrise and sunset on Mauna Kea (not the moment when the Sun appears above the ridge). The grey lines show the measurements of the individual sensors on some days. The dashed line is the zero level (temperatures are equal), shown for reference.
For comparison also the figures with the temperature changes of centerbeams and coneheads are shown again.

One can see that the narrow backing structure shows much faster (and sometimes larger) temperature variations than the cone and spinebars
The general trend is that at night the backing structure is colder than the more massive bars, while during the day it is warmer. During the night the temperature difference slowly becomes 1-2 degrees less. In June there is a small spike after opening roof and doors, possibly because of warm air coming in. This spike is absent in February. 0.5 to 1 hour before sunset the temperature difference increases, and the nighttime situation is reached at about 20h HST (June) and 18h30 HST (February). Note that the stabilization in the centerbeams/conebars plot is reached several hours later.
The reverse happens at sunrise: the temperature difference increases about 1.5 hour after sunrise (when the Sun is above the ridge) at 6h30 HST (June) or 7h30 HST (February). When observing to the East there is a large maximum (up to 3 degrees more) (however it is mainly so prominent because roof and doors are closed at 9h30 HST and the telescope cools down).
The effect of these temperature differences possibly can be seen in RxH3 maps made after sunrise (such a map is planned for 20050415). However to maximise the number of hours that the telescope surface is good, it seems most important to find ways to cool the centerbeams in the cabin.
The possibility and the effect of opening the doors early could be investigated. When the louvers can be opened again, tests should be made to check the effect of opening them during the day on the temperatures of the different structures.