February 01, 2013

Technology Informs Design Decisions at Historic Bottling Plant

Data readings from wireless sensors display in a web interface.

The first external deployment of our Wireless Sensor Network on an existing project took place in an exposed masonry building in Philadelphia, Ortlieb's Bottling House, which is presently being transformed into a new studio for KieranTimberlake. Given the building's historic significance, one of the critical questions this research sought to address was whether to add perimeter insulation or to retain the exposed terra cotta tile that gives the interior its distinctive character.

To probe the thermal properties of the existing wall section, we embedded temperature sensors at multiple locations in the outside and inside faces of the wall: near the floor slab, below the band of factory windows, and in the engaged columns. These readings were accompanied by ambient interior and exterior air temperature readings, which served to establish boundary conditions for our data analysis. 
 
Where other sensor technology had proven problematic for installation in masonry construction, due to sensors exceeding the dimension of common mortar joints, our specially retrofitted sensors allowed for rapid installation with a ¼-inch masonry drill bit. The deployment interface also proved its value in this first experiment. We were able to name each sensor on the fly using its printed QR code and subsequently confirm that all of our sensors were recording data at the proper time intervals.

Over the 17-day period of this study, in which the building was unoccupied and unheated during winter, we found that the masonry construction introduced a thermal lag of several hours and mediated diurnal temperature swings that often exceeded 10°C. We developed a custom Grasshopper plug-in to determine the thermal diffusivity of the wall section and visualize the predicted thermal behavior alongside the measured behavior. 
 
Using this linear regression analysis, we were able to obtain quite good agreement (±0.26°C) between predicted and measured results and to conclude that the thermal diffusivity of the wall section was approximately 16.2 hours, a figure that is half of what one would expect to find using published values for the heat capacity and thermal conductivity of brick. We concluded that variations in bulk material properties were insufficient to account for this discrepancy. Likely, this is evidence of significant infiltration rates, which are not accounted for in conductive transfer models but nonetheless have a significant impact on heating loads in the winter. 
 
These findings confirmed our decision to insulate the majority of the exposed exterior walls and replace the leaky factory windows with tighter, insulated glazing. The small fraction of exterior wall that lies below the factory windows, which was deemed most significant to the historic character of the space, will remain exposed. However, future sensor deployments following the window replacement will allow us to reexamine this condition and determine the value of considering other, non-permanent methods of insulation.