http://people.csail.mit.edu/jaffer/convect/History
	Convection Project History

Having grown up in the subtropics, I have long had an interest in radiative cooling. in the 2000s I worked out the mathematics of radiative transfer of thermal infrared transmission through Earth's atmosphere.

With the assumption that tropical roofs are at ambient temperature, only radiation needs to be considered when simulating the roof. SimRoof runs a stateless simulation driven from Typical Meteorological Year (TMY3) files containing hourly quantities for meteorological variables for locations throughout the United States, Puerto Rico, Virgin Islands, and Guam. Locations in Hawaii, Puerto Rico, Virgin Islands, and Guam are tropical.

Cooling of buildings is also important in the southwestern deserts of the US, but the assumption that roofs are at ambient temperature no longer holds. The convective heat transfer must be considered. Most roofs are rough; the scant prior work on convection from rough surfaces was clearly insufficient to calculate reasonable estimates of cooling.

This prompted the fabrication of a wind-tunnel and rough plate to make measurements. Having worked 30 years designing electronics, and having a wood shop in the basement, I was able to construct the wind-tunnel and electronics. My friend John Cox (1957-2022) milled precise roughness into an aluminum plate.

Forced convection measurements from this apparatus found that the Prandtl and Schlichting 1934 formula (based on the pipe-plate analogy) predicted roughly twice what was measured, confirming the unexplained factor of 1/2 adopted by the Pimenta, Moffat, and Kays measurements and the Mills and Hang study. With the pipe-plate analogy invalidated for roughness, its assumption of a continuous boundary layer adjacent to surface roughness was also jettisoned, paving the way for a fresh analysis based on boundary-layer disruption.

Natural convection measurements of the plate outside of the wind-tunnel over the full range of inclinations found that moderate surface roughness did not appreciably affect natural convection.

Being a mathematician, I had to learn how to write for physicists. The combination of a defensive style, testing against all prior works' data, and quantifying all theory-data comparisons using root-mean-squared relative-error calculations may have contributed to peer-reviewed journal acceptance.

Four of these articles are published; two more are anticipated pending measurements from the apparatus being modified.