I made a short LES simulation
There is a hot patch 100 m x 100 m in the middle. There is 0.1 K.m (~110 W/m^2) sensible heat flux elsewhere but 0.3 K.m (~330 W/m^2) in the that square. The wind speed is very low, only 1 m/s (~2 knots), still any significant long-term stationary plume is hardy visible. The BL height is about 1 km (900 m at the start, but the gif starts later). The case is completely dry for simplicity. With 30 fps it is 2 minutes of real time per every second of the animation.
Even after 1 hour averaging, the increased temperature is only visible quite localized close to the ground and the vertical velocity field is just determined by those large convective cells, any rising current from the patch cannot be identified in the 1-hour-averaged vertical velocity field at all.
Those sensible heat fluxes I applied are not low but rather typical. Remember that normally a large part of the total heat flux is taken by the latent heat flux.
Firstly, please note that surface heterogeneity is not necessary for thermal convection. Convection happens because of the surface heat flux and thermodynamic instabilities, not because of heterogeneities. Heterogeneities may introduce secondary circulation and various currents.
According to the simplemost particle method, an air particle will raise roughly to the height where its temperature will equal the temperature of the surrounding atmosphere. Your 5.4 K per 1000 ft corresponds roughly to the dry adiabatic lapse rate, that means that the potential temperature is constant. That happens in the mixed part of the convective boundary layer and after the capping inversion with some step rise it will then increase with some gradient. If you heat your thermal say 1 K above the mixed layer potential temperature, then it will rise until it equalizes with the surroundings. If your pot. t. gradient above the mixed layer is 1 K / 100 m, the inversion is weak, it may rise roughly another 100 m. 1 K is already a lot. It won't get you kilometers above the capping inversion. Only large convective clouds can do that.
Even regular thermals do enter the stably stratified layer above the capping inversion and cause entrainment. This is the entrainment layer.
The real thermodynamic temperature is less convenient because it decreases in the mixed layer (9.8 K per 1 km) and then may either increase or decrease above the capping inversion (end still be stable stratified, the sign is not that important).
Then you can approximately compute how high can a heated particle rise. If it is a really isolated heated source, it will work as a chimney with some plume rise. Due to the horizontal size it may be more similar to a cooling tower (without the moisture).