Recently, the flow of fluids into a fracture from a point source has been the subject of several different papers. Cleary and Fonseca first suggested that convective transport should play a major role in the placement of proppant when the flow into a fracture was from a point source, Clark and Courington presented data showing that for non-viscosified fluids convection was the dominant mechanism of transport. However, they showed, that for uniform fractures, viscosifying the fluid made a large difference in the transport mechanism. In a later paper, Clark and Zhu presented data for non-uniform fractures and viscosified fluids weighted with either salt or silica flour that showed that the presence of minor non-uniformities serve to negate the effect of convection even more than viscosifying the fluids.
In this work, we have extended the work presented in the previous two papers to high viscosity Newtonian fluids and crosslinked fluids. The experiments have all been done with various concentrations of silica flour to simulate added proppant Both changing the nature of the non-uniformities and crosslinking the polymer solution have a profound affect on the flow into the fracture and the convective process.
In their 1992 paper, Cleary and Fonseca introduced the concept of convection as a mechanism for proppant transport. They based their conclusion on an "order of magnitude" analysis of settling velocity versus convective transport. This analysis yields estimates of convective settling that are 100 to 1000 times the Stokes type settling. It is not the purpose of this paper to critique this analysis, but a survey of the literature on fracture modeling indicates that there has been little or no evidence of convective settling being dominant 3 in any model study reported prior to 1994. Clark and Courington showed that, with simple fluids, the viscosity of the fluid in the slot made a big difference in the flow pattern in the slot. This was confirmed by data presented by Clark and Zhu. In this study, they presented data that showed that a thickened fluid in the slot lessens the tendency for fluids to flow convectively. Although, early in the experiment there is clear evidence of convective type flow, the convective flow is not sustained during the experiment. In the fall of 1994, Barree and Conway presented a paper that indicated that convection could dominate some flows but it is not entirely clear that boundary effects were eliminated.