There are always five questions that must be answered for hydraulic fracture design; fracture length, height, and width, where proppant/ acid is placed, and fracture direction/ azimuth. Four of these five critical parameters come from fracture modeling; thus, fracture modeling is 80% of the answer.
A qualitative history of fracture modeling is seen in Fig. 1. The industry used 2D models, where the user specified fracture height and the model calculated the fracture width & length, for only 6 or 8 years. These were quickly supplanted by Pseudo-3D models with the first of these models commercially available in 1981. These Pseudo-3D models approximately calculated fracture height, then calculated fracture width/ length much the same as the original 2D models did.
Calculating the fracture height was a big step forward. Net pressure inside the fracture is a strong function of fracture height, and fracture height is a function of net pressure. Thus, mathematically, fracture height is a strong function of fracture height! Thus there is no basis for estimating fracture
height as required for 2D models. Pseudo-3D models were an important advance at that time. However, it is difficult to believe that these approximate, pseudo models are still used over 25 years later! Particularly as more robust 3D models, with a rigorous solution to mathematical fracture propagation equations, have been available since the mid-80s.
Whatever the reasons for this technology stagnation – it is time to move on! P3D models may be useful for preliminary designs and scoping studies, but except in very simple geologic environments they give wrong answers and cannot be used for final fracture analysis/ design.
“We don’t need 3D models as we don’t have the data anyway”
This is a commonly heard excuse for not using rigorous fracture models. But, what about cases (not all that rare) where we do have the data? Do we still use models that give wrong answers for our designs?
The following example is a hard rock case of a propped fracture pumped from a horizontal wellbore (drilled parallel to the expected fracture azimuth). Extensive data was collected prior to the treatment including dipole sonic log, cores for lab stress-strain (modulus) testing, in situ stress tests in several zones, and a gel minifrac in the actual completion interval. All of this data was then used in a Pseudo-3D model to design a frac.
This showed the fracture mostly growing down (into the pay) from the perforations (red circle in Fig. 2), with sufficient width over the perfs. Basically, the design showed “All is Fine”. A different P3D model was run for post-analysis, and while details differed, major results were similar, i.e., growth down & “Everything is Fine”.