Using Laboratory Results from New Methods of Measuring Proppant Conductivity to Model Hydraulic Fractures in Reservoir Simulation

Purpose: Hydraulic fracturing processes are conducted to create new fractures in a rock to increase the size, extent, and connectivity of existing fractures. The American Petroleum Institute (API) developed two testing procedures for measuring conductivity of proppants in a laboratory setting, namely; the Short-Term Proppant Conductivity Testing Procedure and Long-Term Proppant Conductivity Testing Method.  However, these laboratory testing methods have produced inconsistent results, with a significant coefficient of variance of ±80% from one test to the other even with the use of the same proppants and procedures. Thus, this work seeks to use an improved laboratory variance from Montana Tech conductivity measurements to model hydraulic fractures in reservoir simulation to evaluate how it performs or compares with field performance. Methodology: Montana Tech researchers have developed new proppant conductivity testing methods to lower this variance. These testing procedures showed more consistent results with an average variance of ±7.6% and ±14.3% in ceramic and sand proppants respectively. These tests were all done at laboratory conditions and therefore this work used field production data obtained from the Willison Bakken Formation and an arbitrary high permeability value as a benchmark against the fracture models built using laboratory results from the new methods of measuring proppant conductivity testing by Montana Technological University. Findings: The conductivity values corresponding with 6,500 psi closure stress obtained for sand and ceramic were 2,133.5 md-ft and 4,870.3 md-ft respectively. The high permeability model recorded an incremental recovery increase of 42% over the unfractured model. Similarly, the laboratory sand and ceramic models had an incremental recovery increase of 12.9% and 33% respectively over the unfractured model. The dimensionless fracture conductivity for the laboratory sand, laboratory ceramic and high permeability models were 1,246, 2,844 and 233,577 respectively. Generally, laboratory conductivity overestimates field performance, however, this work did not show an improvement in modeling fractures using laboratory data as a result of the extremely low porosity and permeability values of the Bakken wells used for the study and the limitedness of the software package used. Simulation of low permeability reservoirs is still an area in development as traditional models often fail to produce results that match the physics. It is possible that as simulation methods for these types of reservoirs improve, the new laboratory data for fracture conductivity will prove beneficial in modeling. Unique contribution to theory, practice and policy (recommendation): A sensitivity analysis should be performed in Petrel that starts with the laboratory fracture conductivity and ends with infinite fracture conductivity. This would help determine the effect of correctly measuring fracture conductivity. Again, a better technique in Petrel such as using a tartan grid is encouraged to better assess the performance of each of the fractures and lastly, more well data with associated measured porosity and permeability data is suggested for future works.