Automatic well test interpretation based on convolutional neural network for a radial composite reservoir

LI Daolun,LIU Xuliang,ZHA Wenshu,YANG Jinghai,LU Detang

Table 5 Comparison of interpretation parameters between the method presented in this paper and analytical method.

Field case Parameter Interpretation value of the CNN method Interpretation value of analytical method Relative error/% 1 K 1.277×10-3 μm2 1.297×10-3 μm2 1.54 S -2.693 6 -2.627 6 -2.51 C 0.044 6 m3/MPa 0.044 3 m3/MPa -0.68 M 8.766 3 8.562 7 -2.38 F 1.624 6 1.534 6 -5.86 2 K 1.530×10-3 μm2 1.489×10-3 μm2 -2.75 S -3.506 3 -3.542 8 1.03 C 0.615 8 m3/MPa 0.619 7 m3/MPa 0.63 M 10.058 3 10.204 6 1.43 F 5.214 7 5.421 3 3.81 3 K 6.079×10-3 μm2 5.877×10-3 μm2 -3.44 S -1.401 3 -1.421 1 1.39 C 0.095 1 m3/MPa 0.093 3 m3/MPa -1.93 M 2.524 2 2.423 8 -4.14 F 0.375 6 0.373 6 -0.54 4 K 10.234×10-3 μm2 10.335×10-3 μm2 0.98 S 6.766 3 6.848 7 1.20 C 0.009 9 m3/MPa 0.009 3 m3/MPa -6.45 M 9.454 6 9.271 0 -1.98 F 0.602 6 0.541 3 -11.32 5 K 1.773×10-3 μm2 1.783×10-3 μm2 0.56 S -2.580 2 -2.580 6 0.02 C 0.165 6 m3/MPa 0.162 1 m3/MPa -2.16 M 11.060 9 11.593 9 4.60 F 10.198 1 10.744 8 5.09 6 K 17.935×10-3 μm2 17.732×10-3 μm2 -1.14 S 1.843 0 1.851 3 0.45 C 0.015 2 m3/MPa 0.015 3 m3/MPa 0.65 M 10.974 4 10.838 8 -1.25 F 0.163 3 0.175 3 6.85