ͳͳͳ ͳǡͳǡͳǡʹǤ ǡ Ǧ ȋǡ Ȍ ǡ Ǥ ǡ Ǥ ǡ Ǧ Ǥ Ǧ Ǥ Ǧ Ǧ Ǥ ǡ Ǧ Ǥ ǡ Ǥ Ǧ Ǧ Ǥ Ǧ Ǥǡ ȋǤǤ ȋȌ ȌǤ ȋ ͳͻͶͻȌ Ǥǡ ȋͳͻͺͲȌ Ǥ ǡ Ǥ ǡ ȋ Ƭ ͳͻͻȌǤ Ǧ ͳ ǡǡǡǤ ʹǡǡǤ ͳͳʹ ȋ ͳͻͻͲǢ Ǥ ͳͻͻ͵ȌǤ ǡ ǡ Ǥ ǡ Ǧ ǡ ȋȌǤ Ǥ Ǥǡ Ǥ Ǥ ȋͳͻͶͻǢͳͻͺͳǢͳͻͺȌǤ ȋƬ ͳͻͻȌǡ ȋƬ ͳͻͶǢ ͳͻͺͲȌǤ Ǧ ȋǯ Ƭ ͳͻͶǢǤͳͻͻͶǢͳͻͻͷȌ ȋȌ ȋ ͳͻͻʹȌ Ǥǡ ǡ ȋ Ǥ ͳͻͻͶǢ Ǥ ʹͲͲ͵Ǣ Ǥ ʹͲͲͶǢ Ǥ ʹͲͲȌǤ ȋͳͻͷͳȌ Ǧ ǤǤ ȋʹͲͳͲȌ ǡ ǡ ǦǤǡ ǡǤ Ǥ ǡ Ǧ ǡ Ǥ Ǧ Ǧ Ǥ Ǧ Ǥ ȋ Ǥ ʹͲͲͶȌ Ǧ ȋǤǤǡ Ȍ Ǥ ǡ Ǥ ͳ Ǧ Ǥ ȋ Ƭ ʹͲͲͳȌ ͳͳ͵ ȋǤʹͲͲͷȌ Ǧ ȋ ͳͻͻʹȌǤ ʹ Ǧ ȋǤ ͳͻͺȌǤ ǡ Ǥ Ǧ Ǥ ǡ ͳ Ǥ ǤȋʹͲͲͶȌ Ǥ ͳ Ǥ ͵ ʹǤ Ǥ ǡαȋφǡǡȌǡ φǡ ǡ ǡ Ǥ ȋφǡǡȌǡ φǡ Ǥ͵ Ǧ Ǥ Figure 1 Resampling of a rock-physics model from velocity-porosity to lithology-porosity space. Ci are model results for various clay contents. ͳͳͶ Ǧ Ǥ φǡ Ǥ ǡ ȋ Ǧ ǡ ǡ ρȌǡ Ǧ Ǥ ȋǤǤ ρȌ Ǥǡ Ǧ ȋ ͵ȌǤ ǡ ǤǤǡ Ǥ Ǥ ǡ Ǥ ǡ Ǥ ǡ Ǧ ǤǡǦ Ǥ δ Ǥ Figure 2 Bulk modulus constraint cube in the porosity, lithology, fluid space (PLF). The vertical axis is water saturation. The lithology axis varies from pure quartz (zero) to pure clay (one). Figure 3 Three observations (Vp, Vs and density isosurfaces) intersecting in the PLF space. Solutions exist at the two indicated points. ͳͳͷ Ǧ φǡǤ Ǥ δ/2Ǥ ͵Ǧ δαͲǤͲʹǡ ͶǤ Figure 4 Three observations (Vp, Vs and density) intersecting in the PLF space. Uncertainty is implicitly handled by the proximity based implementation of the inverse modelling, providing a cloud of solutions as opposed to points as in Figure 3. Ǧ ǡǤǤ ǡ ǦǤ ǡ Ǥǡ Ǥ ǡ ǡ Ǥ ǡ Ǥ ǡ ǡ ȋ ǡǡ ǤȌ Ǥ Ǧ Ǧ Ǥ ͳǣ Ǧ Ǧ ȋͳͻͻʹȌǤ ͳȀʹͲǡ Ǥ ǡ ȋǦ ͳͳ ȌǤ Ǥ ͷͲ ͷ ǤǦ ȋǤͳͻͻʹȌǤ Ǧ ǡǦ ȋ ǦǦȌ ǦǤ ǡ Ǥ ȋ Ƭ ʹͲͲͳȌ Ǧ ǦȋȌȋƬͳͻ͵Ȍ Ǥ Ǧ ȋ ͳͻͶͻȌǤ Ǧ Ǥ Ǥ ȋ ͳͻͻʹȌǤ Ƭ ȋʹͲͲͳȌ ȋȌ Ǥ Ǥǡ Ǧ ȋǦȌǡ ǡ ȋǦȌǡǤ Ǧ Ǥ ǡ Figure 5 P-wave velocity versus porosity trend of clay-sand composites at constant pressure prepared by Yin (1992). Dashed arrows indicate increasing clay content. The V-shaped trend of increasing clay content, producing an increase of velocity (and a decrease in porosity) until clay content equals sand porosity, has been attributed to a pore filling clay topology. Figure 6 Dispersed clay model (Dvorkin-Gutierrez) calibrated to Yin’s data. Note porosity is modelled well, but velocity is in general over-estimated. The model predicts that clay mineral becomes load-bearing (porosity minimum and velocity maximum) at about 30 % clay content. ͳͳ Ǥ Ǧ ȋȌǣ −1 K MIX 1 − C φ ss C φ ss 4 = + − μ ss , 3 K ss + (4 / 3) μ ss K cc + (4 / 3) μ ss ȋͳȌ −1 1 − C φ ss C φ ss + μ MIX = − Z ss , μ ss + Z ss μ cc + Z ss Z ss = μ ss 9 K ss + 8μ ss 6 K ss + 2μ ss , ȋʹȌ ȋ͵Ȍ ǡ Kcc μ cc Ǧ ǤǡɊ φss Ǥ ǡ ǡ Ǥ Ǧ Ǧ ȋǤʹͲͲͷȌǤ Ǥ ǯ Ǧ Ǥ ȋ Ƭ ͳͻͻȌ Ǥ ǯȋͳͻͷʹȌ Ǥ ǦǤ Ǧ ȋ ȌǤ ǡ Ǥ Ǧ ǡ Ǥ Ǥ ǡ ǡ Ǧǡ ǡ ͵ʹΨǤ ͳͳͺ Ǥ ǡ ǡ ǡ ǡ Ǥ ǡ δ͵ͲΨǤ Ǧ ǡǦǦ ǡ ǡ Ǥ ȋȌǤǡ Ǧ ǡ Ǧ Ǥ ͺǡ ǡ Ǥ ǡ Ǣ Ǥ ΪȀǦͷ Ǧ ǡ ȋͻȌǤ Ǧ Ǥ Ǥ Figure 7 Conventional calibration of two shaley-sands models; structural clay (a) and dispersed clay (b) models. Both models have some success, but the accuracy of each is difficult to estimate. ͳͳͻ Figure 8 Quantitative calibration of dispersed clay (left) and structural clay (right) models. Vertical axis is sample number, empty circles are lab measurements and filled circles model predictions. Only 7 samples with clay content below 0.3 were used. Of the two models, the structural clay model produced more solutions. Figure 9 Quantitative calibration of the dispersed clay model (left) and structural clay model (right), using 5% uncertainty in Vp. Vertical axis is sample number, empty circles are lab measurements and filled circles model predictions. More solutions were found relative to the results (Fig. 8) in which uncertainty was not included. Figure 10 Quantitative calibration of dispersed clay (grey) and structural clay (black) models, using a 5 % of uncertainty in Vp. Vertical axis is percentage of predictions matching the data, versus various tolerance ranges in porosity and lithology predictions (+/− − 0.01, 0.02, 0.03, 0.04, 0.05) in the horizontal axis. ͳʹͲ ͳͲǡ ǡ ȋΪȀ−ͲǤͲͳǡ ͲǤͲʹǡ ͲǤͲ͵ǡ ͲǤͲͶǡ ͲǤͲͷȌ Ǥ ͳͲ ǡ Ǥǡͺͻ Ǧ Ǥ ʹǣ ǦǦ ͺͲǡ ȋǤͳͻͺȌǤ Ǥ Ǥ ǦȋȌ ǦȋͳͻͶͻȌ ǡǦ Ǥ Ǥ Ǧ ȋǤʹͲͲͷȌǤ Ǧ ǡ Ǥ ǡα͵ ǡɊ αͶͶ ǡ αʹͷ Ɋ αͻ ȋǤͳͻͺȌǤ ͶͲǡ ȋ Ȍǡ ͲǤͶ ͲǤʹ ͳǤ ǡ Ƭ ȋͳͻͻͷȌ Ǧ ǡ ȋȌ ȋ Ȍǡ Ǥ Ǧ ȋ Ƭ ͳͻͺͻǢ Ǥ ͳͻͻͶȌ ȗ Ͳ ǣ N ( ) S* = S 0 − vn S 0 C n − I K n , n =1 [ ( ( K n = C0 I + G n Cn − C0 ))] −1 . ȋͶȌ ȋͷȌ Ͷͷǡȗ ǢͲ Ǣ Ǣ ͳʹͳ Ǥ Ǧ ȋͲȌȀ Ǥ ȋȽȌ ȋȽȌǤ ȽαͲǤͳȽαͲǤͲͷ Ǥ Ǥ ǯ ȋͳͻͷͳȌ Ǥ Ǥ ȋȌȋͳͻͻʹȌ ȋȽαͲǤʹͷȌ Ǥ ǣ (1 − y ) d dy [K ( y )] = (K * 2 ) − K * P (*2 ) ( y ) , (1 − y ) d [μ * ( y )] = (μ 2 − μ * )Q (*2 ) ( y ) , dy ȋȌ ȋȌ ȗɊȗ ǡȗȋͲȌαͳɊȗȋͲȌαɊͳ ȋ Ȍǡ ʹǡ Ɋʹ Ǥ ȗ ȗ Ǥ ǯȋ ͳͻͷͳȌǤ Ǧ Ǧ Ǧ ȋ ͳͳȌǡ Ǥ ͳʹǤ ǡ ǦǤ ǡ ȋȌǤ ǡǦ ͳ͵Ǥ ȋȌǡ Ǥ ȋ ȌǤǡ ǡ ǡ Ǥ ͳʹʹ Ǥ ͳʹ ͳ͵ ǡǤ ǡ Ǥ ͺ ͳͲ Ǥ ȋȌǡαͺȋͳͶȌǤ ȋͳͷȌ ȋȽ ǡ Ƚ Ȍ ǣ ͳ α ȋͲǤͳǡ ͲǤͲ͵ͷȌǢ ʹ α ȋͲǤͳǡ ͲǤͲȌǢ ͵ α ȋͲǤͳʹǡ ͲǤͲ͵ͷȌǢͶαȋͲǤͳʹǡͲǤͲͷȌǢͷαȋͲǤͳʹǡͲǤͲȌǣͳͷǤ ǡ Ǥ Ƚ α ͲǤͳʹ Ƚ α ͲǤͲ͵ͷ ȋ ͵ȌǤ ǡ ȋͶȋȽ α ͲǤͲͷȌͷȋȽ αͲǤͲȌͳͷȌǤ Figure 11 Conventional calibration of the modified HS upper bound (MHS) model (top left); Xu and White (XW) model (top right) and Differential effective model (DEM) (bottom right), for P and S velocities on Han’s dataset. Note that the three models seem to provide satisfactory calibrations, particularly for the P-wave data. ͳʹ͵ Figure 12 Results of the quantitative calibration of the modified Hashin-Shtrikman (MHS), Xu and White (XW) and differential effective modelling (DEM) models on Han’s data set. Vertical axis is percentage of predictions matching the data, versus various tolerance ranges in porosity and lithology predictions (+/- 0.01, 0.02, 0.03, 0.04, 0.05) in the horizontal axis. Note the poor performance of the DEM model in terms of lithology. Figure 13. Quantitative calibration of the modified HS upper bound (MHS) on the left; Xu and White model (XW) on the right panel. Vertical axis is sample number, empty circles are lab measurements and filled circles model predictions. Note that both models provide a large number of accurate predictions. ͳʹͶ Figure 14 Improved calibration of the modified Hashin-Shtrikman model (MHS), on Han’s data set for various coordination numbers (C). Vertical axis is percentage of predictions matching the data, versus various tolerance ranges in porosity and lithology (+/- 0.01, 0.02, 0.03, 0.04, 0.05) in the horizontal axis. Results are stable, and the calibration is not significantly dependent on coordination number. Figure 15 Improved calibration of the Xu and White model (XW), on Han’s data set for various combinations of aspect ratios. Models 1 to 5 have aspect ratios (sand-pores, clay-pores) as follow: M1 = (0.1, 0.035); M2 = (0.1, 0.06); M3 = (0.12, 0.035); M4 = (0.12, 0.05); M5 = (0.12, 0.06). Vertical axis is percentage of predictions matching the data, versus various tolerance ranges in porosity and lithology (+/- 0.01, 0.02, 0.03, 0.04, 0.05) in the horizontal axis. Note that porosity predictions were optimized by using different aspect ratios whereas lithology maintains relatively stable. ͳʹͷ Ǧǡ Ǧ Ǥ ǡǡ Ǥ ȋȌ ȋ Ȍ ȋǤǤǡ ǡǡ Ȍǡ Ǥ ǡ Ǥ Ǧ ǦǤǦ ȋͳȀʹͲȌ Ǧ Ǥǡ Ǥǡ Ǧ Ǥ Ǥ ͳͲ Ψ ǡ Ǧ Ǧ Ǥ Ǧ ǡ ͳͲΨ Ǥ ǡ Ǧ Ǥ Ǥ Ǧ Ǥ ȋǡ Ȍ ȋȌǤǡ Ǥ ǡ ǡ ǡ Ǥ ʹȋǯȌ ȋǡ Ȍ ǡ Ǧ Ǥǡ ǡ ǡ ǡ ǡ ǯ Ǥ Ǥ ͳʹ Ǥ ǡ Ǥ ǡ Ǥ ȋεͻͲΨȌǤ ǡǡ Ǥ Ǧ ǡ ǡ ǡǤ ǡ ǡ Ǧ Ǧ Ǥ Ǧ ǡ ȋ Ȍ Ǥ ǡ Ǥ ǡ Ǥ ǡ ǡ Ǥ ǡ ǡ Ǥ Ǧ ǡ ǡ Ǥ Ǥ ǡ Ǥ ͳʹ ȋȌ ȋǡǡ ȌǤ Ǥ ǡǤǡǡǤǡǡ ǤƬǡǤʹͲͳͲǤ Ǧ ǡ ǡǦ Ǧ Ǥ ͷǡ ͵ͳǦͶǤ ǡǤǡǡǤƬǡ ǤʹͲͲͷǤ ǣ ǤǤ ǡǤǤǡǡǤ ǤǡƬǡǤǤͳͻͻ͵Ǥ Ǥ ʹͲǡʹͳͻǦʹʹʹǤ ǡǤ ǤͳͻͺͲǤǦ Ǥ Ǥ ͺǡͳͺͲͻǦͳͺͳͻǤ ǡǤ ǤͳͻͻʹǤǦ Ǥ ͻͳǡͷͷͳǦͷͳǤ ǡǤ ǤͳͻͻͷǤ Ǥǣ ǡ ǡ ǡǡǡͶͻǦ;ȋǤǤǤǤȌǡ Ǥ ǡǤǤͳͻͺͳǤ Ǧ Ǥ Ǧ ͶͺǡͺͲ͵ǦͺͲͺǤ ǡǤǡǡǤƬǡǤǤʹͲͲǤ Ǥ ͳʹǡͶͻǦͷǤ ǡǤƬǡǤͳͻͻǤ Ǧǣ Ǥ ͳǡͳ͵͵Ǧͳ͵ͲǤ ǡǤƬ ǡǤǤʹͲͲͳǤ ǡǣ Ǥ ʹͲǡͳͶǦͳǤ ǡǤͳͻͷͳǤ Ǥ ͳǡ͵ǦͺͷǤ ǡǤǡǡǤƬǡǤͳͻͺǤ Ǥ ͷͳǡʹͲͻ͵ǦʹͳͲǤ ǡǤƬǡǤͳͻ͵Ǥ Ǥ ͳͳǡͳʹǦͳͶͲǤ ǡǤͳͻͷʹǤ Ǥ ͷǡ͵ͶͻǦ͵ͷͷǤ ͳʹͺ ǡǤǤǡ ǡǤǤƬǡǤǤͳͻͻͶǤ Ǧ Ǥ ͷͻǡͳͷͲǦͳͷͺ͵Ǥ ǡǤǤƬǡǤͳͻͺͻǤ ǦǦ Ǥ ͳ͵ͳǡͷͷͳǦͷǤ ǡǤǡǡǤǤƬǡǤǤʹͲͲ͵ǤǦ Ǥ ͳͷͶǡͷ͵͵ǦͷͷͺǤ ǡǤǤǡǡǤǤƬǡǤʹͲͲͶǤ Ǥ ͷʹǡͳ͵͵ǦͳͶͻǤ ǡǤǤǡǡǤƬǡǤʹͲͲͶǤ Ǥ ʹ͵ǡ ͳʹǦͳʹͻǤ ǡ ǤǤƬǡǤǤͳͻͶǤ ǦʹǦǤ Ǥ ͵ͻǡͷͺǦͲǤ ǡǤǡǡǤǡǡǤƬǡǤͳͻͻʹǤ Ǧ Ǥ ͷǡͷͷͶǦͷ͵Ǥ ǡǤǤͳͻͶͻǤ Ǥ Ǧ ͳǡʹͷͻǦʹͺǤ ǯǡǤǤƬǡǤͳͻͶǤ Ǥ ͻǡͷͶͳʹǦͷͶʹǤ ǡǤͳͻͻͲǤ ǦǦ Ǥ ͶͳǡͶͷͲǦͶͷͳʹǤ ǡǤͳͻͺǤ Ǧ Ǥ ͵ͷǡʹͳ͵ǦʹʹǤ ǡǤǤƬǡǤǤͳͻͻͷǤ ǦǤ Ͷ͵ǡ ͻͳǦͳͳͺǤ ǡǤͳͻͻʹǤ ǡǡ ǡ ǤǤǤǡǤ
© Copyright 2026 Paperzz