Analysis of Experimental Data for Flow Thorough Fractures using Geostatistics

1. Analysis of Experimental Data for Flow Thorough Fractures using Geostatistics DICMAN ALFRED Dr. ERWIN PUTRA Dr. DAVID SCHECHTER

2. W = 2b Fracture model Cubic law of fractures From the experiments knowing pressure drop and flow rate , the aperture b can be calculated.

3. Actual core surface

4. e 2b Actual fracture surface Modified cubic law Louis (1974) proposed that when e/D < 0.033, then f = 1 e/D is defined relative roughness, where D is the hydraulic diameter = 2*2b e/D >0.033, then f = (1 + 8.8(e/D)^1.5)

5. Previous Research • Tsang (1990) chose a statistical description of a fracture with variable apertures by means of three parameters , performed numerical flow and transport experiments with them with particular emphasis of correlate the fracture geometry parameters. But concluded that the correspondence between observations and the hydrological properties isSTILL AMBIGUOUS. • Detailed measurements of fracture apertures have been obtained by joint surface profiling (Bandis et al. 1981, Brown and Scholz 1985, Gentier 1986), low melting point metal injection (Pyrak-Nole et al. 1987, Gale 1987), and resin casting technique (Hakami 1988, Gentier et al. 1989).BUT THEY ARE EXPENSIVE AND THE DATA MAY NOT BE A TRUE REPRESENTATIVE OF THE FRACTURE. • Work by researchers, such as Neuzil and Tracy (1981), Brown (1987), Tsang and Tsang (1987), Tsang et al. (1988) and Moreno et al. (1988), have shown that the flow through a fracture follows preferred paths or flow channels due to the variation in fracture aperture.

6. Our Approach • Multiphase flow • Upscaling to outcrop studies Expermental data analysis b,Kf Qf, Qm Simulate and match the pressure drop from experimental data Fracture surface generated randomly through geostatistics Include the friction factor to derive fracture permeabilty Study the effect of variance and friction factor on flow Simulation model with varying permeability distribution Experimental data-DP,K,Q,Kavg

7. If is the mean and is the variance Probability Density Function for Log Normal Distribution To standardize this , Similar to the normal distribution

8. How do we apply this to our research ??????

9. Variogram and Kriging Variogram : summarises the relationship between the variance of the difference between measurements and the distance of the corresponding points from each other. Kriging : uses the information from a variogram to find an optimal set of weights that are used in estimating a surface at unsampled locations.

10. Co- variance Lag distance Sill : describes where the variogram develops a flat region, i.e. where the variance no longer increases. Range : the distance between locations beyond which observations appear independent i.e. the variance no longer increases. Nugget variance : when the variogram appears not to go through the origin.

11. Kriging We can use the variogram to estimate values at points other than where measurements were taken. This process is termed kriging.

13. Variance 1800 Variance 2200 What is the effect of changing variance on permeability ???? Variance 2320

14. So lets get started !!!!

15. Experimental Data Pressure Drop 5 cc/min 10 cc/min 15 cc/min Flow through fracture

17. Core40 var 100 Core56 var 200 VARIOGRAM MODELING Core20 var 30

18. CORE 56.4 VAR 200

19. CORE 40 VAR 100

20. CORE 20 VAR 30

21. PERMEABILITY DISTRIBUTION CORE 56.4 VAR 200

22. PERMEABILITY DISTRIBUTION CORE 40 VAR 100

23. PERMEABILITY DISTRIBUTION CORE 20 VAR 30

24. The volume of the core is maintained constant Grid definition 31*15*15

26. RESULTS • Sensitivity studies • Pressure Drop match • Rate comparisons between theoretical and • simulated flow • Permeability comparison • Variance vs Overburden pressure • Comparison between cubic law and modified • cubic law

34. Future Considerations • Extending it to outcrop studies • Modeling 2-phase flow.