Hydro-frac Source Estimation by Time Reversal Mirrors Weiping Cao and Chaiwoot Boonyasiriwat Feb 7, 2008
Outline ■ Motivation ■ Methodology ■ Numerical Examples ■ Conclusions
Outline ■ Motivation ■ Methodology ■ Numerical Examples ■ Conclusions
Motivation Hydro-frac is important for oil recovery operations Only a local velocity model near the well is needed to locate hydro-fracs by TRM Potential for super-resolution and super-stacking properties from TRM
Outline ■ ■ Motivation ■ Methodology ■ Numerical Examples ■ Conclusions
Methodology TRM imaging Apply TRM to locate hydro-fracs by wavefield extrapolation Detailed implementation
TRM Imaging Time Reversal Mirror Time Primary Multiples Image source location with natural Green’s functions (GF) No velocity model needed
Apply TRM to Locating Hydro-fracs : passive data generated by hydro-fracs Problem: to find Solution: to extrapolate VSP or seismic while drilling (SWD) data. TRM imaging s g
Obtain by Wavefield Extrapolation g gogo x gogo g x Forward extrapolation: : convolution Backward extrapolation: : crosscorrelation Semi-natural GFs obtained with a local velocity model
Summary for the implementation ■ Record VSP or SWD data ■ Extrapolate VSP or SWD data to obtain semi-natural GFs between surface and image points using the local velocity model near the well ■ Crosscorrelate these semi-natural GFs to the passive seismic data generated by hydro-fracs
Outline ■ Motivation ■ Methodology Numerical Examples ■ Conclusions
Numerical Examples Synthetic Tests with SEG/EAGE Salt Model: TRM locating hydro-fracs with correct source excitation times TRM locating hydro-fracs in the presence of strong background noise Sensitivity of TRM image to source excitation times
Synthetic Data Generation Z (km) X (km) SEG/EAGE Salt Model 4 (km/s) 2 (km/s) Synthetic data: RVSP or SWD data, passive seismic gathers
Z (km) 812 X (km) km/s Image with Correct Source Excitation Times TRM imaging with forward extrapolation Actual hydro-frac location: (10 km, 3.4 km)
Z (km) 812 X (km) km/s Image with Correct Source Excitation Times TRM imaging with backward extrapolation Actual hydro-frac location: (10 km, 3.01 km)
Strong Background Noise Synthetic Passive Gather Receiver X (km) Time (s) Noisy Gather: S/N = 1/10, Receiver X (km) Time (s) Actual hydro-frac source location: (10 km, 3.01 km)
Strong Background Noise Z (km) 812 X (km) TRM Image from the Noise-free Gather
Strong Background Noise TRM Image from the Noisy Gather: S / N =1 / Z (km) 812 X (km)
Incorrect Source Excitation Times 20 ms advance Z (km) 8 12 X (km) Exact source excitation time Z (km) 8 12 X (km) 20 ms delay Z (km) 8 12 X (km)
Outline ■ Motivation ■ Methodology ■ Numerical Examples Conclusions
■ TRM accurately locates hydro-fracs from VSP or SWD data using correct source excitation times. Only a local velocity model is needed. ■ TRM images show strong resilience to white noise. ■ TRM images are sensitive to source excitation times. ■ 2-D media assumption.
Acknowledgments We thank the 2007 UTAM sponsors for their support