Sie sind auf Seite 1von 15

TUGAS UJIAN AKHIR SEMESTER MATA KULIAH SEISMIK INVERSI RESERVOIR

Band Limited Impedance Inversion

Oleh : Yudhawastu Wijaya 12309037

PROGRAM STUDI TEKNIK GEOFISIKA FAKULTAS TEKNIK PERTAMBANGAN DAN PERMINYAKAN INSTITUT TEKNOLOGI BANDUNG 2013

The following is a step by step description of the BLIMP method: 1) Compute the linear trend of the impedance estimate and subtract it (this reduces edge effects during subsequent frequency domain operations). 2) Compute the Fourier spectra of (1). 3) Apply a band-limited integration filter to each seismic trace and exponentiate the result. 4) Compute the Fourier spectra of (3). 5) Determine a scalar to match the mean power of (4) and (2) over the seismic signal band. 6) Multiply the spectra of (4) by the scalar from (5). 7) Low-pass filter (2) and add to (6). 8) Inverse Fourier transform (7). 9) Add the low-frequency trend from (1) to (7)

Script Matlab :
clear % Menghapus semua variabel yang terdefinisi sebelumnya %Generate data input %Generate beberapa data Well untuk sintetik acoustic impedance TWT=[1:1:250]/1000; %detik ndata=length(TWT); %jumlah data, didapat dari AI terhadap kedalaman offset=8000*ones(1,250); %penentuan offset dengan nilai awal 8000 Awal=offset(1:50)+1500.*randn(1,50); %rentang offset 50 Tengah=offset(51:150)+1000.*randn(1,100); %rentang offset 100 Akhir=offset(151:250)+2500.*randn(1,100); %rentang offset 100 AI_N=[Awal Tengah Akhir]; trend_sint=TWT*12500+250; %penentuan nilai awal trend AI=AI_N+trend_sint; %data sintetik log impedance plot(TWT,AI) title ('Impedance Log') ylabel('Acoustic Impedance') xlabel('TWT detik') % Koefisien Refleksi for i=1:length(AI)-1; Rc(i)=(AI(i+1)-AI(i))/(AI(i+1)+AI(i)); end RC=[0 Rc]; % Wavelet W=mexihat(-50,50,250); %-50, 50 menunjukkan nilai min dan max % Gambar Wavelet dengan tipe Mexican Hat figure(2) plot(W) title('Wavelet Mexican Hat') % Sintetik Seismik SS=conv(RC,W); SS(length(SS)-125+1:length(SS))=[]; SS(1:125-1)=[]; % Grafik AI, RC, dan Synthetic Seismic figure(3) subplot(1,3,1):plot(AI,TWT) xlabel('AI') ylabel('TWT') set(gca,'YDir','reverse') subplot(1,3,2):plot(RC,TWT) xlabel('RC') ylabel('TWT') set(gca,'YDir','reverse') subplot(1,3,3):plot(SS,TWT) xlabel('Synthetic Seismic')

ylabel('TWT') set(gca,'YDir','reverse') % Trend regresi=polyfit(TWT,AI,1); % mencari koefisien polinomial derajat n=1 trend_AI=polyval(regresi,TWT); % mengembalikan nilai polinomial derajat n=1 figure(4) subplot(3,1,1):plot(TWT,AI) title('Impedance'); ylabel('impedance'); subplot(3,1,2):plot(TWT,trend_AI) title('Linier trend Impedance'); ylabel('impedance'); %Impedance-Trend IM_Trend=AI-trend_AI; subplot(3,1,3):plot(TWT,IM_Trend) title('Linier trend-Impedance'); ylabel('impedance'); xlabel('TWT') figure(5) % frequency domain-Spectral Analysis Using FFT time=TWT; % time sampfrek=200; % determine the frequency sampling fr=sampfrek*(1:ndata)/ndata; % set the frequency %fft data log aud=fft(IM_Trend,ndata); % aud = FFT from the UD data (Impendace Log) udfft=aud.*conj(aud); % udfft = resultan from real and imajiner subplot(4,1,1):plot(fr(1:ndata/2),udfft(1:ndata/2)); title('Magnitude vs Frequency') ylabel('|A|'); % lowpas-filter 10hz % spectrum of low-cut filter(10hz) n_lowfr=10*(ndata/sampfrek)+1; sisa_lowfr=ndata-2*n_lowfr+1; low_freq=[ones(1,n_lowfr) zeros(1,sisa_lowfr) ones(1,n_lowfr)]; subplot(4,1,2):plot(fr(1:ndata/2),low_freq(1:ndata/2)); title('Low-Cut filter 10 Hz') %filtered-lowpass for log filtered_log=low_freq.*udfft; subplot(4,1,3):plot(fr(1:ndata/2),filtered_log(1:ndata/2)); title('filtered 10 Hz') ylabel('|A|'); %lowpass 10hz real-imajiner filtered_realimag=low_freq.*aud; subplot(4,1,4):plot(fr(1:ndata/2),filtered_realimag(1:ndata/2)); title('filtered 10 Hz') xlabel('frequency'); ylabel('A');

figure(6) %fft data trace seismik seismogram aud2=fft(SS,ndata); % aud = FFT from the UD data (sintetik seismogram) udfft2=aud2.*conj(aud2); % udfft = resultan from real and imajiner subplot(4,1,1):plot(fr(1:ndata/2),udfft2(1:ndata/2)) title('Magnitude vs Frequency Seismogram Sintetik') ylabel('|A|'); %band pass data seismik 10-80 hz %spectrum of band-pass filter(10-80hz) n_lowfr=10*(ndata/200); n_highfr=80*(ndata/200)-n_lowfr; sisa_bandfr=ndata-2*n_highfr-2*n_lowfr+2; band_freq=[zeros(1,n_lowfr) ones(1,n_highfr) zeros(1,sisa_bandfr) ones(1,n_highfr) zeros(1,n_lowfr)]; subplot(4,1,2):plot(fr(1:ndata/2),band_freq(1:ndata/2)); title('Bandpass filter 10-80 Hz') %filtered-bandpass seismic trace filtered_seistrace=band_freq.*udfft2; subplot(4,1,3):plot(fr(1:ndata/2),filtered_seistrace(1:ndata/2)); title('Seismic Trace Limited Frequency') ylabel('|A|') %filtered-bandpass seismic trace real-imajiner filtered_seistrace_realimag=band_freq.*aud2; subplot(4,1,4):plot(fr(1:ndata/2),filtered_seistrace_realimag(1:ndata/2)); title('Seismic Trace Limited Frequency') ylabel('A') xlabel('frequency') %estimate scale factor lamda=0.3*10^9; scaled_seis=filtered_seistrace*lamda; scaled_seis_realimag=filtered_seistrace_realimag*sqrt(lamda); spc_log_scaled=scaled_seis+filtered_log; spc_log_scaled_realimag=filtered_realimag+scaled_seis_realimag; figure(7) subplot(2,2,1):plot(fr(1:ndata/2),filtered_log(1:ndata/2)) title('Spectrum of filtered log') ylabel('|A|') xlabel('frequency') subplot(2,2,2):plot(fr(1:ndata/2),filtered_seistrace(1:ndata/2)) title('Spectrum of Integrated RC') ylabel('|A|') xlabel('frequency') subplot(2,2,3:4):plot(fr(1:ndata/2),spc_log_scaled(1:ndata/2)) title('Spectrum of filtered log + Integrated RC') ylabel('|A|') xlabel('frequency') figure(8)

subplot(2,1,1):plot(TWT,AI) title ('Impedance Log') xlabel('TWT (second)') ylabel('Acoustic Impedance') inverse_fft=ifft(spc_log_scaled_realimag,ndata); %inverseFFT result=inverse_fft+trend_AI; %plus trend subplot(2,1,2):plot(TWT,result, '-r') title ('BLIMP') xlabel('TWT (second)') ylabel('Impedance')

Figure yang dihasilkan dari tiap bagian script :

Menghasilkan Figure 1 berupa Impedance Log :

Menghasilkan Figure 2 : berupa wavelet tipe Mexican Hat :

Menghasilkan Figure 3 berupa Acoustic Impedance, Koefisien Refleksi, dan Synthetic Seismic :

Menghasilkan Figure 4 berupa Impedance, Linier trend impedance, dan Linier-trend impedance :

Menghasilkan Figure 5 :

Menghasilkan Figure 6 :

Menghasilkan Figure 7 :

Menghasilkan Figure 8 :

Grafik warna merah menunjukkan hasil akhir dari algoritma BLIMP yang telah dijalankan, dan bias dilihat perbedaan dengan plot impedance log awal yang berwarna biru.

Das könnte Ihnen auch gefallen