|CASE STUDY: A TOWED EM SYSTEM TEST SURVEY|
A newly developed towed EM system has been tested offshore in the North Sea. We show that the measured electric field data are of sufficient quality and signal-to-noise ratio for successful detection and inversion of the high resistivity reservoir area including distinction of some of the shallow gas accumulations above the reservoir. 1D inversion in the frequency domain is performed on individual common mid points (cmps) along a survey line across the reservoir with robust results as well as 2.5D inversion. A 3D resistivity model is also built from seismic data and interpreted horizons. This model is manually fine tuned by comparing resulting 3D forward modeling data with the measured data. Finally, the estimated resistivity structure is investigated with respect to available vertical resolution from the data. This is accomplished by reformulating the inverse problem to a boundary value problem with solutions that approximately give the vertical resistivity structure at each cmp.
The motivation for developing a towed EM system is to significantly increase the acquisition efficiency compared to existing stationary systems. Efficient EM data acquisition increases the range of applications as better spatial coverage can be achieved at lower cost. In the test survey, an electric current dipole source and an EM streamer were simultaneously towed along a 12km long survey line from one vessel in a speed of 4 knots.
1D and 2.5D inversions are performed on the frequency response data along the survey line. In both cases, the reservoir is clearly observed, which agrees well with the seismic information. At shallower depths, there is an increase in resistivity above the reservoir, which probably originate from the thin gas pockets above the reservoir. This is also supported by the 3D modeling. The estimated sea-water resistivity also agrees well with the values from in-situ measurements.
In order to separate the high resistive anomaly from the background and airwave parts of the electric field, a target response (TR) is constructed. The frequency response functions at every cmp are normalized with the frequency response at the first cmp. This means that the target response is a measure of the resistivity change along the line. In this case, the lateral termination of the high resistivity region along the survey line can be observed directly from the TR where the lateral extent of the anomalies corresponds well with the expected target width interpreted from the seismic data.
It can also be concluded that the airwave and sub-bottom contributions to the electric field are fairly constant outside the reservoir region. From the higher frequencies, it is seen that the primary airwave part is constant along the whole line. Any variations would be revealed in the TR otherwise.
An initial 3D resistivity structure of the Peon area has also been built by using a visualization and model building software. A basic velocity model based on the well log data is used to convert the time horizons to depth. A total of five depth horizons is then utilized to generate the 3D structure. This model together with the 1D background from the 1D inversion constitute a starting model in forward modeling. The resulting TRs after successively refinement of the reservoir resistivity value agrees well with measured TR as well as with the modeled data using the model from the 2.5D inversion. The presence of the thin gas areas above the reservoir is revealed in the TR data. The vertical resolution calculated from the boundary value problem strengthen the inversion results.
In conclusion, analysis shows sufficient quality and signal-to-noise ratio of the electric field data for successful detection and inversion of the highly resistive reservoir area including distinction of some of the shallow gas accumulations above the reservoir. Furthermore additional data acquired in 2010 has shown consistent results and indicated that the outline of the Peon gas accumulation is clearly visible from a swathe of densely acquired EM profiles.
|THE PEON GAS RESERVOIR AND 2009 SURVEY PLAN|
The test was conducted in the Norwegian sector of the North Sea over the Peon field, a very shallow but sizeable commercial gas discovery operated by Statoil. It was very important to select a target with a suitably large EM response for this early test.The measurements were made along a 12km-long survey line crossing the reservoir as shown in the left plot in Fig. 1. The nominal towing speed was 4kn. A total of 12 runs, each consisting of 48 shots, were conducted on the line. Each shot sequence had a length of 120 s and consisted either of a Pseudo Random Binary Sequence (PRBS) of order 10 and 10 bits/s or a 0.1Hz square wave (SQR) signal.
Figure 1 The initial 3D resistivity grid model (left) of the Peon reservoir (dark red) with shallow gas accumulations (blue) in map view. The survey line is crossing the reservoir as indicated by the grey plane. The location of the 35/2-1R well is marked in green. To the right, a seismic cross section along the survey line with the 3D resistivity model overlaid. The main reservoir is shown in red (250Ωm) and the shallow gas pockets in blue (100Ωm).
To build the initial 3D resistivity structure of the Peon area, a visualization and model building software was used. A basic velocity model based on the sonic log was used to convert the time horizons to depth. A total of five depth horizons were then used to generate the resistivity grid as shown in Fig 1.
In-line electric field data resulting from the electric dipole source was measured in the configuration shown in Fig. 2. The towing depths of the source and the EM streamer were 10m and 100m, respectively. Four offsets were used (1325, 1850, 2025 and 2545m) and the data were monitored and quality controlled in real time.
Figure 2 The in-line towing configuration of source and receivers along the survey line.
The electric field data were then deconvolved with the transmitted source signal to obtain the frequency response function for each shot point. The transient method using PRBS sequences is described in (Wright et al., 2002, and Wright et al., 2005). The data was sorted into common midpoint gathers (cmps) with separations of 250m.
1D frequency domain Differential Evolution (DE) inversion (Storn and Price, 1996) was performed at every cmp on the frequency response data along the survey line. All offsets and 16 frequencies in the range 0.1 - 4.3Hz were used. The inversion results are shown in Fig. 3.
Figure 3 The 1D inversion results plotted together as 2D resistivity cross sections along the survey line for SQR (left) and PRBS data (middle). To the right, the 2D resistivity inversion of SQR data is overlain in colour on the seismic cross section.
The reservoir (570-590m) is clearly observed at cmps 20-40 for both data sets, which agrees well with the seismic information. At shallower depths, there is an increase in resistivity above the reservoir, which could originate from the rightmost gas pocket, or could be a response from the reservoir that is influencing the shallower layer as well, or possibly both. However, at cmps 18-20 a spot of increased resistivity can be noted at a depth of ~480m, a feature that has been observed for several different setups of the inversion. This resistivity increase may originate in the leftmost gas pocket, which is not directly overlying the reservoir and is therefore easier to detect.
The finite lateral extent of the high resistivity gas field surrounding the survey line causes an underestimation of the resistivity values from 1D inversion (Wright et al., 2009). This is also seen from the 3D modeling discussed below and the well log analysis. However, the background resistivity layering consisting of the sea-water, overburden and underburden are accurately estimated with 1D inversion outside the reservoir. The estimated sea-water resistivity also agrees well with the values from in-situ measurements.
|TARGET RESPONSES AND 3D MODELLING|
The lateral termination of the high resistivity region along the survey lines can be observed directly from the normalized frequency response function, target response (TR). This is exemplified in Fig. 4 for offset 2025m. The lateral extent of both the PRBS and SQR anomalies correspond well with the expected target width interpreted from the seismic data. The TR is defined as
where F is the frequency response function, s the cmp along the line and f the frequency. In this case the first cmp is used for normalization.
It can also be concluded that the sub-surface structure is fairly flat along the survey line outside the high resistive region. The TR would reveal any changes along the line. Hence, using the first recorded cmp to normalize the data is in this case a good way to present the electric field data.
Figure 4 Measured normalized frequency response amplitudes, i.e. target responses (TR) along the survey line for the offset 2025m based on PRBS data (left) and SQR data (right).
To model the structure of the Peon area more accurately the initial 3D-model in Fig. 1 is used. The resulting TR from forward modeling (Mattsson 2006; Andersson et al. 2007), after decreasing the reservoir resistivity to 100Ωm, is presented in Fig. 5.
Figure5 Modeled TR along the survey line for offset 2025m using the 3D resistivity model based on seismic data but with a reservoir resistivity of 100 Ωm (left). The right plot shows the TR when the thin overlaying gas pockets have been removed.
The amplitude and shape of the computed TR in the left plot of Fig. 6 agrees well with the measured TR in Fig 4. The maximum difference between measured and computed TR is below 1.2dB and the rms difference is 3.36%.
The effect of removing the shallow gas pockets from the 3D model is visualized in Fig. 5. (right). It is clearly seen that the amplitude below 1Hz is decreased and that the spatial shape of the TR above 1Hz is changed. The lateral extent is shortened when the shallow gas pockets are removed. Hence, it can be concluded that there is a series of high resistivity thin gas pockets on top of the reservoir visible in the measured TR as well as a thin gas accumulation just outside the reservoir. The short distance between each cmp, 250m, the wide bandwidth and the TR in the analysis, make this conclusion possible.
|WELL-LOG ANALYSIS: MODEL APPRAISAL|
The 35/2-1R log data were analysed to better understand the reservoir architecture as shown in Fig. 6 below.
Figure 6 Selected Peon 35/2-1R well logs. Gamma Ray (GR; API); Resistivity (R; Ωm); Vp/Vs (unitless); Neutron porosity – Density Porosity (N-D; unitless). Depth [m] is relative to the mud-line.
The total thickness of the reservoir interval, indicated by the Gamma ray log (red), is about 33m, but the Vp/Vs ratio (green) shows that the gas column is about 28m thick. The Neutron porosity – Density porosity log (black) shows that there are three levels of gas saturation (about 95%, 80% & 10% saturation; Sw not shown) in the column but only the high and medium saturation intervals have an appreciable resistivity signature (blue). The resistivity model of the hydrocarbon charged interval can be described by two consecutive layers: the upper layer is 12m thick with an average resistivity of 280Ωm and the lower layer is 6m thick with an average resistivity of 30Ωm. The combined transverse resistance for the two layers is then 3500Ωm2. This agrees well with the total transverse resistance in the refined 3D model which is 3600Ωm2.
|UPDATES FROM 2010 SURVEY|
During the summer of 2010 the area was re-visited and a more comprehensive data volume was acquired. In addition to showing consistency with previously acquired data it is clear from preliminary displays prior to processing and inversion, that the reservoir outline is clearly visible in Fig 7. The additional data acquired has confirmed that even in rough weather “Unprocessed” frequency responses are generally stable, amplitude trends and amplitudes levels are generally consistent from line to line. The “main” Peon reservoir can be seen in the raw frequency responses and where Line 1(the test line from the 2009 survey) was repeated, and extended, target responses in Fig 8. compare well. Continuing analysis is ongoing to estimate the volume of gas in place.
Figure 7 mean electric field amplitude per shot for channel 6 (2150 m offset) mapped to Common Mid Point position
Figure 8 Line 1 target responses from 2009 and 2010 surveys
A newly developed towed EM system has been demonstrated over a known hydrocarbon accumulation. Analysis shows sufficient quality and signal-to-noise ratio of the electric field data for successful detection and inversion of the highly resistive reservoir area including distinction of some of the shallow gas accumulations above the reservoir. It is concluded from the 3D modelling that the resistivity values in the reservoir are in good agreement with the well-log data. As expected the 1D inversions result in lower transverse resistance as a compensation for the exaggerated lateral extent. Furthermore on the basis of the 2010 results we see consistent results, good repeatability and early indications of reservoir detection in the raw data.
We thank Petroleum Geo-Services and Statoil for the permission to present this work. We also thank Statoil for the permission and support to carry out this survey over the Peon discovery.
Wright, D., Ziolkowski, A. and Hobbs, B., 2002, Hydrocarbon detection and monitoring with a multicomponent transient electromagnetic (MTEM) survey: The leading Edge, 21, 852-864.
Wright, D., Ziolkowski, A. and Hobbs, B., 2005, Detection of subsurface resistivity contrasts with application to location of fluids: US patent 6,914,433.
Storn, R. and Price, K., Minimizing the real functions of the ICEC'96 contest by Differential Evolution, IEEE Conference on Evolutionary Computation, Nagoya, 1996, 842 - 844.
Wright, D. (PGS), Ziolkowski, A. (PGS), Parr, R. (BP), Limond, C. (PGS) and Morris, E. (PGS), 2009, 3-D Time –lapse Modeling and Inversion of Multi-transient EM Data over the North Sea Harding Field: EAGE conference, Amsterdam.
Mattsson, J., 2006, Modeling of low frequency electromagnetic fields in range dependent marine environments, Marelec conference, Amsterdam.
Andersson, B. L., Hall, J. O., Otto, K., 2007, Parallel computer algorithms for the solution of volume integral equation models in marine electromagnetics, Technical report, FOI-R—2326—SE.