Application of Reservoir Seismic Inversion to the Damintun Sag in the Liaohe Oilfield

INTRODUCTION

The Damintun sag is a first-order tectonic unit in the northeast of the Liaohe basin, surrounded by boundary faults (Zhou et al., 2006).Because of its special geological conditions, such as complicated structure and intense lateral variations of thin-interbed reservoirs, the reservoir prediction is difficult in this area.

The reservoir seismic prediction is a process tha uses seismic data acquired on the surface, with constraints of available geological knowledge and logging data, to image (or solve) the spatial structure and physical property of subsurface media.It is also a special technique using seismic data to make inversion of wave impedance of strata (Shen et al., 2003).Wave impedance is an important physical parameter of rock structure, which can be compared directly with drilling data, and used to make lithologic interpretation and physical property analysis of reservoirs (Dai et al., 2006; Xiong et al., 2006; Cui et al., 2005;Huang, 2004).The acoustic equation of time curves from logging is the most important routine data, which represent reservoir characters.These curves have been applied to wave impedance inversion of reservoirs, yielding remarkable geologica effects (Fan et al., 1998).At present, this method is in a very prominent position in the technical research of reservoir prediction.Therefore seismic inversion is often called wave impedance inversion.Li (1994) claimed that wave impedance inversion is the final expression of high-resolution seismic data processing.In this work, we use the seismic data of prestack high-resolution processing and logging constraints to make multi-attribute inversion of reservoirs.Our purpose is to raise the lateral precision of reservoir prediction and enhance the ability of identifying thin-interbed sand bodies.

WAVE IMPEDANCE INVERSION OF RESERVOIR

In general sense, reservoir inversion is classified into two kinds, prestack inversion and poststack inversion, according to their processing of original seismic data.During the last 20 years, poststack inversion has gained great progress, and generated many advanced methods.On the basis of the role of logging data, poststack inversion includes 6 types: direct seismic inversion (track integration processing), logging-constraint seismic inversion, seismic-logging joint inversion, seismic-constraint logging inversion, model-constraint random lithologic inversion, and wave impedance inversion with nerve network training.According to implementation approaches, it can be divided into recurrence inversion and model-based seismic inversion.

Poststack inversion is a processing method, which analyzes seismic tracks and attempts to reconstruct velocity and acoustic impedance structures of strata.The basic model of inversion is the one-dimensional convolution model (Huang, 2004)

where r(j) is the time series of stack-migration reflection coefficients of strata, W(i) is seismic wavelet, which is assumed constant, and n(i) is noise.In this model, multiple waves are neglected.

The inversion is to determine reflection coefficients r(j) for given seismic trace T(i).As the reflection coefficients are related with acoustic impedance of strata, we have

where I(j) =ρ(j)v(j).It also helps to determine the impedance of underlying strata when the impedance is known.

Logging-Constraint Inversion

This is an inversion method of poststack seismic records.It integrates seismic data, logging data, and available geological information to produce a broadband model of impedance, which makes i possible to describe characters of reservoirs and predict new reservoirs in lateral direction.The logging-constraint inversion can preserve basic features (faults and occurrences of strata) of seismic reflection.The wave impedance solved by this method has good stability in both horizontal and vertica directions, and it is easy to implement constrain conditions.Hence it overcomes the multi-solution problem of inversion, and is able to reveal spatia variations of lithologic phases and characters.In the case of relatively stable lithologic characters, this method can exhibit changes in physical property of reservoirs.

In the inversion process, a macro-scope geological model is constructed using seismic geological, and logging data.The wave impedance is derived from logging data.Next, using the macro-scope model to proceed integration, yields an initial model of wave impedance as the input of seismic inversion under the exact model constraints Then, the initial model is revised by iteration to find an optimal geophysical model, in which the residua errors of responses with respect to the observationa data are minimum in the least square sense.The inversion principle is to transform the input initia geological model into an output result model under the constraints of seismic data.The output model mus meet the following conditions: the output mode should match the logging data used for construction of the initial model, the trend of the initial model is no changed, the output model closely follows the initia model, and the synthetic seismic records by the outpu model should match the seismic records.

Advance Processing of Logging Data

Processing and interpretation of logging data is the key for reservoir prediction.In this work, we have made full use of the software PetroWorks to proceed with the standardization and environmental correction to the logging curves of over 50 important wells in the five target areas Shen 230, Shen 225, Shen 257, Shen267, and Shen 258.

Because the basic data from many years, various logging instruments, and varied logging series are different, there exist major systematic errors between logging curves, i.e.basic value shifts.If these curves are directly used to build the initial inversion model, the systematic logging errors are carried over into the inversion result, producing lateral abrupt changes of strata, which are not in accord with real geological data.Therefore, prior inversion of all logging curves should be standardized (or normalized) to remote systematic errors between wells.Then, environmental correction is made to the equation of time curves using calliper curves, which creates conditions for constructing a correct initial model of inversion.

Seismic data inversion is a systematized method of processing and interpretation.The quality of every step can influence the inversion result.So we should monitor high-resolution and high signal/noise ratio data during the whole process.To ensure the accuracy of seismic reservoir inversion, the seismic data sets on whichamplitude-preservation processing and zero-phase correction have been made should be used.After environmental correction and standardization processing to logging data, we perform the logging-constraint inversion for the work area.

Resampling of Seismic Data

In the work area, there are numerous nonhomogeneous fissures in the buried hills, which are of Archean metamorphic rock.The analysis of high-frequency information in logging data can reveal these fissures in detail.Thus, this information should be preserved though the seismic data and logging data do not match each other in sampling rates.Therefore, before extracting wavelets, denser sampling on seismic records is performed.By fine interpolation, the sampling rate of seismic data is raised from 4 ms to 2 ms at the cost of proper reduction of resolution.It helps in matching the seismic data and logging data.Although denser sampling of seismic data does not provide any new information, it makes the logging data retain information about fissures after it is resampled at the seismic sampling rate.This procedure provides seismic inversion with a high-resolution constraint, and enhances the capability of predicting fissures by using seismic inversion data.

Wavelet Extraction and Record Synthesis

The quality of wavelet extraction is influenced by the matching relationship between well logs and seismic records.Calibration of seismic data by logging curves is also related with inversion result of wave impedance.During the process of matching logging data and seismic data, synthetic seismic records are the media linking seismic data and wel log curves.After matching synthetic records and seismic records, we should choose time windows of high single/noise ratio and good matching degree of the seismic data to extract seismic wavelets, so tha waveforms and phases of wavelets are reasonable.The specific procedure is to pick out wavelets from side-well seismic tracks of all wells.When the synthetic records and side-well seismic tracks are adjusted to a certain resemblance, new wavelets are extracted from well logs of all wells.These wavelets should have concentrated energy and bigger main-peak side lobes.Figure 1 shows the relevan results of synthetic records for wells Shen 229 and Shen 628, for which the correlative coefficients are more than 76%and 79%, respectively.

Figure  1.  The correlation of synthetic seismogram in wells Shen 229 and Shen 628.

DownLoad: Full-Size Img PowerPoint

Estimation and Generation of Model

Model estimation includes redistribution of priority and micro-adjustment of interpreted horizons It makes the generated model meet the following conditions: roughly matching the well logs and interpreted horizons, and comparable with synthetic records and seismic data.This procedure is to make the weight distribution between well follow changes of seismic data, and the generated parameter model is the integration of seismic information, well logs, and geological data.Model estimation does not generate a realistic model of strata, instead it produces a parameter model, which contains parameters of weight distribution, horizons, and contact relationships.Based on these parameters, a real strata model can be generated (Fig. 2).

Figure  2.  Low frequency model of wave impedance inversion (wells Shen 232, Shen 635).

DownLoad: Full-Size Img PowerPoint

Seismic Inversion under Logging Constraint

After the synthetic records and side-well seismic tracks are calibrated, on the basis of generated impedance model and constraints of the geological model and seismic characters, the whole 3D impedance data sets can be calculated by inversion.For the five target areas of the Damintun sag, i.e.the Shen 230, Shen 225, Shen 257, and Shen 267 on the western slope zone, and Shen 259 on the eastern slope zone, totally about 500 km² in size, we have conducted seismic inversion under logging constraint separately.

Figure 3 shows the impedance cross-sections through wells Shen 223, Shen 232, and Shen 635.It seems to reveal the two oil-bearing layers in the Es₄ of strata in the wells Shen 232 and Shen 635.But as both the compaction of Es₄ mudstone and isochronous sedimentary interface in the sag zone can produce clear impedance interfaces, the impedance of the two oil layers of Es₄ differs only slightly from that of mudstone compaction, making discrimination difficul to some extent.

Figure  3.  The impedance inversion profile from wells Shen 223 to Shen 635.

DownLoad: Full-Size Img PowerPoint

Wave Impedance Inversion of Prestack HighResolution Seismic Data

To strengthen the fine assessment of hidden reservoirs of Es₄ in the western slope zone and depression zone of the Damintun sag, we have made a further seismic inversion by using prestack time migrationandhigh-resolutiondata.These high-resolution seismic data can reveal the inner features of Es₄ sandstone reservoirs and buried hills with clarity.On the other hand, the structural configuration displayed by these data is greatly different from that of routine data processing.Therefore, based on the fresh structural interpretation and assessment using high-resolution seismic data, we conduct a study of high-resolution impedance inversion and reservoir prediction to the focused areas mentioned in the last section.Figure 4 shows the low-frequency model constructed on the basis of high-resolution seismic data.Figure 5 displays the impedance inversion result for the depression area Compared with the inversion results of routine data the impedance cross-section through wells Shen 223Shen 232, and Shen 635 has dramatically enhanced resolution in both horizontal and vertical directions Moreover, the impedance interferences from mudstone compaction and isochronous sedimen interface have been suppressed to some degree.

Figure  4.  The lower frequency impedance model of high resolution data.

DownLoad: Full-Size Img PowerPoint

Figure  5.  The impedance inversion profile of high-resolution data from wells Shen 223 to Shen 635.

DownLoad: Full-Size Img PowerPoint

Logging Multi-attribute Inversion

As mentioned earlier, both the mudstone compaction and isochronous sediment interface can produce distinct impedance boundaries.Hence it is somewhat difficult to identify the Es₄ reservoir at the Shen 232 and Shen 635 well areas only by impedance To solve this issue, we have made a logging multi-attribute inversion of acoustic equation of time to the Es₄ of strata.As shown in Fig. 6, the origina equation of time on the inversion cross-section accords well with the predicted equation of time Meanwhile, it displays that the I group oil layer is relatively continuous in lateral direction, and the three sets of sand bodies in the II group oil layer are no linked laterally.Using the visualization technique, we have analyzed and calculated the inversion result, and made the final prediction of all existing sand bodies in the work area.

Figure  6.  Cross-section of acoustic inversion and sedimentary facies through wells Shen 223 and Shen 232.

DownLoad: Full-Size Img PowerPoint

CONCLUSIONS

In the Damintun sag of the Liaohe oilfield, we have carried out studies of logging-constraint inversion, wave impedance inversion of prestack highresolution seismic data, and logging multi-attribute inversion.The results of these inversions have raised vertical resolution of reservoir prediction, enhanced the ability of identifying thin-interbed sand bodies, and made the reservoir prediction more reliable.This work has also helped confirmation of favorable lithologic traps, achieved good geological effects, and is of significance for exploration of hidden reservoirs in this area.