The cost of oil production needs to be lowered substantially in high-cost areas like the North Sea for multinational oil companies to maintain activity levels. Many fields have yields no greater than 45%. When they are closed down, more than half the original reserves are still in place.
It should be a top priority for national or international energy policies to find the missing oil prior to field closure. However, to achieve a real breakthrough in improved oil recovery, detailed information is needed about whole reservoirs at the 1-meter scale. This is beyond the capability of conventional seismic methods, including vertical seismic profiles (VSP), which yield data from the target zone with an upper frequency of 50 Hz, and with resolution limited to roughly 20 meters.
To obtain the data throughout the reservoir at the 1-meter scale, we need to explore in the 50-2,000 Hz seismic frequency range. This is possible using multi-well imaging, whereby multiple multi-component receivers are installed permanently in the wells in or near the reservoir, and borehole seismic sources are used to generate broad bandwidth seismic waves in the reservoir.
Time-lapse multi-well seismic surveys can image changes, in particular the movement of fluids. Bypassed hydrocarbons will show up as volumes within the reservoir in which no changes occur.
This technology could provide a step-change in our ability to exploit oil reserves. The receiver and recording instrumentation is available, and different borehole seismic sources also exist. Furthermore, the strategy for processing the data is well-known and understood, although 10 man-years of software development may be needed for it to become an industry-standard technique.
Detailed rock properties such as porosity, permeability, and fluid saturation are required for reservoir parameterization and simulation. These can be determined precisely only at the wells and in cores taken from the wells. Away from the wells, conventional seismic data is not much help. Even the stratigraphic relationships are not well determined. Rarely can both the top and bottom of the reservoir be imaged on seismic data, and often it is not possible to tie wells with a sufficient precision to interpret strati graphy between them.
As conventional seismic data is incapable of providing the required resolution, the industry has resorted to geostatics to perform the interpolation. This procedure also has problems. Consider, for instance, permeability. Most detailed knowledge of permeability comes from measurements on core samples. If the measurements were truly representative of the whole rock volume, the distribution of permeabilities from these measurements could then be pinned down, allowing realization of the permeability based on this distribution.
Unfortunately, permeability is not like this. There never is a representative sample of the rock volume - the bigger the sample, the greater the variability in measured permeability values. In a real reservoir we do not know how permeability is distributed though rock volume and fluid flow cannot be predicted by extrapolating from measurements taken from a few core samples. It needs to be measured in-situ.
Despite all the technology available to surface seismic reflection, there remain severe limitations on the resolution of the data, determined by the highest frequency signal returning from the target that can be seen above the noise level. For targets at depths of 2-3 km, this upper frequency is normally less than 50 Hz and nothing, apparently, can be done about it. The limit is controlled by elastic and anelastic attenuation in the earth, the travel path length over which seismic waves must propagate, and by the ambient noise level at the receivers.
When imaging a 50-Hz wavelength in relation to a reservoir, the resolution is typically tens of meters, but features of interest in the production of hydrocarbons may be in the order of just a few meters. At least an order of magnitude increase in resolution is required to resolve such reservoir features if geophysics is to make any impact in improved oil recovery.
The accompanying figure shows (upper part) a log-log plot of the number of faults resolvable per unit volume against the size of the faults in meters (lower part). The data reveal a fall-off, more or less on a straight line, but there is a gap of several orders of magnitude between what can be seen with core and well logs in the narrow vicinity of a well, and what can be resolved with 3D seismic data. The lower half of the figure shows the seismic bandwidth associated with the structures that can be resolved.
Distribution of faults
The implied straight line between core and well-log data on one hand and seismic data on the other is a power law that can be used to estimate realizations of the distribution of sub-seismic faults in the reservoir. Within the bandwidth of the data there are very large numbers of possible realizations, each of which is equally probable in the absence of further constraints. So, the likelihood of any one of them being true is actually tiny.
To increase the resolution, the attenuation effect has to be reduced by moving the source and receiver closer to the target. Boreholes or existing wells must be used. VSPs use the wells for receivers, but the travel path from the source to the receivers includes the most attenuating near-surface rocks, and the resolution is therefore generally no better than for seismic reflection.
The cross-well technique has been used onshore in the US, but has been considered prohibitively expensive for offshore use. Single-well imaging is also a possibility, but the geometry imposes obvious constraints. If the benefits are great enough, single-well and crosswell techniques can be extended to multi-well imaging.
It needs to be established that attenuation of seismic energy is much less at depths of 2-3 km than near to the earth's surface. Crosswell was applied to two deviated wells in the Groningen onshore gasfield in The Netherlands in 1992 (see figure). The two wells were approximately 350 meters apart at a depth of 2,400 meters.
In one well, a SouthWest Research bender source was installed at a depth of 2,370 meters, while in the other, a hydrophone string measured arrivals from the source. Timing lines on the recorded data were 10 ms apart (a transmission rate of 1,000 Hz, or around 100 wavelengths between the wells).
Analysis of the data, bearing in mind the input and output spectra, indicated that the attenuation was very low or Q - very high. In this case, Q had to be at least 1,000. It was concluded that at depth, in typical southern North Sea rocks, high frequency seismic waves may be propagated over hundreds of meters without significant attenuation, except by geometrical spreading on the wavefront.
A fourth option involves use of permanent instrumentation, something which should be considered for all existing and new wells, particularly those close to potentially unswept hydrocarbons. The instrumentation would comprise perhaps hundred of broad-bandwidth, three component geophones at approximately 1 meter spacing, delivered via coiled tubing in the annulus between the well casing and liner.
The University of Edinburgh is leading the project to build the required multi-level multi-component receivers necessary for this technique, known as Uniwell, with sponsorship from Amerada Hess, Conoco, the UK Department
of Trade and Industry, Exxon Production Research, Mobil, and Oyo Instruments. The last was subcontracted to build the receiver. This instrumentation is available now. Recording of the signals from all geophones is done at the surface with very high precision (24 bits) and very high sample rates (4,000 Hz).
A variety of borehole seismic sources exist such as air guns, imploders, piezo-electric transducers, sparkers, and vibrators. Most have been proven in onshore cross-well surveys, but have yet to be implemented offshore. The source is inserted through the liner and does not interfere with any receiver instrumentation installed in the same well. Thus any well can act as the source. When many wells are instrumented with receiver strings, there are a range of methods for exploring the reservoir:
- Passive seismic monitoring
- VSPs and time-lapse VSPs
- Uniwell and time-lapse Uniwell imaging
- Cross-well and time-lapse cross-well imaging
- Multi-well and time-lapse multi-well imaging.
Synchronization between the source in one well and received data from others - perhaps on different platforms - can be done with global positioning system clocks. There is no need for a direct link between the source and receivers. In the absence of borehole sources, the instrumented wells can be used for passive seismic monitoring or for VSPs.
The data processing of multi-well seismic data is currently being developed at the University of Edinburgh. The strategy is based on conventional seismic and VSP processing, but it incorporates the measured vector motion at the three-component receivers to constrain the imaging solution. Since these surveys will take place in fields where 3D seismic data already exist, the background velocity model for wave propagation, especially of P-waves, is already known.
Imaging of the data can be performed using these known velocities. In time-lapse mode, moving oil-water contacts can be observed by taking the difference between two identical surveys made at different times. With a permanent installation of receivers, the recording parameters can be made very repeatable.
Anton Ziolkowski is a member of the Department of Geology and Geophysics at the University of Edinburgh, Edinburgh, Scotland.