LOGGING WHILE DRILLING KEEPS HORIZONTAL WELL ON SMALL TARGET

Jack Leake
Oxy U.S.A. Inc.
Houston

Frank Shray
Schlumberger Well Services
Houston

The logging-while-drilling (LWD) measurement of two resistivities of different characteristics has led to a new interpretation method for the analysis of horizontal wells.

By logging deep and shallow resistivity in real-time, marker beds were identified to help maintain well bore trajectory. The resistivity measurements were split into vertical and horizontal components to provide additional information for formation evaluation.

In 1945, Ark Fuel Co. discovered and began developing the Olla field on the crest of the La Salle arch in La Salle Parish, La. Oil production comes from the Wilcox formation from alluvial sand packages that range in thickness from 3 ft to 120 ft.

Now operated by Oxy U.S.A. Inc., Olla field was chosen in 1990 for a horizontal well pilot project. It was hoped that a horizontal well could alleviate water coning in one of the field's more productive sand packages the 40-ft Cruse sand.

Although some of the vertical wells in the Cruse produced at rates of more than 300 b/d, most had dropped to below 50 b/d because of water coning up to the perforations.

The Cruse was an attractive target for this pilot project because it had been extensively drilled, providing excellent control of all pay horizons and remaining oil reserves. Furthermore, this zone has exceptional porosity, permeability, and production capacity, and it had been recently unitized with the state. It was an ideal candidate.

The pilot program tested the feasibility of drilling a horizontal well bore above the water leg of the Cruse sand to eliminate water coning prevalent in the field's vertical wells. From a logging perspective, the objectives were to get the well bore to the target and keep it there.

The planning team included personnel from Oxy U.S.A. Inc. as well as the directional drilling, measurement while drilling (MWD), and logging companies.

LOGGING REQUIREMENTS

There were three fundamental requirements for the use of LWD in the horizontal well:

Evaluation of the target formation, the Cruse sand
Correlation of the marker beds above the Cruse sand and identification of the Cruse sand
Steering of the horizontal well bore through the oil zone.

Because the Cruse sand is a shaly sand reservoir, evaluation of its specific petrophysical characteristics was just as important in a horizontal setting as in any of the field's vertical wells. Resistivity and porosity measurements were necessary. In addition, confidence in the correct location of the horizontal well bore could be increased significantly if its trajectory mould be correlated with resp(..ct to lithology markers and the top of the Cruse sand.

More specifically, operational objectives required that the well bore stay within the top 10-1 2 ft of the Cruse sand to ensure that water would not be encountered.

PREPARATION

MWD tools have become standard equipment for drilling many horizontal wells. These devices transmit borehole inclination, azimuth, and tool face orientation to the surface through a mud telemetry system. In this case, a real-time, drill-collar-type logging system accompanied the MWD tool to acquire the needed resistivity and porosity measurements while drilling.

The resistivity and gamma ray measurements were obtained by an electromagnetic, borehole-compensated tool that can investigate multiple radial depths. It alternately broadcasts 2-megahertz electromagnetic waves between pairs of receivers and puts out a phase-shift-based, shallow resistivity (Rps) reading and an attenuation-based, deep resistivity (Rad) reading.

The Rps measurement has a sharper vertical response than the Rad measurements It was believed that the different vertical response characteristics of these measurements could help to identify the approach of the well bore to the target zone, the actual target top, and the departure of the well bore from this top.2 To confirm this, a series of model logs were computed with an electromagnetic modeling program .3 The program uses model characteristics of the particular resistivity tool to generate predictive log data at any degree of bed dip or borehole inclination, given multiple inputs of formation bed resistivities and thicknesses.

MODELING PROCEDURE

Based on the drilling plan, the path of the well bore would pass close to two vertical wells. One of these vertical wells would be close to the end of the curved (build) section of the horizontal well, and the other would be close to the end of the horizontal section.

Because each vertical well would give a different perspective of what might be encountered by the horizontal well, model logs were prepared based on data from both vertical wells.

In the following discussion, dip is defined as the relative angle between formation bedding planes and a plane normal to the axis of the borehole. The first step in creating a series of deviatedwell model logs involved finding a good match of the modeled log data to the data from an offset vertical well. Fig. 1 compares the model log to the induction log from the offset well.

Next, a model log was computed at relative dip angles of 750, 80, 85, and 89. This procedure gave an idea of what each marker bed and the target zone would look like at various angles of borehole deviation. Because of the inherent uncertainties in drilling, model logs were prepared to cover a variety of possible situations.

Fig. 2, for example, is a model log of the 2-megahertz resistivity tool in the vicinity of the top of target at a relative dip angle of 85.

At 85, one vertical foot of formation thickness is equivalent to 11.5 ft measured depth (MD). Furthermore, two important features appear on this log. First, there is a polarization horn at the shale/target sand boundary, at about 735 ft. The dip angle, the resistivity contrast between the two beds, and the level of resistivity of the beds contribute to the occurrence of this horn.3

The model log indicates that the shallow and deep resistivity readings will be influenced differently by the approach of the target zone and the departure from the shale that lies above it.

In Fig. 2 the model log indicates that the shallow Rps curve and the deep Rad curve will cross over several vertical feet (about 20 ft MD) above the bed boundary, or top of target. The two will remain separated until the too[ is about 4 vertical ft (40 ft MD) below the boundary. Then, once the resistivity sensor is 5 vertical ft within the target oil sand, the Rps and R,d curves are predicted to overlap.

This sequence provided some confidence that the approach to and departure from the top of target zone could be pinpointed.

Upon completion of the modeling, it appeared that the polarization effect and the separation of the two resistivity curves within the top of target sand would be key resources in the effort to keep the well within its narrow target zone. Additionally, the model logs would provide an advance picture on a measured depth resistivity log of the marker formations above the target. This feature would prove useful in picking the markers. Prior to the operation, it was difficult to imagine what the various formations would look like logged at a high degree of deviation.

LWD DRILLING

As the well was drilled, an early problem affected the logging outcome. A shallow, unconsolidated formation took fluid. This forced the drillers to start the build portion of the well 60 ft below the planned kickoff point and to drill at a higher build rate to compensate.

As planned, the first one third of the curve was drilled with only the MWD tool. Then a pipe trip was made to install the resistivity and neutron density/porosity tools on the bottom hole assembly (BHA) to start the LWD phase.

The porosity tool, equipped with a stabilizer, could not get through the higher build rate curve and had to be removed from the BHA. The tool's stabilizer has since been modified to allow it to negotiate high build rate curves. But on this well, LWD porosity measurements were not taken.

The resistivity tool was positioned above the drill bit and mud motor, which placed the sensor measuring point about 36 ft behind the bit. The resistivity tool measures the conductivity of a spatial volume, which is most simply described as being concentric with the axis of the tool and centered at the two receivers.

The depths of investigation of Rps and Rad are about 45 in. and 65 in., respectively, at 10 ohm-m. As the tool inclination increases, it is possible for the sensor to read down or up into a formation that the bit is not cutting. Thus, the sensor's investigative volume will include information that is at a true vertical depth (TVD) both above and below the sensor.

Fig. 3 shows the relative layout of the BHA with calculation of vertical distance above the bit for various borehole inclinations. The plans called for the borehole to be inclined about 85 at the top of target. This would place the resistivity sensor only 3.0 ft vertically above the bit and the target formation.

The closer the BHA would come to horizontal, the closer the sensor would come to the formation drilled by the bit. In this way, the approach of a bed boundary would be sensed by the resistivity tool measurement, as predicted by the model logs.

Drilling progressed with the MWD and resistivity tools.

The directional, resistivity, and gamma ray data were transmitted to the surface computer and presented in both MD and TVD formats. It was found useful to display hole deviation on the resistivity log. This made it easier to select the most appropriate model log for each particular section of borehole.

An updated set of MD and TVD logs was generated after every 50 to 100 ft of new borehole. Comparisons of the MD logs to the model logs and the TVD logs to the offset vertical well logs simplified the correlation process. While the well was drilled, correlation of the marker beds confirmed that the target and the well bore were going to intersect as planned.

The Lower Matthews sand, from 2,680 ft to 2,730 ft, caused some concern (Fig. 4). The resistivity readings in the lower half of this sand began to increase on the logging unit monitor (at about 2,710 ft), leading to speculation that the target had been reached. This characteristic of the Lower Matthews had not been recognized in the vertical wells and, therefore, was not included in the well models. This resistivity increase is believed to be a tight sand stringer that was too thin vertically in the offset wells to have been fully resolved by a wire line induction tool.

The real target marker was about to be reached when the Lower Matthews anomaly arose. From the model logs, the signal of the approach of the pay zone would be the separation of the Rps and Rad resistivities, and at 2,746 ft this separation began. The bit was at 85 deviation, about 4 vertical ft below the resistivity sensors, and just entering the top of target. The model log and the true log showed good correlation into the top of the target. However, the Rad deep resistivity increased above the expected 4 ohm-m to a value approaching 10 ohm-m.

At 2,960 ft the resistivity began to decline. Combined with the bit's TVD, this decline indicated that the well bore was returning to the shale above the target. The directional driller indicated that the bit was building angle and that a pipe trip might be required to change a key stabilizer in the BHA. Although some success was made at dropping angle, a pipe trip was made at 3,118 ft.

When drilling and logging resumed, the resistivity and gamma ray logs confirmed that the upper shale/target boundary had come close to the well bore. The log from 3,090 ft to 3,130 ft indicated that that shale was probably close above the well bore, and sand was just below it.

Both the gamma ray and resistivity log values were between the values recorded earlier in the shale and the sand. No polarization horn appeared, and this indicated that the well bore did not actually go through the shale/target interface.

It is interesting to note that the TVD of the shale interval was 1.5-3 ft below the previously seen bottom of the shale (Fig. 5). It seems reasonable to believe that bed boundary undulations, or changes in true vertical thickness of beds, are to be expected. This tends to further justify the use of real-time petrophysical logging in horizontal wells.

The drilling and logging operation continued rather uneventfully until about 800 ft of the target had been drilled. Fig. 5 shows that about 12 vertical ft of the target sand had been drilled. At that time, it was noted with some interest that the pay zone resistivity values had continued to be greater than those observed in the nearby vertical wells.

OFF SITE PROCESSING

The drilling and production programs for this horizontal well were deemed successful, yet from a well logging perspective the relatively high resistivities in the pay zone had to be understood. Why were the measured resistivity values higher than had been expected in the pay zone?

Core analyses from Olla field vertical wells showed that the top few feet of the Cruse pay zone often contained intervals of highly cemented sand grains, and the other Cruse sand intervals were likely to be finely laminated with shale.

Additionally, the response of the two resistivity curves was different. It is common to interpret separation of deep and shallow resistivity curves as a result of the effect of filtrate invasion. However, this log was made while drilling. Although invasion effects have been observed in LWD, the amount of the separation in this log seemed anomalous.

The resistivity log modeling program was used to help confirm that the increased horizontal resistivity and vertical resistivity were unequal. A new model represented the upper interval of the pay zone as a series of high and low resistivity streaks. This reasonably represented the effects of formation anisotropy, as shown by Anderson, et al .3 Two logs were computed using this model: one at O dip (vertical), and another at 89 dip.

These model logs, although not representative of a unique solution, did successfully recreate the actual log responses of both the vertical wells and the horizontal well. They show that a laminated formation can create a 4 ohm-m log in a vertical well and a 9 ohm-m log in a horizontal well.

It is important to mention several comments concerning the assumptions with this kind of interpretation.

Resistivities measured by different sensors (varying in vertical resolution and depth of investigation, for example) can vary for numerous reasons, including borehole effects, bed thickness, shoulder beds, fluid invasion, or a combination of all these. With high relative dip angles between resistivity sensors and formation beds, anisotropy and polarization effects must be considered.

However, with LWD, several assumptions can simplify the interpretation task. For example, the borehole may be close to gauge, which will minimize the borehole effects. Additionally, the resistivity sensors will generally log a formation within 30 min after the bit drilled it, which tends to minimize invasion effects. If shoulder beds are more than about 2 vertical ft from the logging too[, then only the formation anisotropy effect remains as a potential problem.

HORIZONTAL RESISTIVITY

From the results of the laminated, anisotropic model log, a method was sought to solve for the horizontal component of the resistivity measured in this well. Fig. 6 illustrates the difference between a measurement of horizontal resistivity in a vertical well and the measurement of some combination of vertical and horizontal resistivity in a horizontal well. If horizontal resistivity could be obtained in the horizontal well, then it would appear to be the proper measurement to compare to the vertical well logs, as well as for quantitative formation analysis.

Based on the resistivity tool's engineering code, an anisotropic software package was developed. Given the resistivity measurements and the relative dip angle between the tool and the formation beds, this software provided a continuous log of the horizontal and vertical resistivity components of the given measurements.

The continuous horizontal resistivity log obtained for this well is shown in Fig. 5. The results indicate that the horizontal resistivity shows good correlation to the resistivity measured in a nearby vertical well (Fig. 1). In the interval from 2,785 ft to 2,970 ft, which traverses 2 vertical ft and which was likely to be more cemented than other intervals of the Cruse, the horizontal resistivity reads between 5 and 6 ohm-m.

Practically the entire remaining interval of the Cruse exhibited a horizontal resistivity of 3-4 ohm-m, exactly as predicted from the vertical well logs. In the interval where the well bore approached shale, at about 3,100 ft, the influence of the shoulder bed created a tool response that the software recognized as being unrealistic for anisotropic effects, and hence did not output any values. The anisotropic software tested in this project promises to provide several advantages to the log interpretation process. In real time, the horizontal resistivity component can be ascertained while drilling. A more direct correlation with offset well logs can be made.

If a well deviates into a water zone that could be masked by the anisotropic effect, then the operator would have the ability to quickly identify the true nature of the problem by seeing the horizontal resistivity component of the measurements. To calculate formation water saturation, porosity values combined with horizontal resistivity should provide a more accurate analysis than has been previously available.

The most important result of any well drilling operation is the production of hydrocarbons, and in this case the results were excellent. After 1 month of production with a submersible pump through a slotted liner, the well was producing at a rate of about 400 b/d with no water. As a result of the success with this horizontal pilot project, Oxy will drill more horizontal wells in Olla field.

ACKNOWLEDGMENT

The authors thank Kirk Sparkman, Oxy U.S.A. Inc., for his expertise as a project geologist; Barry Cross, Schlumberger Well Services, and Chris Lenamond, Anadrill, for their well site logging expertise; and David Allen, David Best, and Richard Rosthal, Schlumberger Well Services, for their contributions to the interpretation.

REFERENCES

Clark, B., et al., "A New Resistivity Measurement for FEWD," Paper A, Transactions of the SPWLA 29th Annual Logging Symposium, San Antonio, June 5-8, 1988.
Allen, D.F., and Luling, M.G., "Integration of Wireline Resistivity Data with Dual Depth of Investigation 2-mHz MWD Resistivity Data," Paper C, Transactions of the SPWLA 30th Annual Logging Symposium, Denver, June 11-14, 1989.
Anderson, B., et al., "Response of 2-mHz LWD Resistivity and Wireline Induction Tools in Dipping Beds and Laminated Formations," Paper A, Transactions of the SPWLA 31st Annual Logging Symposium, Lafayette, La., June 24-27, 1990.