During periods when a well is in a reservoir zone with gas components, gas molecules will start to diffuse from the formation into the drilling fluid in the wellbore. The amount of gas that can diffuse into the drilling fluid is dependent on the type of drilling fluid, the solubility of the gas in the drilling fluid, and the volume of drilling fluid in the open hole section of the reservoir. The diffusion rate is rather slow which means that the drilling mud must have spent much time in the reservoir zone before any measurable amount has entered the mud. The solubility of gases in water is orders of magnitude less than the solubility in oil, thus water based muds will be ignored in the rest of this paper. In an oil based drilling fluid, it is mainly the base oil component that will retain the gas, so the maximum amount of gas that can be dissolved is dependent on the solubility of the reservoir gas in the base oil component, and the fraction of base oil in the drilling fluid.
The aim of this topic is to provide a methodology for quantifying the effects of hydrocarbon gas molecule diffusion into oil based mud (OBM) when the well is in “safe” overbalance. Situations due to the wellbore being in underbalance with respect to the reservoir pressure is not a topic in this discussion.
The discussion here aims at describing the diffusion process, and predicting the potential dangers that the well operation could face because of the diffused gas. By using the program tools discussed in this report, it will be possible to predict the level of danger as a function of operational delays in the drilling process, both foreseen and unforeseen, such delays due to bad weather, waiting for equipment, etc. This could provide improved planning and better preparedness which again will result in improved safety and optimal well operation.
Measurements of diffused gas are compared with the amount expected from the theoretical diffusion simulations. Two positions in different parts of the well are used to examine the diffusion mechanism. A cook-book style methodology is developed for determining the expected amount of methane that would be absorbed in the base oil as a function of the delay periods, i.e. the period with no movement of the OBM in the well.
Methane Diffusion In Oil Based Mud
Measurements from a North Sea well were obtained from Aker BP. The well was nearly vertical, and had used advanced hydrocarbon detectors in-line. The detectors could detect hydrocarbons: C1, C2, C3, iC4, nC4, iC4 and nC5. A schematic diagram of the detector system is given in Figure 1.
OBM and air is moved into a gas trap chamber. The hydrocarbon gas is released by mixing the air vigorously with the mud, causing the hydrocarbon gas to diffuse out of the OBM and into the air until the two are in equilibrium with respect to hydrocarbon concentration. The air flow is typically 5 l/min and the mud flow rate is typically 3 l/min. These flow rates may differ. Mud flow rates as high as 11 l/min was used.
The air with the hydrocarbon gas obtained from the gas trap flows through a Flame Ionization Detector (FID) (or “FID Total Gas and Chromatograph detector”). They are optimal for this type of application since they are highly sensitive (over a linear range of 7 orders of magnitude) and “filter” out nitrogen, water vapor, CO and CO2. This well had two such systems installed:
- One connected to the Pumps-Up Gas System in the returns flowline to evaluate the gas in the mud as it comes out of the well.
- One connected to the Pumps-Up Gas System in the shale shaker pit to determine the gas resident in the mud as it is pumped into the well. The shale shaker pit is after the degasser. This is the “background gas” in the mud.
The mud degasser is unable to eliminate hydrocarbon gas from the OBM. Since the FID is so sensitive it is important to be able to determine how much of the gas detected is due to the influx during circulation and not the hydrocarbons already in the mud.
These types of measurements make proper analysis of the influx mechanism possible.
The discussion that will follow is a description on how the measured values could be used to obtain some of the parameters needed for a diffusion study, and thus providing a basis for the diffusion computations needed to investigate if long delays could grow to a well control problem. This discussion is only meant as a recipe on how to do the computation, and not as a “serious” example. There are too many uncertainties in the measurements to do this, as will be discussed later.
We will investigate two positions in well, at 1850 m measured depth, and 4575 m measured depth. For simplicity, the mud will be represented as a fluid with only three components: water phase, base oil and weight material. The base oil element description has not been made available, so a base oil used in previous computations, available at IRIS, will be used instead. The mud densities used in the operation are available as well as the Oil Water ratio. The densities were measured at 50°C. Using these assumptions, it is possible to determine the amount of base oil present in the mud.
1″6 hole at 1850 m MD. The mud density provided by the mud logs were 1.55 sg while drilling the 16″ hole at 1850 m MD. The pressure at 1850 m was approximately 281 bars. The formation temperature was approximately 50°C. Using the assumed base oil components, the saturation level of methane at 281 bars and 50°C turned out to be 64 mole percent in the base oil, computed using Calsep’s PVTsim. While drilling into the formation at 1850 m MD, the hydrocarbon measurements were:
Table 1—Measured hydrocarbon at 1850 m MD.
|Rel. to C1||1||0.00106||0.0002||0||0||0||0|
This section of the well was drilled at about 40 m/h, while the mud flow rate was 3900 l/min. In other words, in one minute, 2/3 m of formation was drilled. This results in 0.0865 m3 cuttings. The relative formation volume in the returning mud was 0.0865 / (0.0865+3.9) = 2.167%.
The 15110 ppm is the methane measured in the air-hydrocarbon measurement of the FID in the hydrocarbon gas detector system. We will assume that the amount of mixing done by the gas trap in Fig. 2 has been vigorous enough over long enough time to obtain methane equilibrium in the air-methane mixture and the base oil-methane mixture. The method to determine the methane mole % of the base oil goes as follows:
- Assume the mud flow rate into the gas detector system was 3 l/min. The greatest contribution to methane absorption is the base oil, thus the base oil volume is the only part of the mud volume needed. Since the base oil fraction of the clean mud was 63%, the volume needed is 3 l/min × 0.63 = 1.9 l/min. Or 1.9 volume units.
- The air flow rate was 5 l/min, i.e. 5 volume units.
- Determine the ratio of moles of base oil to the number of moles in the air. Air occupies 24368 cm3/mol at 20°C and 1 atm, and the base oil occupies 274.6 cm3/mol. The ratio of base oil moles to air moles in the degasser becomes 35.5.
- Mix 35.5 moles of base oil and 1 mole of air in a PVT simulator.
- Add methane to the base oil and perform a Flash calculation.
- Iterate until the methane component in the gas (or vapor) becomes 15110 ppm (the amount measured by the FID) after the Flash calculation.
- This results in 0.056% methane in the base oil.
If we assume that all the methane in the drilled-through reservoir ended up absorbed in the base oil, then the 15110 ppm converts to 0.056% mole fraction methane in the base oil. Since this came from 2.167 % of the mud and cuttings mixture, the methane volume fraction in the formation only is approximately 2.58% (= 0.056%/0.002167).
The porosity in the formation at 1850 m is unknown, but it is possible to make educated guesses from the methane computation. If the reservoir was pure methane, then the porosity must be 2.58%. If we guess that the reservoir fluid had a methane content of 65%, then the porosity would become 4%. Iterating, using Calsep’s PVTsim, the saturation limit of the base oil under the conditions at 1850 m was found to be 64% mole fraction methane.
Assuming the reservoir fluid was an oil, the diffusion coefficient of methane in oil was estimated to be roughly 1.5×10−9 m2/sec.
At this point, all parameter needed for the diffusion calculations have been established except the depth of the invasion zone.
After drilling the 16″ section was finished, the 13 3/8″ liner was tripped in. The delay between pulling the drill string out of the 16″ hole and the liner circulating the mud at 1850 m was approximately two days. The hydrocarbon measurements taken from the 1850 m depth are given in Table 2.
Table 2—Measured hydrocarbon at 1850 m MD after two days delay.
Table 2 shows no C2 and C3, as opposed to Table 1. This indicates that the methane entered the base oil through molecular diffusion rather than as a “small” kick. In a kick situation, the whole reservoir fluid would enter the well bore. In a diffusion situation, individual molecules diffuse at different rates. Methane, the smallest molecule, diffuses faster than ethane, which diffuses faster than propane etc.
The 12136 ppm converts to approximately 0.046% mole fraction methane in the base oil. Using the diffusion model described earlier, and the parameters above, the modeled diffused methane mole fraction for different invasion depths are listed below, Table 3.:
Table 3—Computed diffused methane mole fraction for different invasion depths.
|Invasion Depth||2.4 cm||2.7 cm||3.0 cm|
6″ hole at 4575 m MD. The mud density provided by the mud logs were 1.9 sg while drilling in the 6″ hole at 4575 m MD. The base oil volume fraction in the mud was computed to be approximately 54.4%. The pressure at 4575 m was approximately 852 bars. The formation temperature was about 150°C. Under these conditions, methane is infinitely soluble in the base oil.
While drilling into the formation at 4575 m MD, the returned mud measurements on hydrocarbons are shown in Table 4.
Table 4—Measured hydrocarbon at 4575 m MD.
|Rel. to C1||1||0.2334||0.041||0.0047||0||0||0|
This section of the well was drilled at about 4.9 m/h, while the mud flow rate was 826 l/min. In other words, in one minute, 8.17 cm of formation was drilled. This results in 0.00149 m3 cuttings. The relative formation volume in the returning mud is 0.18%.
The 634 ppm is the methane measured in the air/hydrocarbon measurement of the FID in the hydrocarbon gas detector system. Using the same method as for the 1850 m MD section, we obtain a methane content in the mud/cuttings mixture to be 0.002% mole fraction. This computes to 1.3% methane in the drilledthrough formation.
The formation porosity was measured to 5%, thus each pore contains about 25% methane. The C2, C3 and iC4 content of Table 4, was relatively much greater than that of the formation at 1850 m depth. It can be argued that this reservoir is of a heavier oil than the reservoir at 1850 m. Heavier oil has a smaller methane diffusion coefficient than lighter oil. The diffusion coefficient was roughly estimated to be 0.8×10−9 m2/sec.
The delay between pulling the drill string out of the 6″ hole, tripping in and circulating was 4 days. The hydrocarbon measurements taken from the mud that was circulated from 4575 m depth after 4 days delay is given in Table 5.
Table 5—Measured hydrocarbon in drilling mud at 4575 m MD after 4-day delay.
|Rel. to C1||1||0.0427||0.006||0||0||0||0|
Table 5 shows relatively much smaller measurement of C2 and C3, i.e. by a factor of 5.5, than that of Table 4. This indicates again that the methane entered the base oil through molecular diffusion rather than as a “small” kick.
The 14460 ppm converts to approximately 0.054% mole fraction methane in the base oil. Using the diffusion model described earlier, and the parameters above, the “expected” diffused methane for different invasion depths are listed below, Table 6.
Table 6—Computed diffused methane mole fraction for different invasion depths
|Invasion Depth||2.4 cm||2.7 cm||3.0 cm|
These examples are using values that are very uncertain. To use the “FID Total Gas and Chromatograph detector” for analyzing the hydrocarbons in the formation quantitatively, many improvements should be made:
- The proper base oil description should be used.
- The mud sample should be taken prior to the bell nipple, i.e. before having been mixed with air.
- The background hydrocarbon measurements should have been a part of the analysis.
- The mud and air flow rates should be recorded. The new “standard” seems to be 5 l/min and 3 l/min respectively.
- Better estimation of invasion zone.
- Improved diffusion constants of methane in the hydrocarbon fluids in the well.
However, it seems to be possible by using rough measurements to obtain good enough estimates to assess the danger due to diffusion when a well is left without circulation for some time while in overbalance.
Oil based mud has many positive characteristics, such as it being a lubricant. This makes it the obvious choice for inclined and far reaching wells. It is also the choice for drilling through shale formations and salt. One of the disadvantages lies in gas kick detection. Hydrocarbon gas gets absorbed in the base oil, reducing the pit gain when the kick is taken.
Another less dramatic characteristic of oil based mud is that when the well is in overbalance, gas can still get absorbed due to diffusion. For most wells, this is not a problem, but ignoring it can cause severe problems such gas-in-riser without much warning. One of the advantages of oil based mud, its ability to create thin mud cakes, is a disadvantage when it comes to the rate methane is dissolved in the oil based mud when the well is in overbalance.
This report has covered the physics of diffusion in a well bore, and produced a pseudo code to determine the amount of methane that will be absorbed in the mud due to diffusion. There is a lack of good measurements and theory for determining the diffusion constant of methane in hydrocarbon fluids.
A better understanding of the static and dynamic invasion depth as a function of mud composition and formation characteristics would be welcomed. Paradoxically, the well operations that seems optimal, such as drilling quickly, and avoiding filtration into the reservoir, give the shallowest fluid invasion and the largest methane diffusion rate into the well.
In the Case Study, measurements from a well in the North Sea was used to test the diffusion process model and show how this code could be used to compute this quantitatively. The Case Study pointed out a great deal of uncertainty, and how this could be mitigated. The measurements could be of far greater quality if the mud measured could be transported from the riser prior to the bell nipple, i.e. without being exposed to air prior to measurements.
More case studies would be useful in obtaining a greater insight in the methane diffusion process during drilling. Especially if important parameters were made available, such as mud composition, well parameters, temperatures, and FID measurements of OBM which has not been exposed to air prior to entering the gas trap.
Finally, software developed to dynamically circulate oil based mud and reservoir fluids was used to show the potential danger of ignoring the diffusion effect when the well has been in a “quiet” state for some time.
The methods used in this report ought to be used in planning a well and during drilling, especially for deep water wells. The more one understands about the interaction between oil based mud and reservoir fluid, the less surprises one gets.