TECHNOLOGY Condensate kicks pose additional problems for deep well control

R. Szczepanski, B. Edmonds, T.K. Yerlett
Infochem Computer Services Ltd.
London
N.P. Brown, T.A.P. Hamilton
Offshore Safety Division, Health & Safety Executive
London

Modeling condensate and gas kicks in a deep North Sea well helps drillers understand what to expect after prompt kick detection in a straightforward kick circulation program with competent choke handling.

Deep condensate wells pose a difficult problem for well control, because if a condensate kick is not detected promptly, the gas could evolve and expand rapidly from the kick very near the surface. By then, the well may also be loaded with a condensate mixture throughout its length. Riser unloading and high gas concentrations on the rig floor may result before the well can be shut in.

The surface pressures in a condensate kick and expansion in the well bore and around the shoe will nearly always be less than that for a methane kick of the same volume. In some circumstances, kick detection may be delayed for condensate kicks relative to that for methane kicks.

Gas kicks have been recognized as a potential hazard in North Sea operations for many years. With an increasing number of the new developments on the U.K. continental shelf involving condensate fields, the question arises whether the models originally designed for simulating methane kicks are still applicable.

Concerns have been raised in the industry that as a condensate kick is taken and subsequently circulated out of the well, unexpected surface indications may be present. Blowouts, resulting from uncontrolled kicks, could be particularly hazardous for condensate fields as these are often high-pressure, high-temperature operations.

A deeper understanding of the differences, and similarities, between gas and condensate kicks has resulted from an investigation by Infochem Computer Services Ltd., funded by the U.K. Health & Safety Executive (HSE). This study has led to an assessment of the practical significance of any differences for the personnel involved in well control.

Background

Normally, gas or gas condensate enters a well bore only through production equipment under controlled conditions. During drilling, however, unexpected conditions may cause an influx of the reservoir fluid into the well bore. Such an occurrence is known as a "kick" and is potentially dangerous, leading to an uncontrolled release of fluids (blowout) if not detected and controlled.

It is important to be able to detect a kick as early as possible and before it has progressed a long way up the well bore, so that remedial action may be taken. Kick control usually involves release of the accumulated gas through a choke valve (circulation from the well).

A key indicator of a kick is pit gain-an increase in the level of drilling mud in the surface mud pit from the mud in the well bore being displaced by the kick. Delta flow, the difference between the mud flow into the drill pipe and that returning from the annulus, may also be used for early kick detection.

The properties of condensates differ from gases. Condensates contain small but significant amounts of heavier components. An important aspect of this study was to determine if this difference in properties was sufficient to lead to different surface indications and the subsequent sequence of events seen by the drilling operator.

Simulation and modeling

The simulation of the gas and condensate kicks was carried out in two stages.

First, the effects of pressure, temperature, and solubility on the kick volume as it rises up the well bore were modeled.
In the second stage, this fluid was combined with a dynamic kick simulator to study the effects on kick detection and circulation from the well.

PVT modeling

The volumetric behavior of the kick was modeled using a modified Peng-Robinson (PR) cubic equation of state to calculate the volume as a function of temperature, pressure, and fluid composition.1

For modeling condensate kicks, four representative fluids from North Sea wells were selected from data made available by operating companies. These fluids correspond to a light, a medium, and a heavy condensate and a black oil. The fluids had surface gas/oil ratios (GORs) of 1 million, 40,000, 5,000, and 600 million standard cu ft/bbl, respectively. Compositional and pressure-volume-temperature (PVT) laboratory data for each of the fluids were used to fit the parameters for the modified PR equation, which was tuned to match density and vapor pressure.

A typical deep North Sea well configuration was chosen, rather than a specific well (Fig. 1 [17302 bytes]). The temperature profile in the well bore during drilling was modeled using the Holmes and Swift method.2 A mud flow rate of 325 gpm was used, with an assumed geothermal gradient of 2 F./100 ft. Again, the objective was to obtain representative temperature profiles rather than to simulate known wells.

Water-based muds (WBMs) and oil-based muds (OBMs) are commonly used in drilling, although there is an increasing trend towards WBMs and pseudo OBMs for environmental reasons. Both WBMs and OBMs were considered in this work to provide an opportunity to isolate the effects of gas solubility on kick expansion.

When a reservoir fluid mixes with a WBM, there is very little mutual solubility between the water and the hydrocarbon phases. The expansion of the kick may be treated as independent of the mud.

OBM consists of a hydrocarbon-based oil, brine, and various solids such as barite, sand, and clay. Methane and condensates are miscible with the base oil at high pressures, and the solubility of the reservoir fluid in the oil element is composition dependent. The solubility of methane in OBM and its effect on kicks has been extensively studied in previous work.3-5

The volume of a drilling mud hardly changes with depth (because of pressure), but the volume of a gas, once it comes out of solution, changes rapidly with pressure. It is the expansion of the gaseous part of the kick that causes the surface changes in annulus pressure and flow from the well.

The PVT studies took a calculated volume of 1 bbl of kick for methane and each of the different condensates and analyzed how this volume changes as it rises up the well bore.

For WBM, Fig. 2 [22065 bytes] shows the kick volume for each of the condensates relative to the volume of methane at the same depth in the well bore. Although there are differences between the methane and condensate kicks, the surface volume for a condensate kick is always less than or equal to the volume of methane. The heavier the condensate (i.e., the lower the GOR), the smaller the surface volume in relation to methane. Thus, condensates should represent a smaller gas flow at the surface than methane. It was also noted that the majority of the kick expansion for condensates occurs close to the surface, similar to earlier observations for methane.

Modeling kick expansion in OBM is not as simple as in WBM, because the mud is miscible with both gas and condensates. At high pressures, such as bottom hole conditions in a deep well, both methane and condensates behave more like a liquid than a gas and are completely miscible with OBM. As pressure decreases up the well bore, the solubility of the lighter components (mainly methane) also decreases until the gas comes out of solution in the mud at the mixture bubble point. The gas then expands rapidly as the pressure decreases further.

Because kick solubility in OBM depends on composition as well as pressure and temperature, it is necessary to consider the makeup of the mixture formed when the kick enters the well. This mixture makeup will depend on the composition of the condensate, the OBM, and on how much kick enters the well compared to the circulation rate of the drilling mud. Four comparative rates were considered, which for the PVT studies equate to kick sizes of 1.5, 14, and 50 vol % of kick mixed with mud at bottom hole conditions and the condensate alone. A larger kick size equates to a higher relative rate of kick entry.

At high concentrations of the heavy condensate (50%), the presence of the OBM makes no difference to the kick expansion when compared to the pure condensate. For the 14% case, the kick volume remained almost constant until the kick reached 2,000 ft, where there was a sharp change as the gaseous components came out of solution at the bubble point.

This effect was even more marked at low (1.5%) concentrations. At low concentrations, the effective volume decreases slightly as the kick fully dissolves in the OBM and then remains constant until the bubble point of this mixture is reached at about 300 ft. There is then a 100-fold increase up to the surface.

The effective volume of methane kicks of various sizes in OBM shows the same general characteristics; light and medium condensates show intermediate behavior. In all cases, higher concentrations (i.e., higher rates of influx from the formation relative to the mud pumping rate) give earlier warning.

Fig. 3 [23476 bytes] shows the heavy condensate kick volumes for different kick sizes relative to the volume of the same size methane kick. At higher concentrations the behavior of condensate kicks in OBM and WBM are very similar. For small kicks in OBM, however, the condensate and methane volumes remain close only until the gas comes out of solution. This will happen at a higher pressure/greater depth for methane; the condensate also contains heavier components, and the gas will not separate until lower pressures nearer the surface.

Consequently, for a given kick size, surface pressure rise for a condensate kick will be delayed relative to that for a methane kick.

Contamination of OBM is a potential problem with condensate kicks. When a condensate kick occurs in OBM, the condensate dissolves in the base oil. When the surface pressure is reached, the light components in the condensate will have been released in the gas phase, but the heavier components will tend to remain in the base oil. This effectively changes the composition and properties of the OBM, which may need to be treated or disposed of. Contamination of an OBM by condensate may lead to an increased flammability risk.

Dynamic kick modeling

The PVT modeling concentrated on what happened to a single slug of kick fluid as it expanded up the well bore with decreasing hydrostatic pressure. For these results to be useful to drilling engineers, they must be translated into observable surface effects. These surface observables include pit gain, choke pressure (the pressure at the valve where the kick is released), and the shoe pressure.

The shoe pressure is the pressure at the point at which the end of the casing is attached to the rock formation. Below this point the fluids are not contained by the casing, and too high a pressure may result in damage to the formation.

A simple but realistic model, capable of calculating what happens when a kick is taken, has been developed by BP Exploration.6 The model was made available as a computer program, Kicker, which determines kick volume, casing shoe pressure, and choke pressure as a function of time (or volume of mud pumped). The program combines the speed of operation of simple models with most of the physical effects included in packages such as SideKick, software produced as a result of earlier involvement of the HSE in encouraging the development of tools for understanding gas kicks.7

As originally written, Kicker was restricted to gas (methane) kicks in water-based mud. For this work Kicker was extended to include OBM and the Holmes & Swift model, allowing direct comparison of kick simulations with the PVT studies. All calculations relating to solubility and phase behavior were carried out with Infochem's Multiflash program, again as used in the PVT modeling.8

Simulations were performed to observe the differences between methane and condensate kicks for water and oil-based muds as a function of various input parameters, such as influx rate and fluid composition. A simple injection model was used for the gas influx rate. Two rates were considered: a rapid influx of 2 bbl/min (equivalent to 25% by volume) and a slower influx of 0.5 bbl/min (6% by volume). The mud circulation rate was 325 gpm during influx entry and 150 gpm during kick circulation.

With Kicker to model the kick detection stage, a minimum detectable pit gain of 10 bbl was used, and it was assumed that well control measures began immediately when this level was reached. For a deep well, pit gain for a given kick size is insensitive to the type of mud or reservoir fluid present; the pit gain is about the same as the volume of kick injected in all cases. In a deep well the bottom hole pressure (at 1,100 bar for example) is high enough for both methane and condensate to act like incompressible liquids. In either case, 1 bbl of kick displaces 1 bbl of mud, be it WBM or OBM.

After the kick is detected, the blowout preventer is shut, and the kick is circulated out through the choke. The surface pressure as a function of time is monitored by the drill operator during this post-detection phase.

Four cases were examined for each influx rate comparing methane and the heavy condensate in WBM and OBM. The results for the higher influx rate are plotted in Fig. 4 [20765 bytes], which also includes calculations for a black oil influx. The time shown is from the start of the kick, but the pressure is only plotted for the post-detection phase. For all cases detection occurs about 5 min after influx. Detection is followed by an initial drop in pressure as the kick passes the bottom hole assembly, with the consequent volume change in the annulus.

There is little difference between WBM and OBM cases, except that the rise in pressure occurs later for OBM as the gases remain dissolved longer. The time from kick detection to the maximum surface pressure is similar in all cases. The black oil behaves like the heavy condensate, with an increased delay in pressure rise when the bubble point is passed, particularly for OBM. Pressures are higher for methane because the volume of volatile gas is greater. Results for a slower influx, 0.5 bbl, indicate a steeper pressure rise, consistent with the PVT study (Fig. 5 [19927 bytes]). The time for kick detection is significantly longer (20 min), as might be expected, but the time from the start of the kick to the rapid rise in surface pressure is shorter because the kick has traveled farther up the well bore before being detected.

Another important consideration in well control is the casing shoe pressure, as excessive pressure at this point may lead to formation failure. The Kicker simulation suggests that, while this pressure is always slightly lower for condensate kicks than for methane, differences are insignificant.

Simple condensate model

The simulations described above were carried out using full compositional models for the reservoir fluids, which contained 20 or more components. Clearly, computer time for such simulations increases rapidly with the number of components used, and some investigations were carried out to determine if it were possible to represent the condensates with fewer components but still reproduce the essential features of kick behavior.

Initial trials with methane plus one heavier hydrocarbon failed to match the volume and pressure profiles predicted by Kicker for the full compositional models. The use of a ternary mixture of methane, ethane, and hexane met with more success, provided the mixture was made up to match the surface GOR of the particular condensate being modeled.

The three-component model predicts the change in volume of the kick as it rises in the annulus, but overpredicts the pressure as the kick approaches the surface. Further work is required to develop a general condensate model, particularly for kicks in OBM. Preliminary results indicate that it is feasible.

Results

In actual situations, kick fluids are distributed over a greater length of the well bore than considered here.

Free gas bubble rise in mud plays an important part in this distribution, and near the surface the amount of free gas and bubble rise becomes important. How they behave cannot be predicted with the simple model used because Kicker is not designed to model dynamic effects rigorously when the kick is near the surface and does not include bubble rise.

Nevertheless, this study helps to understand what to expect after prompt kick detection in a straightforward kick circulation program with competent choke handling. If a condensate kick is not detected promptly, the first warnings may be rapid evolution and expansion of the gas near the surface. By that time, the well may also be loaded with condensate mixture throughout its length. Riser unloading and high gas concentrations on the rig floor may result before the well can be shut in.

The surface pressures in a condensate kick and expansion in the well bore and around the shoe will nearly always be less than that for a methane kick of the same volume. In some circumstances, kick detection may be delayed for condensate kicks relative to methane kicks.
There is a smooth continuity of kick behavior, moving from a methane kick in WBM to a heavy condensate or oil kick in OBM. As the kick/base oil mixture becomes heavier, there is a corresponding decrease in volatility and bubble point pressure (i.e., rapid expansion happens closer to the surface).
For WBM, the volume of a condensate kick in the well bore is always less than or equal to the volume of a methane kick for a given influx volume. The surface pressure is therefore usually less than that for a methane kick of the same initial volume. For kicks in OBM, the condensate kick volume follows the same trend.
Significant expansion of a kick occurs nearer the surface for condensates than for methane, and when it does occur, due to gas evolution, it is more rapid. The rise in surface pressure shows the same trend. The heavier the condensate and, for OBM, the lower the concentration, the nearer the surface the rapid expansion and pressure rise occurs, as the bubble point pressure is lower.
When a condensate kick occurs in OBM, contamination of the mud with condensate occurs, as only volatile components are released at the surface. This may have consequences for mud/gas separator design and arrangements for disposal of contaminated mud.
It is probably not necessary to use a full compositional model for condensates to predict the behavior of a kick for engineering purposes. A relatively simple three-component model based on GOR appears to be adequate for this purpose, at least for kicks in WBM.

Acknowledgment

This work was funded by the Offshore Safety Division of the Health & Safety Executive. The authors would like to thank the following individuals who assisted during the course of this work: M. Grist of the DTI and C. Lockyear, Y. Lou, A. Steven, and B.W. Swanson of BP Exploration.

References

1. Peng, D.Y., and Robinson, D.B., "A new two-constant equation of state," Industrial Engineering Chemistry Fundamentals, Vol. 15, No. 59, 1976.

2. Holmes, C.S., and Swift, S.C., "Calculation of Circulating Mud Temperatures," Journal of Petroleum Technology, June 1970.

3. Infochem Computer Services Ltd., "Review of Thermodynamics in the Gas Kick R-Model," final report for the HSE, Contract MATSU/

8568/3084, July 13, 1994.

4. Thomas, C., Lea, J.F., and Turek, E.A., "Gas Solubility in Oil-Based Drilling Fluids: Effects on Kick Detection," Journal of Petroleum Technology, June 1984.

5. Swanson, B., Gilvary, B., and McEwan, F., "Experimental Measurement and Modelling of Gas Solubility in Invert Emulsion Drilling Fluids Explains Surface Observations During Kicks," Society of Petroleum Engineers paper 18371, 1988.

6. Jackson, C., Report on Industrial Placement Year at BP Research Center, Sunbury-on-Thames, August 1992-July 1993.

7. Hamilton, T.A.P., Swanson, B., and Wand, P., "Use of New Kick Simulator Will Increase Wellsite Safety," World Oil, September 1992, pp. 71-77.

8. Counsell, J.F., Moorwood, R.A.S., and Szczepanski, R., "Calculating Multiphase Equilibria," paper presented at VLE 90, Aston, June 7, 1990.

The Authors

Richard Szczepanski is a director of Infochem Computer Services Ltd., a London-based consultancy and software house specializing in physical properties for the oil and gas sector. Before joining Infochem in 1988, he was leader of the operational thermodynamics project at the British Petroleum Research Centre, Sunbury-on-Thames. Szczepanski holds BSc and PhD degrees in chemical engineering from Imperial College, London University.

Beryl Edmonds is a director of Infochem Computer Services Ltd. in London. He was previously the director of the Physical Property Data Service of the Institution of Chemical Engineers. She holds BSc and PhD degrees in chemistry from Sheffield University.

Trevor Yerlett holds a PhD in chemical thermodynamics from the University of Bristol. He has worked for both British Gas and BP Research, developing engineering software for oil and gas applications. He is currently an independent consultant specializing in computer modeling.

Nigel Brown is a technology development manager in the Offshore Safety Division of the U.K. Health and Safety Executive in London. He holds BSc and PhD degrees in chemical engineering from Imperial College, London University.

Terry Hamilton is a principal inspector in the Offshore Safety Division of the U.K. Health and Safety Executive in London. He holds a BSc degree from Bristol University and an MSc degree from Cranfield Institute of Technology.