Uncertainties remain in predicting paraffin deposition

Alex Hunt
Texaco Ltd.
London

Current methods for predicting paraffin deposition in flow lines are useful for providing a design engineer with a feel for the magnitude of the potential problem, but a number of uncertainties remain.

This conclusion of a two-part series, that began in OGJ July 15, describes paraffin, asphaltene, and multiphase flow problems in subsea flow lines, and the available computer simulators for predicting paraffin formation.

Predicting paraffin

When predicting the onset of hydrate formation in flow lines, no effort is generally made to estimate the quantity of hydrates that could be formed. This is because blockages tend to be localized, with the crystals being carried along until reaching a point at which agglomeration starts.

In principle, therefore, once there is sufficient hydrate to form a plug, then potential problems could arise and the quantity of hydrate produced is unimportant.

Wax or paraffin production is a different matter, however, because it may be deposited on any cooled surface that falls below the wax appearance temperature or cloud point.

Deposition and build-up may occur anywhere within the production system that is cool enough, while hydrate plugging tends to be a local phenomenon. Thus, unlike hydrates, the quantity and nature of the wax that may be deposited is important, because this impacts the operating strategy that may be needed, such as chemical injection or routine pigging.

WAT

Wax is not a single compound (Fig. 1 [17739 bytes]).^{1 3} The term covers a wide range of different components. In general, waxes are higher molecular weight paraffinic compounds with carbon chain lengths ranging from C15 to C70+.

These compounds are soluble in the lighter paraffinic and aromatic ends present in a typical hydrocarbon liquid (oil or condensate) with the high molecular weight compounds being the least soluble.

Unlike hydrates, paraffin solubility is dominated by temperature, with water cut having very little effect. As the liquid mixture cools, each paraffin component becomes less soluble until reaching a temperature at which the highest molecular weight paraffins precipitate. In the past, this temperature has been referred to as the cloud point, but the current term is wax appearance temperature (WAT).

Previously, it was always considered acceptable to determine the WAT by laboratory measurement using "dead" samples. The relatively simple procedure involved cooling a sample until paraffins started to appear. This point was called the cloud point because the samples tended to become opaque at this temperature.

The DeepStar project surveyed different cloud point techniques used by oil companies and commercial laboratories³ and sent out three different samples for analysis. Results showed cloud point differences of at least 10° F. This was felt to be unsatisfactory.

To date, a "right answer" does not exist, although certain procedures have been abandoned. The DeepStar participants are now trying to standardize on a procedure and on what point should be regarded as the WAT.

Pressure can have some effect on WAT, because at higher pressures the lighter fractions are compressed proportionately more than the heavier ones. With less apparent solvent, the paraffins become less soluble. Therefore, the WAT tends to be higher at elevated pressures.

Nature of deposits

Another uncertainty now being addressed concerns the nature of the paraffin deposits. This becomes important in formulating a wax management strategy.

Soft deposits may readily be sheared from the pipe wall and carried along in the flowing fluid, such that the deposits may not cause significant problems during normal operation. Chemical injection may well be adequate for control.

In multiphase systems, the flow regime may also help provide some shear forces for detaching deposits from the pipe wall. Hard deposits may, however, need regular pigging to remove them.

Because the oil industry has been unable to agree on what constitutes the WAT, it is not surprising that few good predictive tools are available for estimating WAT. A further complication is the tendency of some paraffins to "age," becoming harder over time.

Predictive programs

In general, all current predictive programs use a thermodynamic approach based on regular solution theory, and some then tune the results based on proprietary experimental data. However, PVT report data are inadequate for good predictions. A typical report gives compositional data up to C10+ or, in some cases, up to C20+.

Because of the nature of the wax components, good predictions are only obtained if characterization data are available for at least C50+ and possibly as high as C70+. If these components are only a small part of a typical crude oil, a relatively large crude sample is required to obtain a characterization.

Overall, available predictive tools for estimating WAT are acceptable within the limits of current experimental accuracy if suitably detailed characterization is carried out; however, current experimental accuracy is not acceptable at present. Because a tool is only as good as the data on which it is based, a great deal of work is needed to improve the predictions.

To predict deposition, a different approach has to be taken, although results from the thermodynamic models form part of the input data.

Different oil companies are developing proprietary products, and two packages will soon be available commercially. These are: Parasim from AEA Technology PLC, which it hopes to release in the next few months, and D-WAX from Fantoft, launched at the end of 1995.

It is no coincidence that these programs are being marketed by two companies that specialize in transient multiphase flow simulation, because the elements of the models are as follows:⁴

Multiphase flow calculations
Heat transfer and thermal transient calculations
Fluid and wax thermodynamic and physical property calculations
Wax deposition rate calculations.

The Parasim program has been developed from work by a Joint Industry Project sponsored by Brasoil U.K. Ltd., the U.K. Department of Trade & Industry, Elf Enterprise Caledonia Ltd., Lasmo North Sea PLC, Oryx U.K. Energy Co., Phillips Petroleum Co. U.K. Ltd., and Shell U.K. Exploration & Production. AEA Technology is currently seeking approval from the project sponsors to release the program commercially.

All current models, both proprietary and commercial, tend to work in the same way. The deposition rate calculations are based on the molecular diffusion theory with a laboratory scale-up function to account for flow effects.

Transient multiphase flow simulation provides input data on pipe wall and fluid temperature profiles, and the thermodynamic technique for predicting WAT is then used to determine the solid-wax phase behavior.

It has already been highlighted that thermodynamic model accuracy is questionable. There is also some debate on the applicability of deposition rate calculations, because these are based primarily on single-phase homogeneous systems.

The key factor becomes the scale-up function. At present, the industry has very little laboratory and field data on wax deposition rates in multiphase flow regimes, and obtaining these data will take some years, due to the complexity and expense involved.

Given these uncertainties, one might be forgiven for questioning whether there is any point in trying to predict paraffin deposition. However, it is considered that the current predictions are useful in that they provide the design engineer with a feel for the magnitude of the potential problem.

Where they are best able to add value at present is in modeling sensitivities, which enable cost-effective wax management strategies to be developed.

Paraffin management

Three different methods can control paraffin deposition: thermal,⁵ mechanical,⁶ and chemical.⁷ Thermal methods were discussed in Part 1 of this series.

Mechanical methods revolve around pigging, either as part of normal operations or following a solvent wash when a flow line has been shut down and depressurized. However, such techniques are not generally applicable to multiphase flow lines, because they would require subsea pig launchers or dual flow lines to run pigs.

One exception is the foam pig developed by Petrobras⁸ for introduction into a flow line via a gas lift line. This pig passes through variable line diameters, enabling flow lines to be pigged. The pig can be run when the system is producing, but flows need to be reduced to keep the pig within the required velocity range.

Field trials show some interesting results, which Petrobras has confessed to being unable to interpret fully. Most pipelines cleaned with this technique have been shut down beforehand with water serving as the drive fluid.

Almost all deposits removed came out after the pig had been retrieved and the line was being flushed. This result was a surprise, and research is continuing to try to explain this phenomenon. The technique works, even if nobody currently knows exactly why.

Chemical methods to control paraffin deposition are based on injecting low dosages of crystal growth modifiers, which work in a similar manner to chemicals used for hydrate inhibition.

Service companies normally supply chemical inhibitors, and the selection of a suitable inhibitor is usually based on "dead" fluid sample laboratory tests. Choices are made on a case-by-case basis.

A chemical suitable for one duty may not work well in a different application. However, there seems to be little science involved in making these selections, the majority being done by trial and error.

BP Research has recently started the "wax attack" Joint Industry Project to develop an inhibitor package which aims to prevent the paraffin deposition on pipe walls and tubulars. Sponsors are Amerada Hess Ltd., Amoco (U.K.) Exploration Co., BP Exploration Operating Co. Ltd., Chevron U.K. Ltd., Conoco (U.K.) Ltd., Exxon Production Research Co., Marathon Oil U.K. Ltd., Mobil North Sea Ltd., Statoil, and Texaco Britain Ltd. The large number of sponsors gives an indication of the perceived importance of this issue.

Asphaltenes

Similar to paraffins, it is important to predict when asphaltenes are likely to form and also the quantities that may be deposited. However, it should be noted that asphaltenes pose less of a general problem than paraffins.

Asphaltenes only tend to deposit in certain types of crude oils, where the ratio of resins to asphaltenes is low.

Typically, as part of an initial fluid property evaluation, a SARA (saturates, aromatics, resins, and asphaltenes) analysis of a crude oil sample will be performed. If the ratio of resins to asphaltenes is high, then asphaltenes are unlikely to deposit, because they are stabilized by the resins that keep them in solution.

This is because a resin has a polar group at one end and an aliphatic side-chain at the other end of the molecule (Fig. 2 [16190 bytes]). An asphaltene molecule is a rigid plate structure with a number of polar sites and only short aliphatic side-chains (Fig. 3 [16980 bytes]).

As asphaltene molecules agglomerate, the increasing number of polar sites offers locations for the resins to bond. With their longer aliphatic side-chains, they promote stabilization by enabling the resultant colloid to remain in solution (Fig. 4 [19781 bytes]).

Based on the SARA analysis, it is possible to predict whether asphaltenes are likely to cause problems. In most cases, they do not. However, should a potential problem be identified, it is important to quantify it.

As with hydrates, asphaltene solubility is a function of both temperature and pressure. It is therefore possible to produce a hydrate-type asphaltene formation curve based on experimental results. An optical densitometer is normally used to identify the crystallization onset. This technique has been used for several years and is now well understood.

A number of proprietary and commercial computer packages predict asphaltene appearance conditions. One of the best known is CMG-PROP from Computer Modeling Group in Canada. All of these packages use a thermodynamic approach based on polymer solution theory.9 It is this same approach that is being adapted for paraffin prediction programs.

Similar to hydrate predictions, for best performance the tuning of the equations of state is recommended, but detailed fluid characterization is not required. Experience shows that satisfactory predictions for asphaltene appearance conditions are contained in a typical PVT report.

Currently, no model is available for predicting asphaltene deposition, and estimates are based on laboratory measurements. The techniques used are established and compare well with actual field performance; therefore, there has been no demand for such a model. The general view is that such a program would only be needed for a very limited number of crude oils.

For asphaltene management, there are relatively few choices. Obviously, the prime way to prevent deposition is by adding a resin-type solvent that assists in maintaining the asphaltenes in solution. Deposition, when it occurs, tends to be in well bore tubing rather than in the flow lines. Therefore, it is treated in a similar manner to scale.

Once asphaltenes are deposited, they can be very difficult to remove. A single solvent wash often is less than 15% effective. Multiple washes are generally required, making this an expensive and time-consuming process.

Multiphase flow simulation

In the previous discussion on possible problems, highlighted was the importance of establishing the matrix of different operating scenarios. Also, it was shown that multiphase flow may have a significant effect on deposition.

Tools are available to predict the behavior of multiphase flow systems.

For steady-state modeling, a number of PC-based flow line simulation packages are available, such as Pipephase from Simulation Sciences Inc. and Pipesim from Baker Jardine. These have been available for some time and may be considered as relatively mature products.

Transient modeling is a newer area that is important during start-up and shutdown conditions. Two transient flow line simulation packages are currently commercially available: OLGA from Scandpower AS and PLAC from AEA Technology.

OLGA, currently has only a workstation version, although PLAC has both PC and workstation versions. Both programs are considerably less user-friendly than the steady-state simulation programs, primarily because of complex solution algorithms.

These programs have been available commercially for a much shorter time, so that the technology must be considered as significantly less mature. It should also be noted that a new transient program, Tacite from IFP, soon will be commercially available.

Simulating flow line cooldown is a relatively simple problem for these programs because there is no net flow through the system. It is therefore primarily a heat transfer problem. However, restart and warm-up following a shutdown is more complex because flow now has to be taken into account. This results in a changing temperature profile along the flow line with time. These programs were developed to solve such problems.

Data from these simulations enable cooldown and warm-up times to be estimated (Fig. 5 [21986 bytes]). When these results are combined with the predictions for required inhibitor concentrations, it is possible to estimate volume requirements and hence costs. The results also enable development of operational restart procedures based on different shutdown durations.

Future needs

By now, it should be evident that much progress has been made, both in the development of predictive tools and in improved inhibitors. However, predictive computer programs are only as good as the data that have been used to calibrate them.

The models that are currently available for hydrate prediction are reasonably accurate with correctly tuned equations of state. Only a fundamentally different approach is likely to improve them significantly and this is unlikely in the near future.

Work is currently under way to validate the performance of threshold hydrate inhibitors. Field trials with gas systems have been successful, and inhibitors for such applications are now available commercially. For oil systems, if the current validations are successful, suitable inhibitors should become available over the next year or so.

Paraffin is a different matter. The predictive tools are really only suitable for identifying trends at present, primarily because the quality of base data is poor. Similarly, the inhibitor performance is variable. A great deal of work is in hand to try and turn this technology from an art into a science, but it is unlikely that any major changes will be seen for at least 2 years. However, at least the oil industry is now aware that this is a major problem area and is taking concerted action to try to address this.

The prediction of asphaltene appearance conditions is considered acceptable at present and there are few plans for further work in this area, primarily due to the limited number of relevant applications. It is considered to be a "niche market" at present and is unlikely to see much further progress unless a significant demand arises within the industry.

Multiphase flow simulation tools are commercially available, although, for transient analysis, they are not easy to use. Future developments are focusing on making them more user-friendly, an aim which is being helped by the ever-increasing power available on desktop computers.

A substantial amount of work is also scheduled for gathering field data for comparing actual transient behavior with that predicted by programs.

Benchmarking of the programs will also form part of this work. Statoil is working closely with Scand power, while BP, among others, is supplying data to AEA Technology.

The programs also require further development in the areas of three phase (gas/oil/water) systems, slug tracking, and horizontal/vertical flow transitions, although, once again, very little field data are available to verify simulation results.

Three-phase models, such as W-Olga will reach the market in the next few months and slug tracking models, such as the Fantoft D-Spice program, are already available. Texaco is using D-Spice for modeling the Erskine field multiphase pipeline to gain some experience with the product and its capabilities.

One final area that is not currently being addressed and will need to be considered in the future concerns the development of overall inhibition strategies. To date, hydrates, paraffins, and other depositional problems are looked at and dealt with in isolation.

It is now becoming apparent that injected chemicals may interact and that injection of one can seriously degrade the performance of another.

Until the specific problems are better understood, it is rather ambitious to suggest that interactions should be examined, but these are likely to be the goals for long-term work in this area.

Acknowledgments

The author would like to thank Texaco Inc. for permission to publish this two-part series, and to thank all those who have supplied background information on research and development efforts and their results and current status.

The views expressed in this article are those of the author. They do not represent a Texaco corporate viewpoint.

References

1. Forsdyke, I.N., "Wax Deposition-Design and Management in Multiphase Systems," Advances in Multiphase Operations Offshore Conference, London, Nov. 29-30, 1995.

2. Paraffin Waxes in Subsea and Floating Production Systems, DeepStar II Project Report No. DSII CTR 934-1, June 1994.

3. Cloud Point Round Robin, DeepStar II Project Report No. DSII CTR 902-1, July 1995.

4. Dawson, S., "Simulation Of Wax Deposition-Case Study," Advances in Multiphase Operations Offshore Conference, London, Nov. 29-30, 1995.

5. Thermal Methods to Control Paraffin Deposits, DeepStar II Project Report No. DSII CTR 932-1, June 1994.

6. Paraffin and Asphaltene Control by Mechanical Means, DeepStar II Project Report No. DSII CTR 932-1, June 1994.

7. Chemical Methods to Control Paraffin Deposition, DeepStar II Project Report No. DSII CTR 931-1, June 1994.

8. Lima, P.C.R., et al., "Cleaning the World's Deepest Flowline Using Foam Pigs," 8th Deep Offshore Technology Conference, Rio de Janeiro, Oct. 30-Nov., 1 1995.

9. Burke, N.E., et al., "Measurement and Modeling of Asphaltene Precipitation from Live Reservoir Fluid Systems," Paper No. SPE 18273, SPE Annual Technical Conference, Houston, Oct. 2-5, 1988.