Novel centrifugal process removes gas contaminants

Nov. 13, 2006
A new technique based on centrifugal separation can economically process gas from fields contaminated with large amounts of carbon dioxide (CO2) or hydrogen sulfide (H2S).

A new technique based on centrifugal separation can economically process gas from fields contaminated with large amounts of carbon dioxide (CO2) or hydrogen sulfide (H2S).

The process could help mitigate expected shortages in global natural gas supply during the next few decades in an environmentally responsible manner. Current energy-intensive methods often cannot economically remove CO2 or H2S in existing or newly discovered gas fields if the gas contains more than 15% CO2 or H2S.

Selective absorption in an aqueous solution is the standard technique for removing these gaseous contaminants from methane.1 High contamination levels, however, require unacceptable levels of energy consumption for purification. These existing technologies require more energy than the energy in the purified gas.

Absorption and membrane technologies are understood processes that offer no economic prospects for these fields.

Shell research project

A Shell research program investigated the application of novel rotational separation technologies starting with centrifugal gas separation.2-4 It established that applying this process to clean natural gas differs from that used in uranium enrichment, its typical use. A modeling and experimental program, however, also showed that building small compact units for a gas centrifuge device was not feasible because of long separation times required because gaseous diffusion is slow.5

A phase change is the only way to accelerate separation.

The program thus developed the condensing centrifuge, a novel concept based on the gas centrifuge.6 In this device, unlike other devices, the centrifuge operates at elevated pressure so that the CO2 condenses at the centrifuge periphery.

Although this improves separation compared to the purely gas-phase centrifuge, it was too slow to make the process attractive for production sites producing 100-500 MMscfd. The slowness is caused by the interior part of the condensing centrifuge still operating in the gas phase.

Gas-phase centrifugal processes are restricted by the fact that the product of density and diffusion coefficient is constant. Although mass flows increase, operation at high pressures leads to slower separation without improving net overall performance.

A way to accelerate the separation process of the two components was to include a phase change. This involves changing the thermodynamic state of the mixture so that the CO2 becomes a liquid.

Click here to enlarge image

Fig. 1 shows this novel two-step method that includes:

  1. Cooling the gas to a temperature whereby the gaseous contaminant becomes liquid in the form of a micron-sized droplet mist.
  2. Separating the mist from the gas by the rotational particle separator (rps), a device already used in health care and environmental emission control.7-9 The process is a spin-off of the gas centrifuge and is the core innovation in the Shell process.10

High energy demands do not hamper this method, which has short residence times and compact units. It potentially can boost recoverable world gas reserves by amounts that are energetically equivalent to many billion barrels of oil.

Thermodynamics

For economic gas-gas separation, the process has to transform one of the components into a phase capable of forming particles. Cooling and condensation will do this.

Because gas in reservoirs is compressed, for example at 130-450 bar, expansion can attain the necessary cooling for contaminant condensation. Even the reduced pressures available top hole, 80-130 bar, are sufficient to drive expansion cooling.

An expansion turbine is in most cases preferred to techniques employing expansion by acceleration such as the Joule-Thomson process. The turbine can drive a compressor to bring the gas back to system pressure. In addition, after the turbine, the gas velocity can remain relatively slow. Withdrawing power from the gas rather than from gas speed causes the cooling.

This process avoids the risk of heating the gas with internal friction, which would cause the ultrafine condensed droplets to evaporate.

Separation of components with condensation to form a phase boundary is a well established process.11 The method used is temperature reduction, although usually with heat exchange rather than the direct expansion method in the Shell process.

The latter method is standard in such processes as producing cryogenic gases or LNG. The main problem, however, is the spatial separation of the two phases that often are intermixed and difficult to split into separate streams.

The low-pressure side of the expansion refers to a condition that provides sufficient cooling for forming and separating the two phases. The product phase is gaseous, enriched in methane, with depleted CO2. The waste phase is liquid, enriched in CO2, and depleted in methane.

This process, of course, does not spontaneously lead to a nicely separated liquid and gas phase. A mixture of fine CO2-rich droplets forms in a methane-rich gas. Between the expansion and separation steps, the microdroplets will increase in size sufficiently for separation.

This so-called “induction” process also takes place in knockout vessels for removing such components as condensate components with cyclones. These devices, however, need high-volume throughputs for droplet sizes greater than about 15 µm and require much lower mass loadings than are in contaminated gases. Removal of smaller droplets is possible but only for extremely low throughputs, so-called microcyclones.

In the Shell process, droplet sizes will be smaller than this with mass loading much higher than condensate in gases.

Consider a feed flow (Qf) of contaminated gas that is split into a cleaned-up stream (Qc) and a waste stream (Qw) of CO2 rich liquid. Conservation of mass requires that input is the same as output (Equation 1 in accompanying equations box).

Click here to enlarge image

If xi is defined as the concentration of methane in each of the three streams (i = f, c, w), then a mass balance on the methane component yields Equation 2.

The most obvious condition is that the process should have the highest concentration xp of methane possible in the product stream. Simultaneously, the process needs to minimize the loss of incoming feed methane into the waste stream so that the maximum number of molecules of methane in the feed are in the product stream. This corresponds to maximizing the recovery r given by Equation 3.

The recovery is a function of the methane concentrations xi. This is seen by dividing Equation 1 and 2 by Qf and solving for the two variables Qc/Qf and Qw/Qf. Because the input concentration is specified, the recovery is then purely a function of the product and waste concentrations derived from thermodynamics.

The methane content in the clean and waste streams (xc and xw) is obtained from the pressure and temperature (p and T); however, the p and T conditions have a large feasible range with corresponding solutions for xc and xw.

The calculations need to find realistic values that also optimize the recovery r at the same time.

Consider a 50/50 mixture of CH4/CO2, for example xf = 0.5, which is representative for contaminated gas and used as a basis of design. For the optimum values of product concentration xc and methane recovery r for a variety of pressures and temperatures,12 the lower bound for these conditions can be obtained from an evaluation of the phase boundary for pure CO2 in the freeze-out curve in Fig. 2.

Click here to enlarge image

Most materials have a triple point below atmospheric pressure with normal melting and boiling points corresponding to the temperature points at 1 atm where solid-liquid and liquid-gas transitions occur, respectively. For CO2, however, the triple point is above atmospheric pressure.

For CO2, pt is 5.1 bar and Tt is -56.6º C. The subscript t refers to the triple-point conditions.

Fig. 2 shows that operation in the liquid regime requires that the process is greater than 5 bar and -56° C. This forms the minimum value for the thermodynamic conditions.

The maximum value is set by reasonable pressures for the expansion and bearing in mind that this is after the cooling expansion phase.

Click here to enlarge image

Fig. 3 shows the methane recovery r plotted against methane product concentration xc for a range of pressures and temperatures, as obtained from an extended cubic equation of state simulation based on the Soave-Redlich-Kwong model. The figure shows that a single separation step will obtain high methane recoveries; however, the problem requires maximizing the methane concentration in the product stream.

At the ideal point xc ≈ r ≈ 0.85, the turbine inlet pressure would be 600 bar, which is unrealistic. In general, given the restrictions for pipe WT and corresponding safety and handling considerations, the process should have an inlet pressure less than 200 bar.

The question then is to choose the optimal realistic p and T values for the separator operation. With too much expansion, the temperature may be sufficiently low but the pressure will be too low for liquefaction to take place. A restricted expansion, on the other hand, may have sufficient pressure but not enough cooling and hence insufficient yield.

An examination of the various p and T conditions coupled to the parameters previously noted shows that an expansion to 25-30 bar, providing the inlet pressure is greater than 100 bar, gives significant phase separation.

From a practical engineering standpoint, a 102-bar expander inlet pressure is sufficient for recovering about 95% of the methane into the product gas stream with a concentration of about 67%. Note that phase separation is initiated by the expansion and is only complete by the end of the induction period in which the liquid state is materialized in droplets of a few microns in size. Subsequently, the spatial separation of the dispersed waste and purified product takes place in the rotational particle separator.

The foregoing analysis assumes that the CO2-rich liquid is already in the dispersed waste material of flow rate QW that will be spatially concentrated and separated to the waste stream in the rps.

Separation

The previous discussion evaluated the equilibrium conditions required for forming a dense waste phase. The other concern is the kinetics of droplet removal that needs to occur within a very short time and for very small droplet sizes.

In principle, a standard cyclone is capable of doing this; however, this requires a long residence time for very small droplets, something difficult to attain with high throughputs.

The main advantage of the rotational particle separator (rps) is that it enables this to happen much more quickly, at much higher throughputs, and for much smaller droplets than with other centrifugal methods such as cyclones or gas centrifuges.

The core of the rps is a cylindrical body that has a large number of axially oriented channels, each with a 1-2 mm diameter.8 The swirling high gas flow solely induces the rotation of the assembly, which is mounted on bearings. An external motor drive is unnecessary.

A cylindrical housing contains the entire assembly of channels. The unit does not require rotating shafts perforating stationary elements and the associated rotating seal mountings.

Inside the channels, the centrifugal force will move the particles in the gas radially to the outer walls. Because the channels are narrow, particles can arrive at the collecting outer walls during a short residence time. The process squeezes the liquid films formed at the outer walls from the channels and provides exit ports for the liquid to leave the rps (Fig. 4).

Click here to enlarge image

The distance particles travel is much shorter than in a cyclone so that the process can collect much smaller particles. The basis of the separator is thus the use of centrifugal force to separate small, condensed droplets from the gas flow.

The particle radial velocity is calculated from a balance between the centrifugal force and the fluid force exerted on the particle in the case of motion relative to the surrounding carrier fluid.10 We can express the critical separable droplet size as a function of three variables.13 These are:

  1. Flow rate Qf, the feed flow rate of the gas stream.
  2. Residence time in the separator τ, which is effectively a measure of size and thus capital cost.
  3. Specific energy consumption ε, which gives the operating costs.

Fig. 5 compares this last parameter for several CO2 gas stream decontaminations. It is clear that the process compares favorably in terms of energy consumption.

Click here to enlarge image

Currently, the rotational particle separator is proven technology for the spatial separation of micron-sized solid and liquid particulate matter dispersed in gases. It has been used for removing aerosols from domestic air, filtering fly ash from flue gases in coal fired power stations, water removal from natural gas, and inlet gas conditioning in gas turbines.7-9 14 15

Current development programs address scale-up to levels and conditions associated with producing well gas flows.

Pressures and temperatures are designed so that dissolved methane in particles and gaseous pollutant in methane are at a minimum ensuring maximum enrichment and depletion of product and waste streams. Cases of high contamination may require compressing the single-stage product gas and repeating the process to obtain greater purity.

The process occurs under pressure so that the size of the unit as a whole is small. Energy consumption is small, costing a few percent of the calorific value of the gas at most. Capital and operating expenditure are therefore low so that they form no obstacle for profitable exploitation.

A unit for a typical gas field producing methane at 100 kg/sec with 50% CO2 contamination would have a 10-m length and 1-m radius. The rps would have the same radius and a length of 0.5 m. It would rotate at a 50 m/sec peripheral speed. Pressures and temperatures downstream of the turbine are typically 25 bar and -50° C. These values illustrate the attractiveness of the process.

Another major advantage is that whereas standard processes produce CO2 or H2S contaminant at low pressures, the Shell process automatically generates the waste stream at high pressures enabling reinjection back into the gas reservoir from which it originally came, while yielding clean natural gas with a much lower net production of polluting gases.

References

  1. Kohl, A.L., and Nielsen, R.B., Gas Purification, Houston: Gulf Publishing Co., 1997.
  2. Van Wissen, R.J.E., Golombok, M., and Brouwers, J.J.H., “Separation of carbon dioxide and methane in continuous countercurrent gas centrifuges,” Chem. Eng. Sci., Vol. 60 (2005), No. 16, p. 4,397.
  3. Golombok, M., and Morley, C., “Thermodynamic factors governing centrifugal separation of natural gas,” Chem. Eng. Res. Des., Vol. 82 (2004), No. A4, p. 513.
  4. Golombok, M., and Chewter, L., “ Centrifugal separation for cleaning well gas streams,” Ind. Eng. Chem. Res., Vol. 43 (2004), No. 7, p. 1,734.
  5. Golombok, M., and Bil, K., “Removal of CO2 from a gas stream using an experimental centrifuge,” Ind. Eng. Chem. Res., Vol. 44 (2005), No. 13, p. 4,721.
  6. Van Wissen, R., Golombok, M., and Brouwers, J.J.H., “Gas centrifugation with wall condensation,”A.I. Chem. E. J., Vol. 52 (2006), No. 3, p. 1,271.
  7. Brouwers, J.J.H., “Rotational particle separator: a new method for separating fine particles and mists from gases,” Chem. Eng. Tech., Vol. 19 (1996), No. 1.
  8. Brouwers, J.J.H., “Particle collection efficiency of the rotational particle separator,” Powder Tech., Vol. 92 (1997), No. 5, p. 89.
  9. Brouwers, J.J.H., “Phase separation in centrifugal fields with emphasis on the rotational particle separator,” Exp. Thermal & Fluid Science, Vol. 26 (2002), p. 381.
  10. Brouwers, J.J.H., “On compressible flow in a gas centrifuge and its effect on the maximum separative power effect,” Nuclear Tech., Vol. 39 (1978), No. 3, p. 311.
  11. Buonicore, A.J., Theodore L., and Tchobanoglous, G., Waste Management in Perry’s Chemical Engineers’ Handbook, New York: McGraw-Hill, 1984.
  12. Shavit, A., and Gutfinger, C., Thermodynamics: from concepts to applications, London: Prentice Hall, 1995.
  13. Van Wissen, R., Golombok, M., and Brouwers, J.J.H., “In-line centrifugal separation of dispersed phases,” submitted to AIChE Journal.
  14. Mondt, E., van Kemenade, H.P., Brouwers, J.J.H., Bramer, E.A., “Rotating sorbent reactor,” 3rd International Symposium on Two Phase Flow Modelling and Experimentation,” Proceedings, Part III, 2004, Pisa, Italy, p. 1845.
  15. Laagland, G.H.M., Brouwers, J.J.H, and Poorting, J., “The rotating air filter for gas turbines,” Proceedings Power-Gen Europe, Vol. 2 (1996), p. 413.

The authors

Click here to enlarge image

J.J.H. Brouwers (j.j.h.brouwers @tue.nl) is professor and head of process technology in the department of mechanical engineering at Technische Universiteit Eindhoven. He previously ran isotope separation research for URENCO, a joint UK, Dutch, and German ultracentrifuge development project, worked for Shell International, and was a professor of thermal engineering at Twente University. Brouwers has an MSc and a PhD from TU Eindhoven.

Click here to enlarge image

Ralph van Wissen ([email protected]) is finishing his PhD at Technische Universiteit Eindhoven from where he obtained an MSc in mechanical engineering with a specialty in process technology.

Click here to enlarge image

Mike Golombok ([email protected]) is a principal scientist with Shell Exploration and Production. He has previously worked in downstream research in Shell on fuel combustion, gasoline component manufacture, and on steam cracking. Golombok holds a BSc from University of Glasgow and a PhD from University of Toronto. He is a member of the Institute of Chemical Engineers and also a part-time professor in mechanical engineering at Technische Universiteit Eindhoven.