1

Paper prepared for SOGAT2010, Abu Dhabi, UAE 28-03/01-04 2010

Condensed Rotational Separation to upgrade sour gas

J.J.H. (Bert) Brouwers H.P. (Erik) van Kemenade

Eindhoven University of Technology P.O. box 513, 5600 MB, Eindhoven, the Netherlands

e-mail: ptc@tue.nl, www.tue.nl/ptc 1. INTRODUCTION A steadily increasing amount of newly located natural gas fields is severely contaminated with CO2 and/or H2S. Percentages of 30 % CO2/H2S – or even larger – are not uncommon. Fields with such high degrees of contaminant can not be economically exploited by conventional techniques based on amine treatment, as implementation would lead to a huge energy consumption. Moreover, capital costs involved with the erection of the installations would be very large [1]. Thus, there is a need for alternatives which do not suffer from these shortcomings. Condensed rotational separation, abbreviated as CRS [2] forms such an alternative. In addition to ‘‘upstream’’ applications, for purifying natural gas [3], the technique also has potential for removing CO2 from combustion effluent, CO2 from syngas and condensates from natural gas. CRS is a simple and straightforward method to upgrade heavily contaminated gases to light degrees of contamination of only a few percent. The size of the installation is small; the energy required to do the job is only a small fraction of the heating value of the produced methane. CRS is a cheap process which makes heavily contaminated gas fields almost as economic as lightly contaminated fields. The working principle of the patented CRS process is as follows. The sour gas is rapidly lowered in temperature (-50/-80oC) and/or reduced in pressure (20/40 bar) by applying compact coolers and expanders. A mixture forms which consists of predominantly gaseous methane with in it a mist of small micron-sized droplets consisting of predominantly liquid CO2/H2S. The liquid droplets are separated by applying the patented apparatus of the rotational particle separator, abbreviated as RPS (Figure 1). The process is further enhanced by introducing a second step of regeneration. The collected liquid is reduced in pressure (~10/20 bar). In this way most of the methane which was dissolved in the liquid evaporates. The gas has a composition roughly equal to the incoming feed gas. It is re-fed into the gas stream in the first part of the process.

2

Figure 1 Condensed rotational separation

Essential in the process is the availability of the rotational particle separator (RPS). Although not widely known, rapid cooling of binary or multi component mixtures of gases to temperatures where one, or some of the components preferentially condense, leads to a mist of very small droplets with diameters of 1 to 10 micron. The phenomenon is known to occur by aerosol formation in flue gases of biomass combustion installations [4], condensate droplets resulting from cooling of wet natural gas [5] and fog created downstream of windmills, Figure 2. It has also been measured in experiments with Ch4/CO2 mixtures and CO2/N2 mixtures [6,7]. For a process which relies on fast phase change as a means of separation to be economical and practical, it is necessary to have a device capable of capturing micron-sized droplets with high collection efficiency, low pressure drop and small building space. The RPS is such a device In this paper we shall address the important aspects of CRS. In section 2 a description is given of the thermodynamics of the process. Section 3 deals with the rotational particle separator, while in section 4 a presentation is given of the sizing of an installation capable of handling a field of 100 MMscf/day with composition 59 %mole CH4, 27 %mole H2S and 14 %mole CO2 (this corresponds to the Huwaila field in Abu Dhabi). Other applications of the CRS technology are discussed in section 5, conclusions are presented in section 6.

3

Figure 2: Mist formation due to expansion cooling at an offshore wind farm

2. PROCESS THERMODYNAMICS CRS involves cooling of the gas mixture followed by formation and growth of droplets rich in the contaminating substance and subsequent separation of these droplets. Because of binary condensation, a mist of small droplets is formed, which quickly reaches thermodynamic equilibrium (Figure 3). The droplets will then only grow by coagulation, a process that relies on the mobility of the droplets. When droplets grow, the mobility rapidly decreases [8]. Overall this results in micron-sized droplets when a growth time of about 1 second is available.

Figure 3: Mist formation

4

The liquid phase will mostly contain molecules of the species with the lowest partial vapor pressure [9]. The vapor phase will mainly contain molecules of the species with the highest partial vapor pressure. The two phases, liquid and vapor phase, finally end up in equilibrium. The concentration depends on pressure, temperature, and start composition. The multiphase feed (z) enters the phase separator and is divided into an equilibrium gaseous (y) and a liquid fraction (x). The (methane) enrichment is the change in mole fraction (methane) between feed (z) and product (y) stream. The enrichments is therefore an important parameter to define the performance of the separation system for it determines the purity of the product. The ratio between the number of moles of methane in the product stream and the number of moles of methane in the feed stream is called recovery and can be defined as [6]:

)()(

111

111

1

11 xyz

xzyQzQyrF

P

−−

== (1)

where Q is the mole flow rate, with F and P denoting feed and product gas, respectively. xi, yi, and zi are the mole fraction of component i in the liquid waste, product, and feed. With the described process, we can choose between high recoveries and low enrichments or high enrichments at the expense of a lower recovery rate (y1 vs. r1). Normally the higher the pressure the higher the enrichment but at the expense of a lower recovery rate, i.e., because more methane will dissolve in the liquid CO2/H2S. With pure CO2 and CH4 mixtures it is difficult, because of the thermodynamic properties, to reach concentrations exceeding 85 %mol CH4 [6]. When H2S is present in the natural gas, the freeze out temperature of the CO2 shifts to lower temperatures, thereby unleashing possible enrichment up to gas concentrations over 95 %mol at high recovery rates [6]. The thermodynamic performance of the system is compared with thermodynamic predictions and is found to be in good agreement [3]. In general, the lower the pressure at constant temperature, the higher the recovery of methane, but the enrichment diminishes due to the high amounts of CO2 that do not condense. The optimal pressures and temperatures can be found in the lower left corner of the ‘‘liquid and vapor’’ phase, close to the freeze out curve. The optimal and most economical process conditions for real applications have to be determined while incorporating the whole gas treatment facility in the analysis. The phase diagram for a mixture of 59 %mole CH4, 27 %mole H2S and 14 %mole CO2, corresponding to known data for the Huwaila field in Abu Dhabi, is depicted in Figure 4 (continuous line). To obtain this diagram, an extended equation of state program based on a cubic equation of state of the Soave-Redlich-Kwong type with pure component parameters fitted to vapor pressures and liquid densities along with a composition dependent mixing rule is used. A freeze out model T\Tsolid CO2 is incorporated. When the mixture is expanded from the gas phase to a pressure and temperature within the liquid and vapor regime, small liquid droplets are formed. The concentration of each phase as well as several other properties of both phases can be calculated using isothermal flash calculations.

5

Figure 4: Phase diagrams for CH4, H2S and CO2 mixtures.

Calculation of the methane recovery r against methane product concentration xc for a range of pressures and temperatures shows that in a single separation step it is possible to get 95 %mole methane recovery by a combination of isenthalpic cooling and adiabatic expansion to 30 bar and -85 oC (point 1 in Figure 4). The compositions of the resulting vapor and liquid streams are presented in Table 1. Table 1: Compositions after the first separation step Liquid concentration

[%mole] Vapor concentration [%mole]

Total vapor fraction [%mole]

CO2 23.1 3.2 45.7 H2S 48.6 1.3

CH4 28.3 95.5 A considerable amount of the CH4 is still dissolved in the liquid phase however. The best of both worlds, i.e. high enrichment and high recovery is achieved by adopting a two step process. First cooling and expansion is applied to the heavily contaminated gas such that point 1 in Figure 4 is reached. The separated liquid is subsequently treated by a process such that point 2 is reached (Figure 4, dotted line). The gas that evaporates from the liquid when going from 1 to 2 has a composition which is not far from that of the untreated original gas. It is re-fed to the original gas entering the installation. The result is a maximum concentration of methane in the product gas and a minimum concentration of ‘lost’ methane in the liquid waste. The process scheme is shown in Figure 5, constructed using the Aspen Plus modeling tool.

6

It can be seen in the flow diagram of Figure 5 that the regeneration loop also doubles as a refrigeration cycle with HE3 as the condenser and HE2 as the evaporator. The required heat fluxes almost balance each other so they can be integrated in one heat exchanger. The product streams contain ample cooling power (HE5 and HE4) to cool down the feed gas stream (HE1) to the required temperature. An external cooling machine is not necessary. The compressors (COM1 and COM2 in Figure 5) both operate at a pressure ratio <4, implying that standard single stage compressors, as commonly used in the LNG industry, can be procured. It is noted that the addition of the regeneration loop hardly affects the size of the installation as equipment for an external cooling cycle would be of the same size. The energy costs of the cycle can be expressed in percentages of the (higher) heating value of the incoming methane (5.5.107 Jkg-1), with the assumptions that heat to power efficiency is 0.4 and one stage compressor efficiencies are 0.85. The results are presented in Table 2. It should be noted that the result of 3.7 % is conservative as the cycle produces excess cooling energy, indicating that the thermodynamic optimum is not reached Table 2: Energy costs of the cycle in percentage of the heating value of methane

methane in liquid 1.9 % regeneration compressor 0.4 % methane compressor 1.4 % total 3.7 %

3 SEPARATION TECHNOLOGY A key feature of the described setup is the rotational particle separator, designed to separate large amounts of liquid CO2/H2S droplets larger then 1 micron from a semi-cryogenic natural gas stream. The rotating particle separator is basically an axial flow through cyclone with in it a rotating element: Figure 6. The preseparator is constructed with a tangential inlet to provide the rotating flow. To calculate the efficiency of this preseparator, we first look at the efficiency of a gravitational separator. In case of a gravitational separator, the terminal settling velocity (vT) of the particle is used, i.e., the velocity of the particle when drive and drag force are in equilibrium, to predict the performance. The vT for a small particle where viscous forces are dominating (a so called Stokes particle) is described by [8]:

7

Figure 5: Flow diagram of the two-stage separation process.

μρρ

18)( 2gd

v pgpT

−= (2)

with ρp, the density of the particle/droplet; ρg the density of the gas; dp the diameter of the particle/droplet; g, the gravitational acceleration; and μ, the dynamic viscosity of the fluid. The particle diameter that this cyclone can collect with a 50 % probability is called dp,50%, and is analogous to the also used dp,100% which is the droplet size that is collected with a 100 % probability. The dp,50% can be calculated with a relation based on the vT of a particle under influence of a centrifugal force [6]:

dzv

rrvd

tL

gp

waxp 2

0

2%50

2

%50, )()(9

∫−−

=ρρ

μ (3)

with vax the gas velocity in axial direction; vt, the gas velocity in tangential direction; rw, the radius of the wall; r50% the radius half way between the inner and outer wall of the cyclone volute; and L, the length of the separator.

8

Figure 6: Droplet catcher based on the rotational particle separator

The centrifugal acceleration has been described by vt2/r which replaces g in Eqn.2 The tangential velocity is assumed constant within the whole preseparator, because of the stabilizing influence of the rotating barrel. The velocity at the inside of the cyclone is equal to the barrel velocity, Eqn. 3 results in:

Lv

rrvd

tgp

waxp 2

2%50

2

%50, )()(9

ρρμ

−−

= (4)

this yields a dp,50%, which varies between 10 and 30 μm depending on the tangential speed of the separator element and gas flow rate. We use the dp,50% definition instead of the dp,100% definition because the turbulent flow will interfere with the particle collection [9]. As a result of turbulence, it is hard to define an efficiency of 100 %, due to the fact that a small fraction of the droplets theoretically will never reach the wall. From Direct Numerical Simulation (DNS) calculations [10,11], it follows that to achieve 98 % separation efficiency the mono disperse particle diameter dp should be around 3dp,50% or the dp,50% should be chosen 1/3 dp.

9

Figure 7: Coagulation element

In Figure 7, a single channel is depicted rotating around a central axis at a radius r. A droplet that enters the channel on the left side is forced to the outer wall of the channel with a velocity equal to the vT. The distance the droplet has to travel to achieve 50% separation equals the radius of the channel, therefore in that case:

cax

cT dvL

v21

= (5)

where Lc is the length of the channels and dc is the width of the channels. This can be rewritten:

rL

vdd

gp

axcp 2%50, )(

9Ω−

=ρρμ

(6)

where Ω2r is the magnitude of the centrifugal acceleration with Ω the rotational speed of the element (rad/s). Assuming a optimal axial velocity field ( rvax ÷ ) Eqn.6 can be integrated between the inner radius of the coalescer element Ri and the outer radius Ro:

)()1()(

2/2733250iocpf

cf%p, RRLρρ

dηd

−−Ω−=

επφ

(7)

The dp,50% varies between 0.5 and 1 μm depending on the rotational speed of the separator. Summarizing, the preseparator collects droplets from 20 μm upward and the coagulation element collects droplets as small as 0.5–1 μm. Additional separation at stationary conditions is caused by the impactor effect because of the small space between the rotating element and housing at the entrance of the separator. This impactor effect results in good separation efficiencies even at static conditions. Many RPS’s have been designed and tested over the past 15 years [12–19], also for other areas of application: e.g. ash removal from flue gas of combustion installations, air cleaning in domestic appliances, product recovery in pharmaceutical and food industry and oil/water separation. The advantage for the food/pharmaceutical industry is that the separation takes place within a stainless steel environment which is easy to clean. Stainless steel can also operate at high temperatures which makes different processes viable. The oil water separation is mainly aimed at complying with increasing environmental legislation. The air cleaning device is designed for people with

10

allergic/respiratory problem. Illustrations of designs applied in these areas, as well as CRS, are shown in Figure 8.

Figure 8: RPS designs Separation efficiencies have been assessed for a number of separation elements of different size (length, radius, channel, height, etc.) subject to different conditions (angular speed, flow rate, particulate matter, etc.). Particle collection efficiencies were determined by measuring distributions at inlet and outlet using cascade impactors and laser particle counter techniques. For each of the cases the value of dp,50% according to Eq. (7) was calculated. These were subsequently used to generate efficiency distributions as a function of dimensionless particle diameter dp,50%. Results are shown in Figure 9. For reasons of comparison the theoretical curve is shown as well. It is seen that results of measurements are consistent with each other and compare well with theory, also in the case of the CRS demonstration model [20]

Figure 9: Efficiency of the Rotating Particle Separator

11

4 COMPONENT SIZE To get a feeling for the size and costs of the process a conceptual design is presented for a plant treating 30 kg/s (~100 MMscf/day) of contaminated gas with the composition 59 %mole CH4, 27 %mole H2S and 14 %mole CO2. The design procedure for a RPS under those conditions is comprehensively documented in [6,7]. The RPS is placed inside a standard certified pressure pipe. This pipe can then be adapted by welding on all the supply and outlet pipes (Figure 10). The top and bottom side of the RPS use standard blind flanges, which are adapted to fit the bearing housing.

Figure 10: Rotational particle separator design

In [19] it is identified that performance for rotational separations can be expressed in terms of three variables which determine capital and operating costs. These were the flow rate, residence time and specific energy consumption. The results are a function of volume flow and not of mass flow. For the conditions of the first RPS (-85 oC, 30 bar, 55 kg/s) this leads to outer dimensions including outer casing of Ø 800x2000 mm (approximately 1 m3). Although the massflow through the second RPS is smaller (-60 oC, 10 bar, 26 kg/s) also the pressure is lower leading to a higher volume flow and consequently a larger volume (1.5 m3)

12

Estimation of the dimensions of the heat exchangers is based on a multistream, spiral wound type (Figure 11, [21]) as commonly applied in LNG plants. Typical dimensions of the tubes are 10 mm with a spacing (centre to centre) of 15 mm. Reported Reynolds numbers vary between 2000 and 10000’with heating surfaces between 50 – 150 m2/m3. The heat transfer coefficients were calculated as described by [22], typically in the order of 100 Wm-2K-1. For the heat transfer rate we can write

[ ] [ ] )()()( ,, incinhcoldinoutphotinoutp TTkATTcmTTcmQ −=−=−= ε&& (8)

m& denotes the mass flow, cp the heat capacity, k the overall heat transfer coefficient and A the heat exchanging surface. The quantity ε is the effectivity of the heat exchanger defined as

)()()()(

)()()()(

,,min

,,

,,min

,,

incinhp

incoutccp

incinhp

outhinhhp

TTcmTTcm

TTcmTTcm

−

−=

−

−=

&

&

&

&ε (9)

Figure 11: Sketch of a multi-stream spiral heat exchanger [20]

where the denominator, or maximum possible rate of heat transfer, is based on the stream with the smallest (mass-flow rate)(specific heat) product, also known as the minimum thermal-capacity rate and indicated by the subscript “min”. In this case the incoming gas has the minimum thermal capacity rate. The effectivity is a function of the ratio

min)( pr cmC &= / max)( pcm& and the number of transfer units min)/( pcmkANTU &= . For a counterflow heat exchanger we can obtain the relation

13

])1exp[(1

])1exp[(1NTUCCNTUC

rr

r

−−−−

=ε (10)

This relation is plotted in Figure 12. The NTU number relates to the size and thus capital costs of the heat exchanger, the effectiveness to the energy consumption and thus operating costs. In this study we fix the maximum for the NTU number at 4.

Figure 12: relation between NTU and effectiveness

Heat can be exchanged between HE2 and HE3 of Figure 5 in a single heat exchanger. This heat exchanger requires a volume of 30 m3. The cooling energy in the liquid product stream is sufficient to cool the incoming feed gas stream. This heat exchanger has a volume of 50 m3. This leaves the cooling energy in the methane stream available for other uses. If desired this stream could be used to reduce the size of the heat exchangers. An impression of the size of the installation is presented in Figure 13.

5 OTHER APPLICATIONS CRS is also investigated for the following applications:

- removal of condensate from natural gas - removal of CO2 from coal gasification producer gas - removal of CO2 from coal combustion gas

These applications are shortly discussed below.

Natural gas condensate is a low-density mixture of hydrocarbon liquids that are present as gaseous components in the raw natural gas produced from many natural gas fields. A conventional configuration for processing is depicted in Figure 14. The incoming gas is cooled to below the dewpoint of the hydrocarbon’s of the feedstock pressure. In the separator a good part of the hydrocarbons are removed. Subsequently the gas condensate is throttled to a low pressure separator. The reduction in pressure across the throttle valve causes the condensate to undergo a partial vaporization where even more hydrocarbons can be removed. The CRS technology as described above can directly be applied to this

14

process with as compelling advantages: higher separation efficiency, smaller installation and lower energy usage.

Figure 13: conceptual design of a 100 MMscf/day installation (on scale)

Figure 14: Flow configuration for condensate removal.

15

The flow diagram for CRS application in coal gasification plants is schematically shown in Figure 15. Due to the lower operating pressures, the main cooling of the gas in this case is by condensing heat exchangers. A feasibility study did show however that CRS technology becomes competitive with conventional technology at gasifier pressures of approximately 40 bar and larger. This is in line with current developments in gasification technology. Application of CRS to conventional coal combustion power plants show that large improvements can be made in separation efficiency when combined with a relatively small oxygen enrichment of the air. The energy costs of separation are below those of existing carbon capture technologies, although low energy oxygen enrichment technologies need to be found. The key factor of the attractiveness of this technology is finding cheap methods for slight oxygen enrichment. 6 CONCLUSIONS The capability of collecting micron sized droplets from gas streams by the rotational particle separator enables new process designs for

- upgrading sour gas fields - collecting valuable condensate from wet gas - CO2 removal in coal conversion

The key elements of the process have been demonstrated at lab scale and semi-industrial scale. The next step is to start field tests.

Figure 15: Flow configuration CRS application in a coal gasifier

16

LITERATURE CITED 1. Kohl A.L, Nielsen R.B., Gas Purification, 5th ed. Houston, TX: Gulf Professional

Publishing, 1997. 2. Brouwers J.J.H., van Wissen R.J.E., Golombok M. Novel centrifugal process removes

gas contaminants, Oil Gas J. 2006;104:37– 41. 3. Willems, G.P., Golombok M., Tesselaar, G., Brouwers J.J.H., Condensed Rotational

Separation of CO2 from Natural Gas, AIChE 2010;56;1:150-159 4. de Best C.J.J.M., van Kemenade H.P. , Brunner T., Obernberger I., Particulate

Emission Reduction in Small-Scale Biomass Combustion Plants by a Condensing Heat Exchanger, Energy and Fuels, 22(1), 587–597, (2008)

5. Austrheim A.; Experimental Characterization of High-Pressure Natural Gas Scrubbers, PhD thesis, University of Bergen, Norway, 2006

6. van Wissen R.J.E. Centrifugal Separation for Cleaning Well Gas Streams: From Concept to Prototype, PhD thesis, Eindhoven University of Technology, 2006, the Netherlands

7. Willems, G.P., Condensed rotational cleaning of natural gas, PhD-thesis, Eindhoven University of Technology, 2009

8. Hinds W.C. Aerosol Technology. New York: John Wiley & Sons, 1999. 9. Perry R.H. Perry’s Chemical Engineers’ Handbook, 7th ed. New York: McGraw-Hill,

1999. 10. Kuerten J.G.M. ,van Esch B.P.M.,. van Kemenade H.P, Brouwers J.J.H., The effect of

turbulence on the efficiency of the rotational phase separator, Int. J. Heat Fluid Flow, 28, 630-637, (2006)

11. van Esch B.P.M, Kuerten J.G.M. Direct numerical simulation of the motion of particles in rotating pipe flow. J Turbulence. 2008;9(4):1–17.

12. Brouwers J.J.H. Particle collection effciency of the rotational particle separator. Powder Technol. 1997;92:89–99.

13. Brouwers J.J.H.. Rotierende partikelabscheider als neues verfahren für die abscheidung von feinstaub und nebel aus gasen. Chem Ingen Techn. 1995;67:994–997.

14. Brouwers J.J.H.. Secondary flows and particle centrifugation in slightly tilted rotating pipes. Appl Sci Res. 1995;55:95–105.

15. Brouwers J.J.H.. Rotational particle separator: A new method for separating fine particles and mists from gases. Chem Eng Technol. 1996;19:1–10.

16. Brouwers J.J.H.H. Phase separation in centrifugal fields with emphasis on the rotational separator. Exp Therm Fluid Sci. 2002;26: 325–334.

17. van Kemenade H.P. ,Mondt E., Hendriks A.J.A.M., Verbeek P.H.J., Liquid-Phase Separation with the Rotational Particle Separator, Chem. Eng. Techn., 26(11), 1176-1183, (2003)

18. Mondt E. ,van Kemenade H.P., Schook R., Operating performance of a naturally driven Rotational Particle Separator, Chem. Eng. Techn., 29(3), 375-383, (2006)

19. van Wissen RJE, Brouwers JJH, Golombok M. In-line centrifugal separation of dispersed phases. AIChE J. 2007;53:374–380.

17

20. Willems, G.P., Kroes J.P., Golombok M. , van Esch B.P.M., van Kemenade H.P., Brouwers J.J.H., Performance of a Novel Rotating Gas-Liquid Separator, J. Fluid Eng. 2010, in print

21. Linde A.G., Catalogue entitled Rohrbündel-Wärmeaustaucher, Linde A.G., Werksgruppe, TVT, Munich, Germany.

22. Neeraas, B.A., Fredheim, A.O., Aunan, B., Experimental data and model for heat transfer, in liquid falling film flow on shell side, for spiral-wound LNG heat exchanger, Int. J. Heat and Mass Transfer, 2004;47; 3565-3572

Where available, references to the cited literature are provided at: http://tinyurl.com/ptcsogat The authors are indebted to the management of Romico Hold to present and publish the work on the process of condensed rotational separation.