BTEX Removal From Natural Gas
BTEX Removal From Natural Gas
BTEX Removal From Natural Gas
Executive Summary
BTEX (benzene, toluene, ethylbenzene, and xylene) is present in natural gas streams and is being picked up in amine and glycol dehydration units, being present in the exit CO2 and Water streams. As a result, incineration temperatures need to be increased and hence the cost of compliance with EPA emission limits (25 Tons/year) is large. Specifically, in the presence of BTEX the appropriate temperature is 1500 F, whereas when other organics (like methane) are present one can incinerate at 1350 F. In order to cut down the cost of incineration, the removal of BTEX from various streams in the amine unit was researched. Different removal techniques such as minimizing absorption in the amine unit, removing in the amine unit, and removing prior to the amine treatment were researched. The best technique from an economic standpoint was the use of adsorbents in the acid gas stream. This was determined by comparing its cost with the savings produced by reducing the incineration temperature. Adsorbents with the largest adsorption cost to extra incineration cost ratio were silica aerogels (SAG) and macroreticular resins (ionic resins). The aerogels were able to remove 8 ppmv of BTEX in a 575 MMSCFD stream of natural gas from the acid gas stream saving $100,000 when comparing the economics of removal with that of incineration. This was achieved by reducing the buying price of SAG from $37/kg to $34/kg at 14 cycles. Similar savings were found for ionic resins used in the acid gas stream by reducing the price of the resins from $43/kg to $35/kg. Both of these reductions could be achievable with long-term contracts with Cabot and Dow respectively. If the number of cycles is larger, then savings will further increase.
Table of Contents
Executive Summary ........................................................................................................................ 1 Introduction ..................................................................................................................................... 3 Background ..................................................................................................................................... 3 Ability of Simulators to Make Predictions in Amine Units ............................................................ 8 Ability of Simulators to Make Predictions in Glycol Units .......................................................... 20 BTEX Removal Solutions............................................................................................................. 22 CONCLUSIONS........................................................................................................................... 42 References ..................................................................................................................................... 44
Introduction
BTEX components are listed by the EPA in the Clean Air Act of 1990 as some of the 188 hazardous air pollutants. The EPA sets a standard of 25 tons per year for total aromatic compounds emitted in any given plant. There also exists 10 tons per year limit on each individual aromatic compound emitted. The reason for the restrictions are because benzene is a human carcinogen (promotes leukemia), toluene exposure can lead to reproduction or developmental effects, ethylbenzene affects the blood, kidneys, and liver and finally, xylene exposure can affect the central nervous system leading to respiratory and cardiovascular problems. All gas companies must handle the emission of BTEX (benzene, ethylbenzene, toluene, and xylene) and C5+ which are absorbed in both the amine and dehydration units and released to the atmosphere if not incinerated properly. In order to reduce the emission of such compounds, one must incinerate these compounds at elevated temperatures. This report evaluates more economical ways to remove BTEX other than incineration.
Background
We first briefly review how amine units work, then discuss the solubility of BTEX in amines and its estimation, the emission limits of BTEX, and the current cost of incinerating the BTEX emissions from amine unit. In the next section we discuss means of capturing the BTEX instead of incinerating it. Amine Treating Units: Typical natural gas has the compositions of Table 1. Table 1: Typical Composition of Natural Gas
Compound Methane CH4 Ethane C2H6 Propane C3H8 Butane C4H10 Carbon Dioxide CO2 Hydrogen Sulfide H2S Nitrogen N2 Oxygen O2 Rare gases A, He, Ne, Xe Composition 70-90% 0-20% 0-8% 0-5% 0-5% 0-0.2% trace
In order to achieve pipeline dry gas quality from natural gas off the wellhead, the compounds must be separated from the methane. This is done by processing the gas through a series of processing plants that separate (when needed) the carbon dioxide and hydrogen sulfide, the water, the nitrogen, and finally, natural gas liquids (C2+). Carbon dioxide is removed from natural gas to increase the BTU value of the gas as well as to avoid the formation of carbonic acid, which will corrode pipelines while transporting the natural gas. The standard value for 3
dioxide in tre eated gas is 2% by volu ume, unless it is transpo orted in LNG form, in w G which carbon d case the value is a lot lower. In turn, hyd drogen sulfi is remov to prev ide ved vent corrosio in on pipelines and reduce health risk s, ks. We foc mostly o amine un and we touch briefl on cus on nits fly glycol un where similar solutio can be im nits ons mplemented. Amine tr reating plants are the m most popular processes used to rem r move the car rbon dioxide and e hydrogen sulfide. Th process of amine trea n he ating is show in Figure 1: wn
Figure 1: Amine Tr reating Unit (Bullin (20 004)) Gas with CO2 and hy h ydrogen sulf contents above 5.7 m per cubic foot of nat fide s mg tural gas is c called sour gas It enters the amine u contactor where is exposed to a lean amin s. unit ne-water mix xture, which is virtually CO2 and H2S free. Thi amine-wa mixture absorbs ma C is ater ajority of ca arbon dioxide a hydroge sulfide. The natura gas leaves the contac and en al ctor as swe gas, fre of eet ee hydrogen sulfide and with low C n d CO2 contents while the rich (in C and H2S amine sol s, CO2 S) lution is sent to a regenerati unit. ion The rich amine is first flashed at a lower pressure se h d r eparate the v vapor. The liquid mixtu is ure then sent the regener t rator where i is heated to high tempe it o erature to re emove the co ontaminants. The liquid am mine is then sent back to the contact while the CO2 and H rich va o tor e H2S apors are rele eased from the regenerator as acid gas. This acid g is then se to an incinerator whe it is burn to gas ent ere ned remove v volatile orga anics and eve entually harm contam mful minants such as BTEX or it can be se to r ent a Claus u to recover the sulfur unit r. The type of amines are used in industry are monoethan es e nolamine (M MEA), diethan nolamine (D DEA), di-isopro opanolamine (DIPA), d diglycolamin (DGA), and methy diethanola ne yl amine (MD DEA). These am mines react with carbon d w dioxide and hydrogen su ulfide as show in Figure 2 wn e
Figure 2: Amine R Reactions (Sh hiveler, (200 05)) On top o the above mentioned reactions, a of e d amines also participate in some sid reactions with de other com mponents (C Sulfur c CO, compounds, Strong acid present in the gas) (A ds n Abdi) and le to ead the degr radation of amines. T f Thermal deg gradation h happens as a result o high reb of boiler temperatu ures. Finally reactions with carbon monoxide a sulfur ca form form compoun y, n and an myl nds The react tion with str rong acids, w which enter t through inlet stream (Ab t bdi), cause h stable sa to heat alts form. T These salts can remain i the system and inter c in m rrupt the abs sorption pro ocess. An a amine reaction w strong acids is show in Figure 3: with wn e
Incinerat tion: Incine eration invo olves the com mbustion of the acid gas and fuel. The oxidiz is f . zer shown in Figure 4. This gas is then contacted with a flame at the appropriat temperatu to n s te ure fully com mbust the gas. These ox xidizers run a temperatu from 100 at ures 00F up to 1 1500F for h highly toxic VO OCs. The cl lean exhaus gas contai mostly C 2 and H2O with some r st ins CO residual CO.
Combustion is a sequence of exothermic reactions between fuel and oxygen, usually in the form of air. These reactions break down the fuel into carbon dioxide and water releasing mass amounts of heat. This same type of combustion happens in acid gas incineration in amine plants. The reactions that take place are the following:
CH
4
+ 2 O 2 CO 2 + 2 H 2 O
C 2 H 6 + 3 . 5 O 2 2 CO 2 + 3 H 2 O C 3 H 8 + 5 O 2 3 CO 2 + 4 H 2 O C 4 H 10 + 6 . 5 O 2 4 CO 2 + 5 H 2 O C 5 H 12 + 8 O 2 5 CO 2 + 6 H 2 O C 6 H 14 + 9 . 5 O 2 6 CO 2 + 7 H 2 O C 7 H 16 + 11 O 2 7 CO 2 + 8 H 2 O C 6 H 6 + 7 . 5 O 2 6 CO 2 + 3 H 2 O C 7 H 8 + 9 O 2 7 CO 2 + 4 H 2 O C 8 H 10 + 10 . 5 O 2 8 CO 2 + 5 H 2 O
Figure 5: Reactions in oxidizer In order for VOC to combust, certain temperatures must be reached. These temperatures are given in Table 2: Table 2: Auto ignition temperatures Compound Benzene Toluene Xylene Methane Ethane Propane Butane I-Pentane N-Hexane AIT (F) 1097 997 924 997 859 842 788 788 437
Ethylbenzene 870
N-Heptane 419 * (Lewandowski, 2000) In order to combust at these auto ignition temperatures (AIT), long residences times must be used in the flame. In order to reduce this residence time to around .5 seconds for 99% 6
combustion, an additional 400F is needed for about 99% combustion of VOCs. This combustion is done in thermal oxidizers. Thus, the temperature for incinerating BTEX needs to be near 1500F. The temperature of incineration for acid gas when no BTEX is present is within the normal operating temperatures of the oxidizer since the ignition temperature for other compounds such as I-Pentane is 788F. Therefore, this temperature is around 1350F. The EPA also monitors the amount of carbon monoxide and NOx admitted into the air by certain facilities. Carbon monoxide affects the respiratory system and in high levels can cause fatality. The EPA is trying to reduce these emissions but has set no real limits on it. NOx are monitored due to its affect of ground level ozone which can cause acid rain, affect global warning, and cause health problems such as respiratory issues. The overall goal of EPA is to reduce total NOx emissions to below 1.2 million tons in the US. In order to restrict formation of these compounds, the amount of oxygen in the outlet of the oxidizer must be between 1-3%. The lower lime of 1% is set to reduce the amount of carbon monoxide that forms below this oxygen content. The upper limit of 3% is set to reduce the amount of NOx formed since these compounds are readily formed above this level of oxygen. Plant data from Williams Ignacio plant are shown in Table 3: Table 3: Ignacio Plant Data for Oxidizer Air Fuel 355,888 23,162 Flowrate (ft3/hr) 0% 100% Methane 0% 0% Carbon Dioxide 78.11% 0% Nitrogen 20.95% 0% Oxygen 0.93% 0% Argon 0% 0% Water
This plant only incinerates the Acid Gas. The exit from the glycol unit is not burned but instead it is separated, and the BTEX is then sold. This data was validated with Pro-II and hand calculations to see if they are reproducible. Running Pro-II with the same inlets gave a temperature of 1421.21F which is a 1.3% difference between the 1440F reported in Ignacio data. The temperature calculated from this was 1430.72F which has a 0.64% difference between real data and calculated data.
AmineCalc and Pro II use the same set of chemical reactions to calculate the equilibrium constants. These are:
H2O K1 H + + OH
8
(3)
H2 S K2 H + + HS
HS K3 H + + S2
Both AmineCalc and Pro II use different equations of state (EOS). Aminecalc uses the PengRobinson EOS:
P=
RT a V b V V +b +b V b
) (
) (
(9)
Some of the problems with the EOS are the mixing rules used for the value of a. AmineCalc uses simple cubic mixing rules for calculating the values of a, which have been shown to be incapable of modeling real systems. In turn, ProII uses the SRK EOS:
P=
( ) ( )
cij xi 2 ( k ) + (k k ) 1 ij ij ji xi + xi
RT a V b V V +b
(10)
This equation uses a different method for calculating the values of a, and these come from the study performed by Mathias-Klotz-Prausnitz. The equation used to calculate the value of a is:
aij = (ai a j )
(11)
This mixing rule has been shown to perform better than simple cubic mixing rules, but it is still not considered to be the best. Finally Pro-II and AmineCalc recommend certain ranges for pressure, temperatures and amine concentrations (Tables 4 and 5). In addition Pro-II does not handle certain amines with high accuracy. Mixtures of amines are discouraged with this program. DIPA amine data is not good enough to scale up equipment around.
Table 4: Ranges of Pro-II for amine units MEA DEA DGA DIPA MDEA 25-500 100-1000 Pressure, psig <275 Temperature, F Concentration, 15-25 25-35 55-65 50 30 wt% amine Acid gas loading, 0.5-0.6 0.45 0.5 0.4 gmole gas/gmole amine
Contactor Stages Temperature, F Concentration, 0-30 wt% amine Acid gas loading, gmole gas/gmole amine *(AMINE Calc Users Manual (1999))
Table 5: Ranges of AMINECalc MEA DEA DGA 22 77-260 0-50 50-75 < 1.0
TEA
MDEA
0-50
0-50
Simulation Results: Data from a Williamss plant was used for the simulations. The inlet raw gas had a flow rate of 575 MM SCFD, 85 F, and 500 psia. The amine flow rate was set at 702,176.88 gal/hr. The results from the simulations are shown in table 6, and this table will be used to compare the ability of the simulators to predict the ability of amines to pick-up carbon dioxide. Field data from amine units from Williams has reported that the acid gas stream contains approximately 92 wt% of carbon dioxide. As seen in table 6, AmineCalc predicts that the acid gas stream contains 99 wt% of carbon dioxide, whereas Pro II predicts that the acid gas stream contains 94 wt% of the carbon dioxide. Thus, AmineCalc is over predicting the amount of carbon dioxide present in the acid gas stream. Also, as mandated by the EPA the sweet gas should contain less than 2 volume% of carbon dioxide present, and from table 6, it can be seen that the carbon dioxide present in the sweet gas stream is below the pipeline design guidelines.
10
CO2 Methane Ethane Propane i-Butane n-Butane i-Pentane n-Pentane Hexane Heptane Octane Nonane Benzene Toluene Ethylbenzene Xylene N2 H2O MDEA
9.37 89.57 0.746 0.13 0.025 0.025 0.046 0.005 0.009 0.005 0.01 0.008 0.0004 0.0005 0 0.0002 0.05 0 0
3.12E-09 9.83E+01 0.816 0.143 0.0275 0.0272 0.0505 0.0055 0.00992 0.00552 0.011 0.00884 2.40E-05 0.000311 0 9.14E-05 9.14E-05 9.14E-05 9.14E-05
1.330 96.710 0.804 0.140 2.698E-02 2.694E-02 4.964E-02 5.397E-03 9.722E-03 5.403E-03 1.081E-02 8.650E-03 4.020E-04 5.219E-04 0 2.064E-04 5.384E-02 0.814 1.154E-04
99.9500 5.00E-02 0.0004 0.00006 0 0 0 0 7.11E-07 0 0 0.00E+00 5.00E-05 1.00E-05 0.00E+00 0.00E+00 0.00E+00 0.00E+00 0.00E+00
87.050 0.392 8.521E-03 1.364E-03 1.708E-04 4.588E-04 3.659E-04 3.537E-05 2.961E-05 1.042E-05 1.235E-05 2.308E-06 2.850E-04 1.655E-04 8.99E-05 1.04E-03 12.55 7.15E-17
We conclude that Pro II and actual BTEX show different liquid concentrations, but further experiments should be included in order to determine which simulator has a better prediction since no hard data is available. AmineCalc, on the other hand, is overestimating the amount of BTEX. Therefore, it can be concluded that AmineCalc and Pro II, or for that matter no simulator contains the right thermodynamics to deal with the presence of BTEX in the amine unit. BTEX Solubility calculation in Contactor and Regenerator Tower: To obtain credible BTEX concentrations obtained experimental data is needed. Coquelet et. al. 2007, has calculated the values of the activity coefficients for BTEX, and these values will be used to calculate how much BTEX should be present in the contactor and regenerator tower. Figure 8 shows the solubility of benzene, toluene, and ethylbenzene in DGA as a function of amine concentration and temperature. Figure 9 shows the activity coefficient of the benzene also as a function of amine concentration and temperature. By lowering this activity coefficient, the absorption onto the amine is raised causing more benzene or any other aromatic compound to be adsorbed more rapidly.
11
hows the act tivity coefficients as a function of temperature for toluene and e e Figure 10 and 11 sh zene respect tively: ethylbenz
12
Figure 10. Activity coefficients as function of temperat y s n ture for tolue ene.
to ty ray ontactor, the liquid and v vapor In order t calculate the solubilit of BTEX in the last tr of the co streams c coming out of trays 6 a 5 respec and ctively were calculated. These flow were obta e . ws ained from Pro II by using pseudo-str o reams (L2 a V). Figu 12 show how this was done in the and ure ws simulator and the ps r, seudo stream are shown in the pink d m dotted lines. Data obtain from Pro II is ned o reported in table 7. Now, the pr N rocess taking place at the last tray is essentially a flash as sh g e s hown in figure 12:
13
L2 G V
Table 7. Contacto pseudo-str e or ream results from Pro II. Co ontactorDat ta575MMSC CFDFeedan nd702MGal l/hrMDEA Tray5T(F) T 145 Tra ay5P(psia) 250 G L2 V L1 Ben nzene(mol% %) 4.00E04 8.67E06 4.00E04 9.13E06 4 6 4 6 Tolu uene(mol%) 5.00E04 5.42E06 5.02E04 5.65E06 ) 4 6 4 6 Eben nzene(mol% %) 0 0 0 0 Xyl lene(mol%) 2.00E04 3.16E06 2.02E06 2.97E06 4 6 6 6 FlowRate F 2.86E+07 8.87E+07 2.85E+07 8.88E+07 7 7 7 (gmol/hr) (
14
Fig gure 13. Las tray of con st ntactor tower. By doing a component balance, t following set of equa g the g ations was ob btained:
L1 G Gy BTEX + L2 xBTEX = Vy BTEX + L1 xBTEX
(12) (13)
sat
sat
L1 G Gy BTEX + L2 xBTEX = V
BTEX (T )xBTEX P
P
+ L1 XBTEX
(14) (15)
xBTEX
xBTEX
L1 G Gy BTEX + L2 xBTEX = (T )P sat + L V BTEX 1 P L1 G Gy BTE + L2 xBTEX EX = 1203.53 3 6.89 89 BTEX (T )*10 T + 219.8 + L1 V P
(16)
The last e equation is used directly by taking th values of flows (G, V L1 and L2) from Pro II. The u y he f V, . reason w this is leg why gitimate is b because the B BTEX conce entrations are very low a their chan e and nges will not a affect these flows. The flow of xBT f TEX is found and used to solve yBT d o TEX. n c e n d vapor In order to calculate the solubility of BTEX in the second tray, the liquid and v streams c coming out of trays 2 an 3 respectiv were ca o nd vely alculated. Figure 13 sho how this was ows s done in t simulato and the ps the or seudo stream are shown in the pink dotted line Data obta m n k es. ained from Pro II is report on table 8. The process taking place at the second tray is essentia a o ted e e y ally flash with the form in h ndicated in F Figure 14:
15
V V L1
Figure 14. Regenerato pseudo-str or reams from simulator. Tab 8. Regene ble erator pseud do-stream res sults from Pr II. ro Co ontactorDat ta575MMSC CFDFeedan nd702MGal l/hrMDEA Tra ay2T(F) 211 Tra ay2P(psia) 15.5 V' L1 V Ben nzene(mol% %) 2.0 01E05 8.67E06 2.85E04 2 Toluene(mol%) ) 1.17E05 5.42E06 1.66E04 1 Ebe enzene(mol% %) 0 0 0 Xyle ene(mol%) 6.3 37E06 3.16E06 8.99E05 8 FlowRate (g mol/hr) 3.7 78E+07 8. .87E+07 2 2.66E+06
16
By doing a component balance, as for the contactor tower, the following set of equations was obtained: (17) V ' yV ' = Vy BTEX L1+BTEX x BTEX Py BTEX = BTEX (T )xBTEX P sat
V' BTEX
V'y
=V
+ L1 xBTEX
(18) (19)
xBTEX =
xBTEX =
(20) + L1
BTEX (T )*10
P
6.89
1203.53 T + 219.89
In tables 9 and 10, the concentrations of BTEX obtained using our equations and experimentally obtained activity coefficients are compared with the results obtained from Pro II for the contactor and regenerator trays respectively. The same verification could not be done in AmineCalc because the flows in tray 5 from the contactor tower and tray 2 in the regenerator tower could not be obtained. Table 9. Contactor Results. BTEXConcentrationfromExperimental Results Contactor Component Calculated ProII Benzene 1.17E06 8.67E06 Toluene 5.99E07 5.42E06 xi EthylBenzene 0 0 Xylene 2.78E07 3.16E06 Benzene 2.46E04 2.85E04 Toluene 1.46E04 1.66E04 yi EthylBenzene 0 0 Xylene 8.10E05 8.99E05
17
Table 10. Regenerator results. BTEXConcentrationfromExperimental Results Regenerator Component Calculated ProII Benzene 4.79E05 9.13E06 Toluene 9.96E05 5.65E06 xi EthylBenzene 0 0 Xylene 5.44E05 2.97E06 Benzene 2.80E04 4.00E04 Toluene 9.55E04 5.02E04 yi EthylBenzene 0 0 Xylene 2.35E04 2.02E06 It is our belief that Pro II produces good answers for flows and CO2 concentrations in the amine unit. It is our suggestion to trust Pro II for the CO2 concentrations. As far as the BTEX, we have proven that neither Pro II nor AmineCalc contains the right thermodynamics for the amine unit. From solubility data, we were able to reproduce the correct concentrations of BTEX in the contactor and regenerator towers. These values will have to be compared to data obtained from the plants.
18
Pro II uses a special thermodynamics package called GLYCOL which uses SRK EOS, with the modification from Mathias-Klotz-Prausnitz added to calculate the binary interaction coefficients; (this is the same EOS used in the amine package). One of the limitations with this thermodynamic package is that some of the binary interaction coefficients for some species are not present, even though some of these components might not be present in significant quantities. 19
The manual recommends a temperature range from 80-400 F, and a pressure limit of 2000 psia. It also uses rigorous tray by tray calculations with appropriate enthalpy calculations. . Glycalc, in turn, makes some modeling assumptions: Around the absorber, the water mass balance can be calculated in two ways: either the number of theoretical stages is given or the amount of dry gas can be specified. If the number of stages is given, the program uses the Kremser-Brown approximation for the absorption of each compound. This approximation is done by using the following equation:
Yn1 Y1 A n1 A = Yn1 Yo A n1 1
(21)
where A is the average absorption factor, Yn-1 is the mole fraction of compound in wet gas entering absorber, Y1 is the mole fraction of compound in dry gas exiting absorber, Yo is the mole fraction of compound in the vapor phase that is in equilibrium with lean glycol, and n is the number of theoretical stages. In turn, the average absorption factor A is given by:
A= L KV
(22)
where L is the average liquid glycol flow rate and V is the average gas flow rate in column. Thus, the contactor tower is not modeled rigorously by using a stage by stage flash calculations, average values of K are used. . For the regenerator, the manual notes that to avoid complex heat and material balances that would be needed if the regenerator were rigorously modeled, a simple empirical calculation is used. Furthermore, the manual adds that the use of stripping gas can also reduce the amount of organics remaining in the lean glycol, however, for the sake of simplicity, GlyCalc does not attempt to determine the impact of stripping gas on the removal of organics from the rich glycol. Therefore, the program assumes that the fraction of each component that is removed from the glycol in the regenerator is assumed to be the same, regardless of whether stripping gas is used or not. Simulator Results: Data obtained from the Williamss Milagro plant was used for the simulations. The inlet gas had a flow rate of 49 MM SCFD, 104 F, and 887 psig. The glycol circulation rate was set at 10 gal/min, at a temperature of 382F. The results from the simulations are shown in table 11. Pro II and GlyCalc produce good answers for flows and close results for H2O concentrations in the glycol unit. Since no real data is available at this moment, no comparison can be drawn at this moment as to what simulator more accurately predicts the concentrations of H2O. GlyCalc 20
produces better results for BTEX concentrations in glycol dehydration unit in comparison to data in table 11. As far as the regenerator results, one would be able to accurately predict how much BTEX is present, since GlyCalc can reproduce the results of the contactor tower. What GlyCalc would not be able to predict is the amount of duty needed in the regenerator since the simulator does not solve the energy balance, but instead uses a simple correlation. Table 11. Results from GlyCalc and Pro II without BTEX. Feed Dry Gas Components (mol%) GlyCalc Pro II Milagro Data Methane 98.8880 98.9 98.880 98.9160 Ethane 0.8164 0.816 0.816 0.7994 Propane 0.1605 0.16 0.160 0.1556 Isobutane 0.0263 2.630E-02 2.619E-02 2.51E-02 n-Butane 0.0262 2.620E-02 2.603E-02 2.53E-02 Neopentane 0.0003 N/A 5.213E-03 3.00E-03 Isopentane 0.0086 N/A 8.581E-03 8.30E-03 n-Pentane 0.0053 5.290E-03 5.286E-03 5.10E-03 2,2-Dimethylbutane 0.0003 N/A 2.838E-04 3.00E-04 2,3-Dimethylbutane 0.0006 N/A 5.524E-04 6.00E-04 2-Methylpentane 0.0017 N/A 1.571E-03 1.60E-03 3-Methylpentane 0.0009 N/A 8.094E-04 9.00E-04 n-Hexane 0.0016 1.600E-03 1.584E-03 1.50E-03 Heptanes 0.0051 5.070E-03 5.029E-03 5.20E-03 Octanes 0.0003 5.920E-03 2.895E-03 5.00E-03 Nonanes 0.0006 N/A 5.582E-04 3.00E-04 Decanes plus 0.0004 N/A 3.524E-04 1.10E-03 Nitrogen 0.0569 5.690E-02 5.687E-02 5.29E-02 Carbon Dioxide 0.0000 0 0 0 Oxygen 0.0000 0 0 0 Water N/A 4.860E-03 5.804E-03 N/A Benzene 3.000E-04 2.720E-04 2.189E-04 3.000E-04 Toluene 5.000E-04 4.220E-04 3.281E-04 4.000E-04 Ethylbenzene 0 0 0 0 Xylene 6.000E-04 3.930E-04 2.459E-04 3.00E-04 2,2,4-Trimethylpentane 1.000E-04 9.980E-05 9.912E-05 1.00E-04 Cyclopentane 0 0 0 0 Cyclohexane 9.000E-04 8.880E-04 8.744E-04 9.00E-04 Methylcyclohexane 1.000E-03 9.840E-04 9.851E-04 1.10E-03
21
All three of these are investigated and discussed below in detail. But first the economics of incineration is discussed. Incineration Economics: We use Williams Milagro plant that processes 575 MMSCFD of gas containing 8 ppm BTEX. We compute the savings from the reduction of temperature from 1500F to 1350F. Using the knowledge of a 2% goal for oxygen in what is released to atmosphere, the amount of methane and cost on a $5/MMBtu basis was calculated for 1500F and 1350F. This reduction can be done due to the lower AIT for the trace hydrocarbons in the acid gas stream. Table 12 shows these amounts: Table 12: Reduction of Temperature Results Thermal Oxidizers Analysis With BTEX Amount of CH4 (MMft^3/year) Cost per year 221 $1,117,000 Without BTEX (Constant Air Excess Assumed) Amount of CH4 (MMft^3/year) Cost per year 161 $814,000 Saving per Year ($) $303,000 We now discuss removal solutions and compare with this cost. Adjusting Amine Treating Parameters: In order to reduce the absorption of BTEX into amine, a few of the parameters of the treating process can be looked at and modified. These parameters consist of temperature of rich amine, pressure of contactor column, circulation rate type of amine, and concentration of amine. These parameters were tested and researched in Bullin (2004). They are shown in Figures 18 and 19. In order to compare effect of temperature, Pro-II simulations were run to back up Bullins discoveries (Figure 18). Bullin concludes that the higher the rich amine temperature, the lower absorption until a CO2 cutoff is reached where anything above this temperature does not adsorb enough CO2. In addition, running at higher temperature causes the reboiler and heat exchangers to have higher heat duties. The pressure 22
dictates the partial pressure of BTEX, but it is more affected by inlet conditions and amine conditions so it is hard to control directly.
Figure 19: Absorption of toluene amine type and concentration (Bullin 2004)
23
BTEXEmissionVersusAmineTemp
165 160 155 BtexEmission(tons/year) 150 145 140 135 130 125 120 50 70 90 110 130 150 170 BTEX CO2Cutoff
AmineTemperature(F)
Figure 20: Temperature trials for benzene absorption Regarding amine circulation, the higher the rate then the more BTEX is absorbed. This circulation is already reduced to lowest possible to prevent excessive corrosion in the system. The lower the concentration of amine, the less BTEX it picks up. In addition, the higher the amine in terms of tertiary, secondary, or primary, then the higher the concentration can be used to achieve same absorption rate. This does not solve the problem due to high temperatures of amines are needed in order to reduce the BTEX emissions below the 25 tons per year limit. This high amine temperature will exceed the carbon dioxide cutoff of 2% required in outlet gas. Use of Water: In order to remove benzene from the feed gas, one can take advantage of the solubility of benzene in water. The process flow diagram is shown in figure 21. Tsonopoulos (1983) found an empirical correlation for Henrys constant as a function of temperature. This expression is shown below:
lnH = A +
B + CT 2 + DlnT T
(23)
LIQUID
GAS
Figure 21. Process flow diagram for addition of water Now, we proceed with the mass balance. We have:
(24)
where FH2O is flow of water into mixer, xB is concentration of benzene in water, FF is flow of feed gas, yB is the concentration of benzene in feed gas, FL is the flow of liquid out of mixer, xB is the concentration of benzene in the stream out of the mixer, FG is the flow of the gas out of the mixer, and yB is the concentration of benzene in gas out of mixer. The composition of benzene in the feed is usually around 0.4-100 ppm. Therefore, if we simplify our example and say that most of the BTEX is benzene, we have our composition in the feed stream. The composition in the water feed stream depends on how the liquid stream coming out of the flash drum is dealt with. If the benzene is removed from the water, and then the water is recycled, some benzene will be present when mixed with the feed gas. If the liquid stream containing the benzene is incinerated, then the water mixed with the feed gas will contain no benzene. Finally, the composition of benzene in the liquid stream will depend on the Henrys constant of benzene in water. The equation relating the concentration of benzene in water is shown below:
xB , w =
PB H B,w
(25)
where PB is the partial pressure of benzene. Now, we proceed to present how the problem would be solved if there was no recycle stream present.
25
If there is no BTEX in the recycled water, then the concentration of benzene in the water will be zero, so we will get:
FF yB = FL x'B + FG y'B
(26)
(27)
equation 16 will tell us how the concentration of benzene in the gas changes as the flow of water is increased or decreased. Trials were run on a inlet gas stream with the following composition: Table 14: Composition of gas
Methane Ethane Propane Butane Heptane Hexane CO2 Benzene Total 83% 6% 4% 2% 0.55% 0.42% 4% 0.03% 100%
To obtain results, we used gas was flowing at 100 MMSCFD at 25C and 200 psia. Using the method described earlier and solving for amount of benzene in the outlet gas stream assuming complete separation of benzene of water or no recycle used is represented below:
26
BenzeneCompositionVs.WaterFlow
0.03500 0.03000 BenzeneMolarComposition 0.02500 0.02000 0.01500 0.01000 0.00500 0.00000 0 50 100 150 WaterFlow(MMSCFD) 200
Figure 22: Composition of Benzene with Water Flow The above figure shows how for a flow rate of 100MMSCFD of natural gas, about double the water flow is required to reduce the amount of benzene in the gas from .03 mole percent to negligible amounts. On top of that, the water would have to then be treated to remove benzene. This would require some sort of adsorbent such as a surfactant modified zeolites as in Ranck (2005). This would be an additional cost on top of the solvent. This was also solved assuming best case scenario of nothing else absorbing into the water besides benzene. This would not be true for methane does absorbs in water although not as readily as benzene since Henry constant of methane is about 100 times the Henry constant of benzene (http://www.handpmg.com/lustline29-oh-henry.htm). This trial proves that the use of water as a solvent is not a viable option. In addition, most solvents would require the removal of BTEX in order to reuse or dispose of the solvent. Use of Activated Carbon: Activated carbon (AC) is an adsorbent used in many industries such as water filtration and vent gas emissions. Many companies produce activated carbon making it readily available for purchase at low cost. Activated carbon has the following properties:
*(http://www.activated-carbon.com/1-3.html)
In addition, activated carbon adsorbs VOCs by the following amounts: Table 16: AC Adsorption of certain compounds
*(Shepherd (2001)) BTEX is on the high end of adsorption available for activated carbon. Activated carbon must be replaced after 2 runs due to low regeneration ability. This regeneration and adsorption after can be represented by the following table: Table 17: Absorption of AC after 1 cycle regeneration
*(Standecker (2008)) In order to evaluate the cost, the amount of absorbent to remove concentrations of BTEX from a Milago type plant for natural gas flow from 575MMSCFD is calculated. This mass of adsorbent was then evaluated as cost of the absorbent to run the two pressure swing adsorption columns in 28
T n ng sure of the c column not b being 12 hour intervals. There was an assumption of lowerin the press y mospheric pr ressure in ord to de-ad der dsorb the BT TEX from ea column. This ach currently used to atm was then sent to one of two desig gns. The sch hematic for t setup is as follows: this
23: S ica s Figure 2 System Setup for Sili Aerogels This syst will then be sent to a possibility of two sepa tem n arate designs. These designs are shown in figures 24 and 25. s
Design 1 consists of desorbing o BTEX in a 12 hour p f of period by a s simple tempe erature relea in ase the case of acid gas and a pressu release i case of fe gas. Th BTEX is then mixed with ure in eed his d air v xed ethane (fuel) into the oxi idizer enough a to be in vapor form. This mixture is then mix with me or flare a raised to 1500F to in and ncinerate BT TEX.
29
Design 2 is the same concept, except that after mixing the temperature is dropped right below the dew point and some of the BTEX is retrieved. This mixture is then flashed and the vapor is mixed with methane (fuel) in the oxidizer. This combustion occurs at 1500F. The columns were priced and sized using the amount of adsorbent needed +10% in order to compensate any absorbent lost during the adsorption and desorption process. The formula for this material is represented below:
lb molinletgas 4536 g mol 24hour Xmolethylbenze 106167 . ne g . grams 100 adsorbent 1kg * * * * * * hr lb mol day 1000000 inlet g molEB 93.98EB 1000 grmas Adskg 2 columns Ads needed day of Ads 52 weeks $37 $SAG day * * = * * * *1.2 for10% extra= year day 2 12 hour day last week years kg
Once the volume needed was obtained the height to diameter ratio of 2-3 was used. This allowed the calculation of the vessel cost (Peters and Timmerhaus, Figure 15-11). The labor was also calculated using the amount of BTEX processed assuming a batch process using Figure 6-9 (P&T). The cost of the blower was done using the flow rate of air needed to bring the desorbed BTEX in vapor phase using Figure 12-32 (P&T). The flash tank volume processed was used to size using same figure as the column cost with a 2-3 height to diameter ratio. The piping and instrument cost was based off of 68% of equipment cost as established by Table 6-6 (P&T). The fuel cost was based off $5/MMBtu and 1010 Btu/scf. These designs were compared with that of the extra cost of incinerating at excess temperatures. The price for activated carbon was found to be $4 per kilogram. In order to estimate worst case scenario of absorption, the aromatic that adsorbed the least was used which is benzene. The following table for the first design was found adding activated carbon in acid gas stream of amine unit:
30
Columns Blower Piping Total FCI Materials Labor Fuel Total Operating Cost Total Annualized Cost
The above table shows that the activated carbon would cost $163,000 more than using incineration at the current price of $4/kg. In order to save $100,000, a cost of 22 cents per kg of activated carbon must be used. Since this cost reduction is great, another absorbent such as silica aerogels must be cost analyzed as well. Use of Silica Aerogels: Silica aerogels (SAG) have emerged in the past five-ten years as a viable adsorbent to separate contaminates from air streams. The physical properties of silica aerogels can be seen below: Table 19: Physical properties of SAG
*(http://eetd.lbl.gov/ecs/aerogels/sa-physical.html) These gels could be used on natural gas streams for removal of BTEX. Silica aerogels have shown a high proximity for aromatic compounds. The following chart shows how much adsorption of these aromatics can take place per 100 g of SAG: Table 20: Adsorption capacity of BTEX 31
*(Standecker (2008)) We note that silica aerogels have the ability to absorb more BTEX in the second cycle than in the first with all compounds except ethylbenzene. Silica aerogels have also been shown to be reused for up to 14 cycles (Standecker (2008)). In addition, silica aerogels also show low adsorption of other components in natural gas. These adsorptions are as following: Table 21: Adsorption of hydrocarbons in SAG
Component Ethane Propane Adsorption .02gramseth/gramsSAG .07gramseth/gramsSAG
The low adsorption of ethane and propane shows an increase of adsorption as the hydrocarbons increase in number of carbon atoms. This would be that methane would have an even lower adsorption than ethane. In addition, silica aerogels are hydrophobic, meaning they repel water. An example of this can be seen in the TMOS block of SAG with water droplet on it:
*(Standecker (2008))
The way these aerogels are produced have been expensive in the past, but are becoming cheaper to manufacture. The above reactions take place in carbon dioxide enriched atmosphere. This accelerates the formation of the aerogels. One way in which SAGs are produced is the following:
32
*(Hwang (2007))
With such high adsorbent of BTEX, silica aerogels show promise in becoming a cost effective solution in removing BTEX from natural gas prior to processing the gas. The small adsorption of hydrocarbons and zero adsorption of water make SAG usable in natural gas streams. The cost analysis of use of silica aerogels will be reviewed further in the report.
The incineration savings was then used to cost the two designs with the current cost of silica aerogels of $37/kg given by Cabot. This resulted in columns of 3X8m and flash for design 2 of 1X3m. These sizes and cost were done using Peter & Timmerhaus using Marshall and Swift Index to cost the equipment at a today price. The cost results are as follows: Table 22: Design 1 Cost for 7 cycles at $37/kg Desorb and Burn Columns Blower Piping Total FCI Materials Labor Fuel $1,143,000 $38,000 $18,000 $1,199,000 $1,240,133 $384,000 $7,000 $226,000 $657,000
33
Table 23: Design 2 Cost for 7 cycles at $37/kg Desorb and Burn Columns $384,000 Blower $7,000 Piping $226,000 Flash Tank $2,000 Total FCI $659,000 Materials $1,143,000 Labor $38,000 Fuel $17,500 Total Operating Cost Revenue $500 $1,240,000
$1,199,000
As shown above, the cost of both designs exceed saving from lower temperature by about $900 thousand dollars. The distribution of cost can be seen in the following graph based on ppm treated at 500MMSCFD gas flow:
CostbyusingSilicaAerogelsComparedtoFlameTemp Reduction
$5,000,000.00 CostusingSAG $0.00 $5,000,000.00 $10,000,000.00 $15,000,000.00 $20,000,000.00 $25,000,000.00 ppmofBTEXat500MMSCFD 1 5 10 20 30 40 50 60 80 100 135 140 145
Design2 Design1 34
In order to justify using silica aerogels, a reduction of cost per kg will be needed. This can be done with further bulk orders and possible contract with Cabot. The tables for this solved reduce cost is as follows: Table 24: Design 1 at Reduced SAG Price 14 cycles Desorb and Burn Columns Blower Piping Total FCI $384,000 $7,000 $226,000 $657,000
$203,000
Table 25: Design 2 at Reduced SAG Price 14 cycles Desorb , Flash, and Burn Columns Blower Piping Flash Tank Total FCI Materials Labor Fuel Revenue $114,000 $38,000 $17,500 $170,000 $500 $202,700 35 $384,000 $7,000 $226,000 $2,000 $660,000
Design 1 would yield a saving of $100,000 per year, and design 2 would yield a saving of $100,300 per year. This cost was found to be about $9 per kg using the 14 cycles instead of 7 cycles. If we assume the gel has enough adsorption to go 21 cycles without significant losses, the following table would result: Table 26: Design 1 at Reduced SAG Price 21 cycles Desorb and Burn Columns Blower Piping Total FCI Materials Labor Fuel $115,000 $38,000 $18,000 $170,000 $203,000 $384,000 $7,000 $226,000 $657,000
Desorb , Flash, and Burn Columns Blower Piping Flash Tank Total FCI Materials Labor Fuel $114,000 $38,000 $17,500 $170,000 $384,000 $7,000 $226,000 $2,000 $660,000
$202,500
This would result in a cost of about $14/kg for the cost of silica aerogels. Considering the low cost of aerogels that must be used in order to meet a $100,000 a year savings, putting such column design in acid gas stream was investigated. The first design of incinerating directly after desorption is the best design for this case, due to the low amount of BTEX in the stream is not enough to make back money from condensing the desorbed vapor. The first trial was run at 7 cycles with a cost of $37/kg. The results for this are as follows:
Desorb and Burn Columns Blower Piping Total FCI Materials Labor Fuel $347,000 $37,000 $5,000 $390,000 $422,000 $373,000 $7,000 $258,000 $638,000
By using the aerogels in the acid gas instead of up front of the column, lower flow has to be processed as well as lower amount of BTEX. The data used for these flows and BTEX amount were taken from Pro II simulation from Milagro Plant data. This amounted in an extra cost of $119,000 in comparison with lowering flame temperature. This cost was found using a $37/kg cost of SAG. The optimal cost to save $100,000 was $17/kg. Below is the same analysis done at 14 cycles with $37/kg: Table 29: Design 1 Cost in Acid Gas at $37/kg 14 cycles Desorb and Burn Columns Blower Piping Total FCI Materials Labor Fuel $164,000 $37,000 $5,000 $206,000 $227,000 $373,000 $7,000 $258,000 $638,000
This led to a savings of $76,000 per year in savings. In order to reach the $100,000 dollars in savings, a cost of $34/kg is needed. 38
Below is the same analysis done at 21 cycles with $37/kg: Table 30: Design 1 Cost in Acid Gas at $37/kg 21 cycles Desorb and Burn Columns Blower Piping Total FCI Materials Labor Fuel $115,000 $37,000 $5,000 $158,000 $190,000 $373,000 $7,000 $258,000 $638,000
This leads to a savings of $143,000 per year for 21 cycles at $37/kg. Due to the savings possible by implementation of silica aerogel columns in acid gas streams, these columns should be looked at it as a possible solution to BTEX emission issues. Use of Macroreticular Resins: Macroreticular resins have shown signs of selectivity of BTEX in adsorption columns. These columns adsorption of BTEX can be seen in Langmuir factors below:
39
These adsorption isotherms are similar to those of the Amberlites, Amberlite XAD4, a product of ROHM & HAAS, subsidiary of DOW, has the highest selectivity towards BTEX. The physical properties of this ion exchange resin can be seen below: 40
Figure 29: Physical Properties of Amberlite XAD4 (DOW) In addition to this, these have shown to lose 5% adsorption capacity after 20 cycles (Chen (1990)). This allows for the product to be used for up to 40 cycles without a major loss in adsorption of 10%. The same procedure for the silica aerogels and activated carbon was used on the resins. These resins turned to have lower adsorption ability and higher initial cost which results in the following table:
Table 31: Design 1 Cost in Acid Gas at $42/kg 42 cycles Desorb and Burn Columns $165,000 41
This would have a savings of $61,000 in comparison to incinerating at higher temperature. In order for this to save $100,000 a year, price reduction to $35/kg would need to be arranged with Dow Chemical.
CONCLUSIONS
Three possible solutions in removing BTEX from amine plant acid gas emissions were researched. These included minimizing absorption in the amine, removing within amine unit, or removing before amine unit. The first option is one of the least viable options due to minimizing BTEX absorption in the amine unit only sends the BTEX problem downstream. This would not be bad if tweaking the parameters of the amine unit only affected BTEX. It could have a similar affect on the absorption of carbon dioxide and hydrogen sulfide too. The second option of removing BTEX within the amine unit can lead to similar issues as removing it minimizing the absorption in the unit. In order to use this technique of removing within unit, maximizing the absorption would have to be achieved by tweaking parameters. This would in turn affect the absorption of carbon dioxide and hydrogen sulfide. This type of removal could also result in loss of amines due to adsorption into a column costing extra to produce correct weight percent of the amine. The third option of removing BTEX prior to entering column consist of looking into use of solvents such as water and adsorbents such as activated carbon, modified sand, and silica aerogels. As explained earlier, high flow rates of water are needed in order to separate BTEX from the natural gas. In addition, an adsorbent would be needed to separate the BTEX from the water. The water also has competing adsorption of hydrocarbons such as methane with Henry constant almost 100 times that of benzene. This competition would lead to methane loss, which affects price of gas sold. The use of modified sand shows promise but is used mostly in the case of liquid adsorption of BTEX. This becomes less useful for removing in natural gas unit. The use of solvent such as water added more cost since separation after dissolving BTEX in solvent
42
would be needed. This extra cost led adsorption to be the front runner. The adsorption solutions consisted of using activated carbon, silica aerogels, and macroreticular resins. All three adsorbents were priced as two pressure swing adsorption columns. With the current prices of AC, SAG, and ionic resins, activated carbon appeared to be clear solution. This was not the case since silica aerogels showed a 6 times better adsorption rate then did activated carbon while seemed to be the worse. The use of AC showed an extra price of over $163 thousand in comparison to incineration when system was placed in acid gas stream of gas plant with 8 ppm BTEX and flow of 575 MMSCFD. This led to activated carbon being discarded as a possibility. Overall pricing of the SAGs in comparison to incineration cost led to a cost being able to treat about a plant with 8 ppm BTEX inlet at a flow of 575MMSCFD to be more expensive then incineration in the front of the unit. When the column was placed at the back of the unit, this savings was easily made $100,000 by simply reducing cost of gels from $37/kg to $34/kg. Due to contacts made with Cabot Corporation, higher concentrations of BTEX could be treated if contracts and bulk discounts were discussed. The pricing and lower adsorption of ionic resins led to the assumption of not a possible solution. However, macroreticular resins showed promise in being able to perform 42 cycles without significant reduction in adsorption, less than ten percent. This led to a saving in the acid gas of $100,000 by reducing the original price of resins from $43/kg to $35/kg. This might be possible through DOW but not as certain as Cabot.
43
References
Abdi, M.A., Meisen, A., Amine Degradation: Problems, Review of Research Achievement Recovery Techniques, Institute of Petroleum Industry. AMINECalc Version 1.0 Users Manual, Health and Environmental Safety Department, API Publication Number 4679. Bullin, J. A., Brown, W. G., Hydrocarbons and BTEX Pickup and Control from Amine Systems, Proceeding of the 83rd Annual Gas Processors Association Convention, 2004. Chen, T., Pinto, N.G., Stability and Equilibrium Properties of Macroreticular Resins for Flue Gas Desulfurization. Industrial and Engineering Chemistry Research, 1990. Collie, J., Hlavinka, M., Ashworth, A., An Analyis of BTEX emissions from Amine Sweetening and Glycol Dehydration Facilities, Laurance Reid Gas Conditioning Conference Proceedings, 1998. Coquelet, C., Richon, D., Solubility of BTEX and Acid Gases in Alkanolamine Solutions in Relation to Environment, CNRS FRE2861 CEP/TEP. Gitipour, S., Bowers, M., Bodosci, A. The Use of Modified Bentonite for Removal of Aromatic Organics From Contaminated Soil, Journal of Colloid and Interface Collisions, 1997. Hartman, B. Oh Henry!, a constant, Mobile Geometry Inc. http://www.handpmg.com/lustline29-oh-henry.htm. Hwang, S-W., Jung, H-H., Hyun, S-H., Ahn, Y-S., Effective preparation of crack-free silica aerogels via ambient drying, Springer Science, 2007. Lewandowski, D. A., Design of thermal oxidizer systems for volatile organic compounds, Lewis Publishers, 2000. Lin, S. H., Huang, C. Y., Adsorption of BTEX from aqueous solution by Macroreticular resins. Journal of Hazardous Material, 1999. Morrow, D.C, K.M. Lunsford, Removal and Disposal of BTEX Components from Amine Plant Acid Gas, Proceedings of the 76th Gas Processors Association Convention, 1997. Peters, Max S, Klaus D Timmerhaus, and Ronald E West. Plant Design and Economics for Chemical Engineers. 5th Edition. St. Louis: McGraw-Hill, 2003. PRO-II Manual Volume 1: Chapter 3, Simulation Sciences Inc., 1994-2002. Properties of Activated Carbon, CPL Carbon Link, http://www.activated-carbon.com/1-3.html. 44
Ranck, J.M., Bowman, R.S., Weeber, J.L., Katz, L.E., Sulivan, E.J., BTEX Removal from Produced Water Using Surfactant-Modified Zeolite, Journal of Environmental Engineering, 2005. Science of Silica Aerogels, http://eetd.lbl.gov/ecs/aerogels/sa-physical.html Shepherd, Austin, Activate Carbon Adsorption For Treatment of VOC Emission, Proceeding of 13th Annual EnviroExpo, 2001. Shiveler, G., Solis, G.S., Gonzalez, L.H.P., Bueno, M.L., Retrofit of a H2S Selective Amine Absorber Using MellapakPlus Structured Packing, AICHE Spring Meeting, 2005. Standeker, S., Novak, Z., Knez, Z., Removal of BTEX vapours from waste gas streams using silica aerogels of differenthydrophobicity, Journal Hazardous Material, (2008). Thermal Oxidizer Review, Edge Engineering. http://www.engineersedge.com/manufacturing/about_thermal_oxidizers.htm. Tsonopoulos, C., Wilson, G.M., High-Temperature Mutual Solubilities of Hydrocarbons and Water, AICHE Journal, 1983. Yua, J.S., Tsai, F.N., Correlation of solubilities of carbon dioxide in aromatic compounds, Fluid Phase Equilibria, 1992.
45