PROCESSING

Performance testing of natural gas enters a two-phase separator

an advanced glycol dehy- that divides liquid hydrocarbons from
drator design showed the gas stream. Liquid products feed a
that the unit, compared condensate storage tank for sale and
with a conventional de- wet gas flows to an absorber.
hydration technology, Lean, dry triethylene glycol (TEG)
Gas Processing saved the equivalent of directly contacts the wet gas and ab-
more than 35 million sorbs water vapor, methane, HAPs in-
standard cu ft/year of natural gas val- cluding benzene, toluene, ethylben-
ued at more than $173,000. It also zene, and xylene (BTEX), n-hexane,
lowered hazardous air pollutant (HAP) and VOCs.
and volatile organic compound (VOC) Dry
emissions. natural
During this test, the new unit pro-
duced by Engineered Concepts LLC,
gas exits
the ab- Advanced dehydrator design
Farmington, NM, also reduced CO2
emissions by nearly 3,000 tons/year.
A 1996 study of CH4 emissions
sorber as
pipeline-
quality
recovers gas, reduces emissions
from the natural gas industry, conduct- gas ready
ed by the US Environmental Protection for sale. The rich, wet glycol exiting the
Agency and the Gas Research Institute, absorber feeds a regeneration reboiler
estimated that active glycol dehydrators that removes the absorbed constituents
in the US collectively emitted about resulting in a lean glycol mixture suit- David A. Kirchgessner
18.6 billion standard cu ft/year of able for reuse in the absorber. US Environmental Protection Agency
Research Triangle Park, NC
CH4.1 The regeneration process is typically
These units also reportedly produced the primary source of emissions from a Robert G. Richards
85% and 81% of the production sec- dehydrator. The process contains a Southern Research Institute
tor’s HAP and VOC emissions, respec- process pump, reboiler still, and a vari- Research Triangle Park, NC
tively.2 3 These considerations make de- ety of heat exchangers.
hydrators ideal candidates for both gas The process pump moves glycol Forrest Heath
savings and emission reductions. through the system and is either elec- Engineered Concepts LLC
Houston
Testing of the Quantum Leap natural trical or, most commonly, a gas-assisted
gas dehydration technology (QLT) oc- Kimray pump. Robert D. Smith
curred at the Kerr-McGee Corp. gather- High-pressure natural gas powers Kerr-McGee Corp.
ing station in Brighton, Colo., in April the Kimray pumps; spent pump gas is Brighton, Colo.
2003. usually dumped into the rich glycol
EPA’s Office of Research and Devel- stream, flashed off in the regenerator,
opment, under its Environmental Tech- and vented through the still column. The still column vent emits stripped
nology Verification program, conducted Units with a glycol flash tank upstream water, methane, HAPs, and VOCs to the
the study. The program operates six of the reboiler can recapture some of atmosphere unless a combustion or
centers that focus on testing technolo- the spent gas for reboiler fuel. condensation device controls the
gies designed to mitigate a broad range The reboiler strips absorbed water, stream. Combustion devices include
of environmental problems. HAPs, VOCs, and CH4 out of the glycol flares or, in the case of the Kerr-McGee
This test was conducted by the and into the still column. Regenerated site, thermal oxidizers. Condensers in-
Greenhouse Gas Technology Center that lean glycol exits the reboiler, is cooled clude water-knockout systems and oth-
is comanaged with EPA and operated via cross exchange with returning rich er separation systems that produce
by Southern Research Institute, Re- glycol, and enters a surge tank. From saleable condensate products.
search Triangle Park, NC. there, it is pumped to a glycol-gas heat The QLT system contains these com-
exchanger and back to the absorber. ponent modifications and additions
Technology description This heat exchanger controls the (Fig. 1): electric glycol circulation
The QLT can be configured as a lean glycol temperature to the absorber. pump, eductor, still column effluent
retrofit to, or a replacement of, existing High glycol temperatures relative to the condenser, glycol cooler, three-phase
conventional dehydrators. gas temperature reduce TEG’s moisture- vacuum separator, three-phase emis-
Fig. 1 shows a schematic of the tech- absorption capability. Conversely, tem- sions separator, electric process pump,
nology as installed at the Kerr-McGee peratures that are too low promote gly- and two high-efficiency glycol filters.
plant and Fig. 2 shows the installed sys- col loss due to foaming and increase Circulation of glycol via the circulat-
tem. the glycol’s hydrocarbon uptake and ing pump (A) through the eductor (B)
In conventional dehydrators, wet potential still-vent emissions. creates a vacuum. This vacuum is con-
Reprinted with revisions to format, from the July 26, 2004 edition of OIL & GAS JOURNAL
Copyright 2004 by PennWell Corporation
PROCESSING
A S-INSTALLED SYSTEM SCHEMATIC Fig. 1
Makeup fuel Fuel to burner 9
7 C D
Effluent condenser Glycol cooler
Effluent

Fuel accumulator
Glycol-gas 8
heat exchanger

Absorber H
Filter

B
Eductor
Still column

F
Glycol Emissions
1 separator
Gas in 4 reconcentrator Condensate
to storage
Gas to
sales
6 Condensate
Water to
3 disposal

5
E A
Vacuum separator Circulating pump
Glycol-glycol heat
G exchanger
Electric, positive-
displacement pump
Level control valve
Glycol controlled by level controller
storage in emissions separator
2

H
Condensate Rich glycol Sidestream filter
Water Mixed vapors, condensate, water
Gas Vapors
Lean glycol

trolled at about 4-in. water column (in. losses or weathering occur. processes about 26 MMscfd of natural
WC) and pulls the still column over- The vacuum pulls all noncondens- gas in the dehydrator. The facility in-
head vapors through the still column able vapors to the eductor where they stalled the new technology because ex-
effluent condenser (C). are compressed to about 30 psig. The cess moisture in the still vent caused
The circulation pump provides more eductor outlet stream—a mixture of persistent problems with thermal oxi-
glycol than is needed for the eductor; glycol from the circulation pump and dizers.
excess glycol cools the effluent con- noncondensed vapors from the vacuum Table 1 summarizes the test site’s key
denser. A fan-driven air cooler (D) re- separator—feeds the emissions separa- design and operating parameters.
jects the heat accumulated in glycol cir- tor (F) where the glycol and vapors are After the new system was installed,
culated through the effluent condenser. separated. operators discovered that the burner
Still column overhead vapors con- Vapors are sent to the reboiler fuel and reboiler were operating at an in-
dense under a vacuum at about 120- system. An electric-powered process consistent temperature. The reboiler
130° F. The vacuum separator (E) sepa- pump (G) and high-efficiency filters was equipped with an electronic ther-
rates the resulting liquid hydrocarbon- (H) complete the process. mostat that responded rapidly to small
water mixture. changes in temperature with large in-
Water is disposed of and hydrocar- Test setup, equipment creases or decreases in output to the
bon liquids are sent to storage. Because modification control valve. This resulted in wide
the hydrocarbons are collected under a The Kerr-McGee gathering facility, temperature swings with alternating
vacuum, they are stable and no vapor 14 miles northwest of Brighton, Colo., periods of heavy firing, which required
 remained closed dur-
ing the tests.
Gas processed in
the dehydrator had
an unusually high
btu content. Conse-
quently, the process
Reboiler stack occasionally recov-
ered more high-btu
vapors than it could
consume during
Still
normal operations.
This resulted in a
higher fuel pressure
and subsequent
problems in main-
Housing for taining an adequate
pumps, vacuum.
Effluent The accumulator
separators,
condenser helped dampen the
etc.
system response; but
Reboiler as an added meas-
ure, Kerr-McGee in-
stalled a water-injec-
tion system.

Water-injection
system
A compressed-air-
driven pump was in-
stalled to inject some
of the vacuum sepa-
rator’s recovered
waste water back in-
to the reboiler. This
would increase the
reboiler load when
This photo shows the system’s features installed at the test site in Brighton, Colo. (Fig. 2). necessary, which en-
abled the burner to de-
makeup gas, and corresponding off cy- closed during normal operations. mand more fuel.
cles during which the recovered fuel The fuel-gas pressure and effluent
gas pressure would increase to a point Fuel accumulator vessel condenser temperature control this
that the eductor was unable to pull an A 430-gal accumulator vessel was pump. The effluent condenser tempera-
adequate vacuum. installed to dampen the effects of large ture is a key control point because hot
A pressure-relief valve and fuel accu- swings in reboiler temperature. vapors cause inefficient hydrocarbon
mulator were installed to prevent this The accumulator vessel increased the condensation.
pressure buildup and to enable proper fuel gas system’s reserve volume during The pump operates when the fuel-
burner control. high-recovery periods. This allowed the gas pressure is 20 psig or more and the
burner to fire with small amounts of overhead temperature is 120° F. or less;
Pressure relief valve makeup fuel during high-firing cycles otherwise, the pump automatically
A pressure-relief valve (PRV) was in- and provided a pressure cushion to ac- shuts down. The water pump was de-
stalled in the vacuum separator. It cumulate fuel during low-firing peri- signed with a reserve capacity sufficient
would open to atmosphere when the ods. to handle all reasonably expected gas
recovered fuel gas pressure reached 30 A pressure-activated valve would compositions at the test facility.
psig and close when the pressure was open to the atmosphere if the gas pres-
less than 30 psig. sure in the vessel exceeded 28 psig. This Performance verification
The PRV is a safety device. The sys- could produce air emissions, but this We verified the operational perform-
tem uses this feature only during initial PRV is a safety device that would actu- ance for sales-gas moisture and produc-
system start-up; the vent remains ate during abnormal conditions only. It tion rate, glycol circulation rate, and
PROCESSING
makeup natural-gas fuel flow
rate. We also verified the en-
TEST SITE DESIGN, OPERATING CONDITIONS Table 1
cy is ±1.0% of reading and
can be used on 0.25-360 in.
Natural gas production,
vironmental performance of MMscfd at 14.7 psia, 60° F. 26 diameter pipes with fluid
reboiler-stack emission rates Sales gas moisture content, flow rates of 0-60 fps.
lb H2O/MMscfd gas <7
and HAP destruction efficien- Electric 5-hp motor pump
cy. circulation rate, gpm
Glycol for absorption, regeneration 5
Makeup natural gas
Glycol for condensation, eductor power 72 The new reboiler burner
Glycol-glycol heat exchanger
Sales gas Duty, 1,000 btu 325 can accept up to 166 scfh of
Kerr-McGee continuously Shell operating conditions (lean glycol) Atmospheric pressure, 400° F. makeup natural gas as sup-
Tube operating conditions (rich glycol) 30 psig, 300° F.
monitors sales-gas moisture Reboiler still plemental fuel.
Duty, 1,000 btu/hr 600
using a GE Panametrics meter Operating conditions 0-2 in. WC (vacuum) A Halliburton Co. MC-II
with a moisture measure- Reboiler burner EXP turbine meter installed
Total heat input required, MMbtu/hr 1.2
ment range of 0-20 lb Fuel gas from emissions separator ~233-388 scfh (70-80 vol %) on the 1-in. ID gas line up-
H2O/MMscf, a lower detec- Specific gravity = 0.75 stream of the reboiler meas-
Lower heating value = 1,410
tion limit of 0.2 lb H2O/ btu/scf ured makeup gas flow. The
Makeup natural gas ~0-166 scfh (0-30 vol %)
MMscf, and a rated accuracy Specific gravity = 0.65 meter includes an integral-
of ±5% of reading. Lower heating value = 950 signal display and transmitter
btu/scf
Panametrics calibrated the Stack dimensions 10-in. diameter, 20-ft high with a linear flow range suf-
meter before installation. We Glycol condenser, ficient to measure gas flows
glycol-air heat exchanger
used the 1-min average mois- Duty, 1,000 btu/hr 225 if the reboiler operates on
Rich glycol operating conditions 30 psig, 150° F.
ture data. Emissions separator makeup gas only (0-600
Kerr-McGee uses an Emer- Dimensions 30-in. diameter, 6.5-ft high scfh).
Operating pressure, psig 15
son MVS205 multivariable- Vacuum separator The manufacturer used a
sensor orifice meter to docu- Dimensions 20-in. diameter, 5.5-ft high piston-type volume prover to
Operating pressure,
ment sales-gas production. in. WC (vacuum) 0-5 calibrate the meter. It is tem-
Water discharge rate,
The sales-gas meter contains gal/1.5-in. change in liquid level ~1.89 perature and pressure-com-
a 4-in. orifice plate and is Condensate discharge rate, pensated, and provided a
gal/1.5-in. change in liquid level ~1.89
temperature and pressure Effluent condenser, mass flow output accurate to
compensated to 60° F. and vapor-glycol heat exchanger 1% at standard conditions.
Duty, 1,000 btu/hr 100
14.7 psia (gas industry stan- Tube operating conditions (still vapors) 0-5 in. WC (vacuum), 212° F. We used the 1-min average
Shell operating conditions (rich glycol) 30 psig, 110° F.
dard conditions). data from this meter.
The meter’s operating
range is 0-2 million standard Reboiler stack
cu ft/hr with a rated accura- EMISSIONS TESTING Table 2 emissions
cy of ± 1% of reading. Site US EPA Cubix Corp., an independ-
personnel calibrated the Measured reference Analyzer Instrument ent stack testing contractor in
variable method type range
flowmeter before testing. We Austin, performed reboiler
NOx 7E Chemiluminescence 0-100 ppm
used the meter’s 1-min aver- CO 10 Nondispersive infrared 0-100 ppm stack emissions testing to de-
ages. Total hydrocarbons 25A Flame ionization termine concentrations and
detector 0-100 ppm
O2 3A Paramagnetic 0-25% emission rates for: CO, total
Glycol circulation CO2
CH4
3A
18
Nondispersive infrared
Gas chromatograph,
0-20% hydrocarbons, greenhouse
A Controlotron Corp. flame ionization gases (CO2, NOx, and CH4),
detector 0-100 ppm
1010EP1 ultrasonic meter BTEX,* n-Hexane 18 Gas chromatograph, BTEX, and total HAPs, which
measured the glycol circula- flame ionization are BTEX plus n-hexane.
detector 0-100 ppm
tion rate. The meter is a digi- Exhaust gas 1A and 2C Differential 9,000-11,000 Cubix conducted three
tally integrated flowmetering volumetric (modified) pressure scfh 90-min (nominal duration)
flow rate
system that consists of a Moisture 4 Gravimetric 0-100% test runs for each parameter
portable computer and ultra- *Includes separate benzene, toluene, ethylbenzene, and xylene quantification. while the system was operat-
sonic fluid flow transmitters. ing at normal conditions.
The meter determines flu- Emission rates reported in
id velocity by measuring ultrasonic computer. ppm (vol) dry (ppmvd) are correlated
pulse transit times between the trans- The flowmeter determines sonic ve- with the stack volumetric flow rates in
ducers. A precision-mounting jig se- locity based on the known distance be- dry standard cu ft/min to yield lb/hr
cures the transducers to the pipe at a tween the transducers for zero-flow emission rates for NOx, CO, CH4, VOC,
known distance. conditions with the pipe full of fluid. It hexane, BTEX, and HAPs.
The operator enters the fluid com- multiplies fluid velocity by the internal VOC emissions are all organic com-
position (100% TEG for this test), pipe pipe area, and reports 1-min average pounds minus methane and ethane
diameter, material, wall thickness, and volumetric flow rates. emissions according to Colorado De-
expected sonic velocity into the meter’s The flowmeter’s overall rated accura- partment of Public Health and Environ-
OPERATING DATA, NORMAL CONDITIONS1 Table 3
Sales gas
Valid moisture content,2 Sales gas Makeup natural gas Glycol circulation
Date in data, ––– lb H2O/MMscf ––– –––– flow rate,2 MMscfd –––– ––––– flow rate,2 scfh ––––– ––––– rate, gpm ––––––
2003 hr Range Average Range Average Range Average Range Average

Apr. 23 15.05 0.80-1.69 1.02 28.67-31.39 29.31 0.11-345.98 16.32 1.55-6.04 3.63
Apr. 24 24.00 0.79-1.03 0.89 26.18-32.02 28.63 0-220.19 1.22 1.47-4.00 3.30
3
Apr. 25 20.73 0.91-1.44 1.12 26.09-29.96 28.38 0-190.44 0.63 0-4.71 3.00
Apr. 26 24.00 0.73-1.99 1.28 26.13-29.97 28.15 0-317.04 1.68 0.64-5.34 3.21
Apr. 27 23.95 0.95-1.69 1.27 25.69-28.83 26.88 0-3.92 0.83 1.79-4.23 3.67
Apr. 28 24.00 0.85-1.76 1.24 23.13-29.96 26.81 0-706.33 5.41 1.68-4.61 3.68
Apr. 29 24.00 0.89-1.64 1.18 25.20-29.96 27.38 0-3.61 0.83 1.87-4.43 3.77
Overall average 1.14 27.90 3.85 3.47
Normal operating
conditions1 0.89-1.50 26.54-29.26 0-1.76 3.14-3.93
1
Normal operating conditions is the range represented by 75% of individual 1-min measurement values. 2Source: Kerr-McGee operations. 3The flowmeter occasionally reported zero on
this date due to aeration in the pipeline. When the operator added makeup TEG to the system, the aeration ceased and the flowmeter resumed normal operations.

ment regulations. particularly interesting.

All the test procedures are REBOILER STACK EMISSIONS We expected the new re-
Table 4
documented Title 40 CFR 60 –––––––––– Test run ––––––––––– boiler to use up to 166 scfh
1 2 3 Average
Appendix A reference meth- of makeup gas to supplement
ods. Exhaust O2, % 6.4 6.7 6.8 6.6 its fuel supply, but the overall
Stack gas velocity, fps 23.64 23.72 24.27 23.87
Table 2 summarizes refer- Stack flow rate, dscfh 10,793 10,369 10,359 10,507 average flow rate was 3.85
ence methods performed for Emissions scfh. This showed that the
NOx, ppmvd 67.8 66.0 61.6 65.1
emissions testing supporting NOx, lb/hr 0.0873 0.0817 0.0761 0.0817 unit could use high-btu, wet
CO, ppmvd 0.3 1.0 0.6 0.6
this verification. CO, lb/hr 0.0003 0.0007 0.0004 0.0005 hydrocarbon vapors as a pri-
VOCs, ppmvd 0.4 0.8 0.5 0.6 mary fuel.
VOCs, lb/hr 0.0002 0.0004 0.0002 0.0003
HAP destruction CH4, ppmvd <0.1 <0.1 <0.1 <0.1 Table 4 shows the reboiler
efficiency CH4, lb/hr
CO2, %
<0.00004 <0.00004
9.5 9.2
<0.00004
9.1
<0.00004
9.3
stack emissions results for the
Destruction efficiency is CO2, lb/hr 117 108 107 111 three test runs. A continuous-
Total HAPs, ppmvd <0.6 <0.6 <0.6 <0.6
the net HAPs entering the Total HAPs, lb/hr <0.0016 <0.0016 <0.0015 <0.0016 ly extracted stack-gas sample
system from the glycol mi- periodically injected into a
nus those leaving the system gas chromatograph provided
in emissions sources divided by the net HAPs dissolved in the condensate the material for organic (CH4, HAPs)
HAPs entering the system. stream are considered to be “con- concentration determinations.
Testers determined the HAPs inputs trolled” or “sequestered” and not emis- Test personnel performed six injec-
via the Atmospheric Rich-Lean Method sions. tions, each about 15 min apart, during
for Determining Glycol Dehydrator We conducted performance testing each test run.
Emissions.4 in two stages: operational testing oc- The analyst determined that each
HAP emission sources at this site in- curred over 7 days both to obtain re- HAP constituent was consistently below
clude: fugitive leaks, reboiler burner portable flow rate data and to ensure the instrument’s detection limit of <0.1
exhaust, waste water, and PRVs. We de- that the plant was operating normally; ppmvd. This equates to an average
termined that fugitive leaks are negligi- environmental testing occurred on the emission rate of <0.0016 lb/hr, which
ble because the fabricator certified the following day in three test runs of 70- is well below the site’s permit require-
system to be leaktight. The burner stack 85 min each. ment.
may emit unburned HAPs to the atmos- All CH4 results were also below the
phere; HAPs dissolved in waste water Test results gas chromatograph, flame ionization de-
can release during disposal. Table 3 shows the operational re- tector’s detection limit of <0.1 ppmvd.
Consistent with 40 CFR Part 63,5 sults. Makeup natural gas flow rates are Table 5 summarizes HAP destruction
efficiency for each test run and the
HAP DESTRUCTION EFFICIENCY Table 5 overall average. The calculation method
90%
for total HAP destruction efficiency to
confidence sum HAPs in the waste water (Table 6)
––––––––––––– Test run ––––––––––––– Average interval
and reboiler exhaust (Table 7) and di-
HAPs in, lb/hr 1 2 3 vide by HAPs inputs from glycol
Rich glycol 9.83 8.37 10.19 9.46 1.62
Lean glycol 0.33 0.37 0.4 0.37 0.06 streams (Table 8).
Net inflows 9.50 8.00 9.79 9.09 1.62 The operational performance data
HAPs out, lb/hr
Waste water 0.0209 0.0220 0.0232 0.0220 0.0020 showed that:
Stack <0.0016 <0.0016 <0.0015 0.0016 0.00015
Vented 0 0 0 0 – • The moisture content of dry natu-
Net emissions 0.0226 0.0236 0.0245 0.0236 0.002 ral gas was well below the 7 lb H2O/
Destruction efficiency, % 99.76 99.70 99.75 99.74 0.01
MMscf limit that the operator required
PROCESSING
HAPS IN WASTE WATER Table 6
––––––––––––––––––––––––––– Concentration in waste water, µg/ml –––––––––––––––––––––––––– Waste water HAP in
–––––––– Average ––––––– production, waste water,
Sample 1 Sample 2 Sample 2a Sample 3 µg/ml ppg gpm lb/hr

Run 1
n-Hexane 0.801 (ND) 1.001 (ND) 1.001 (ND) 1.001 (ND) 0.951 0.000008 0.101 0.00005
Benzene 200.489 227.145 313.737 289.688 257.764 0.002151 0.101 0.01298
Toluene 104.976 113.377 175.446 157.164 137.741 0.001150 0.101 0.00693
Ethylbenzene 0.971 (J) 1.279 (J) 1.918 (J) 1.642 (J) 1.453 0.000012 0.101 0.00007
m- and p-Xylene 8.352 9.058 16.928 15.516 12.463 0.000104 0.101 0.00063
o-Xylene 2.829 (J) 3.434 (J) 5.570 5.212 4.261 0.000036 0.101 0.00021
Total HAPs 318.417 355.294 514.600 470.223 414.634 0.003460 0.101 0.0209
Run 2
n-Hexane 0.400 (ND) 1.001 (ND) 1.464 (J) 1.001 (ND) 0.967 0.000008 0.119 0.00006
Benzene 186.603 271.942 146.598 284.810 222.488 0.001857 0.119 0.01328
Toluene 96.426 165.647 78.479 168.820 127.343 0.001063 0.119 0.00760
Ethylbenzene 0.855 (J) 1.635 (J) 1.001 (ND) 1.803 (J) 1.324 0.000011 0.119 0.00008
m- and p-Xylene 8.629 15.951 7.379 16.304 12.066 0.000101 0.119 0.00072
o-Xylene 2.671 5.173 2.485 (J) 5.408 3.934 0.000033 0.119 0.00023
Total HAP 295.584 461.350 237.407 478.147 368.122 0.003072 0.119 0.0220
Run 3
n-Hexane 1.001 (ND) 1.001 (ND) 1.001 (ND) 1.001 (ND) 1.001 0.000008 0.098 0.00005
Benzene 275.285 272.485 291.060 307.250 286.520 0.002391 0.098 0.01407
Toluene 156.729 157.039 165.044 168.717 161.882 0.001351 0.098 0.00795
Ethylbenzene 1.609 (J) 1.555 (J) 1.706 (J) 1.677 (J) 1.637 0.000014 0.098 0.00008
m- and p-Xylene 15.815 15.510 16.673 16.276 16.068 0.000134 0.098 0.00079
o-Xylene 5.391 5.090 5.367 5.323 5.293 0.000044 0.098 0.00026
Total HAP 455.829 452.680 480.850 500.245 472.401 0.003942 0.098 0.0232
Overall average
total HAPs 356.610 423.108 410.952 482.871 418.386 0.003 0.106 0.0220

ND = nondetect or the analytical result is below the minimum detection limit (MDL). J = analytical result is between the MDL and the limit of quantification.

throughout the monitoring period. Ac- sure load of 1,070 psig, which worked
tual daily averages were 0.89-1.28 lb REBOILER EXHAUST STREAM Table 7 against the absorber and the glycol
H2O/MMscf. Run HAP, lb/hr flash separator. Kimray literature indi-
• Average sales-gas flow rates were 1 n-Hexane <0.000241 cates that, at this pressure, the pump
26.8-29.3 MMscfd. Benzene <0.000218 consumes 5.95 scf/gal of glycol
Toluene <0.000258
• The system burned all noncon- Ethylbenzene <0.000297 pumped or a total of 34,200 scfd based
p-Xylene <0.000297
densable hydrocarbon vapors without o-Xylene <0.000297 on a 4 gpm glycol circulation rate.
venting them to the atmosphere and Total HAP <0.00161 The reboiler in the original configu-
2 n-Hexane <0.000232
used little or no makeup gas. Average Benzene <0.000210 ration consumed some of this gas. It
flow rates of makeup natural gas were Toluene <0.000247 burned about 520,000 btu/hr assum-
Ethylbenzene <0.000286
0.63-16.32 scfh with an overall average p-Xylene <0.000286 ing a 50% efficiency.
o-Xylene <0.000286
of 3.85 scfh. Total HAP <0.00155 The approximate net heating value of
• Daily average glycol circulation 3 n-Hexane <0.000232 gas from the glycol-gas separator was
Benzene <0.000210
rates were 3.0-3.77 gpm. Toluene <0.000247 about 1,160 btu/scf according to a process
The environmental testing proved Ethylbenzene <0.000285 model.This is higher than the plant fuel
p-Xylene <0.000285
that: o-Xylene <0.000285 gas value of 1,107 btu/scf because the
Total HAP <0.00154
• Overall average emission rates for Overall average 0.00157 gas contained some of the heavier com-
NOx, CO, CO2, and VOCs from the re- ponents absorbed by the glycol.
boiler stack were 0.0817, 0.0005, 111, The reboiler therefore consumed
and 0.0003 lb/hr, respectively. 2.88 gph, respectively. 448 scfh and the rest was vented to the
• HAP and CH4 concentrations in thermal oxidizer. Total gas no longer
the reboiler stack were undetectable. Cost savings, wasted is 34,300 – 448(24) = 23,500
Maximum HAPs leaving the system in emission reductions scfd. At 1,160 btu/scf this equals 1.14
the reboiler exhaust and waste water The four main sources of cost sav- MMbtu/hr.
were 0.0016 and 0.022 lb/hr, respec- ings for the new dehydrator result from
tively. replacing the Kimray pump, eliminat- Gas stripping
• HAP destruction efficiency was ing gas stripping, reducing still column Because a condensing water ex-
greater than 99.74% ± 0.01%. overhead emissions, and lowering ther- hauster was incorporated into the new
• PRVs did not operate at any time mal oxidizer fuel use. design, the gas stripping system (sparg-
during the entire test. No releases are er) was no longer needed. The mini-
anticipated during normal operations; Kimray pump mum design consumption rate using a
therefore, no expected emissions were An electric pump in the QLT replaced sparger was 4 scf of gas/gal of glycol
assigned to PRV operations. the Kimray pump, which is pressurized circulated; we assumed that consump-
• Average waste water and conden- with natural gas in typical dehydrators. tion rate for these tests.
sate-production rates were 6.36 and The Kimray pump developed a pres- At this rate, the sparger used 23,040
HAPS IN GLYCOL STREAMS Table 8
––––––––––––––––––––––––––––––– Concentration, µg/ml ––––––––––––––––––––––––––––––
Run 1 ––––––– Average ––––––– Lean glycol HAP,
Lean 1 Sample 1 Sample 2 Sample 2a Sample 3 µg/ml ppg flow, gpm lb/hr

n-Hexane 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.769 0.01
Benzene 69.4 54.0 68.2 59.1 62.66 0.0005229 3.769 0.12
Toluene 89.6 66.5 87.6 69.4 78.26 0.0006531 3.769 0.15
Ethylbenzene 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.769 0.01
p-Xylene 16.6 15.6 14.8 19.9 16.73 0.0001396 3.769 0.03
o-Xylene 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.769 0.01
Total HAP 193.58 154.14 188.63 166.28 175.66 0.0014659 3.769 0.33
Difference
––––––– Average ––––––– Rich glycol HAP in net HAPs
Rich 1 Sample 1 Sample 2 Sample 2a Sample 3 µg/ml ppg flow, gpm –––––– lb/hr ––––––

n-Hexane 140.13 107.27 144.50 137.00 132.22 0.0011034 3.916 0.26 0.25
Benzene 1,660.5 1,424.8 1,704.8 1,394.5 1,546.13 0.0129031 3.916 3.03 2.91
Toluene 2,744.7 2,393.9 2,843.6 2,293.2 2,568.85 0.0214381 3.916 5.04 4.89
Ethylbenzene 58.98 51.47 62.31 48.01 55.19 0.0004606 3.916 0.11 0.10
p-Xylene 614.6 545.7 647.7 511.2 579.80 0.0048386 3.916 1.14 1.11
o-Xylene 137.67 120.91 144.49 113.18 129.06 0.0010771 3.916 0.25 0.24
Total HAP 5,356.52 4,644.02 5,547.40 4,497.10 5,011.26 0.0418210 3.916 9.83 9.50
––––––––––––––––––––––––––––––– Concentration, µg/ml ––––––––––––––––––––––––––––––
Run 2 ––––––– Average ––––––– Lean glycol HAP,
Lean 2 Sample 1 Sample 2 Sample 2a Sample 3 µg/ml ppg flow, gpm lb/hr

n-Hexane 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.604 0.01
Benzene 82.5 74.4 68.6 75.5 75.25 0.0006280 3.604 0.14
Toluene 102.5 94.5 86.6 92.8 94.11 0.0007853 3.604 0.17
Ethylbenzene 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.604 0.01
p-Xylene 20.6 16.1 24.8 14.3 18.96 0.0001582 3.604 0.03
o-Xylene 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.604 0.01
Total HAP 223.66 202.98 197.99 200.64 206.32 0.0017218 3.604 0.37
Difference
––––––– Average ––––––– Rich glycol HAP in net HAPs
Rich 2 Sample 1 Sample 2 Sample 2a Sample 3 µg/ml ppg flow, gpm –––––– lb/hr ––––––

n-Hexane 135.27 122.23 119.66 134.68 127.96 0.0010679 3.772 0.24 0.23
Benzene 1,476.1 1,461.7 1,332.5 1,467.7 1,434.50 0.0119715 3.772 2.71 2.57
Toluene 2,197.3 2,355.2 2,133.1 2,275.6 2,240.32 0.0186964 3.772 4.23 4.06
Ethylbenzene 44.96 47.96 42.70 43.75 44.84 0.0003742 3.772 0.08 0.07
p-Xylene 474.4 508.0 458.7 470.2 477.84 0.0039878 3.772 0.90 0.87
o-Xylene 106.65 112.41 100.60 103.05 105.68 0.0008819 3.772 0.20 0.19
Total HAP 4,434.72 4,607.53 4,187.29 4,495.03 4,431.14 0.0369797 3.772 8.37 8.00
––––––––––––––––––––––––––––––– Concentration, µg/ml ––––––––––––––––––––––––––––––
Run 3 ––––––– Average ––––––– Lean glycol HAP,
Lean 3 Sample 1 Sample 2 Sample 2a Sample 3 µg/ml ppg flow, gpm lb/hr

n-Hexane 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.887 0.01
Benzene 74.3 74.3 71.6 86.0 76.57 0.0006390 3.887 0.15
Toluene 91.1 88.7 86.5 105.6 92.98 0.0007760 3.887 0.18
Ethylbenzene 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.887 0.01
p-Xylene 17.1 21.6 16.5 18.4 18.41 0.0001536 3.887 0.04
o-Xylene 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 (ND) 6.00 0.0000501 3.887 0.01
Total HAP 200.51 202.67 192.57 228.09 205.96 0.0017188 3.887 0.40
Difference
––––––– Average ––––––– Rich glycol HAP in net HAPs
Rich 3 Sample 1 Sample 2 Sample 2a Sample 3 µg/ml ppg flow, gpm –––––– lb/hr ––––––

n-Hexane 144.62 128.69 144.76 135.22 138.32 0.0011544 4.047 0.28 0.27
Benzene 1,652.1 1,566.6 1,576.0 1,592.6 1,596.84 0.0133263 4.047 3.24 3.09
Toluene 2,609.8 2,517.0 2,529.8 2,665.7 2,580.59 0.0215361 4.047 5.23 5.05
Ethylbenzene 50.25 51.93 51.89 53.51 51.90 0.0004331 4.047 0.11 0.09
p-Xylene 524.0 542.0 544.3 550.5 540.19 0.0045081 4.047 1.09 1.06
o-Xylene 115.01 121.90 121.59 120.08 119.64 0.0009985 4.047 0.24 0.23
Total HAP 5,095.74 4,928.21 4,968.39 5,117.57 5,027.48 0.0419563 4.047 10.19 9.79

Avg. Difference
mass in net
Overall Sample 1, Sample 2, Sample 2a, Sample 3, ––––––– Average ––––––– Avg. flow rate HAPs
average µg/ml µg/ml µg/ml µg/ml µg/ml ppg rate, gpm –––––– lb/hr ––––––

Lean total HAP 205.92 186.60 193.06 198.33 195.98 0.0016355 3.753 0.37
Rich total HAP 4,962.33 4,726.59 4,901.03 4,703.23 4,823.29 0.0402523 3.912 9.46 9.09

ND = nondetect or analytical result below the minimum detection limit.

scfd of 1,107 btu/scf gas or 1.06 sisting mostly of HAPs valued at about this fuel because that is where the old
MMbtu/hr. All of this gas was routed to 0.13 MMbtu/gal, or 0.375 MMbtu/hr. dehydrator obtained it; therefore, 0.52
the thermal oxidizer. All these condensates previously went MMbtu must be added back into the
to the thermal oxidizer. still column calculation.
Still column Additionally, the still column over- Total still column overhead emis-
overhead emissions head emissions provided all the process sions no longer wasted is 0.375 +
During testing, the new dehydrator fuel for the new system. The Kimray 0.520 = 0.895 MMbtu/hr.
recovered 2.88 gph of condensate con- pump calculation already accounts for
PROCESSING
About 60% of the vapors burned and Natural Gas Transmission and Stor-
Thermal oxidizer fuel in the thermal oxidizer, however, age—Background Information for Pro-
We did not meter the fuel flow to were methane from gas used by the posed Standards,” EPA-453/R-94-079a,
the thermal oxidizer on the tested unit; Kimray pump, gas stripping, and fuel US Environmental Protection Agency,
however, a parallel dehydrator similarly for the thermal oxidizer. Office of Air Quality Planning and Stan-
equipped with a thermal oxidizer that CO2 emissions from the methane dards, Research Triangle Park, NC, 1997.
processed the identical gas stream was portion of the incinerated gas are less 3. “Preliminary Assessment of Air
metered. That dehydrator circulated 6 than that for higher-btu gas compo- Toxic Emissions in the Natural Gas In-
gpm of glycol. We assumed that the hy- nents. Less CO2 emissions to account dustry, Phase I,” topical report GRI-
drocarbon pickup is similar and that for the methane yields a value of about 94/0268, Gas Research Institute,
the thermal oxidizer duty is directly 680 lb/hr, or a reduction of about Chicago, 1994.
proportional. 2,980 tonnes/year. ✦ 4. “Atmospheric Rich/Lean (ARL)
Fuel consumption in the metered Method for Determining Glycol Dehy-
thermal oxidizer, according to Kerr- References drator Emissions,” Gas Research Insti-
McGee, is 34,700 scfd of 1,107 btu/scf 1. “Methane Emissions from the tute, Chicago, 1995.
gas. Fuel for the thermal oxidizer that Natural Gas Industry,” Vol. 2, technical 5. “National Emission Standards for
the new system replaced was therefore report EPA-600/R-96-080b, US Envi- Hazardous Air Pollutants for Source
(34.7/6) 4 = 23,100 scfd or 1.067 ronmental Protection Agency, National Categories, Subpart HH—National
MMbtu/hr. Risk Management Research Laboratory, Emission Standards for Hazardous Air
The total previously wasted gas in Research Triangle Park, NC, 1996. Pollutants from Oil and Natural Gas
the pumps, gas stripping, still column 2. “National Emissions Standards for Production Facilities,” 40 CFR 63, US
overheads, and thermal oxidizer fuel is Hazardous Air Pollutants for Source Cat- Environmental Protection Agency,
1.14 + 1.06 + 0.895 + 1.07 = 4.16 egories: Oil and Natural Gas Production Washington DC, June 17, 1999.
MMbtu/hr.
At a value of $5.00/MMbtu, the gas
saved is worth more than $182,000/
year.
The authors in industrial technology (welding engineering)
Electric utility costs David A. Kirchgessner from Utah State University, a BS in Art (ceram-
The new process required 11.9 kw of (kirchgessner.david@epa.gov) ics) from the University of Wisconsin, and a 2-
electricity to drive the motors for the has been a senior research sci- year certificate (diesel mechanics) from the Mon-
process pump, circulation pump, and gly- entist for the US Environmen- tana College of Technology. He is a licensed profes-
tal Protection Agency’s Officer sional engineer in Montana.
col cooler. The conventional dehydrator of Research and Development in
required 0.8 kw of electricity to power Research Triangle Park, NC, for Forrest Heath is an engineering
the blower on the thermal oxidizer. 28 years.The last 11 years manager at Engineered Con-
At $0.09/kw-hr, the additional elec- have been spent on research di- cepts LLC, Houston. He has 25
tric consumption of the new process rected toward the quantification and mitigation of years’ experience in research
was about $8,800/year. greenhouse gases from the fossil fuel industries. and development, application,
Kirchgessner received a bachelor’s in economics and specification, design, and man-
Payback a master’s in geology from the University of Buf- ufacture of high-efficiency oil
falo. He also holds a PhD in geology from the and gas processing equipment
Overall savings attributable to the University of North Carolina and a master’s in and combustion systems. Heath
QLT process is about $173,000/year. public health administration. He is a registered holds a BS (1979) in chemical engineering from
The process cost about $300,000 but professional geologist in North Carolina. Texas A&M University. He is a registered profes-
replaced equipment valued at about sional engineer in Texas.
$225,000. The payback on the incre- Robert G. Richards (bob-
mental difference in capital cost is richards@sri-rtp.com) is a Robert D. Smith (rdsmith@
therefore less than 6 months. senior engineer for Southern kmg.com) is the compression &
Research Institute’s Greenhouse process manager for Kerr
Gas Technology Center in Re- McGee Corp., Brighton, Colo.
CO2 emissions search Triangle Park, NC. He He has 25 years’ experience in
The new reboiler consumes about has more than 11 years’ expe- installation, manufacturing,
0.520 MMbtu/hr of a high btu, mixed- rience in environmental engi- and field servicing of gas com-
hydrocarbon stream and produces 111 neering, 4 years’ experience in pression facilities and produc-
lb/hr of CO2. If the thermal oxidizer manufacturing engineering, and 7 years’ experience tion equipment. He currently
on the old dehydrator consumed 4.16 in heavy equipment mechanics. Richards designs oversees the operations of seven gas gathering com-
MMbtu/hr of a similar mixture of and manages test campaigns, commissions a wide pression stations, which include a CO2 amine
high-btu hydrocarbon fuels, it would variety of field and in-house equipment, drafts as- plant and two gas processing plants. Smith holds a
sociated documentation, and is directly responsible BS in general engineering from Kennedy-Western
proportionally produce about 888 for conceiving and implementing testing, data ac- University and a 2-year certificate (diesel me-
lb/hr CO2. quisition, and analysis procedures. He holds a BS chanics) from Lamar University, Beaumont,Tex.