Society of Petroleum Engineers

SPE 26588

API Tank Vapors Project

M.S. Choi, Conoco Inc.
SPE Member

Copyright 1993, Society of Petroleum Engineers, Inc.

II
This paper was prepared for presentation at the 68th Annual Technical Conference and Exhibition of the Society of Petroleum Engineers held in Houston, Texas, 3-6 October 1993.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are sUbject to pubiication review by Editorial Committees of the Society
of Petroleum Engineers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment
of where and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A. Telex, 163245 SPEUT.

Abstract to quantify its emissions. These emissions inventories

must be broken down by components due to the distinction
The E&P industry has tens of thousands of field storage placed upon the category of Hazardous Air Pollutant (HAP).
tanks in produced oil service. Most of the tanks do not Of the 189 HAPs listed by the CAAA, the compounds most
have vapor recovery and the evolved gases are vented to commonly found in produced fluids are: n-hexane, benzene,
the atmosphere. With the implementation of the 1990 toluene, ethyl benzene and xylenes.
Clean Air Act Amendments (CAAA), the industry is faced
with the need to quantify its emissions. These emissions
Tanl:: Vaporg
inventories must be broken down by components due to Flash Qas
the distinction placed upon the category of HazardOUS Air Ols;pl aClCiiNI'MiiIInt Vapor-
BrGat,",ll1g L~ Hl.P enr""I"""
Pollutant (HAP). Vapor Rate?
C<ln'po5ill t. Ion?
(8TEX, N-e6, H2S)
In addition to HAPs, Volatile Organic Compound (VOC)
emissions also need to be quantified for purposes of VENT

regulatory reporting and as design basis for vapor recovery

units (VRU). Many operators have attempted to measure
emissions directly. However high cost and erratic results
that can be directly attributed to the nature of the tank VENT !
system have discouraged the use of field measurement as !
!
a viable option for estimating emissions. !
!i... .._.._.. .._.._.. _..
01 I Pulp
.. _
In June of 1993, an American Petroleum Institute (API) 50:11 bOlter Tanl<

project was initiated to develop an estimation method for

HAP and VOC emissions from oil field storage tanks. The
method is based on the use of computer process simulation
techniques, predicated on minimal and easy-to-acquire field
data and laboratory analyses. This paper describes the
methodology that will be used and the deliverables that are
expected from the joint industry effort.
Background
The E&P industry has literally tens of thousands of field
storage tanks in produced oil and water services. Most of
the tanks do not have vapor recovery and the evolved
gases are vented to the atmosphere. With the
implementation of the 1990 Clean Air Act Amendments
(CAAA) by the states, the industry is faced with the need
Figure 1
In addition to the HAPs, quantifying VOC emissions from
field storage tanks has also proved to be a problem. The
accepted method of estimating "storage tank" vapor rates
is AP-42. AP-42 was developed for tanks that store
hydrocarbon products (material with Reid vapor pressure of
12 psia or less). In contrast, the fluids that flow into a field
storage tank are usually "live" and have true vapor
pressures upward of 35 psia. Although it is generally
acknowledged that AP-42 method does not provide realistic
emission rates for field tanks, it is nonetheless being used
by most operators for lack of a better alterative.
It was clear that a more appropriate method must be
developed if the E&P industry is to quantify HAP and VOC

791
2 API Tank Vapor Project
emissions for regulatory reporting. In July 1992, Conoco
presented the idea of using process simulation techniques
to estimate tank emissions to API's Clean Air Issues Work
Group. With the help of Chevron, Mobil, Shell and others,
the basic concept was refined and the scope of work for a
joint industry project was prepared for API funding. Upon
execution of the agreement between API and Conoco (with
Conoco as the prime contractor for the project) in June
1993, work on the project began in earnest.
The ultimate goal of the project is to develop and gain
regulatory acceptance for a simplified correlation similar to
AP-42. But before development of a correlation can be
attempted, a data base on the aromaticity of oil (as a
function of its macro properties, i.e. API gravity, molecular
weight, etc.) and the associated HAP emissions must first
be generated. little data is currently available for this
purpose. After refinement and validation, process
simulation will be used to generate a statistically large data
base to search for trends and empirical correlations.
Although it is our intention to develop a simplified
correlation for general use, such a correlation may not
exist. If that turned out to be the case, we would have to
settle on the more costly process simulation method to
estimate HAP and VOC emissions.
Nature of Field Tank Emissions
Many operators have attempted to meter and sample vapor
emissions directly. However the high cost and erratic
results derived from field measurements have discouraged
further development of this method. The poor results can
be directly attributed to the nature of the tank system.
Air emissions (VOC and HAP) from field storage tanks can
be classified in 3 categories:
• Flashing Loss - vapors that are generated when the
pressure of the saturated oil or water is reduced
from that of the low pressure separator to the
storage tank. This source of emission is typically
the largest of the three. The rate and composition
of flash vapors are dictated primarily by the
temperature, pressure and composition of the
incoming fluid.
• Working Loss - vapors that are displaced from the
tank during the fill cycle. Although the emission
rate is essentially the volumetric equivalent of the
incoming liquid (after the flash), the composition of
the vapor is highly dependent on timing in the
system's fill and drain cycles. For example, the
vapors emitted immediately after the tank is
emptied will have a greater concentration of air
(mostly nitrogen and oxygen) than when the tank
is deep into the fill cycle. This is due to air ingress
when liquid is pumped out of the tank.
792
SPE 26588
• Breathing Loss - vapors that are generated when
the tank is heated and cooled during the course of
a day, and liquid hydrocarbon vaporization due to
the stripping effects of air. Typically, the
temperature in a tank will increase during the day
due to radiation heat gain. The energy imparted to
the saturated liquid will cause more vaporization
and emissions. The hotter gas in the air space will
expand in volume and is also vented into the
atmosphere. During the night, the tank will cool
due to conductive and radiation heat losses. The
cooler liquid will reabsorb some of the tank vapor
while the remaining vapors contract. Air is sucked
into the tank, resulting in a new set of equilibrium
conditions.
Difficulties with Direct Measurements
The most rigorous way of determining tank emissions is to
meter the actual flow, sample the stream and analyze the
sample compositionally through Gas Chromatography Mass
Spectrography (GCMS) techniques. Unfortunately,
accurate metering and quality sampling and analysis are
costly to perform and impractical for every location. The
minimum cost of a meter system suitable for the
application is about $30,000, while a sampling event may
require another $30,000 for field technicians and laboratory
analyses.
Not only are field tests costly, many operators have
discovered to their dismay, the data obtained are practically
unusable due to inconsistency in the measurements.
Accurate tank vapor flow measurements are difficult to
achieve due to the low system pressure and often erratic
rates. Since vapor rate and composition are influenced by
ever changing conditions, i.e. temperature, pressoure, fluid
surges, tank level, air ingress, etc., repeatability of "snap
shot" field measurements is poor due to the' inherent
nature of the system. Even if the instantaneous samples
and measurements were .judged to be valid, the
uncertainties introduced by extrapolating the time
dependent data to annual emissions make the results
questionable. To accurately determine tank emissions
through field measurements, continuous meters and
samplers can be utilized to account for system fluctuations.
However due to the expense and practicality of long term
continuous measurements, along with the huge number of
tanks that must be monitored, the industry has ruled out
direct measurements as a viable option for tank emissions
inventory.
Advantages of Computer Process Simulation
Advances in equations-of-state and simple-to-use
computer-based process simulators have provided a low
cost alternative to accurately estimate tank vapors.
Process simulation is not a cure-all. It still requires basic
SPE 26588
knowledge of the tank battery configuration and a limited
number of measured data as input. It does however allow
the use of less expensive and easier to acquire data as
input. while calculating the more difficult to obtain
parameters. Unlike field measurements. a computer model
can be quickly and easily setup and modified to reflect
differences in location and operating conditions. The
procedure being refined and validated in the API project
(fully explained in later sections) can be implemented for
about $500 per tank (or system of tanks in parallel
operation). With timely cooperation from operating
personnel and laboratory. cycle time on an emissions
estimate can be limited to less than one week.
Required Information to Initiate Process Simulation
The information that must be known in order to utilize
process simulation to estimate tank vapor emissions can be
reduced to three data sets:
• Stock tank oil volume.
• One of three commonly used oil characterizations
(RVP, gravity or assay data).
• Complete compositional analysis of the low-
pressure separator oil and the separator operating
conditions (temperature and pressure).
The three were selected on the basis of data availability.
accuracy and cost of acquisition, within the constraints of
process simulation.
__
.. .. .. _--_ .. -.. _-----_ .. _..
SeplrDt.or Oi I R--.......
CcnpoaI't.IQf"I
~~~.
C;1 .. C2.. G3,I-04.. N-C"1 .. ! 011 Tart:::
~:~~:~~CB~ CB.. ~~~._ •• _ •• _ ••_ •• _ •• _. __ •__ •• _ •• _. __ ~~I•• ~= ._j
Figure 2
Sale oil volume is probably the most accurate measurement
maintained by a lease operator. It is the figure by which
sale revenue is calculated. If the oil is sold by pipeline. the
volume may be taken from the LACT meter. For those
facilities where oil is trucked to sales. volume may be
determined from the custody transfer meter located on-site
or at the delivery point. In any event. the official oil volume
M. S. CHOI 3
sold through the tank system could be obtained from oil
accounting and used as the basis for emissions
calculations. In addition. the product price is often
adjusted by the RVP (Reid Vapor Pressure). API gravity or
assay value of the oil. All three parameters are used to
characterize the volatility or stability of the oil. The
stability of the oil could change significantly from one
season to the next, especially if the facility produces a light
oil and experiences extreme seasonal temperature
variations. For these facilities. the most accurate results
are achieved with simulations performed on a seasonal or
monthly basis (if the data are available). and summed to
derive the annual emissions.
Constituents of the separator oil going into the tank is the
primary factor that determines whether HAP vapors will be
emitted. If the oil does not contain any of the HAP
compounds. none can be vaporized and discharged with
the non-HAP vapors. If HAPs are part of the produced
fluid. then their concentrations relative to the other
constituents and process conditions will determine the
amount vaporized. Therefore a compositional breakdown
of the oil is imperative in estimating HAP emissions. The
information can be obtained only through field sampling
followed by laboratory analysis. The bulk of the cost for an
emissions estimate by process simulation is for handling
and laboratory analysis of the sample.
Process Simulation Approach
To accurately estimate vapor emissions. a steady state
model must simulate the dynamic processes occurring
within the system. As discussed in an earlier section
(Nature of Field Tank Emissions). the three categories of
vapors that contribute to the total tank emissions are from:
I'i'\P ~lselone flashing. working and breathing vapors. The compositions
V.par Rote';>;:G:lA) and rates of the three vapors differ significantly and are
c~rtron?
(BTEX, N- CS, H2S) dictated by ever-changing operating and environmental
conditions. Pressure inside the tank will typically vary from
VENT
_-~-_._ .. - -_.-_ .. -.,'""';
.5 oz/in 2 vacuum to 4-8 oz/in 2 pressure. Tank temperature
! and heat gain or loss are highly dependent on the
i temperature of the incoming fluid and ambient temperature.
ori Rot..
ori 0--. Amount of vaporization due to air stripping is a function of
1\1'"
~Vltlf
the frequency and rate of oil removal (pump-out). Since
'!"y none of the dynamic parameters affecting vaporization of
the oil while in the tank are recorded (as normal practice).
_ !
! simulating all the time-related processes occurring in the
tank is impossible. However. by utilizing the abilities of
process simulation to define the vapor-liquid equilibrium of
a system. a steady-state model can be built to approximate
the results of the process phenomenons within the tank.
It is felt that the tank system may be simulated as a two-
step process:
• Adiabatic flash to model the pressure reduction and
subsequent phase separation.
793
4 API Tank Vapor Project SPE 26588

• Fractional distillation to approximate the
"weathering" effects which are responsible for the
working and breathing emissions.
An adiabatic flash (pressure reduction without a gain or loss
of energy) is the process by which flash vapors are
generated. When the saturated low-pressure separator oil
is throttled across the liquid level control valve, certain
gaseous components of the oil will come out of solution.
The phenomenon can be accurately simulated by an
adiabatic flash operation on the computer. Once the
composition and conditions (temperature and pressure) of
the separator oil are known, the amount and composition
of flash gas liberated can be accurately determined. In the
same unit operation, the relative amount and composition
of the "live" stock tank oil (oil that is in equilibrium with the
flash gas) can also be generated.
Calculating the instantaneous working and breathing vapor
compositions and rates are impossible with the limited
information available on the dynamics of the system.
However, those vapors can be estimated by material
balance techniques around the system. Since the material
that enters the tank (separator oil) must exit either as vapor
emissions or product oil, the vapors may be determined by
difference. The steady state material balance equation may
be expressed as:
components "in the "live" stock tank oil (product of the
adiabatic flash) are vaporized so that the resultant oil will
match one of the characterizations (RVP, gravity or assay
value) available for the product oil. The stabilizer operation
not only establishes the liquid composition that will match
the product oil characterization, but also the relative
amount and composition of the vapor that is in equilibrium
with the oil. By working backward from the product oil
characterization and rate, the amount of air ingress, heating
and cooling cycle, etc. that the oil is subjected to in the
tank are eliminated from the calculations.
The components removed from the oil to achieve the
stabilization (stabilizer gas) are the constituents of the
working and breathing vapors. Therefore, total emissions
from the tank are the sum of the flash and stabilizer gases.
Since annual or monthly rather than instantaneous
emissions are of interest, a steady-state simulation model
may be used and accumulation in the tank may be ignored.
Process Simulation Methodology

Material In = Vapor Out + Liquid Out

FI .."h Gae SteDl11zQr GIS
or on a component basis,

The material balance must be done by inference since not

all of the information necessary to calculate the material
balance directly are known. Out of the information
required, the only known parameters are the inlet stream
composition (N i , "live" stock tank oil resulting from the
adiabatic flash) and the liquid rate (product oil rate from oil
accounting). However utilizing process simulation's abilities
to define the vapor-liquid equilibrium of the system, along
with the known oil characterization, the liquid and gas
compositions (Xi and Vi) can be estimated indirectly.
8
""" I
cr API I
N, c:r_1
E-- 1.0
L+VKJ
Figure 3

and The process simulation method for estimating total vapor

emissions (flashing, working and breathing losses) is a two-
step procedure. In step 1, the vapor-liquid equilibrium of
E Y, = E Kl<J = 1.0 the system is determined through process simulation
techniques. Based on the simulation results, the emissions
The unit operation selected for the simulation is an oil corresponding to the volume of oil processed through the
stabilizer. Through fractional distillation, the lighter

794
SPE 26588
tank is calculated in Step 2. An electronic spreadsheet has
been prepared for this purpose.
The process boundaries around a typical field storage tank
are represented by the dash box in Figure 2. The inlet to
the system is the separator oil (pressurized fluid), and the
outlets are the vapor vent (the emissions) and product oil.
In the computer model, the dynamic processes occurring
within the tank system are simulated as two steady-state
unit operations as represented in Figure 3.
• Adiabatic flash to simulate the pressure reduction
imposed on the saturated oil going from the low-
pressure separator (20 + psig) to the near-
atmospheric-pressure storage tank (typically .5
ozlin 2 vacuum to 4-8 ozlin 2 pressure).
• Fractional distillation to stabilize the saturated oil at
atmospheric conditions to match the product oil
RVP, gravity or assay value. This step simulates
the "weathering" effects incurred by the oil while
in the tank.
Inputs to the process simulation model are:
• Low-pressure separator oil composition (obtained
from laboratory analysis; with HAP concentrations
differentiated from the other Ca, C7 , Cs and Cg's)
and an assumed rate.
• Low-pressure separator operating temperature and
pressure.
• Product oil RVP, gravity or assay data.
In the computer model, the pressurized separator oil is first
flashed to atmospheric pressure and then stabilized
(distilled) to match the product oil RVP, or gravity, or assay
value. The ability of the process simulator to define the
vapor-liquid equilibrium of the system will partition the
constituents in the low-pressure separator oil into the
respective phases according to the laws of
thermodynamics. The pertinent results for estimating vapor
emissions are the combined (flash and stabilizer) gas
composition and gas-oil ratio (combined gas to product oil).
Note that the gas-oil ratio is independent of the assumed
separator oil rate.
With the results from the process simulation as input, the
emissions corresponding to the volume of oil processed
through the tank are calculated in an electronic
spreadsheet. The total quantity of gas discharged is simply
the product of the sale oil volume (actual sale volume
during the period in question which may be obtained from
oil accounting) and the GOR (determined by the computer
model). The total vapor rate, along with the gas
composition (result from the process simulation) yield the
emissions from the tank by component.
M. S. CHOI 5
Process Simulators
Most commercially available process simulation software
has the capability to calculate tank vapors generation.
They differ in flexibility and the thermodynamic property
packages available to perform the required unit operations.
Programs designed for the PC's (personal computers) are
more economical to run than the ones for the main frames.
Conoco is currently using Hyprotech's HYSIM process
simulation program. HYSIM is PC based and is relatively
user friendly. Some other simulators commonly used by
E&P companies for this type of calculation are Simulation
Sciences Inc.'s PROCESS or PROII, CHEMSHARE, GPASIM
and ASPEN. To avoid confusion, the API tank vapor
project has standardized on HYSIM and Ping-Robinson's
equation-of-state for the thermodynamic properties.
Validation by Laboratory Techniques
L~boretory Flesh Apper~tus T n--moco"",l ..
(70F)
r----{p Manometer
r----f-+-f- ---.:C~8:.,ps re)
SOCcc
Genter.....,-
1 - - - - - \ ICnown W<>lghi
Cono5t~nt T_er~ture
Hg PLlllP
Figure 4
The laboratory validation is a procedure developed to
simulate the processes that take place in field storage
tanks. Through a specially configured flash apparatus, the
relative amounts and compositions of the vapor that is in
equilibrium with the stabilized oil are measured in the
laboratory under controlled conditions. Core Lab has used
the procedure to generate validation data for 5 of Conoco's
tank batteries. Figure 4 is a schematic of the laboratory
flash apparatus.
The validation method is based on the same separator oil
sample obtained for the process simulation. After a portion
is extracted for compositional analysis, the remaining
sample is used in the validation procedure. Conceptually,
the procedure may be broken down into the following
steps:
1) The flash apparatus (system enclosed by the
rectangle). with its predetermined volume, is totally
evacuated and weighted. The temperature of the
system is maintained at around 70°F by an air
bath.
795
6 API Tank Vapor Project
2) Adequate amount of the sample is introduced into
the apparatus to raise the system pressure to
approximately 8 psia. The sample is kept above its
bubble point pressure by mercury displacement
until it is flashed in the apparatus.
3) After the system has reached equilibrium, the
entire apparatus containing the flashed sample (gas
and liquid) is once again weighted.
4) The flash gas is analyzed by standard GCMS (gas
chromatography mass spectrography). method.
5) The stabilized oil is analyzed by standard GCMS
method.
6) The stabilized oil weight is estimated by system
weight (determined in step 3) minus the evacuated
weight (step 1).
7) Molecular weight of the stabilized oil is estimated
through freezing point depression techniques. Oil
density is calculated from the oil composition
(determined in step 5).
8) Volume of oil in the apparatus is calculated by
dividing the weight of the oil in the apparatus (step
6) by oil density (step 7).
9) Volume of gas in the apparatus is equal to the
known system volume minus the oil volume (step
8).
10) Quantity (moles) of gas in the apparatus is
estimated by the ideal gas law, i.e. N = PV IRT.
Through the above procedure, the relative quantities of gas
and oil (gas-to-oil ratio), and gas compositions are
determined. Those are the same information provided by
process simulation and used in calculating HAP and VOC
emissions. The only difference is that the laboratory
procedure yields a stabilized oil at a fixed set of conditions
(approximately 70°F and 8 psia). Process simulation is
expected to stabilize the oil to the characterization that is
available on the product oil. For validation purposes, the
simulation will be setup to stabilize the oil to the laboratory
test temperature and pressure. The emissions based on
both laboratory and simulated results will be calculated and
compared for discrepancies. An example ofthe comparison
may be found in attachment 4.
The validation procedure described above is judged to be a
practical method to simulate the flashing and weathering
effects that take place in a tank and to obtain quantitative
data on the resultant streams. Like most complex
operations faced by E&P, it is not perfect. The
acceptability of the way some of the data are treated is
disputable. For example, it is assumed that the system
weight minus empty weight is the mass of the liquid in the
796
SPE 26588
apparatus. This, of-course is not quite true. There is flash
gas in the apparatus in addition to the liquid. The
simplifying assumption is that the mass of the gas is
negligible relative to the liquid. Quantity of flash gas is not
measured directly. Instead it is calculated from the ideal
gas la,!" based on the measured parameters, temperature
and pressure, and the gas volume estimated by difference
from the liquid volume. Accuracy in these measurements
has a large impact on the extremely small molar quantity of
gas in the system.
Validation with HAP contents in Product Oil
Although representative tank vapor samples are impossible
to obtain, good product oil samples can be achieved
relatively easy. Therefore in addition to the laboratory
validation discussed above, a number of product oil
samples will be collected along with the separator liquid
samples. The constituents of the product oil will be
determined by laboratory analysis, with special emphasis
on the HAP concentrations. The HAP concentrations in the
product oil calculated by process simulation will be checked
against the laboratory-derived concentrations. The two
independently obtained results should be reasonably close
in value (within the uncertainties of the laboratory
analytical techniques). This material balance method will
be used as an indirect means of validating our vapor
emissions estimated by process simulation.
Need for Simplified Correlations
Although process simulation appears to be the answer for
field storage tank emissions inventories, the estimated cost
of $500 per tank system is still relatively high. Given the
tens of thousands of tanks the E&P industry has in service,
the total cost of a process simulation-based inventory
effort would be substantial. In addition, emissions
inventories are typically compiled by operating personnel
who have little working knowledge of process simulation.
The cost of training these personnel to run the computer
model would increase the cost even further.
Therefore the ultimate goal of the project is to develop a
simplified, empirical correlation or correlations (similar to
AP-42) suitable for general application. By using a
statistically large data base derived from process
simulations, emissions trends could become apparent as a
function of easily obtainable field data such as: stock tank
oil and separator gas specific gravities and rates, ASTM D-
86 distillation results, oil molecular weight, low pressure
separator pressure and temperature, etc.
Problems with Developing a Correlation
Oil and natural gas are not pure compounds, and as such,
their volatility and equilibrium compositions depend on the
SPE 26588 M. S. CHOI
constituents in the original reservoir fluid and process
conditions. Since the CAAA through the HAPs
classification has targeted the aromatic hydrocarbons and
n-hexane for more stringent treatment, these constituents
in oil must be fully quantified.
Conventional characterizations such as specific gravity,
vapor pressure, GOR and water cut do not provide any
indication of the amount of HAPs in the produced fluid or
in the oil. Without knowing the concentrations of HAPs in
the system, determining emissions from storage tanks is
impossible. There is no theoretical basis for predicting
HAPs content in reservoir fluid. Actual compositional
analyses are required to establish the presence of HAPs.
For these reasons, simplified correlation(s) based on readily
available parameters to estimate emission rates of the
components of interest may not be possible.
However, the industry's knowledge of oil aromaticity is
very limited. When an adequate data base covering a range
of process conditions and reservoir fluids is developed,
correlatable trends against easily measured parameters may
become apparent. This effort will have to be purely
empirical. Conoco believes process simulation is the most
cost effective method to generate the data base necessary
to develop the correlation(s).
Data Base Generation
To achieve the goal of developing a simplified correlation
(similar to AP-42) for general application, a large data base
on aromaticity of oil and relationship to HAP emissions is
needed. Since there is no theoretical bases for prediction,
the effort must be achieved through empirical means. A
data base using samples from 200 tank batteries will be
generated by the API project. These 200 sites will be
selected from a list of facilities nominated by the
participating companies with the highest potential of HAP
and VOC emissions.
The project will delivered to a responsible person at the
facility:
• Pressure cylinders (2) for the separator oil samples.
• Sample tags to record facility name, separator
operating temperature and pressure, and any usual
separator configuration that may have an effect on
the tank operation.
• Sampling instructions.
• Sample return shipping instructions.
• Standard Material Safety Data Sheet (MSDS) for
pressurized oil.
797
7
The representative at the facility will collect 2 separator oil
samples according to the sampling instructions. Only 1
sample will normally be analyzed. The second sample is a
backup and will be used in the event the integrity of the
first sample is of suspect. The samples will be shipped
back to Core Lab with the information on sample tags
recorded. If the annual production through the tank system
and oil characterization is available at the site, the
information may be submitted along with the samples. If
not, arrangements may be made to forward the information
from another office.
The separator oil sample will be analyzed by GCMS
methods and its constituents quantified compositionally.
The analysis, along with the separator information will be
use in the process simulation. From the results of the
simulation and the annual oil production, the annual
emissions will be calculated. The emissions data and any
other pertinent information will be submitted to the
company for its use, and become a part of the data base
from which, hopefully a correlation will be developed.
Project Cost & Schedule
The API tank vapor project is estimated to cost $250,000
to complete. The total does not include QAfOC or other
special expenses that may be imposed by EPA or other
regulatory agencies. Conoco is the prime contractor
responsible for implementing the project. Work began in
earnest in June 1993 when the agreement between API
and Conoco was executed. The effort is expected to take
24 months with project completion schedule for mid 1995.
Due to recent changes in EPA's CAAA implementation time
table, the industry may need the estimation method ready
for general use prior to 1995. In that case, the project
schedule may be accelerated.
Conclusion
Process simulation appears to be the best candidate for a
cost-effective method of estimating field storage tank HAP
and VOC emissions. Preliminary review and testing yield
consistent and reasonable results. The API tank vapor
project was created to refine and validate the method, gain
regulatory acceptance for the method as the industry
standard for emission inventory calculations, and if
possible, develop an even lower-cost method in the form of
a simplified correlation.
At the completion of the API tank vapor project, the
industry will have:
• Secured a cost-effective HAP and VOC emissions
estimation method based on process simulation
techniques to calculate emission inventories.
8 API Tank Vapor Project SPE 26588

• Generated a HAP emissions inventory data base

consisting of 200 selected tanks. Owner of those
facilities would have benefitted by having their
emission inventory completed as part of the
project.

• Possibly developed a simplified empirical correlation

for general use.

Nomenclature

Ki Equilibrium ratio of component "i".

Ni Mole fraction of component "in in the inlet
stream.
Xi Mole fraction of component "in in the vapor
phase.
Yi Mole fraction of component "in in the liquid
phase.
l liquid fraction.
V Vapor fraction.
Olnlet Molar rates of the inlet stream.
Ov.par Molar rates of the vapor stream.
0Liquid Molar rates of the liquid stream.

Acknowledgements

The author thanks API and Conoco for permission to

publish this work, and recognizes the important
contributions made by members of the API Clean Air Issue
Work Group.

Reference

1. Choi, M.S.:"Estimation Method Development, VOCand

HAP Emissions from E&P Oil Field Storage Tanks,"
project proposal submitted to API Clean Air Issues
Work Group, October 1, 1992.

798