Statement
he 0.35 m diameter well casing of a sour crude production well breaks off at ground level to
T cause a blowout which spews oil and gas vertically into the air at about 23,000 kg/hr. The temperature of the mixture just before discharge into the atmosphere is 316 K (108 F). It is
r
some time, perhaps several days, before it could be shut OE
The crude assay analysis and flash calculation results of conditions are known. From these, the HsS concentration in
+ expected that the discharge would continue at this rate for
Release Attributes Mateia/: . . . sourcrude oil Method . . . Hole
‘luid state’ . . . Flashing’i~uid Chemical reaction? . . . No Release time type: . . . Steady state Turbulent jet: . . . Vertical
‘‘c/ou# height; . . . . E/e&ed perjet Roughness type: . . . Rural
Stabi’W . . .
Averaging time: . . . 60 s
the vapor phase of the released fluid is about 12 %v.
Dispersion modeling is required to estimate “Worst Case’’
H2S concentrations vs radii from the well. For example, what would be the radius fiom the source for which the H,S concentration would always be less than 10 ppm, for all expected meteorological conditions?
The meteorological conditions typical for the area are hot, humid days with the atmosphere becoming very unstable during the day, with maximum instability occurring in the early afternoon.
At night, the earth can cool enough to sometimes cause stable conditions. Daytime winds as well as cloud cover produce neutral stability.
Analysis
SourcehZelease Parameters
The release to be modeled is a two-phase (vaporfiquid) turbulent jet, and can be treated as steady- state. Because crude oils, in general, contain a large proportion of very low vapor pressure components, the jet would be rising into the air with vapor being disengaged fi-om the liquid, which mostly falls to the ground. The fluid will be intensely mixed with the air in the near-field by the turbulent jet. The wind causes the vertical plume to “feather-out” (which our models cannot handle). Enough of the liquid should be distributed as an aerosol to cause the major part of the plume to S i to the ground. Also, because of their relative densities, the vapor and liquid phases would travel at different velocities (“slipping”) with insufficient mixing time to allow phase equilibrium; this process starts in the well casing pipe. As the fluid is mixed with the air, the liquid and vapor phases will be constantly changing in composition due to evaporation of liquid droplets and mixing with air.
No modeling programs are available which adequately treat ail these phenomena, so simpliGing assumptions must be made. Therefore, assume that the vapor and liquid are in homogenous equilibrium. Given the crude oil assay for the composition of the oil, and the before-release temperature and pressure of the oil, a vapor/liquid equilibrium (“flash”) calculation can be used to estimate the released phase compositions along with physical and thermodynamic properties.
Typical information for a real situation was provided to the author for this example, Table S4-I.
However, the modeling programs being used for these examples cannot directly handle multi- component, multiphase mixtures. SLAB can handle simple phase equilibria for single component e r e ) fluids. DEGADIS has been designed to handle fluid/air mixtures of arbitrary densities; data are “triples,” each of which contains: mole fkaction “contaminant” (basis a contaminadair
Copyright American Petroleum Institute
--`,,-`-`,,`,,`,`,,`---
A P I PUBLm4b28 9 6 = 0732290 0560097 5 2 9
S4-2 Chapter 6
mixture), concentration of contaminant (kg/m3), and total mixture density (see Appendix II).
HGSYSTEM’S HEGADAS can treat pure component evaporating aerosols, but the PGPLüME turbulent jet can only deal with ideal gases. Therefore, the initial modeling was done with SLAB, further simulations made with DEGADIS using the triples option, then HGsYSTEM was used for a vapor- only release.
Pseudo-Pure Component Properties Estimation Refemng to Table S4-I, note that essentially all of the low molecular weight com- pounds, H2S through I-pen- tane, are contained in the va- por phase, with n-pentane distributed about evenly be- tween the phases, while al- most all of the higher molec- ular weight compounds are in the liquid phase at this release temperature and pressure.
The lower part of the table shows parameters for the two phases as well as for the total mixture.
SLAB uses the pure-compo- nent equation for calculating the mass or mole fraction vapor presented in Chapter 3 to calculate the fraction va- porized:
Table S4-1.
CRUDE OIL FLASH RESULTS - TW-PHASE SUCICIARY Temperature: 315.5 K = 108.2 F
Pressure: 0.101325 MPa = 1 atm
Cumla-
-
t t i v e
Ccmonent N5P.K Mol. Ut. Liauid Varxir Total Mass %
H2S 34.15 0.49 11.63 6.59 2.47
CO2 N2 Methane E thane Propane I-Butane n-Butane I -Pen tane n-Pentane Hexanes Heptanes Octanes Nonanes Other
44.01 0.04 2.04 1.14 28.02 0.00 1.40 0.77 16.04 0.18 41.78 22.98 185 30.07 0.28 12.37 6.90 231 44.10 0.95 12.08 7.05 263 58.09 0.46 2.58 1.62 273 58.13 1.53 5.98 3.97 302 72.16 1.48 2.46 2.01 309 72.13 2.24 2.82 2.56 ca. 323 91.98 5.68 2.42 3.89 ca. 372 100.99 12.08 1.69 6.39 Ca. 399 112.99 11.97 0.57 5.72 Ca. 424 129.98 10.83 0.18 4.99 212.00 51.81 0.00 23.42 Total flow r a t e x
Mola 1 , kgmo l / s O. 031 99 O - O3878 O. 07076
Mass, k g / s 5.09867 1.34217 6.44084 V o l m e t r i c , m^3/s 0.00642 1.00383 1.01025 Phase attributes:
Densi t Y . k d m 3 794.152 1.337 6.376 Specific v ~ l u n e , m-3/kg 0.0013 0.7479 0.1569 Mixture mole f r a c t i o n 0.4520 0.5480 1.0000 Mass f r a c t i o n 0.7916 0.2084 1.0000 Volumetric f r a c t i o n 0.0064 0.9936 1.0000
3.02 3.26 7.31 9.59 13.00 14.04 16.57 18.17 20.20 24.13 31 -21 38.32 45.45 100.00
C AT
f = & (3-67)
To obtain the constant heat of vaporization required by SLAB, the heats of vaporization for pen- tane, hexanes and heptanes were found in Perry’s Sixth. These ranged from about 76 to 85 calories per gram at the normal boiling points, and 88 for all at 25 C; so 83 caVg, which corre- sponds to 347.3 Wkg, was selected . Using the release temperature (TI = 3 15.5 K) minus the normal boiling point (TNBP) for AT, and CpI = 21 18 J/[kg.K], TNBp was varied using the above equation so thatf, best agreed with the vapor mass fraction of Table S4-1, 0.208. A TNBP value of 281 K yieldedf, = 0.210 at 3 15.5 K, and O. 128 at 302 K (the ambient temperature to be used).
The latter fraction appears reasonable, considering the fraction of low boiling point compounds shown, and that actually some vapor will be in solution. Therefore, these parameters were used with SLAB. The initial liquid mass fiaction was 0.792 (= 1 -fJ. The SLAB default option for using these input parameters with the Clausius-Clapeyron equation was used to calculate the equilibrium
--`,,-`-`,,`,,`,`,,`---
A P I PUBL*Li628 9 6 0 7 3 2 2 9 0 0 5 6 0 0 9 8 Li65
Scenario 4: Oil Well Blowout s4-3
vapor pressure. Because the flash was made at atmos- pheric pressure and the given temperature, and the liquid phases constituted a large mass fraction of the total, the expanded jet diameter was set to the casing diameter, 0.35 m.
Atmospheric Parameters
Three sets of meteorological conditions were used for the SLAB two-phase simulations: 1) Pasquill-Gifford stability class D (neutral) with a 5 m / s wind speed and 75% humidity; 2) the very unstable class A with a 5 d s wind to demonstrate the increased vertical mixing; 3) for nighttime conditions, stability E with a 2 m / s wind.
A "receptor" or z-plane height of 1 .O m was used unless otherwise noted. Near the source, ground level heights (z = O) may be significantly lower than for a meter or two higher with elevated plumes. At far-field distances, the vertical mixing will tend to make concentrations insensi- tive to this variable.
Simulation
Two-Phase Releases
Figure S4-1 presents the plume centerline concentrations for the three simulations des-
cribed in columns (runs) 1,2, and 3 of Table S4-2 in which the rows under Estimated Values summarize the various distances to selected downwind concentra- tions as well as maximum plume path heights. As might be expected, the A stability case shows greater dispersion than the neutral case, as the distance from the source increases because of increased vertical mixing. Also as expected, the stable case E shows the highest concentrations.
Figure S4-2 compares the plume half- widths, defined * as
MODELING PARAMETER RECAP Pseudo-Pure ComDonent ProDerties
Molecular weight 34.6
Specific heat,
Normal boiling point, K 276 Liquid mass fraction 0.792
Heat of vaporization, Liquid heat capacity, Liquid density, Kg/mJ
c,, KJ4kS.N 1.75
KJlkg 347.3
KJI[kg*)<I 2.12
794 Source
Release rate, kgis 6.44
Temperature, K 315.5
Well diameter, m 0.35
Release height O
AtmosDhenc variables Stability A
temperature, K 305
wind speed, m/s 5
temperature, K 305
wind speed, mis 5
temperature, K 294
wind speed, mis 2
Relative humidity, % 75
Receptor height, rn 1
Stability D Stability E
Roughness length, q, m 0.03
i n
n
c
t
- O c e
C
0 )
O t
o O
3
Downwind H2S Concentrations SLAB Two-Phase Release Simulations
. . '
1 0 0 ' . ' I ' . - ' " " ' I . ' " " "
O 180 360 540 720 900 1080 1260 1 4 4 0 1620 1800
Downwind Distance, m
Figure S4-1.
* In the output file column labeled bbc from SLAB. The relationship is strictly valid only beyond the
downwind distance where gravity slumping becomes negligible.
Copyright American Petroleum Institute
--`,,-`-`,,`,,`,`,,`---
API P U B L X 4 6 2 8 96 0732290 O560099 3 T l
s4-4 Chapter 6
for a 1 m receptor plane height for the three cases. The largest width shown for the E stability cloud can be ascribed to a com- bination of close-in plume im- pact with the ground, more ini- tial gravity spreading and low wind speed. The large air en- trainment and vertical mixing of the A stability case causes the cloud to be wider than that of the D stability case.
Actually, considering SLAB'S
results with respect to plume height and spreading from a
300
E - 250
Y,
f 200
r"
I 150 2
U
o
S
0 O 50
n
= 100
SLAB Cloud Half-Widths
Two-Phase Oil Release
v
O 200 400 600 800 1000 1200 1400 1600 1800
Downwind Distance, m
qualitative standpoint, most of the liquid in the released oil
would fall to the ground and form a pool. Thus, additional emissions and dispersion will result fiom this area source, with the emission rate being controlled by the concentration of H2S in the liquid and its rate of replenishment at the top surface by difision. General purpose programs which can model these phenomena do not appear to be publicly available.
DEGADIS Modeling
For the triples option, a spreadsheet was developed to calculate aidcontaminant (released two- phase fluid) sets for air mole fraction ranging from zero to one in 30 equal steps. The released fluid roperties were taken from the Total column of Table S4-1, and the air density was 1.15 kg/m (300 K, no water vapor). Other parameters were as for SLAB above.
The DEGADIS program failed with an error message which indicated non-convergence of the numerical integration method. The program did run satisfactorily for the triples test case supplied
w i t hthe program. Various convergence tolerances and initial step s i e s were tried, but to no avail;
therefore, this approach was abandoned.
Vapor-Only Release Modeling
Another way of estimating the downwind distance to which hazardous H2S concentrations could travel is to assume that all of the vapor behaves as a single-phase vertical jet, with source diameter equivalent to the area.which the vapor part of the liquid would occupy during HEM discharge.
This would give, perhaps, an upper bound because the same intensive mixing occurs in the turbulent jet in the near field, but the far field would represent dispersion caused by atmospheric phenomena, not with the additional cloud spreading effect on dispersion. This was done, using ali three modeling systems, using only the vapor properties and rates shown in Table S4-1. The simulations were made only for the neutral atmospheric case; results for the unstable and stable cases would be relatively similar to those for the two-phase cases.
The results are summarized in the right half of Table S4-2, and the downwind centerline concen- trations are plotted in Figure S4-3 (the two-phase curve fiom SLAB is also shown for comparison).
Both SLAB and DEGADIs determined that the plume bent over and formed a ground level, dense Figure S4-2.
!
--`,,-`-`,,`,,`,`,,`---
A P I PUBL*i<4628 9 6 m 0732290 0 5 6 0 1 0 0 9 4 3 m
Scenario 4: Oil Well Blowout s4-5
Table S4-2.
Il
SWMARY OF RESULTS FOR SCENARIO 4 O i l U e l l Blowout
Run No. ===>
Released phases Modeling system PG s t a b i l i t y class Uind speed, m/s Ambient temperature, K Ambient temperature, F Relative humidity, X
1 SLAB
D 5 305.4
90 75
- - - _ _ _ Liquid & Vapor 2 3 - - - -
SLAB SLAB
A E
5 2
305.4 294.3
90 70
75 95
Source Parameters
Release rate, kg/s 6 . 4 4 6.44 6.44 O r i f i c e diameter, m 0.35 0.35 0.35 O r i f i c e elevation, m 0.00 0.00 0.00
4 5 6
. - - - Vapor Only - - -
SLAB DEGADIS HGSYSTEM
D D D
5 5 5
305.4 305.4 305.4
90 90 90
75 75 75
1.34 1.34 1.34
0.35 0.35 0.35
0.00 0.01 2.00
&- .
$
Meters* t o 1000 ppm H2S 2 2 ( a ) 14 None(b) Meters t o 100 ppn H2S I sa 4a 31 O Meters t o 10 ppn H2S 669 157 1790 Max. plume path height, 3 . 4 2.6 1 .o
Meters t o maximun height 1.0 1 .o 1.0 Meters t o 100% vapor 4 . 3 3.7 4.3
* Downwind distance along the centerline. A l l concentrations are f o r z = 1.0.
(a) 34 m for z = O. I
(b) 56 m f o r z = O. I
** ***
The distances t o 100 and 10 ppm were about the same as f o r z = 1 f o r both runs 1 and 3 .
[Downwind distancel/[Elevation f o r the concentration1 Plune became n e u t r a l l y buoyant a t t h i s height and distance.
54 None 25/2.7**
205 214 5 4 0 / 1 . 3 726 879 9 0 0 / 1 . 3
1.9 4.5 4.0***
3 . 2 28.0 8.0
Cornnon Parameters
Averaging time, seconds 60 O r i f i c e diameter, m O .35 Roughness length Zr, meters 0.03 Uind speed measurement height, rn 10
gas cloud. HGSYSTJZM’S PLUME determined that the plume remained elevated (centerline at about 4 meters), so it set up the partial input file for PGPLUME, which in turn estimates dispersion with the Gaussian plume model. The results from the latter are shown in Figure S4-3.
However, it was not possible to get PLUME to run with the source height at zero meters; the height of the source had to be greater that 1.5 m. The smooth curve in Figure S4-3 for HGSYSTEM was obtained by linear (x) - log (concentration) interpolation of PGPLUME results to obtain the 1 m z- plane values plotted.
Conclusions
Analysis of the D stability simulations indicates that predicted downwind distances to given H2S concentrations do not differ significantly with respect to the modeling program used or the source characterizations of aerosol vs vapor only. From this example, it appears that the vapor-only release simulations gave adequate estimates compared with the aerosol simulations, and perhaps with a small extra safety factor. However, jets with different momentum, thermodynamic behavior, and other parameters might lead to markedly different conclusions.
Copyright American Petroleum Institute
--`,,-`-`,,`,,`,`,,`---
A P I PUBL*qb28 9 b m 0732290 05bOLOL B B T m
S4-6 Chapter 6
Downwind H2S Concentrations
Vapor-Only vs Two-Phase Simulation
1 O000
. - SLAB, Vapor Only, D Stobilit sWind _ . _ _ _ .
i - - - DEGADIS. Vopor Only, D Sta
O 100 200 300 400 500 500 700 1300 SOO 1000
Downwind Distance, m
Figure 3. --`,,-`-`,,`,,`,`,,`---
A P I PUBLm4628 9 6 D 0732290 0560302 7Lh
Release Attributes
Matenal: . . . Chlorine Method . . . Large hole state: . . . Flashing liquid Chemical reactions? . . . No
~~l~~~~ tirne type: , . ,=inite duration Tank Rupture: . . . Area source
"c'oud"height: . . .
Roughness type: . . . Rural
tank storage temperature equals the ambient temperature, 303 K (86 F). For this locale in the Midwest prairie/farm country, forecasts call for daytime high temperatures of 303 K with winds from 4 to 8 d s with moderate insolation and minimum nighttime temperatures near 293 K with winds ranging from calm to 3 m/s with < 3/8 cloud cover.