Petroleum Refinery Oil

8/18/2019 Petroleum Refinery Oil

1/33

STACK DESIGN OVERVIEW

INTRODUCTION

This chapter explores the design of stacks from the point of view of thedownwind observer whose task is to determine the connection between stackdesign, process emissions, meteorology, and, most important, environmentaleffects. Stacks must be designed to specifications based on meteorologicalconditions and environmental air quallty standards, which may be quite unrelatedto process requirements

The principal factors which must be accounted for when designing a stackfor air pollution control purposes are the dispersion and transport of thepollutants and the performance criteria against which the stack will be compared.These factors include: (1) air quality standards, (2 ) meteorological conditions,and (3) topographical peculiarities.

The problem of designing a stack to exploit its air pollution control potentiallargely reduces to a problem of determining a stack height which will assurenonpolluting performance. This means designing a stack to meet someperformance standard (usually legally binding) given the meteorologicalconditions, topographic influences, and process exit conditions. Usually, the exitgas conditions are unalterable and the topographic influences are unknown orspeculative. This leaves the meteorology and the air quality standards as thegoverning design criteria.

Historically a 2-1/2 design rule, which states that the height of a stack

should be 2-1/2 times higher th n the nearest surrounding structure, representedthe most reliable way to design stacks for avoidance of ground-level pollutionproblems. As regulators became more concerned wth the effects of increasingpollutant concentrations in the atmosphere, ground-level ambient air standards

339


2/33

340 Pressure Safety Design Practices

were adopted which prescribed maximum olerable ambient air concentrations fora variety of substances. The Clean Air Acts authorized the U.S. Environmental

Protection Agency (EPA) to promulgate Primary and Secondary NationalAmbient Air Quality Standards to protect the health and welfare in the UnitedStates.

Source operators must, therefore, be certain that they are familiar with thestandards applicable to plant operations, for the performance of the source mayultimately be compared against a variety of standards. The rule of thumb indetermining which state or federal standard applies when they appear to conflictis that the more stringent standard prevails.

In addition to environmental regulations goveming the concentration ofground-level pollutants, there also exist Federal Aviation Administrationregulations on the maximum permissible stack height at a given distance from anairport or along air corridors. These regulations may limit the air pollutioncontrol potential of a stack by restricting upper limits on stack heights.

The first step in designing a stack for air pollution control purposes is todetermine exactly what regulatory constraints and requirements exist at the

particular site. These constraints and requirements may be so severe thatalternative means of air pollution control may have to be sought. In any case, theregulations specify a performance standard to which the stack must be designed,and against which the design can be evaluated.

METEOROLOGY

Meteorological conditions, as much as any other consideration, determinehow a stack should be designed for air pollution control purposes. Operatingtransport mechanisms are determined by the micro meteorological conditions,and any attempt to predict ground-level pollutant concentrations is dependent ona reasonable estimate of the convective and dispersive potential of the local air.The following are meteorological conditions which need to be determined:

1. Mean wind speed and direction: the air flow is assumed to be horizontal,but the flow may be tilted (to yield a vertical component) due to local topographic

effects. The mean wind speed determines the convection of the stack emissions.

2 . Intensity of turbulence: these factors, represented by the standard

deviations of the horizontal wind direction, 0, , the standard deviation of thevertical wind component, OZ, and the gustiness as measured by the standarddeviation of the wind speed, l have significant bearing on the dispersion ofemissions from a stack.


3/33

Stack Design Overview 341

3 . Vertical temperature gradient: the lapse rate (rate of decrease intemperature with increases in height) must be taken into account because it

affects the fin height to which a buoyant plume rises.

These meteorological parameters, with the possible exception of the meanwind speed and direction, are not generally available for inclusion in calculations.Even wind speed measurements, which are usually taken at 20 ft above grade,must be corrected to the release point elevation. The correction applied to thewind speed depends on the turbulence of the air. The wind speed is the keydeterminant of the convection of pollutant in a plume.

The vertical temperature gradient (the lapse rate) is usually not monitored byroutine meteorological observation, and it, too, must be approximated fromestimates of solar insolation, solar angle, and differential heating due to unevencloud cover. For purposes of diffusion analyses, the lapse rate is usuallyapproximated by a constant.

The parameters about which the least is known are the diffusion parameterswhich govern diffusion transport of pollutants within a plume. These parameters

are not monitored by meteorological stations and must always be approximatedthrough indirect methods. Figure 1 illustrates the role each of these parametershas in the transport of airborne pollutants.

Experimental works have shown that the vertical distribution of diffusingparticles from an elevated point source is a function of the standard deviation ofthe vertical wind direction at the release point. It is known that the standarddeviations of the vertical and horizontal wind directions can be related to thestandard deviations of particle concentrations in the vertical and horizontaldirections within the plume itself. This is equivalent to saying that fluctuationsin stack top conditions control the distribution of pollutant in the plume. Also itcan be noted that the plume pollutant distributions follow a diffusion relation thatcan be approximated by a Gaussian distribution.

The Pasquill and Gifford approach described later, removes the need to

concentrate on determining Oz and Oy (refer to Figure 1) directly fromweather data. In order to do this, Pasquill introduced the concept of theatmospheric stability class.

Pasquill defined six stability classes ranging from highly stable, low-turbulence Class F, to unstable, highly turbulent Class A, and he identified thesurface wind speed, intensity of solar radiation, and nighttime sky cover as beingthe prime factors controlling atmospheric stability. Pasquill then correlatedobservations of the behavior of plumes in terms of their dispersion with the


4/33


Figure 1. Role of meteorological parameters in transport of airborne pollutian.


5/33


6/33


Figure 3. Eddy informarion in the lee of a cliff.


7/33


Figure 4. Diurnal air circulation effects associated wth valleys.

Figure 5. Effects of large water bodies on pollutant transport.


8/33


pollution transport phenomena. Figure 5 shows how thermal convection windcurrents arising from differences in the temperatures of land and water masses

can influence transport properties. Such convection phenomena may be highlydependent on meteorological and seasonal conditions and may defy the availablemethods to estimate them quantitatively. The pattern shown in Figure 5 isappropriate for an occasion when the land mass is warmer t h n the water, afrequently encountered summer air circulation phenomenon which reverses inthe evening as the land cools off faster t h n the lake or ocean.

It is doubtful that stack design would have any significant influence oncontrolling this type of atmospheric transport problem, and it is more unlikelythat such an influence could be quantified. Hence, if sea breeze- induced airpollution effects were found to be objectionable, the only control optionsremaining to the source operator would be: (1) process reduction during periodsof high air pollution potential, or (2) installation of a stack gas cleaning deviceor continuous flaring operations at high burn rates.

SIMPLIFIED CALCULATION METHOD FOR DISPERSION

As noted earlier, atmospheric dispersion of a pollutant largely depends on:

1. meteorological conditions such as ambient temperature, wind speed, timeof day, insulation, cloud coverage; all of which may be classified as atmosphericstability parameters, and

2. pollutant stack emission parameters such as velocity, temperature and themolecular weight of discharging vapors.

The stability of the atmosphere is largely controlled by the atmosphericthermal gradient. This is normally described by the term lapse rate, which isbasically the temperature change of the surrounding air as a function of altitude.The dry adiabatic lapse rate is referred to as a neutral stability and as a reference,is a temperature gradient of -lo C/100 meters (i.e., air temperature decreases loC for every 100 meters of altitude). This condition implies that a volume ofpollutant in air would neither gain or lose its buoyancy upon being emitted to theatmosphere. Unstable conditions with lapse rates greater th n - lo C/100 metersadd to the buoyancy of an emission, and stable conditions or inversions (lapserates less th n -lo C/100 meters) tend to inhibit verticle motion of the pollutantgases or plume.

Any stack should he designed based on a knowledge of prevailingmeteorological conditions and stack emission criteria based on years of operating


9/33


Surface

WindSpeed (at1 0 dm sec-’

2

2-3

3-5

5-6

6

experience. Highly stable atmospheres or inversions can trap waste gases belowthe mass or relatively hot air, thus limting dispersions and allowing pollutantconcentrationsto build up. Of greatest concern from a health risk standpoint, arehigh ground level concentrations (GLC’s) that may occur over the short term andgenerally are a result of unstable atmospheric conditions. Under unstableconditions, it is more likely for atmospheric turbulence and crosswindsto carrythe plume to the ground, thus exposing humansto potentially toxic emissions.These GLC’s canbe referred to as critical values.

DAY

Incoming Solar Radiation

Moderate SlightStrong

A A-B B

A-B B C

B B-C C

C C-D D

C D D

Critical GLC’s can usuallybe calculated based on a unstable atmosphere,thus enabling the designerto determine a worst case scenario. For any given day,typical atmospheric stability data can usuallybe obtained from a local weatherbureau, ormay be estimatedfrom the so-called Pasquill chart for the appropriateAtmospheric Stability Class (referto Table 1 .

ThinlyOvercast or

418 LowCloud

E

Table 1. The Pasquill Chart for Determining the Atmospheric Stability Class

318 Cloud

F

I

Stability Class

A

Class Description

Extremely unstable

NIGHT

B

C

~~~

Unstable

Slightly unstable

+-+cas t conditions duringday


10/33

D Neutral I

Dispersion from an elevated source is effected by the mixing and ddutionof waste gases within the atmosphere. This is generally accomplished by theturbulent action of the existing gases, and the crosswind, turbulent eddycurrents, wind shear, etc. At the effective stack height, pollutant gases arediluted further by increased wind speeds. Higher wind speeds make avadablemore volumes of air to be mixed with the plume in a shorter time period.However, higher wind speeds also tend to bend a plume, retarding theplume's verticle motion and increasing downwind pollutant concentrations.GLC's are greater for higher wind speeds since the plume is forced to groundlevel before the pollutants can be dispersed over a much-broader region andatmospheric volume. The need to increase the area over which a pollutant isdispersed, as well as remove the emissions from harming surroundingstructures results in tall stacks and the desirability for large plume rises.

E

F

The Dispersion Process. The calculation methods to predict ambientpollutant concentrations are based on a two-step process for dispersion. First,the pollutant gases from a stack rise as a result of their own conditions ofrelease, and then they are dispersed approximately in accordance with aGaussian or normal distribution.

Slightly stable

Stable to extremelv stable

Meteorology plays an important role in determining the height to which

pollutants rise and disperse. Wind speed, wind shear and turbulent eddycurrents influence the interaction between the plume and surroundingatmosphere. Ambient temperatures affect the buoyancy of a plume.However, in order to make equations of a mathematical model solvable, theplume rise is assumed to be only a function of the emission conditions ofrelease, and many other effects are considered insignificant.

Plume R ise

The vertical motion of the plume to the height where it becomeshorizontal is known as the plume rise, (refer back to Figure 1). Theplume rise is assumed to be a function primarily of the emissionconditions of release, (i.e. velocity and temperature characteristics). Avelocity in the vertical plane gives the gases an upward momentumcausing the plume to rise until atmospheric turbulence disrupts theintegrity of the plume. At this point the plume ceases to rise. This


11/33


is known as the momentum plume rise. Stack gas exiting temperatures are usuallymuch greater th n ambient making them less denset h n the surrounding air. Thisdifference in densities gives the gases buoyancy, allowing the plumeto rise untilit is cooled by the atmosphere, reducing the density &fferentialto zero. This isknown as the thermal plume raise. The momentum and thermal plume risescombine to produce the plume rise of an emission. These effects are notindependent: gases with a high exiting velocity are cooled faster as a result ofmore atmospheric mixing of the plume. The thermal buoyancy contributiontoplume rise can, therefore, be lessened by increasing exiting velocities. Low exitingvelocities can cause the plumeto become trapped in the turbulent wake along theside of the stack, and fall rapidly to the ground (referred to as fumigation).Fumigation can usually be preventedby keeping the emission velocity greatert h n10 metershecond. An emission velocity that is one and one-half times greatert h nthe atmospheric crosswind is generally accepted as a safety factorto preventfumigation.

In many calculation methods, the momentum contributionsto plume rise areconsidered negligible when comparedto the thermal plume rise, and hence areignored.

Effective Stack H eight

The importance of plume rise isthat it determines the effective stack height,or the height at which most calculation procedures assume dispersionto initiate.The plume rise addedto the actual height of the stack is known as the effectivestack height, described by the following expression:

h = h, Ah (1)

where h effective stack height (m)& = actual stack height (m)Ah = plume rise (m)

Plume rise can be estimated from empirical relationships; some of the morewell known expressions are summarizedin Table 2.

At the effective stack height, the dispersion of thepollutants are assumed tospread out as a Gaussian distribution. The basic dispersion equation considers


12/33

350 Pressure Safety Desigu Practices

Table 2. Maximum Ground Level Concentration forA 5-min. Blow (Worst Case)

Author Expression

1. Holland

h = [ .5 2 . 7 1 0 y)]u,

2 .ConaweS.S3Q,

Ah = ~II 314

3.Stinrgke

h = d l . S V, 6 S d ” ’ [ ]]T. = ambient air lemperahxe K)T. - stack gas temperature K)

4. Lucas-Moare-

Spurr

1 3 5 Q y

Ah =

5 Rauch4 7 . 2 Q r

M i11

6. Stone-Clark

e,”’Ah = ( 1 0 4 . 2 0.171 h,)

h, physical stack height (m)

7. MosesandCamon U

h = ( - 0 . 0 2 9 V, d 5.53 QF)

A - coefficient dependent on atmospheric stabilityStabil i ty

UnstableNeutral 1.08Stable 0.68

Comment

Highly empiricalformulationoflimitedapplicability.

Basicallyaregressionequalion. Suitedmore to largebuoyant plumeappliaticms.

h e n t i a l l y t h esame asHolland’sformula, exceptbuoyancy termdepends on 1/4power.

Regressionequation based

an work ofPriestly.

Same as Lucaset aL butdifferent database

A modificationof the Lucas-Moore-Spurrequation toaccount for theeffect of thephysical stack

Regressionequationdeveloped frommany datasource&


13/33

Stack Design Overview 3 5 1

Table 2 continued.

.Briggs Unstable, neutral conditions

h = 0 . 2 5 Q : h Yu

b. Stable conditions

_ =

I

ae

JZvariations of potential temperature with height -0.03 K/m

Nonempiricalformulation

a continuously emit- point source emanating through a coordmate system withits origin at the base of the source as shown in Figure 1 .

Emissions of gases or particles lessth n 20 microns (larger particles settlemore quickly due to gravitational effects) disperse with anor ig in and plumecenterline at the effective stack height. Pollutant concentrations are greatestwithin one standard deviation of the plume centerline.Thus, he determinationof the value of these standard deviations isan important factor in calculatingambient concentrations.

The standard deviations in the vertical direction0, nd in the horizontaldirection Oy, of the dispersing plume along the centerline are functions ofmeteorological conditions and depending on the evaluation method, downwind

. distance also. These dspers ion parameters and the effective stack height canbecalculated in a number of ways, using various empirical constants. Each mannerof calculating these parameters, Oz, and Oy defines a different calculationmethod without disrupting the basic Gaussian calculation procedures. Thesemethods are summarized as follows: The basic dispersion equationto calculateground level concentrations directly downwind from a point source is:

where: = ground level concentration (gm/m3)Q = pollutant exiting rate (gm/sec)Oz, U y = horizontal and vertical plume standard deviations (m),

hence are a functionof distance x downwind fromthepoint source.


14/33

352 Pressure Safety Desigu Practices

-u s

h effective stack height (m)

mean wind speed at height of stack (m/sec)

The maximum GLCis of major importanceto the refinery in determiningwhether or not emissions meet maximum permissible discharge concentrations.It can be calculatedas follows:

where: e = 2.7181

Th ree commonly used dispersion calculation methods for the prediction ofground level concentrations are based on the above expression. The variance ineach method is the calculation of plume rise, Ah, and the horizontal and verticalplume dispersion parameters. These methods are:

1) The ASME plume rise equations and the ASME dlspersion parameters.

2 )equations.

The Pasquill-Gifford dispersion parameters andBrigg s plume rise

3) The Pasquill-Gifford dspersion parameters and Holland's plume riseequations.

The parameters 0, and '5, are determined through the use of one of theabove calculation procedures and substituted into equations1, 2 andor 3 to obtainGLC's .

The ASME DispersionCalculation Method

ASME Plume Rise Equations

The ASME method is one of the few calculation methodsto consideremissions having relatively low exitmg temperatures and relativelyhigh exitingvelocities. Under these conditions of release the momentum effectsof the plumedominate over thermal, thus the momentum plume riseshould be used in theGLC calculations.

If V, z 10m/sec andTs > 50' K +T, then


15/33

A h = [ 1.4


(4)

where Ah = the plume rise (m)D = stack diameter (m)V,u,

T,T, = ambient temperatureO K

= stack gas exiting velocity (m/sec)= mean wind speed at heightof stack (m/s)= stack gas exiting temperatureO K

The thermal plume rise is based on the relative buoyancyof a plume to thesurrounding atmosphere. When the stack gas exiting temperatures are50 Cgreater th n ambient temperature and relatively large volumes of gases are beingdischarged, the following set of equations shouldbe used to determine plumerise.

For stable atmospheric conditions:A h = 2 . 9 - -)'

where:

F = 2.45 V, D 2 i$-'a

= The atmospheric temperature gradient,%/lOOmA z


16/33


For neutral or unstable atmospheric conditions:

7.4h ,2 /3F1 3h =

us

The ASME Dispersion Parameters. The horizonal and vertical dispersionparameters are represented by the following empirical power law equation

a Z , oy = a x

where a,b= the empirical ASME dispersion constants gven in Table 2.

x = downwind distance from the source

Table 3 . ASME Dispersion Constants, a and b.

The Pasquill-Gifford-Briggs dispersion method:

Brigg's Plume Rise Equation

for x s 3 . 5 W

(7)

1.6 F l 3 x2I3A h =

us


17/33

and x>3.5W

Atmospheric Stability

slightly unstable

1.6 ( F ' . 3 ) (3.5 W)2'3A h =U

K*

1.1

W=14FSt8 if F< 55

unstable

W=34F2I3 if F>55

1.2

w h e r e


(9)

=2.45 VsD 2 r

s -Ta

x = &stance between source and receptor.

The Pasquill-Gifford aspersion parameters arefunctions of downwinddistance and meteorological conditions.The parameters, 0, and Oy may beobtained from Figure 6(a) and 6(b) respectively. The user must know theatmospheric stability as wellas downwind distance from the sourceto select theappropriate dispersion parameters

The Pasquill-Gifford-Holland Dispersion Method.

Holland's Plume Rue equation: for x> 300 meters

where K, = stability class factor givenin Table 4.


18/33

4

: g , a

a

s

z

9

2i

. I

.

684

8

.*t

kz

3 2

a

5 8

s

-%3

TA

s ts

g

2

a

2

2 3

d

k

a

s

d

m

o

ED

t

s

a

-

E


19/33

Stack Design Overview 3 7

neutral

slightly stable

stable

Table 4 continued.

1.0

0 9

0.8

This calculation method uses the same Pasquill-Gifford dispersion para-meters as the Pas@-Gifford-Briggs calculation method. Therefore, the

dispersion parameters0, nd OZ for this method may be obtained from Figures6(a) and 6(b) respectively.

Averaging Time

The calculation procedures just described are methods for predicting short-term GLC's for single stack cases. In the calculation of short-term GLC's it isassumed that meteorological conditions are constant throughout the measuringtime period. Short-term time periods are usually considered to be1 hour or less.The averaging times for the various methods discussed are as follows:

MethodASMEPasquill-Gifford-BriggsPasquill-Gi ford-Holland

Averaging Time60 minutes10 minutes10 minutes

The problem the plant operator faces is that governmental regulations mayspec@ 10, 15, 30 or 60 minute averaging times and one of the above calculationprocedures. A simple means of converting from one averaging time to anotherin order to compare methods, regulations or calculated and measured GLC's ofdifferent averagmg times is given below:

where x1 = GLC of averaging time period t,x = GLC of averaging time periodt, = averaging time period1

= averagingtime period2


20/33


a = an empirical constant equalto 0.17

The above calculation procedures canbe used to assess several dispersionproblems including the determinationof

1) GLC’s effected by the current refinery practices.2 ) GLC’s effected by plant process modifications, i.e. emission

3) GLC’s effected by plant process ad&tions.4) stack blower requirements and process operating conditions.5 ) minimum stack height requirementsto meet safe exposure limits.

quantities, velocities and temperatures.

The maximum GLC as determined by equation3 is dependent on specificmeteorological conditions and downwind &stance from the source. The criticalGLC of a source is the maximum possible GLC that can occur regardlessofmeteorological conditions and downwind &stance.

The determination ofthe critical GLC is atri and error computation of GLC’sdue to various wind speeds, atmospheric stabilities anddownwind distances. Themaximum value obtained from these proceduresis the critical GLC. Because of thenumber of computations involved, calculations shouldbe performed on thecomputer. Software simulationis also necessaryto calculate GLC’s due to multiplestack cases. Wind direction is an addtional variable that mustbe taken into accountwth multiple stact cases.

Long-Term Average GLC’s

Long-term averaging periods are normally consideredto be 24 hours orgreater which is too long a time interval to assume constant meteorology.Therefo re, long-term average GLC’s are calculatedwith the use of actualmeteorologcal data to predict monthly, seasonal or annual average ambientconcentrations.

The weather data are based on thousands of observations ofwind speed, winddirectionand atmospheric stability taken overthe desired averagmg interval at localweather bureau stations.

Software system generally include one or more of the GLC’s calculationmethods previously discussed.Input requirements wouldalso include:


21/33


1) A coordinate systemto identify stack and receptor site positions.2) Stack coordinates.

3) Average stack emission parameters.4) Meteorological data for the desired averaging period.5) Receptor site coordinates.

Hundreds of computations are neededto generate the long-term averageambient concentration at various receptor sites dueto one or more continuouslydischarging stacks.

EXAMPLE OF REACTOR VENTING EPISODE

l h s section illustrates by way of example, the application of simplifieddispersion estimates to assessing a catastrophic venting operation.In thisexample, an analysis was performed to predict the fate of air pollutants,specifically vinyl chloride monomer (VCM ),origmating from an episode typeupset (reactor blow) condition from a reaction vessel.

Principal emphasis of the analysis is an estim ation of ground level conditionswithin proximity of the plant nd the intermediatesurroundmgcommunitywithina 5-mile radius of the plant, and potential commuuity impact.

The basis of the study was to predict ground level concentrations basedona worst/worst case or episode condition scenario, i.e.,th entire contents ofa reactor venting in three (3) nd five (5) minutes and estimating the

concentrations over a broad range of weather conditionsnd surface Windspeeds. Ground level concentrations were calculated by predictive means ofcomputation nd compared on several bases. The figures presented are basedonsuch computationsand best professional judgm ent analysisto show the degree towhich ground level concentrations maybe anticipated under the extremeconditions assumed, and from which a relative risk factor can be assigned.Values computed and worst case conditions summarized are supported in datashown in the discussions below.

In general, ground level concentrations resulting from stacks are a functionof meteorological conditions such as stability of the atmosphere, Wind speednddirection, atmosphericmixing height and ambientair temperature, stack height,stack diameter, exit stack gas speed and tem perature and other factors.The peakemission rate over a three nd five minute venting of an entire reactor contents(extremecondition of failure)is defined in Table 5.


22/33


Item

VCM

V. Acet.

Chain Modifier

HZO

Total

Based on these peak emission rates and other parameters thedata shown andcomputed were developed.

Charge to Reactor Q kg/s)

(Lbs) (kk9 3 Min. Venting 5 Min. Venting

= 13,000 5,900 32.6 19.6

~ 2 , 3 0 0 l,OoO* 5.7 3.5

= 200 100* 0.59 0.35

~21 ,600 9,800* 54.5 32.7

=37,000 16,800 93.4 56.0

Table 5. Initial Data Sheet for Reactor Venting Episode

Discharge Velocity ds)

Stack Height 45 ft. (13.72111) (above grade)

Stack Diameter 8 in. (0.203m)

VCM Specific Gravity = 2.2/Density = 2.467 h / m 3

1,170 700

Direction Wind From) % of Time

Mean Ambient Temp. 50 F (283OK)

Stack Effluent Temp. 155 F (341.3' K)Molecular Wt. VCM 62.50.

To convert mg/m3 multiply by 0.3584 at STP. to ppm.

To convert mg/m' multiply by 0.3584 at to ppm.

*Considered as inerts from an air pollution standpoint.

Distance

Tables 6 provides a summary of the predicted GLC s under worst casemeteorological conditions. Note that the ground level concentrations summarizeddo not consider wind direction nor variation. More specifically,two points wthnthe plant vicinity are presented to summarize wind persistenceas related to knownlandmarks:

NE

NW

5.2 High-school


23/33


Maximum ground level concentrations estimated from the (episode) casecondition assumed herewith within close proximity of the plant (upto 5-miles)

range as follows:

Within one hour of incident 1 .6to 47.0 ppm.Within three hours of incident 1 .3to 39.5 ppm.Within twenty-four hours of incident0.3- 9.9 ppm.This additionally assumes no change in weather conditions or compensatory

diffusiodd ispersion withinthe atmosphere.

Alternative Considerations and Qualifications

Because of extreme venting conditions assumed, effective stack heights andresultant plumes from both 3- and 5-minute discharge conditions attain heightsbeyond the micro-meteorological conditions assumed in accepted computationmodels. It is therefore highly probable there will be considerably furtheratmospheric dispersion and diffusion of theVCM th n predicted in the resultsshown. That is, the ground level concentration canbe expected to be considerablylower t h n the values shownin Table 6 .

Table 6. Summary of Predicted/estimated Ground Level Concentration atWorst (Episode) Case Conditions and Weather.


24/33

3 6 2 Pressure Safety Design Practices

" tY

D

D

C b s Surl ace Wind Worst Csue Ground Level ConceoWi on ppm) Diatance FromDescription Speed (MPH) Plant (miles)

1 hour 3 haun 24 haun

-//- 10.2 23.2 19.5 4.9 2.0

-//- 6.1 12.3 10.3 2.6 5.0

There is some question as to whether the total reactor contents will bedischarged as assumed. If the reaction isin progress, then material alreadywillhave polymerized and will most probably not ventto the atmosphere, thus

resulting in a sigmficantly smaller emission.

In the calculations that were madeto predict ground level concentrationsfrom a VCM reactor blow off, thePasquill-Gifford-Holland dispersion modelwas used as a basis for these estimations. Calculations were made forsixdifferent stability classes and ground level concentrations, and at variousdistances from the point source of emission.

No correlation was made initially asto wind direction, norto probability ofany wind directiodweather condition or percent time of occurrence, however,this is certainly an important factor in the probability of the pollutantbeing at aconcentration predicted. The greatest significance is attachedto predicting anultimate ground level concentration from any potential episode.

The range of ground level concentrations varies widely depending on windspeed, distance from the emission source, duration of the emission.

Assumptions and Model Predictions

The Pasquill-Gifford-Holland dispersion method was used for the aboveexample. Note that vinyl chloride monomer (VCM) constitutes the primary activeingredient in the reactor. It was therefore assumed that:

(a) the density of the gaseous effluent mixture was that of VCM ,

(b) ground level concentrations and maximum GLC's were computed forand

VCM alone.

Calculations were performed for two different venting rates, namely3 and5 minutes. Note that the calculation procedure predicts short-term GLC's for thesmgle stack. This means that meteorological conditions are assumedto be


25/33


constant throughout the measuring timeperiod.

Tables 7 and 8 give the short-termground level concentrations for a 3-minuteventing at 1/2 , 1.0, 1-1/2, 2.0,and 5 miles from the dischargepoint respectively.The calculations assume meteorological conditionsto be constant forapproximately 10-minutes. Valuesin Tables 7 and 8 were computed for thesixstability classes over a range ofwind speeds.

Table 7. Short-term 10 Min. Period) Ground Level Concentration for a3-min. Blow


26/33


Table 7 continued.

5.0 mi.

29.13

21.11

17.53

12.16

9.26

13.25

36.31

46.59

38.61

33.18

23.m

18.46

0.239

8.66

45.13

63.33

62.07

51.72

42.47

30.50

Table 8. Short-term 10 Min. Period) Ground Level Concentration for5-min. Blow


27/33


Table 8 continued.


28/33


C

C

D

Table 8 continued.

2.5 50 3219 2.0 180.1 166.4

2.0 38.8 8049 5.0 221.7 208.0

15.0 15.70 805 0 5 38.9 25.2

Tables 9 and 10 give the maximum ground level concentrations expected forworst case conditions for the3-min. and 5-min. venting rates, respectively; worstcase conditions were obtainedfrom Tables 7 and 8 @e., values were calculatedbased on the largest concentration foundin Tables 7 and 8 for each distancewthn a stability class).

Table 9. Maximum Ground Level Concentration for a 3-min. Blow(Worst Case)

http://14145__11_%20stack%20design%20overview_15.pdf/


29/33


Table 9 continued.

Table 10. Maximum Ground Level Concentration for A 5-min. Blow(Worst Case)


30/33


Table 10 continued.

SAMPLE CALCULATION

We shall now provide a second example to illustrate step-by-stepcalculations. In this example a flare stack is estimatedto be 80% efficient incombustingH,S off-gas. The total off-gas through the stack is400,000 g/hr, ofwhich 7 .0 weight percent isH,S. The physical stack height is250 m, the stackdiameter is 5.5 m, and the stack emission velocity is18 m / s . The stack emissiontemperature is 15°C. The meteorological conditions maybe describedas a brightsunny day with a mean wind speed of3 d s .

a)We Wish to determine theground level concentration of at a point2000meters downwind from the source.

b) Determine the maximumGLC for the above emission parameters andatmospheric conditions.


31/33


Part a) Solution:

Step 1 - Determine the pollutant quantity source strength

288 K h o u r o.80Q = 400,000 k X 0.070 X

r 273 K 3600sec

= 6.564kg H 2 S X 1 0 0 0 g m l k g =6,564gm H 2 S I s e c

Step 2- Determine the atmospheric stability class:

From Pasquill's chart (Table l), for a bright summer day with windspeedsof 3 mls, the atmospheric conditions can be described as Class B, unstable.

Step 3- Determine the plume rise:

In t h ~ s xample, we shall apply the Holland plume rise equation (Equation10). There are, however other expressions that may be used (refer to Table 2 forother formulae):

= 238.0 m

Step 4- Determine the effective stack height

h = A h + h = 238.0 + 250 = 488 m


32/33


Step 5- Obtain the dispersion standard deviation coefficients and compute theGLC downwind from the source.

From Figures 2 and 3, the Pasquill-Gifford dispersion coefficients areobtained for a downwind distance of 2000 meters and for atmospheric stabilityClass B .

cJz =2 30 m and 0,=300m .

The ground level concentration (GLC) 2000 meters downwind from the

source cannow be computed from Equation (2):

- 6 5 6 4 g m l sIT ( 3 m l s ) ( 3 0 0 m ) (2 3 0 m ) 2 (230m) '

= 0 . 0 10 1 X e x p -( 2 . 2 5 0 9 ) X 1 0 6 p g l g m .

Part (b) Solution:The maximum GLC for the above meteorological conditions is estimated byapplying Equation 3:

- 2 ( 6 , 5 6 4 g m / s ) ( 2 3 0 m )( e ) ( I T ) 3 m / s ) ( 4 8 8 m ) 2 (3 0 0 m )

= 0.00 16 5 gmlm H,S

o r x = 1 . 6 5 X 1 0 3 p g / m 3H,S

The Pasqd-Gifford-Holland method predicts GLC's basedupon 10 m inuteaveraging times. However,what is also of interest is the concentrationof S


33/33


based on a 3 hour averaging time. The results maybe convertedas follows:

\

A \ l o m i n u t e s )

= 1 0 0 9 p g / m 3H2 S

This concentration shouldbe compared against permissible exposurelimitsestablished by OSHA (Occupational Safety and Health Act) and allowabledischargeslimits set by federal (EPA) and local emissions regulations.

It should be remembered that the concentrations predicted from the equationsare merely estimatesto be used in guiding the designer. They canbe used as acheck on the stack height and discharge conditions, and certainly canbe used asa basis to justify either a taller stack or higher burning efficiency for the flare.The estimates are not however a guarantee that the designw i l lperform to 100%of the specifications. Some judgement basedon the experience of the designerand the plant's operating history should alsobe applied to the design, andultimately, ambient air monitoring should be viewed as the only methodofaccurately determining the atmospheric effects of emissions.

Petroleum Refinery Oil

Documents

Transcript of Petroleum Refinery Oil