90

SOURCE TES rs - l)UPONT DeNEMOUlS AND COMPANY

~-1 SITE 0VElmiddotVIEW

Summary

feed

Dupont

according to

Antioch produces

the reactions

Freon 11 and 12 from carbon tetrachloride

CC1 HFCCl + HCl (F-11)4 +

CCl + 2HF - CCl c + 2HC1 (F-12)4

Carbon tetrach 1ori de (CT) is off-1 oaded pri mdrily from bottom emptying

ra i 1 road rail cars and stored in a tank vented to the atmosphere Process

feed pumps tranport CT into a reactor which is operated in a continuous

manner Reactor output is fed into a distillation column with recycle back to

the reactor

Emission sources uf CT are expected to be the storage tank and the

fugitive emissions from valves flrnyes and pump seals

Facilities Descrimicrotions

Tank cars containing 20U x 103 lbs of CT are off-loaded into a 570 x 310 lbs capacity storage tank on the order of 250 times per year~ CT is fed

from the tank car by l)ottom un 1oatIi ng and pumped into the storage tank A

feed pump delivers CT to the rectctor wi)ere it is reacted with hydrogen

fluoride Since the HF is nighly corrosive considerable care is taken to

contain all reactants Material is output to the distillation column and

chlorocarbons are recycled to the reactor from the column bottom Beyond this

point there are no enriched CT streams as hydrogen chloride absorbing caustic

scrubbing dhi scrubbing and distillation are accomplished

The single CT storage tank has a 3-inch U-leg vent to the atmosphere

and thus has no vapor recovery system Fittings associated with CT flow are

inspected for leaks and maintained according to plant practices dnd Bay Area Air ijuality District rules on volatile organic emissions The total

number of fittings 1re li~ss than 100

127

It is folt ttldt the sinSJlt~ 10st ir~portdnt source )f CT emissions is

the storage tdnk Tt1ere are two kinds of e1iission train storage tanks (a)

loss due to tcrnk bredti1in9 and (b) middot1orkin i luss due to tank cleaning and

f i 11 i ng Uupont tias made measurements of head spac~ concentrations of CT

during tank filling and also pertori1ed u1or2ticdl calculations oased on vapor

microressure These compliment one awu1er dd ctre 18800 and C(UUlJ lbsyr

respectively Specifically for me calcu dtions tne vamicroor microressure was

taken dt ~odeg C bulk liquid ternmicroerture ccording to f-P-42 the relationship

betv1een v1orling lJss and vapor p2ssure given Dy

wnere M - molecular weight p true vapor prtSSlffP n bu1k liquid conditions (psia) I( turnover fract1on(annual hrougnput)n

tank capacity I

crude oil fractmiddot1on (=l for CT)

working loss (lb)10 3 gal

Uupont did not microredict tne loss due to normal tank breattling during

the year Vapor is expelled from tJO pri11ary mechanisns (1) thermal

expansion of existing vdpors and (2) vctpomiddot expansion caused by barometric

pressun~ changes The AP-42 emission fori ula (EPA 1Y81) is

M( p )U 68 l) 1 7J HO 51 T )5 F C K

p C

147-fJ

where - Lt) = breathing loss lbday

li moleculltlr wt

p true vapor pressure dt bu Imiddot liyuiu condition (microSid)

D tank diameter (ft)

H = dVerafde vamicroor space tll i ~ llt (ft)

T dverage ambient t111p12gtratunmiddot change liurnal (OF I

F = pctint tc1ctur micro

C Sllld 11 tank ddjust111ent fact( r

l( c rude oi l f d c t o r C

128

foe rurrnulct 1s etiildtld to 11e within+ lU~ uf actudl measured values

Fuy1t1ve emissions from leak~ in the re1atiJely feJ (approximately 100)

Vdlves fldnges and microummicro fittings are likely to be of secondary importance to

the stora~e tank emissions rnese however vJill be directly111easured in the

field 1110n i tor nId

9 l 1111ASUKEI1ENT APPRUALH

Top priority is thlmiddot determin1tion of storage tank emissions H1is

Cdn De done for tank workinlJ loss L)_y otdir1ir1q drld drlcil_yLin~ ihnlti smicrodce

Sdrnmicroles duriny tdnk car unloctding lt was proposed to insert a sampling tube

into tanK t1ead space and co 1 Iect a ti me i ntey rated samp1 e over the off - Ioad i ng

period The samp1e would then be transferred to the La Jolla laboratory of

SAI and analyzed directly by gas chromatogrdmicrot1y Normal tank breatning can be

determined sufficiently precisely (_ lC~~) oy utilizing AP-42 with accurate

tank dimensions and meteoroloyy

Since relatively few fuyitive source components exist it unlikely

tndt sucl1 e111issions would be smiddotiynificant ~itn respect to the storage tanks

However it would be cost effectiVt to screen 100 of tne fittings with tt1e

Foxboro OVA since a team would be on site to perform the tank measure111ents

rt1e tugitive screening dmicromicroroacn vwtdltt tol Im- tn~ microrocedure described in

Section 7L

UETtlmiddotMINAT llJN Uf E111SSIUlS

931 Fuyitive Releases

Apmicroroximately llU (lrvice~ were screened and six leaks above the

establisned tt1rest1olltt of LO micropm were recorded Table 9J-1 summarizes tt1e

screeniny ddtd ano 111dss ~mission microrojections Since carbon tetrachloride is

microoorly detected Dy the UVA and TLV instrn1ients fdctors between approximately 5

and 14 ~ere ctpp Ii ed octsed on resiunse t unct i uns to deterrii ne the upper bound

of the leak rate mass emissions Total fuyitive emissions range between S8

and 61U lbyear

93L Storaye Tank tli1issio11s

Head smicroctce sa1i1ples were -caken uuriny tne nearly six t10ur off-loading

interval in order to directly deternine tne CT concentration in the displaced

129

) )

Table 93-1

DUPONT MASS EMISSION PREDICTION FROM OVA SCREENING VALVES

OVA SAi Actual Umicroper Bound Device Response Response Cone Source Leak Rtte Leak Rate Location (ppm) Factor (ppm) a b Se IRc Function lbyr 1byr

--~~--~-

Gate Valve Storage Tank Pump Area

Gate Valve Storage Tank Pump Area

300

700

009

009

3333

7778

466 009 I i

074 I

296

320

D 95

in _ I ~ C~

70

140

Gate Valve Feerl Puir~ Exit Area 350 009 3889 300 97 81 5

Gat e Feed Exit

J l e Pump

700 009 7778 320 l

l0 1 t1 n

Gate Valve 600 009 6667 i I I 315 I 100 127

Valve-Reactor Heat Exchanger 200 009 2222 I I I 285 I 93 52

All measurement are for 100 CT streams

1--- w 0

air smicroace Saturation concentratiun at 20degc is amicroproxiriately 12 Measured

values were bY and 74 These values r~sulted from gas chromatographic

dn a I y s e s ot t rd n s t er re c1 sa111 p I e s fr() m t n e 1U U I i t e r Ted l a r b a y t i 11 e i nt e9r a t ed

sample ror tt1e tank car displdcement of 2008 ft 3 the total CT emitted based

on tile ctverctye is LUU8 x 007L = 144 ft 3bull

Tnen for a density of p =PMRT

P = (1)(1gt4) 64 gL = 04lbft 3

(082)(293)

Tl1erefore tne displacement weiyl1t of CT microer ()ff-load is equal to 144 ftJ x U4

l b ft J = ~ 7 bull 6 1 b bull F o r 2 5 U off - l o ad i n gs an n u c1 1I y we have 14 400 l b s bull Th i s

compares with UuPont I s rnectsure1nents of 9 3 vapor content arid 18800 lbyr

emissions Saturation vapor pressure concentration emission would yield

2400Ulo

l)reathing loss from the tank can be calculated from AP-42 as

Lt3(1bday) = 619 X 10-5 M( p ) 068 D 173 H 051 T 05F CKcmicro 147-P

5 0 68 173 = (619 X 10- )(154( 173 ) bull (155)

147-i73

(lj) 051 (26) 05 (1)(1)(1)

= S 63 lbday

3where a working range of zuu x 10 o 400 X 103 lbs were used to determine an

average l i quid nei grit Jn nu a 11 y e111 i s s ions would be 2057 lb Therefore

working losses dorrii nate tt1e totil elissions from tne microlant

Summary

The upper and lovJer bounos of caroori tetracriloride emission ~vere

determined as Ll467 lbyear (upper bound on fugitives DuPont measurement of

storage tank work i ny erni ss ions nur 1a I tanK oreattli ng) and lbgt 15 loyear

(lower bound on fuyitives s1 111eas--1re111ent of storage tank working emissions

no ri n a I tan k breatt1 i ny ) bull

13]

It is folt ttldt the sinSJlt~ 10st ir~portdnt source )f CT emissions is

the storage tdnk Tt1ere are two kinds of e1iission train storage tanks (a)

loss due to tcrnk bredti1in9 and (b) middot1orkin i luss due to tank cleaning and

f i 11 i ng Uupont tias made measurements of head spac~ concentrations of CT

during tank filling and also pertori1ed u1or2ticdl calculations oased on vapor

microressure These compliment one awu1er dd ctre 18800 and C(UUlJ lbsyr

respectively Specifically for me calcu dtions tne vamicroor microressure was

taken dt ~odeg C bulk liquid ternmicroerture ccording to f-P-42 the relationship

betv1een v1orling lJss and vapor p2ssure given Dy

wnere M - molecular weight p true vapor prtSSlffP n bu1k liquid conditions (psia) I( turnover fract1on(annual hrougnput)n

tank capacity I

crude oil fractmiddot1on (=l for CT)

working loss (lb)10 3 gal

Uupont did not microredict tne loss due to normal tank breattling during

the year Vapor is expelled from tJO pri11ary mechanisns (1) thermal

expansion of existing vdpors and (2) vctpomiddot expansion caused by barometric

pressun~ changes The AP-42 emission fori ula (EPA 1Y81) is

M( p )U 68 l) 1 7J HO 51 T )5 F C K

p C

147-fJ

where - Lt) = breathing loss lbday

li moleculltlr wt

p true vapor pressure dt bu Imiddot liyuiu condition (microSid)

D tank diameter (ft)

H = dVerafde vamicroor space tll i ~ llt (ft)

T dverage ambient t111p12gtratunmiddot change liurnal (OF I

F = pctint tc1ctur micro

C Sllld 11 tank ddjust111ent fact( r

l( c rude oi l f d c t o r C

128

foe rurrnulct 1s etiildtld to 11e within+ lU~ uf actudl measured values

Fuy1t1ve emissions from leak~ in the re1atiJely feJ (approximately 100)

Vdlves fldnges and microummicro fittings are likely to be of secondary importance to

the stora~e tank emissions rnese however vJill be directly111easured in the

field 1110n i tor nId

9 l 1111ASUKEI1ENT APPRUALH

Top priority is thlmiddot determin1tion of storage tank emissions H1is

Cdn De done for tank workinlJ loss L)_y otdir1ir1q drld drlcil_yLin~ ihnlti smicrodce

Sdrnmicroles duriny tdnk car unloctding lt was proposed to insert a sampling tube

into tanK t1ead space and co 1 Iect a ti me i ntey rated samp1 e over the off - Ioad i ng

period The samp1e would then be transferred to the La Jolla laboratory of

SAI and analyzed directly by gas chromatogrdmicrot1y Normal tank breatning can be

determined sufficiently precisely (_ lC~~) oy utilizing AP-42 with accurate

tank dimensions and meteoroloyy

Since relatively few fuyitive source components exist it unlikely

tndt sucl1 e111issions would be smiddotiynificant ~itn respect to the storage tanks

However it would be cost effectiVt to screen 100 of tne fittings with tt1e

Foxboro OVA since a team would be on site to perform the tank measure111ents

rt1e tugitive screening dmicromicroroacn vwtdltt tol Im- tn~ microrocedure described in

Section 7L

UETtlmiddotMINAT llJN Uf E111SSIUlS

931 Fuyitive Releases

Apmicroroximately llU (lrvice~ were screened and six leaks above the

establisned tt1rest1olltt of LO micropm were recorded Table 9J-1 summarizes tt1e

screeniny ddtd ano 111dss ~mission microrojections Since carbon tetrachloride is

microoorly detected Dy the UVA and TLV instrn1ients fdctors between approximately 5

and 14 ~ere ctpp Ii ed octsed on resiunse t unct i uns to deterrii ne the upper bound

of the leak rate mass emissions Total fuyitive emissions range between S8

and 61U lbyear

93L Storaye Tank tli1issio11s

Head smicroctce sa1i1ples were -caken uuriny tne nearly six t10ur off-loading

interval in order to directly deternine tne CT concentration in the displaced

129

) )

Table 93-1

DUPONT MASS EMISSION PREDICTION FROM OVA SCREENING VALVES

OVA SAi Actual Umicroper Bound Device Response Response Cone Source Leak Rtte Leak Rate Location (ppm) Factor (ppm) a b Se IRc Function lbyr 1byr

--~~--~-

Gate Valve Storage Tank Pump Area

Gate Valve Storage Tank Pump Area

300

700

009

009

3333

7778

466 009 I i

074 I

296

320

D 95

in _ I ~ C~

70

140

Gate Valve Feerl Puir~ Exit Area 350 009 3889 300 97 81 5

Gat e Feed Exit

J l e Pump

700 009 7778 320 l

l0 1 t1 n

Gate Valve 600 009 6667 i I I 315 I 100 127

Valve-Reactor Heat Exchanger 200 009 2222 I I I 285 I 93 52

All measurement are for 100 CT streams

1--- w 0

air smicroace Saturation concentratiun at 20degc is amicroproxiriately 12 Measured

values were bY and 74 These values r~sulted from gas chromatographic

dn a I y s e s ot t rd n s t er re c1 sa111 p I e s fr() m t n e 1U U I i t e r Ted l a r b a y t i 11 e i nt e9r a t ed

sample ror tt1e tank car displdcement of 2008 ft 3 the total CT emitted based

on tile ctverctye is LUU8 x 007L = 144 ft 3bull

Tnen for a density of p =PMRT

P = (1)(1gt4) 64 gL = 04lbft 3

(082)(293)

Tl1erefore tne displacement weiyl1t of CT microer ()ff-load is equal to 144 ftJ x U4

l b ft J = ~ 7 bull 6 1 b bull F o r 2 5 U off - l o ad i n gs an n u c1 1I y we have 14 400 l b s bull Th i s

compares with UuPont I s rnectsure1nents of 9 3 vapor content arid 18800 lbyr

emissions Saturation vapor pressure concentration emission would yield

2400Ulo

l)reathing loss from the tank can be calculated from AP-42 as

Lt3(1bday) = 619 X 10-5 M( p ) 068 D 173 H 051 T 05F CKcmicro 147-P

5 0 68 173 = (619 X 10- )(154( 173 ) bull (155)

147-i73

(lj) 051 (26) 05 (1)(1)(1)

= S 63 lbday

3where a working range of zuu x 10 o 400 X 103 lbs were used to determine an

average l i quid nei grit Jn nu a 11 y e111 i s s ions would be 2057 lb Therefore

working losses dorrii nate tt1e totil elissions from tne microlant

Summary

The upper and lovJer bounos of caroori tetracriloride emission ~vere

determined as Ll467 lbyear (upper bound on fugitives DuPont measurement of

storage tank work i ny erni ss ions nur 1a I tanK oreattli ng) and lbgt 15 loyear

(lower bound on fuyitives s1 111eas--1re111ent of storage tank working emissions

no ri n a I tan k breatt1 i ny ) bull

13]

foe rurrnulct 1s etiildtld to 11e within+ lU~ uf actudl measured values

Fuy1t1ve emissions from leak~ in the re1atiJely feJ (approximately 100)

Vdlves fldnges and microummicro fittings are likely to be of secondary importance to

the stora~e tank emissions rnese however vJill be directly111easured in the

field 1110n i tor nId

9 l 1111ASUKEI1ENT APPRUALH

Top priority is thlmiddot determin1tion of storage tank emissions H1is

Cdn De done for tank workinlJ loss L)_y otdir1ir1q drld drlcil_yLin~ ihnlti smicrodce

Sdrnmicroles duriny tdnk car unloctding lt was proposed to insert a sampling tube

into tanK t1ead space and co 1 Iect a ti me i ntey rated samp1 e over the off - Ioad i ng

period The samp1e would then be transferred to the La Jolla laboratory of

SAI and analyzed directly by gas chromatogrdmicrot1y Normal tank breatning can be

determined sufficiently precisely (_ lC~~) oy utilizing AP-42 with accurate

tank dimensions and meteoroloyy

Since relatively few fuyitive source components exist it unlikely

tndt sucl1 e111issions would be smiddotiynificant ~itn respect to the storage tanks

However it would be cost effectiVt to screen 100 of tne fittings with tt1e

Foxboro OVA since a team would be on site to perform the tank measure111ents

rt1e tugitive screening dmicromicroroacn vwtdltt tol Im- tn~ microrocedure described in

Section 7L

UETtlmiddotMINAT llJN Uf E111SSIUlS

931 Fuyitive Releases

Apmicroroximately llU (lrvice~ were screened and six leaks above the

establisned tt1rest1olltt of LO micropm were recorded Table 9J-1 summarizes tt1e

screeniny ddtd ano 111dss ~mission microrojections Since carbon tetrachloride is

microoorly detected Dy the UVA and TLV instrn1ients fdctors between approximately 5

and 14 ~ere ctpp Ii ed octsed on resiunse t unct i uns to deterrii ne the upper bound

of the leak rate mass emissions Total fuyitive emissions range between S8

and 61U lbyear

93L Storaye Tank tli1issio11s

Head smicroctce sa1i1ples were -caken uuriny tne nearly six t10ur off-loading

interval in order to directly deternine tne CT concentration in the displaced

129

) )

Table 93-1

DUPONT MASS EMISSION PREDICTION FROM OVA SCREENING VALVES

OVA SAi Actual Umicroper Bound Device Response Response Cone Source Leak Rtte Leak Rate Location (ppm) Factor (ppm) a b Se IRc Function lbyr 1byr

--~~--~-

Gate Valve Storage Tank Pump Area

Gate Valve Storage Tank Pump Area

300

700

009

009

3333

7778

466 009 I i

074 I

296

320

D 95

in _ I ~ C~

70

140

Gate Valve Feerl Puir~ Exit Area 350 009 3889 300 97 81 5

Gat e Feed Exit

J l e Pump

700 009 7778 320 l

l0 1 t1 n

Gate Valve 600 009 6667 i I I 315 I 100 127

Valve-Reactor Heat Exchanger 200 009 2222 I I I 285 I 93 52

All measurement are for 100 CT streams

1--- w 0

air smicroace Saturation concentratiun at 20degc is amicroproxiriately 12 Measured

values were bY and 74 These values r~sulted from gas chromatographic

dn a I y s e s ot t rd n s t er re c1 sa111 p I e s fr() m t n e 1U U I i t e r Ted l a r b a y t i 11 e i nt e9r a t ed

sample ror tt1e tank car displdcement of 2008 ft 3 the total CT emitted based

on tile ctverctye is LUU8 x 007L = 144 ft 3bull

Tnen for a density of p =PMRT

P = (1)(1gt4) 64 gL = 04lbft 3

(082)(293)

Tl1erefore tne displacement weiyl1t of CT microer ()ff-load is equal to 144 ftJ x U4

l b ft J = ~ 7 bull 6 1 b bull F o r 2 5 U off - l o ad i n gs an n u c1 1I y we have 14 400 l b s bull Th i s

compares with UuPont I s rnectsure1nents of 9 3 vapor content arid 18800 lbyr

emissions Saturation vapor pressure concentration emission would yield

2400Ulo

l)reathing loss from the tank can be calculated from AP-42 as

Lt3(1bday) = 619 X 10-5 M( p ) 068 D 173 H 051 T 05F CKcmicro 147-P

5 0 68 173 = (619 X 10- )(154( 173 ) bull (155)

147-i73

(lj) 051 (26) 05 (1)(1)(1)

= S 63 lbday

3where a working range of zuu x 10 o 400 X 103 lbs were used to determine an

average l i quid nei grit Jn nu a 11 y e111 i s s ions would be 2057 lb Therefore

working losses dorrii nate tt1e totil elissions from tne microlant

Summary

The upper and lovJer bounos of caroori tetracriloride emission ~vere

determined as Ll467 lbyear (upper bound on fugitives DuPont measurement of

storage tank work i ny erni ss ions nur 1a I tanK oreattli ng) and lbgt 15 loyear

(lower bound on fuyitives s1 111eas--1re111ent of storage tank working emissions

no ri n a I tan k breatt1 i ny ) bull

13]

) )

Table 93-1

DUPONT MASS EMISSION PREDICTION FROM OVA SCREENING VALVES

OVA SAi Actual Umicroper Bound Device Response Response Cone Source Leak Rtte Leak Rate Location (ppm) Factor (ppm) a b Se IRc Function lbyr 1byr

--~~--~-

Gate Valve Storage Tank Pump Area

Gate Valve Storage Tank Pump Area

300

700

009

009

3333

7778

466 009 I i

074 I

296

320

D 95

in _ I ~ C~

70

140

Gate Valve Feerl Puir~ Exit Area 350 009 3889 300 97 81 5

Gat e Feed Exit

J l e Pump

700 009 7778 320 l

l0 1 t1 n

Gate Valve 600 009 6667 i I I 315 I 100 127

Valve-Reactor Heat Exchanger 200 009 2222 I I I 285 I 93 52

All measurement are for 100 CT streams

1--- w 0

air smicroace Saturation concentratiun at 20degc is amicroproxiriately 12 Measured

values were bY and 74 These values r~sulted from gas chromatographic

dn a I y s e s ot t rd n s t er re c1 sa111 p I e s fr() m t n e 1U U I i t e r Ted l a r b a y t i 11 e i nt e9r a t ed

sample ror tt1e tank car displdcement of 2008 ft 3 the total CT emitted based

on tile ctverctye is LUU8 x 007L = 144 ft 3bull

Tnen for a density of p =PMRT

P = (1)(1gt4) 64 gL = 04lbft 3

(082)(293)

Tl1erefore tne displacement weiyl1t of CT microer ()ff-load is equal to 144 ftJ x U4

l b ft J = ~ 7 bull 6 1 b bull F o r 2 5 U off - l o ad i n gs an n u c1 1I y we have 14 400 l b s bull Th i s

compares with UuPont I s rnectsure1nents of 9 3 vapor content arid 18800 lbyr

emissions Saturation vapor pressure concentration emission would yield

2400Ulo

l)reathing loss from the tank can be calculated from AP-42 as

Lt3(1bday) = 619 X 10-5 M( p ) 068 D 173 H 051 T 05F CKcmicro 147-P

5 0 68 173 = (619 X 10- )(154( 173 ) bull (155)

147-i73

(lj) 051 (26) 05 (1)(1)(1)

= S 63 lbday

3where a working range of zuu x 10 o 400 X 103 lbs were used to determine an

average l i quid nei grit Jn nu a 11 y e111 i s s ions would be 2057 lb Therefore

working losses dorrii nate tt1e totil elissions from tne microlant

Summary

The upper and lovJer bounos of caroori tetracriloride emission ~vere

determined as Ll467 lbyear (upper bound on fugitives DuPont measurement of

storage tank work i ny erni ss ions nur 1a I tanK oreattli ng) and lbgt 15 loyear

(lower bound on fuyitives s1 111eas--1re111ent of storage tank working emissions

no ri n a I tan k breatt1 i ny ) bull

13]

air smicroace Saturation concentratiun at 20degc is amicroproxiriately 12 Measured

values were bY and 74 These values r~sulted from gas chromatographic

dn a I y s e s ot t rd n s t er re c1 sa111 p I e s fr() m t n e 1U U I i t e r Ted l a r b a y t i 11 e i nt e9r a t ed

sample ror tt1e tank car displdcement of 2008 ft 3 the total CT emitted based

on tile ctverctye is LUU8 x 007L = 144 ft 3bull

Tnen for a density of p =PMRT

P = (1)(1gt4) 64 gL = 04lbft 3

(082)(293)

Tl1erefore tne displacement weiyl1t of CT microer ()ff-load is equal to 144 ftJ x U4

l b ft J = ~ 7 bull 6 1 b bull F o r 2 5 U off - l o ad i n gs an n u c1 1I y we have 14 400 l b s bull Th i s

compares with UuPont I s rnectsure1nents of 9 3 vapor content arid 18800 lbyr

emissions Saturation vapor pressure concentration emission would yield

2400Ulo

l)reathing loss from the tank can be calculated from AP-42 as

Lt3(1bday) = 619 X 10-5 M( p ) 068 D 173 H 051 T 05F CKcmicro 147-P

5 0 68 173 = (619 X 10- )(154( 173 ) bull (155)

147-i73

(lj) 051 (26) 05 (1)(1)(1)

= S 63 lbday

3where a working range of zuu x 10 o 400 X 103 lbs were used to determine an

average l i quid nei grit Jn nu a 11 y e111 i s s ions would be 2057 lb Therefore

working losses dorrii nate tt1e totil elissions from tne microlant

Summary

The upper and lovJer bounos of caroori tetracriloride emission ~vere

determined as Ll467 lbyear (upper bound on fugitives DuPont measurement of

storage tank work i ny erni ss ions nur 1a I tanK oreattli ng) and lbgt 15 loyear

(lower bound on fuyitives s1 111eas--1re111ent of storage tank working emissions

no ri n a I tan k breatt1 i ny ) bull

13]

100

SOUfCE TESTS - STAUFFER CHEMICAL COMPANY

101 SITE OVtRVItw

Summary

Stauffer is located in the Carson area of Los Angeles County It is

a manufacturer of polyvinyl chloride from tl1e monomer (VCM) ll11hich is produced

on sit2 Stauffer is the only California producer of ethylene dichloride

(EUC) which is converted to VCM The VCM is regulated by several standards

including Cal OSHA EPA (emission standard) and CARB (ambient)

During Phase I the plant was inspected and possible emission

sources were identified Th~se cans~sted of EDC storage ta11ks fugitive

emissions from valves flanges and pumps process vat2r content gas

incinerator effluents and th~ loading of outbound tankers

Since the completion of Phase I several events have transpired which

a ffectebullj the test program to deterni ne foe i 1i ty emissions These wer2

~ the completion of a vent ~ldS incineration system tied to all significant EDC storage tdnk breathers

$ the completion of a comprehensive study by SAi staff to develop a nationwide material balance for EOC

bull inactivity in EOC importation for the plant and elimination of EDC exportation from the plant

Based 11pon this input and the plant inspection of February 3 1981

it was expected that a plant emission factor for EDC can be determined with

relatively little uncertainty It is further expected that si 1Jnificant

atmospheric release of EOC du~ to plant operation may not occur at the plant

itself but rather offsite These vmuld drise fro11 two sources - the process

water discharge from the plant and severcil off site EDC storaye tanks

Facilities Description

Ethylene dichloride is being produced at Stauffer by two processes

direct chlorination and oxychlorindtion In the former EDC is produced by

d i rec t ch I o r i nd t i on uf et 11y l t~ n e i n the pr 1~senc e of cHl Fe C 1 J crl t d lys t I n th e

latter process EUC is microroduced by the oxychl)rindtion of ethylene with

hydroyen chloride and oxyuen in tne prescrlCe of a cdtdlyst typically Cuc1 bull2

132

lhe microldrrt fluriti1~s tuC after tne microrimar y reactions by separation and digt ti l I dt i ur1 fgtr-ocJuLL 1 s ctured ctncl umiddot 2d ctS foed for VCM production All

sourclS agre~ tl1dt c1ission of EDC is (f concern in its productionstorage

process stages ctnJ O little importance in VCM production steps (J~B 1980)

At the timt of plant inspection some storage of EOC existed at three

ledsed storage tanks located at tne Port of Los lngeles 1-iaterial has not

been 1-Jithdravm from these tanks during the 1dst year and there had been

discussion to consolidate materictl into a single tank

All microrocess components handling EDC storage are now tied to a closed

ventilation incineration syste11 It is expected that virtually all

chlorinated hydrocarbons includiny EDC will be etfectively destroyed since the

system must demonstrdte VCM concentrations are reduced below 1 ppm Firebox

tempera tu re is aoodegF and residence ti me greater than 1 5 seconds under

heaviest flow conditions Note that by way of comparison combustion perforshy

mance datct for PC8s are 99995 destruction at 1832degF and 1 second residence

time and 9Y9Y9YY4 at 2 seconds dnd 2372degF (N Flynn SAi Personal

Communication)

Fugitive emissions from valves and flanges are monitored by the

plant according to the requirements of SCAQMD Rule 4661 Pumps and

compressor follow ~ule 466 Emission and control requirements for vinyl

chloride are specified by ~ules 1005 and 10051 EOC concentrations in

wastewater are monitored several times daily in tne primary EDC steam stripper

stream and once daily in the composite plant outflow stream Daily water

disct1arge limits are 2~ ppm wit11 typical montt1ly averages being in the 8 ppm

vicinity The discharge limit is ~mbodied in the discharge permit (number

5061) vJith tt1e Los Angeles County ~anitation District

Tnere are a number of points at which tt1e rnc can be released to the

dtmospnere and they are similar for both dir~ct chlorination and oxychlorinashy

tion These include the following along with their emissions factors for direct chlorination

A chlorinator vent 2xliJ- 5 IOdSS per unit mass EDC produced

I) I ight end column vent -52x lt)

C dist i I I at i un column vent lXlU-5

u storage tank breatning 7xlu- 5

133

E storaye tankworkiny loss 27x 0-u 5F fugitiv~ emission bxll -

-JLi ildSteJdter l gtX] U

Em~ssion factors are si~~lJ fer oxychlorination processes

Enlission factor vau-5 (~1rbulle Tdilt2~ frcn tne literature (JR13 198U) and

assume 9b iciency for incineration (A-E) It is also assumed above that

EIJC emissions to water are Lf~ uf ~rihsion to citir ard that in wastewater

treatment 100 of EOC discharged tc1 vater is rel easl~d to a i 1middot ~ These emission

factors 1bull1e1middote used to prioritize r-2ieaies iJut were cldrly uude approximations

to tre planL l)ecause of thlt ncine-xt~on system it was anticipated that

fdctors A-E would be ri2dtKcdft 1-middotactor - 1nii~t1t be reduced since a monitoring

program had been in force al20st o~e y2ar Factor G and the off site storage

tanks~ vni~cn are not tied into lt1n ircinerator system rniyht -iominate emissions

Emissmiddotion limits and concomitant regulations embodied in Rule l00S of

the SC-l)MU have necessitctted tlw incinera-cion system Ninety gas chronatograph

probes are located throughout t11e plant including the stack of the primary

incinerator Con cent rations of vc1v1 are reported es sent id I I y be I ow tt1e r~yu la -

tory limit of 10 ppm and in fact below the limit of detection somewhat less

tl1an 01 micropra

lUl MEASUK~MENT APPROACH

Tne objective of tt1e proyram is to determine a pl ant emissions for

EOC It is not accemicrotabl~ to dtmiddotvelop sucll informatmiddot1on basect upon published

indlllstry Jide esti11dtes of plant control efficiencivs Fortunately it was

possible to desiyn a monitoring and calculational progra1n to determine EUC

emissions for tne site and not ctbsorb a disproportionate share of program

resources

Hie re are tnre~ modes ot re I ease from tt12 micro Iant dnd d fourt11

offsite

bull Post-Process 1nci11er-dtion

Processes A-E of ecc ion lll 1 dre a 11 1ented into the pl ant

incinerdtio11 syste11 rt1e conceritrcttion of VCfmiddotl contir1uously iectsured in till=

stctck cts SJdS output remicroresents d rectsondble UjJper li11it to diJply tor EDC

con cent rat i ons s i nce i ts ef f i c i ency of i nc i nera t i on 1s d t 1 east c l1 u a I to tr1 at

134

ot VCibulll Tlerefore knowledge of the system dirflow rnd VCM concentrcttion are

tt1e necessary and sufficient conditions for determining the bounds of EDC

reledse fhe pldnt expn~sse I wi 11 inyne~)s to provide tnese data in order to

support the calculations

bull Fugitive Emisions - Valves Flanges Seals and Other Sources

Fugitive emission from the valves pumps and flanges can be

determined more precisely tnan was anticipated since Stauffer has completed a

comprehensive leak inventory of all such components in compliance with Rule

466 466 1 and 1005 The Inventory delineates all leaks uncovered by their

three man crew throughout tne year and indicates the screening level in ppmv

and component identity In consultation with Stauffer we will be able to

identify the total number of components associated with EDC handling systems

( 700) the distribution ot substance composition streams the distribution

and incidence of leaks by hardware component type and leak rate We will

utilize this information to provide historical data to compliment our Foxboro

OVA sampling at the site dased upon this monitoring we predict an EDC mass

emission rate for the fugitive releases The plant has agreed to provide the

necessary information We propose to meet witt plant personnel prior to the

start of monitoring and finlize the sdmpliny strategy to accomplish 100

coverage of the lines of hiuhest EDC composition The sampling approach and

calculation of mass emission are described in Section 72

bull ~~as tewater

From examination of the emission factors derived from published

literature (see Section 10~) it is clear that EDC release from wastewater to

air is potentially several 11rders of magnitude higher than any other plant

source l1ased upon microl ant ml~a sured concentrcJ ti ons of 8 ppm EDC in water a more

realistic emission factor wgtuld b 24xlo-4 1-iassunit mass EOC produced

Clearly this could still be the dominant source Furthermore the release

point would be expected to middotgte locctted between the plant and the sanitary

district treatment site whi1 his~ kmfrom the plant at 24501 S Figueroa We

believe it is necessary to ndemicroendently confirm the average 24 hour EDC

concentration in the dischar-qe valer by obt1ininy the refrigerated composite

sample We t1ave identified ttie I 1nes of int~rest and received agrt~ement to

sample and analyze for EDC in the stream ~e propose to draw a duplicate

sample from the compositor nd ani1lyze for its rnc content This will be

135

compared with Stauffers para11eI tnalysis It is not reliable to obtain

direct readings of EDC by survey instrument in the air above the water flow since concentrations at crny single point will be i 1 the near dmbient range

Emissions will be cc1lcJLtecl 1_sing the plants vulumetric daily flm

and ass~m~ng that complete degass40~ of the effluent will eventually occur

This assumption is based umicroon the relative volatility uf EDC and tne distances

involved However it should be noted that no direct experimental or

monitoring data is available with which to confirn this Th2 composition of

the discharge stream is unknomiddotwn since it merges into a larqe multisource flow

Conversation with the LA County Sanitation l)istrict (J f1ilne) reveals the

stream to be both exposed and covered Vents exist where me measurements

could De made to assess gross leaka~e at key points

We will obtain samp1es of several plant process discharge streams in

40 ml bottles witn no head space Transit time fr0m sample collection to

analysis will be minimized and wi l I not exceed 24 nours This time frame is

conservat i v e al though ED C i s a v o l ct ti l e mater i al ar1 ct vJi 11 undergo

concentration degradation and outgdssing lnalysi) will b(~ performed in tt1e

SAI Trace Environmental Chemistry Laborcttory utilizing pur 1Je and trap analysis

FID gas chromatography A trial a11alysmiddotis was conducted and the EOC

characteristic peak was distinctive down to the ppb level Therefore the

analysis should easily confirm concentrations in t1e 8 ppm range

Offsite Storage ranks

Three EOC leased storage tanks are located offsite at the Port

of Los Angeles and are owned by annther firm Annual emissions from the tanks

were very roughly estimated based un AP-42 JRB 1980) as 44 kkgyear

based on preliminary estimates of stored quantites

Considering tne magnitude of this source these storage tanks must

be investigated We will gather information about their configuration and

control i n order to ca I cu l ate their e111 i s s i on s bull

136

lU3 UEfERM I NAmiddot 10~~ OF MISS IONS

1031 Incinerator

Vinyl chl 11ride co11centration is monitored dt approximately 90

locations throughoul the pl trlt (Lanyner 19Bl) including the incinerator

output Conc~ntrations are reported by Stauffer dS less than 0l ppm at a

flow rate between 10000 ant 12000 SCFM Although no direct monitoring of

EDC is conducted it is posible to conservatively bound the concentration of

EDC at U 1 pp111 Tl1at conce 1ltrat ion is a suitable ci10i ce s i nee it represents a

conservative bound on tile VCM levels measured near the stack output

Furthermore the molar volume will be taken as 224 L rather than the higher

value it would nave because of the sl i~thtly elevated temperature at the

detector location

Using 12000 SCFM one has thE annual emission of EDC as 12 x 103

SCFM x 283 LSCF x 526x105 minyr x lmole x 97gmole x 10-7vv 224L

Therefore incinerator emissions of EOG are thought to be bounded by 773 kg or

170 1byear

10J2 Fugitive Emissions

Seven hundred (7UlJ) sources were surveyed comprising nearly 100 of

EDC service All accessibl~ plant areas with streams containing greater than

1 EUC were screened except for a small number of devices located in areas

11lere active maintenance was being condJCted It is believed unnecessary to

p2rf-gtrm any emission factor adjustment since it is estimated that gredter than

95 of the requisite components were screened

Three leaks above an arbitrary OVA reading threshold of 20 ppm were

detected Table 103-1 sumnarizes thei~ screening valves calculation

parameters and mass emission numbers fhe upper bound leak rate was

determined to be close to twice the nominal leak rate which accounted for the

response fltlctor of amicroproximamiddotely 05 by Radirn for TLV detection of EDC

( 13 rown 1980) bull

Stauffer found and reported approximately 10 leaks in the EDC

service during 9 months previous to the pLrnt testing Assuming no111inal leak

137

) )

Table 103-1

STAUFFER CHEMICAL MASS EMISSIONS PREDICTION FROM OVA SCREENING VALUES

OVA SAI Actual Upper Device Response Response Cone Source Leak Rate Bound Location (ppm) Rae tor (ppm) a b Se I Re Function lbyr lbyr

Gate Valve Unit P1559A

Valve Heavy Ends Column Unit 14439

Compressor Seal EDC Storage Unit Tl062

I-- w co

300 11 273 -035 096 0 17 153 D 6

11 I500 114 439 II 241 C 1

r III II II o-is lUI bull2000 vl 1 1818 A 9~

All emissions 100 EDC (12-Dichloroethane)

11

values in the nei~hborhood of 2000 lbs~ total could be emitted annually

Therefore it appears ttiat major reductions in the mass emissions were achieved

by the company run inspection prugram and the fugitive emission source has now

become of seconqary importance as a fraction of total plant emissions

lUJJ Wastewater Discharge

Wastewater samples were collected in duplicate from four sites

within the plant for determination of ethylene dichloride (EDC) concentration

The samples were collected in EPP standard 40 ml VOA vials on July 1 1981

Analyses were performed using standard purge and trap techniques coupled with flame ionization detection gas chromatography The results are given in the

table below

Sample description EDC conc~ntration range (microJml)

EDC concentration average ( 1-19ml)

Approximate fl ow ( gpm)

PVC Interceptor Box 82 - 108 95 200

EDC Stripper number Chlorination area C1404

2

011 - 014 012 25

Final collection site for pH adjustmentP663 349 - 3B2 366 350

Final discharge site sanitary sewer 6 - 254 158 500

The water in the PVC lnterc1~ptor ~ox was warm and represents washings from the

PVC reactor which travels t 1J the Interceptor Box in a concrete drainage ditch

Water from the EDC ctripper was vPry t10t and because of its heat was difficult to collect The ~1eat ot trie wat11 r may in pdrt account for tne

relatively low concentratio11 of ElC here as the EDC would out-gas from the hot

~vater more readily The fir1ctl collection site tor pH adjustment is a large

concrete container middotvith mixers in it located just prior to the final discharge

site Water at the final dischdrg~ site is being constantly aerated due to

the speed that it flows thru~yn the concrete drainage trough This aeration

could account for tne relatively large range in the rnc conotent found here

The average EDC concentraticmiddot1 at tne final discharge site is middotbelow the daily

discharye requirement of 25 19ml EDC

139

Plant oersonnel indicdted monthly iverage readings are typically on

the order of 8 ppm but daily averages can re 1ch greater than 20 microgml In

addition LA County Sanitation Uistrict staff (J Milne 1981) indicated that

an on-site impoundment pond can become laden with EOC during certain abnormal

operating periods and discharge Vdriances arl requested by Stauffer This

mi 9t1t Jccur on the order of O1c0 aCii ecH c1 1d t1erefore is no-c expected to

signifcantly impact averag~ cnnJJI discharg ~ values lt s not expected at

this t1ime tEit evaporative eriss uns from this pond are significant except

infrequently during upset conditions and spi 11 control operations Program

staff ~~ 1er2 una111Jare of any periods ~h 1~n EDC cimtent in the ponds could be

appreciable and therefore no sampling was performed

Utilizing the average uf the two final discharge concentration

1readinJs (158 micrograrnsm~ 1 and 1G 1Ji gpm fl)W one has 34600 lbyear of EDC

released into the sanitary sEwer For the pJrposes of calculating population

exposures from plant releases it will be assJmed that the EDC is locally

emitted from the wastewater streams Tl1ere ire no monitoring data available

with which to develop a more accurate release profile

However emissions of me fron the plant wastewater discharge can be

calculated according to the method of Mcmiddotckay (197S) as modified by Dilling

(1977) It should be noted that this mLst be considered an estimate since

conditions of fl ow and the presence of other substances will influence

emission rates

Using Uilling (1977) for non~erated flow first recalculate

Henry 1 s law constant (dimensionless-my of chmiddot~mical per liter of air divided by

mg of chemical per liter of water) dS

1604 P M 1 Wl

TS l

)

wnere

H Henrys law constant dimensionlessl

Pi tile compounds pur1~ componen- vapor pressure in mm Hg at T

Mwi the 11olecular weiglt

T tile absol ~te temperature of he wastewater in K

Si tt1e compounds ~uluhility in myliter at T

Then for UJC

(l6u4)(7~)(9Y)H = = U0447 with 1

(2lJ4)(8690)

LIii bullul1Jl)i I 1 1y 1 middot UL i11 wt1Ler yiven -1L L0degC is 8u4JU UHI tile vapl)r pressure

1 4 psi ( 7 2mm) at 7 0 degF bull

Tl1e overall liquid mass-transfer coefficient Kil is given by Dilling (1975)

as

(2211)(06) in mhr

-f-0-4~ (M _) 12+ 100 Wl-H-

1

=U108111tH

Then from Mackay the percent desorption is given by

c 1 = exp (-Kil tL) where

c 0

c = the concentrd ti on at time t of EOC 1

co = tile initial concentration

L = the Ii quid depth (Ill)

t = tt1e retention time (in hr) of the liquid in the wastewater

system

Retention time is bc1s~d 11pon a flow velocity range of 3 ftsec as estimated by

the Los Angeles County Sdnitatii)middot1 Dibulltrict (1J Milne) over a middotdistance of

ctmicroprox i 1nate I y S km to the Joi n_t Wdter Po 11 ut ion Contra l Pl ant Sewer fl ow

depth throughout the entire route have been estimated by J Milne as typically

1 foot 030m) to the Davison Pump Plant and 3 feet (09m) to the Joint

Tredtment Plant distances assumed to bt~ 1 mile and 2 miles respectively

Trdnsit tiine l)eco111es between 05 tnd LU hours sequentially Then computing

the net emission reduction as the product of each leg

Therefore 26 of the EOC i erni t ted bdween the pl ant and the Joint Water

Pollution Control Plant Rtgtsfdence ti111e and conditions at the plant account

141

for additional releases and residudl content is forther emitted between the

plant cnd the ocean discharge poiiit ~-Herever the flow is r2tained aerated

or shallow emissions will be accentuated

Offsite Storage Tank Emissions

There are three storage tanks leased by Stauffer for EDC storage

1ocated at 22nd and Gaffey Sts ~ bull San Pedro These tan ks a ~e used for long

term storage rather than providing feed on a routine basis Therefore

breathing loss rather than v1orkir 1J lass is )f concern An tanks are white

two being 67 ft in diameter while t~2 third is 57 ft All are 40 ft 3 inches

high and have capacities of e~thi l0~10000 (2) or 840000 gallons The

tanks are cone roof type and are not tied into vapor recovery systems The

South Coast Air Quality r1dna~J~ment u~ str4 ct has based their estimate of

emissions on the specification of 12 foot vapor level Using the AP-42

formula for breathing loss with the vapor pressure of EDC at 20degc and an

average diurnal temperature variation of 26degF one hds for each of the two

larger tanks

l) 173 H bull51 T bull 5 LlI = (619 x 10-S) M( _P_ tH F CKcp 147-P

68 173 51 5 = (619 X 10-S) (9119~ 116 bull (67) ( I 5) (26) (1) (1)

14 7-1 16)

Thus for the two larger tanks LB= 472 lbday

For the third tank 0=57 and = 178 lbdayL8 The total emissions become 23724 lbyear ~ince he plant is in the process

of shutdown it is not clear what the liquid levels currently are Note that

if the material in the three tanks were combined into one totamiddot1 -~1nissions

would be reduced markedly Exposure to a population from this site was

determined separately since it is lbullgtcated over six miles fron tl1e plant

142

110

ASSESSMENT OF PUPULATlUN EXPUSUR~

111 OVEKVIEW

A major program goal wa~ to compare emissions from the various

sources and identify and rank any 11 tlot spots in California where the general

population was exposed to elevated concentrations of carcinogens A simple

Gaussian dispersion model was therefore used to obtain order-of-magnitude estimates of exposure of the genermiddotal population surrounding each source

Since this was essentially a screening study use of more sophisticated models

was not appropriate

It cannot be emphcsized too stronyly tnat most California residents

are exposed to emissions of hazardous substances from a variety of natural and

rnanmade sources Urban dwellers dre typically exposed to greater concentrashy

tions than rural residents however all are subjected to so called 11 background 11 levels from multiple sources In order to place stationary

source exposures in perspective the typical ambient levels of each substance

were identified from the literature and compdred with the concentrations due

to the emissions from each ~lant Exposures were thus expressed both as

absolute quantities and as increments above background

Comparison of plants presents a further difficulty in that various

substances are being consict~red No attempt was made to evaluate the relative

importance of exposure to two different substances such as chloroform versus carbon tetrachloride other than by ambient concentration

112 DATA SOURCES

1121 Meteorologicdl Llata

The dispersion model to be descritgted in Section 113 required input

of annual average wind speed and frequency of occurrence of wind from each

compass direction These data were obtained for most of the sites from the

South Codst Air wuality Management District (SCAQMD) In other cases (Kaiser

Jotms-Manvi l le l)ow and OuPlt)nt) only clai ly cJVerage wind speed and frequency

data were available In al I cases we used data from the meteorological

station nearest the modeled e11issrnn source Table 112-1 summarizes the

141

Table 112-1 METEOROLOGICAL DATA BASE

Observation Site Data Period of Pldnt (Distdnce Source Observation Remarks

from Plant)

A11 i ed Che11 i Ca 1 Lennox SCA~MD 1Y71-1974 Averaged hourly readings available ~ 1 Segundo (3 111iles)

S ff~r-- Uemical Long l3each SCA()MD 1962-1974 Averaged hourmiddoty leadir19s avai1able offsite Carson (3 miles) storage tanks appruxi1ntL~ly 5 miies from

rneteoro middot1 ogy s I te

I--

-~ Dow Pittsburg Pittsbury lJOVJ 1955-1974 Daily and averacied Ninnd rrJse on1y--no hourly-~

Uu~ont Antiocn (lt5 miles to data uVdilable Antioch)

r~d i ser Stee 1 Fontdnct SCAQMD 1978-1979 Two years of hourlydaily data printout made r - i -- lt3 1niles available (33000 entries) daily averages

estimated as only practical alternative -middot bullA

Jon 1 s 1middot1 an I i 1 l e Stockton NOAA 1941-1970 No hourly data estiamtes based on monthly S-cJc~ton (1 mile) prevalence of directions and speed

R~R City uf Whittier SCAQMD 1969-1973 Averaged hourly readings available lnuustry (4 miles)

1neteoroloyical data 1iase for eden site Wind speed and 1tJind direction

freljuency dcttd us~lti 1r1 Ull~ model are provided in Apmicroendix Li

11LL Populdtion Ddtct

In oroer to assess popu Idt i 011 exposure in tile same way for a 11 the

sources we defined d lLJ-s4u1re mi le impact area arround each ~ant This

size WdS cnosen since it was found in most cdses to include all distances at

whicn incremental ground lev~I concentrations due to plant emissions would

exceed genera I urban ambient Ieve 1s for the microo 11 utant in question In most

cases the plant was placed ltlt the center of the impact area Where winds

were predominantly from one sector or a few adjacent sectors or where an

unpopulated area (ey the Pacific Ucectn) adjoined one side of a plant the

impact drea was defined to li~ imm~didtely downwind of the site

Once the impact areas were defined we obtained Thomas 8rothers maps

of all census tracts within them These maps are provided in Appendix H

Census tract populations were obtained from the 1980 US Census The

population of each tract was assumed to lie dt tne centroid except when tne

tract was large and most of the population was concentrated away from the

centroid in the latter case the best-defined population center was used

fadial and angular distances from the sources to the population centers were

tr1en determined

11~3 OISPERSION MODELING APPROACH

ln urder tu estiindt~ pomicroulcttiun expusures in the census tracts

surrounding each source a simple Gaussian dispersion model was used Use of

a 1nore sopnisticdtect 111odel ~-1as inamicroproJridte yiven the uncertainty in our

emission rate estir~1ates It is questionable v1hether any real gain in accuracy

would nave resulted

ne -Jell-knovm Gaussian oispersion formula is (Porter 1976)

C = 106 Q

iTUOCT z y

(113-1)

C 9rounct leve1 concentration in (i-igm3)

emission rate (gs)

u average vJi nd speed at the microl1ys i ca I stack nei ght (ms) 0 standard deviation of the vertical concentration distribution z

145

--------

a standard deviatinn d the noriz1intal conentration _y

distribution

H effective stack height (m)

y is the crosswind distance froin the plume centerline to the

receptor point (m)

This e4udtion was assumed to microrov1de no 11r1y average ground level concentrashy

tions (Kanzie1~i 1982) Tle vaiues for the standard demiddot-riations o and 0 are y z functions of the downwind distance x

b iJ dX (113-2)

l d

0 ex (113-3)y

where a b c and d dre constants that fit the function to the empirical

curves presented in Turner (t97C) he v~middotJnd -peed at physical stack height is

given by the equation

u (113-4)

where

u wind speed at physical stack height (ms)

llJ = measured wind speed (ms)0

h physical stack neight (m)s h the hei gtlt at which tile known wind speed was measured (m) and

0

p an empirical constant which varit~s with stability class

Lacking data on the heights at which the all known wind speeds were measured

we followed common practice and assumed a value of 10 m for h bull0

Tridl calculations showed the valu~ of tne plume rise for all the

sources except RSR Johns-Manville and the Kdser final cooler cooling tower

to be negligible (ie less than one meter) Plumbull rise formulas developed by

Christiansen (1975) and cited by Porter (1971gt) wer- used for the exceptions

The rise WdS assumed to be momentum-dominated for ltSK Johns-Manville and

Kaiser cooliny tower and bouyrncy-dominated for tne Kaiser coke ovens

As was discussed above the radial and angular distances from a

source to each surroundinlJ census tract were dett~r1ined Wind direction and

See llu s se 1ltJ 7J

146

smicroe~d data meanwt1ile ~ere obtained for eacr1 of the 16 111ajor compass points

To cr1lculdte tile conu~ntrat1on at a giv n microoirt it was first necessary to

d(termine the co1JpdSS s~ctor in Jhic11 Li1e microui11t ldy Fiyure 113-1 gives an

LXdinmicro le for a census tract at r k1~ from a source and at JU degrees from a

reference ctngle ~~t1ich we defined as north (U deyrees) As seen in the

fi9ure this ~oint lil~S in a sector bouided b the NNE (ZL5 deqrees) and NE

( 1i~ de~JretS) rn111microdss dir1~ct1ons lt1e Cttlculc11ion WdS microertormed once t1)r every

11our of the day since annually averctged values of hourly wind smicroeed were

available The followiny schedule of hourly stability clas was determined to

De cons i stent w i th t ne re I d ti on sh i micros s u mrna r i z ( d by Turner ( 1 7) y i v en the

observed distribution of wind speeds at our meteorologoical measurement

stations The schedule vas m1Jdified slightly at the suggestion of the ARl3

(Ranzieri 198l)

Hour Class Hour Class

0 F 13 B 1 F ltl 8 2 F 11) 8 J F l ~ 13 4 F 1 8 5 F 1 ~ IJN 6 F l J UN 7 DU 20 F 8 d 21 F 9 13 22 8 10 13 iJ 8 11 t1 12 13

Using the aoove equations and adj us tnents the concentration at the

point of interest was then cal cu I at~d as the sum of the concentrations

resulting from plumes naviny middot~ne boimdiny compass directions as centerlines

if the anyular distance to th~ point was conuruent with a compass direction

then only one calculcttion was necessary Let C(Y 1ti) and C(U ti) be the2

concentrations calculctted at 1our i for co111pibull)s directions Y and Y 1 2

resmicroectively 1s d1cusseltJ in Section 11L 1ur 111eteoroloyicdl ddtd in most

cases inc I uded t tie f reyuency ot wind direction for each hour of the day Let

f(Y 1

ti) and f(Q t) oe tne microrobctDilities ot occurrence of wind in the2

147

N

~-------------------------E

Figure 113-1 Determination of Compass Position of Census Tract Centroid

118

directions Y dnd w2

resp1bullctive1y at hour i rt1en the expected value of the1 cunce11Lrit1rgtrI it UH point iri questiun tt iiour I is

C(t) = f(G t )C(G t) + f(IJ t )C(q1 t) (113-5)

l 1 l 1 1 2 1 l 1

nw average dnnua I exposure wa then ca I cul ated as the average exposure on

this composite day

23 C = (124) E C(ti)

(113-6)i=U

The model was programmed in Applesoft BASIC on an Apple II

microcomputer having 48 K bytes of random access memory and a disk storage

capability The program which is included as Appendix I was compiled with

an Un-Line Systems Inc Expediter II BASIC compiler in order to decrease

running time

114 POPULATION EXPOSURE FRUM SURVEYED SOURCES

1141 Modeling Kesults

Using the modeling parameters listed in Table 114-1 tne

incremental pop~lation exposure due to each of the stationary sources was

computed Tables 114-2 through 114-12 show the modeled annual average

incremental exposure for each census tract Jround each plant Census tract

numbers appear on tne maps in Appendix H) The cumulative population column

specifies the total population exposed to all concentrations equal to or

greater than the corresponding source weighted concentration entry of the

table Figures 114-1 through 114-11 illustrate the cumulative population

exposure versus incremental concentration above ambient background concentrashy

tions Table 114-lJ lists rangemiddot of typicJI urban ambient concentrations for

eacn substance fhese gtJere used middoto assess tne incremental contribution of the

plant emissions Table 114-14 s 1 1mmarizes the incremental population exposure

due to each source These were bctsed upon annuc1l averaye source strengths and

do not reflect transients in emisions or worst case meteorolgoical

conditions Note thdt no attempt was made to assess the potential health

effects or risks to the public dut to the resultant combined exposure

149

lhe microldrrt fluriti1~s tuC after tne microrimar y reactions by separation and digt ti l I dt i ur1 fgtr-ocJuLL 1 s ctured ctncl umiddot 2d ctS foed for VCM production All

sourclS agre~ tl1dt c1ission of EDC is (f concern in its productionstorage

process stages ctnJ O little importance in VCM production steps (J~B 1980)

At the timt of plant inspection some storage of EOC existed at three

ledsed storage tanks located at tne Port of Los lngeles 1-iaterial has not

been 1-Jithdravm from these tanks during the 1dst year and there had been

discussion to consolidate materictl into a single tank

All microrocess components handling EDC storage are now tied to a closed

ventilation incineration syste11 It is expected that virtually all

chlorinated hydrocarbons includiny EDC will be etfectively destroyed since the

system must demonstrdte VCM concentrations are reduced below 1 ppm Firebox

tempera tu re is aoodegF and residence ti me greater than 1 5 seconds under

heaviest flow conditions Note that by way of comparison combustion perforshy

mance datct for PC8s are 99995 destruction at 1832degF and 1 second residence

time and 9Y9Y9YY4 at 2 seconds dnd 2372degF (N Flynn SAi Personal

Communication)

Fugitive emissions from valves and flanges are monitored by the

plant according to the requirements of SCAQMD Rule 4661 Pumps and

compressor follow ~ule 466 Emission and control requirements for vinyl

chloride are specified by ~ules 1005 and 10051 EOC concentrations in

wastewater are monitored several times daily in tne primary EDC steam stripper

stream and once daily in the composite plant outflow stream Daily water

disct1arge limits are 2~ ppm wit11 typical montt1ly averages being in the 8 ppm

vicinity The discharge limit is ~mbodied in the discharge permit (number

5061) vJith tt1e Los Angeles County ~anitation District

Tnere are a number of points at which tt1e rnc can be released to the

dtmospnere and they are similar for both dir~ct chlorination and oxychlorinashy

tion These include the following along with their emissions factors for direct chlorination

A chlorinator vent 2xliJ- 5 IOdSS per unit mass EDC produced

I) I ight end column vent -52x lt)

C dist i I I at i un column vent lXlU-5

u storage tank breatning 7xlu- 5

133

E storaye tankworkiny loss 27x 0-u 5F fugitiv~ emission bxll -

-JLi ildSteJdter l gtX] U

Em~ssion factors are si~~lJ fer oxychlorination processes

Enlission factor vau-5 (~1rbulle Tdilt2~ frcn tne literature (JR13 198U) and

assume 9b iciency for incineration (A-E) It is also assumed above that

EIJC emissions to water are Lf~ uf ~rihsion to citir ard that in wastewater

treatment 100 of EOC discharged tc1 vater is rel easl~d to a i 1middot ~ These emission

factors 1bull1e1middote used to prioritize r-2ieaies iJut were cldrly uude approximations

to tre planL l)ecause of thlt ncine-xt~on system it was anticipated that

fdctors A-E would be ri2dtKcdft 1-middotactor - 1nii~t1t be reduced since a monitoring

program had been in force al20st o~e y2ar Factor G and the off site storage

tanks~ vni~cn are not tied into lt1n ircinerator system rniyht -iominate emissions

Emissmiddotion limits and concomitant regulations embodied in Rule l00S of

the SC-l)MU have necessitctted tlw incinera-cion system Ninety gas chronatograph

probes are located throughout t11e plant including the stack of the primary

incinerator Con cent rations of vc1v1 are reported es sent id I I y be I ow tt1e r~yu la -

tory limit of 10 ppm and in fact below the limit of detection somewhat less

tl1an 01 micropra

lUl MEASUK~MENT APPROACH

Tne objective of tt1e proyram is to determine a pl ant emissions for

EOC It is not accemicrotabl~ to dtmiddotvelop sucll informatmiddot1on basect upon published

indlllstry Jide esti11dtes of plant control efficiencivs Fortunately it was

possible to desiyn a monitoring and calculational progra1n to determine EUC

emissions for tne site and not ctbsorb a disproportionate share of program

resources

Hie re are tnre~ modes ot re I ease from tt12 micro Iant dnd d fourt11

offsite

bull Post-Process 1nci11er-dtion

Processes A-E of ecc ion lll 1 dre a 11 1ented into the pl ant

incinerdtio11 syste11 rt1e conceritrcttion of VCfmiddotl contir1uously iectsured in till=

stctck cts SJdS output remicroresents d rectsondble UjJper li11it to diJply tor EDC

con cent rat i ons s i nce i ts ef f i c i ency of i nc i nera t i on 1s d t 1 east c l1 u a I to tr1 at

134

ot VCibulll Tlerefore knowledge of the system dirflow rnd VCM concentrcttion are

tt1e necessary and sufficient conditions for determining the bounds of EDC

reledse fhe pldnt expn~sse I wi 11 inyne~)s to provide tnese data in order to

support the calculations

bull Fugitive Emisions - Valves Flanges Seals and Other Sources

Fugitive emission from the valves pumps and flanges can be

determined more precisely tnan was anticipated since Stauffer has completed a

comprehensive leak inventory of all such components in compliance with Rule

466 466 1 and 1005 The Inventory delineates all leaks uncovered by their

three man crew throughout tne year and indicates the screening level in ppmv

and component identity In consultation with Stauffer we will be able to

identify the total number of components associated with EDC handling systems

( 700) the distribution ot substance composition streams the distribution

and incidence of leaks by hardware component type and leak rate We will

utilize this information to provide historical data to compliment our Foxboro

OVA sampling at the site dased upon this monitoring we predict an EDC mass

emission rate for the fugitive releases The plant has agreed to provide the

necessary information We propose to meet witt plant personnel prior to the

start of monitoring and finlize the sdmpliny strategy to accomplish 100

coverage of the lines of hiuhest EDC composition The sampling approach and

calculation of mass emission are described in Section 72

bull ~~as tewater

From examination of the emission factors derived from published

literature (see Section 10~) it is clear that EDC release from wastewater to

air is potentially several 11rders of magnitude higher than any other plant

source l1ased upon microl ant ml~a sured concentrcJ ti ons of 8 ppm EDC in water a more

realistic emission factor wgtuld b 24xlo-4 1-iassunit mass EOC produced

Clearly this could still be the dominant source Furthermore the release

point would be expected to middotgte locctted between the plant and the sanitary

district treatment site whi1 his~ kmfrom the plant at 24501 S Figueroa We

believe it is necessary to ndemicroendently confirm the average 24 hour EDC

concentration in the dischar-qe valer by obt1ininy the refrigerated composite

sample We t1ave identified ttie I 1nes of int~rest and received agrt~ement to

sample and analyze for EDC in the stream ~e propose to draw a duplicate

sample from the compositor nd ani1lyze for its rnc content This will be

135

compared with Stauffers para11eI tnalysis It is not reliable to obtain

direct readings of EDC by survey instrument in the air above the water flow since concentrations at crny single point will be i 1 the near dmbient range

Emissions will be cc1lcJLtecl 1_sing the plants vulumetric daily flm

and ass~m~ng that complete degass40~ of the effluent will eventually occur

This assumption is based umicroon the relative volatility uf EDC and tne distances

involved However it should be noted that no direct experimental or

monitoring data is available with which to confirn this Th2 composition of

the discharge stream is unknomiddotwn since it merges into a larqe multisource flow

Conversation with the LA County Sanitation l)istrict (J f1ilne) reveals the

stream to be both exposed and covered Vents exist where me measurements

could De made to assess gross leaka~e at key points

We will obtain samp1es of several plant process discharge streams in

40 ml bottles witn no head space Transit time fr0m sample collection to

analysis will be minimized and wi l I not exceed 24 nours This time frame is

conservat i v e al though ED C i s a v o l ct ti l e mater i al ar1 ct vJi 11 undergo

concentration degradation and outgdssing lnalysi) will b(~ performed in tt1e

SAI Trace Environmental Chemistry Laborcttory utilizing pur 1Je and trap analysis

FID gas chromatography A trial a11alysmiddotis was conducted and the EOC

characteristic peak was distinctive down to the ppb level Therefore the

analysis should easily confirm concentrations in t1e 8 ppm range

Offsite Storage ranks

Three EOC leased storage tanks are located offsite at the Port

of Los Angeles and are owned by annther firm Annual emissions from the tanks

were very roughly estimated based un AP-42 JRB 1980) as 44 kkgyear

based on preliminary estimates of stored quantites

Considering tne magnitude of this source these storage tanks must

be investigated We will gather information about their configuration and

control i n order to ca I cu l ate their e111 i s s i on s bull

136

lU3 UEfERM I NAmiddot 10~~ OF MISS IONS

1031 Incinerator

Vinyl chl 11ride co11centration is monitored dt approximately 90

locations throughoul the pl trlt (Lanyner 19Bl) including the incinerator

output Conc~ntrations are reported by Stauffer dS less than 0l ppm at a

flow rate between 10000 ant 12000 SCFM Although no direct monitoring of

EDC is conducted it is posible to conservatively bound the concentration of

EDC at U 1 pp111 Tl1at conce 1ltrat ion is a suitable ci10i ce s i nee it represents a

conservative bound on tile VCM levels measured near the stack output

Furthermore the molar volume will be taken as 224 L rather than the higher

value it would nave because of the sl i~thtly elevated temperature at the

detector location

Using 12000 SCFM one has thE annual emission of EDC as 12 x 103

SCFM x 283 LSCF x 526x105 minyr x lmole x 97gmole x 10-7vv 224L

Therefore incinerator emissions of EOG are thought to be bounded by 773 kg or

170 1byear

10J2 Fugitive Emissions

Seven hundred (7UlJ) sources were surveyed comprising nearly 100 of

EDC service All accessibl~ plant areas with streams containing greater than

1 EUC were screened except for a small number of devices located in areas

11lere active maintenance was being condJCted It is believed unnecessary to

p2rf-gtrm any emission factor adjustment since it is estimated that gredter than

95 of the requisite components were screened

Three leaks above an arbitrary OVA reading threshold of 20 ppm were

detected Table 103-1 sumnarizes thei~ screening valves calculation

parameters and mass emission numbers fhe upper bound leak rate was

determined to be close to twice the nominal leak rate which accounted for the

response fltlctor of amicroproximamiddotely 05 by Radirn for TLV detection of EDC

( 13 rown 1980) bull

Stauffer found and reported approximately 10 leaks in the EDC

service during 9 months previous to the pLrnt testing Assuming no111inal leak

137

) )

Table 103-1

STAUFFER CHEMICAL MASS EMISSIONS PREDICTION FROM OVA SCREENING VALUES

OVA SAI Actual Upper Device Response Response Cone Source Leak Rate Bound Location (ppm) Rae tor (ppm) a b Se I Re Function lbyr lbyr

Gate Valve Unit P1559A

Valve Heavy Ends Column Unit 14439

Compressor Seal EDC Storage Unit Tl062

I-- w co

300 11 273 -035 096 0 17 153 D 6

11 I500 114 439 II 241 C 1

r III II II o-is lUI bull2000 vl 1 1818 A 9~

All emissions 100 EDC (12-Dichloroethane)

11

values in the nei~hborhood of 2000 lbs~ total could be emitted annually

Therefore it appears ttiat major reductions in the mass emissions were achieved

by the company run inspection prugram and the fugitive emission source has now

become of seconqary importance as a fraction of total plant emissions

lUJJ Wastewater Discharge

Wastewater samples were collected in duplicate from four sites

within the plant for determination of ethylene dichloride (EDC) concentration

The samples were collected in EPP standard 40 ml VOA vials on July 1 1981

Analyses were performed using standard purge and trap techniques coupled with flame ionization detection gas chromatography The results are given in the

table below

Sample description EDC conc~ntration range (microJml)

EDC concentration average ( 1-19ml)

Approximate fl ow ( gpm)

PVC Interceptor Box 82 - 108 95 200

EDC Stripper number Chlorination area C1404

2

011 - 014 012 25

Final collection site for pH adjustmentP663 349 - 3B2 366 350

Final discharge site sanitary sewer 6 - 254 158 500

The water in the PVC lnterc1~ptor ~ox was warm and represents washings from the

PVC reactor which travels t 1J the Interceptor Box in a concrete drainage ditch

Water from the EDC ctripper was vPry t10t and because of its heat was difficult to collect The ~1eat ot trie wat11 r may in pdrt account for tne

relatively low concentratio11 of ElC here as the EDC would out-gas from the hot

~vater more readily The fir1ctl collection site tor pH adjustment is a large

concrete container middotvith mixers in it located just prior to the final discharge

site Water at the final dischdrg~ site is being constantly aerated due to

the speed that it flows thru~yn the concrete drainage trough This aeration

could account for tne relatively large range in the rnc conotent found here

The average EDC concentraticmiddot1 at tne final discharge site is middotbelow the daily

discharye requirement of 25 19ml EDC

139

Plant oersonnel indicdted monthly iverage readings are typically on

the order of 8 ppm but daily averages can re 1ch greater than 20 microgml In

addition LA County Sanitation Uistrict staff (J Milne 1981) indicated that

an on-site impoundment pond can become laden with EOC during certain abnormal

operating periods and discharge Vdriances arl requested by Stauffer This

mi 9t1t Jccur on the order of O1c0 aCii ecH c1 1d t1erefore is no-c expected to

signifcantly impact averag~ cnnJJI discharg ~ values lt s not expected at

this t1ime tEit evaporative eriss uns from this pond are significant except

infrequently during upset conditions and spi 11 control operations Program

staff ~~ 1er2 una111Jare of any periods ~h 1~n EDC cimtent in the ponds could be

appreciable and therefore no sampling was performed

Utilizing the average uf the two final discharge concentration

1readinJs (158 micrograrnsm~ 1 and 1G 1Ji gpm fl)W one has 34600 lbyear of EDC

released into the sanitary sEwer For the pJrposes of calculating population

exposures from plant releases it will be assJmed that the EDC is locally

emitted from the wastewater streams Tl1ere ire no monitoring data available

with which to develop a more accurate release profile

However emissions of me fron the plant wastewater discharge can be

calculated according to the method of Mcmiddotckay (197S) as modified by Dilling

(1977) It should be noted that this mLst be considered an estimate since

conditions of fl ow and the presence of other substances will influence

emission rates

Using Uilling (1977) for non~erated flow first recalculate

Henry 1 s law constant (dimensionless-my of chmiddot~mical per liter of air divided by

mg of chemical per liter of water) dS

1604 P M 1 Wl

TS l

)

wnere

H Henrys law constant dimensionlessl

Pi tile compounds pur1~ componen- vapor pressure in mm Hg at T

Mwi the 11olecular weiglt

T tile absol ~te temperature of he wastewater in K

Si tt1e compounds ~uluhility in myliter at T

Then for UJC

(l6u4)(7~)(9Y)H = = U0447 with 1

(2lJ4)(8690)

LIii bullul1Jl)i I 1 1y 1 middot UL i11 wt1Ler yiven -1L L0degC is 8u4JU UHI tile vapl)r pressure

1 4 psi ( 7 2mm) at 7 0 degF bull

Tl1e overall liquid mass-transfer coefficient Kil is given by Dilling (1975)

as

(2211)(06) in mhr

-f-0-4~ (M _) 12+ 100 Wl-H-

1

=U108111tH

Then from Mackay the percent desorption is given by

c 1 = exp (-Kil tL) where

c 0

c = the concentrd ti on at time t of EOC 1

co = tile initial concentration

L = the Ii quid depth (Ill)

t = tt1e retention time (in hr) of the liquid in the wastewater

system

Retention time is bc1s~d 11pon a flow velocity range of 3 ftsec as estimated by

the Los Angeles County Sdnitatii)middot1 Dibulltrict (1J Milne) over a middotdistance of

ctmicroprox i 1nate I y S km to the Joi n_t Wdter Po 11 ut ion Contra l Pl ant Sewer fl ow

depth throughout the entire route have been estimated by J Milne as typically

1 foot 030m) to the Davison Pump Plant and 3 feet (09m) to the Joint

Tredtment Plant distances assumed to bt~ 1 mile and 2 miles respectively

Trdnsit tiine l)eco111es between 05 tnd LU hours sequentially Then computing

the net emission reduction as the product of each leg

Therefore 26 of the EOC i erni t ted bdween the pl ant and the Joint Water

Pollution Control Plant Rtgtsfdence ti111e and conditions at the plant account

141

for additional releases and residudl content is forther emitted between the

plant cnd the ocean discharge poiiit ~-Herever the flow is r2tained aerated

or shallow emissions will be accentuated

Offsite Storage Tank Emissions

There are three storage tanks leased by Stauffer for EDC storage

1ocated at 22nd and Gaffey Sts ~ bull San Pedro These tan ks a ~e used for long

term storage rather than providing feed on a routine basis Therefore

breathing loss rather than v1orkir 1J lass is )f concern An tanks are white

two being 67 ft in diameter while t~2 third is 57 ft All are 40 ft 3 inches

high and have capacities of e~thi l0~10000 (2) or 840000 gallons The

tanks are cone roof type and are not tied into vapor recovery systems The

South Coast Air Quality r1dna~J~ment u~ str4 ct has based their estimate of

emissions on the specification of 12 foot vapor level Using the AP-42

formula for breathing loss with the vapor pressure of EDC at 20degc and an

average diurnal temperature variation of 26degF one hds for each of the two

larger tanks

l) 173 H bull51 T bull 5 LlI = (619 x 10-S) M( _P_ tH F CKcp 147-P

68 173 51 5 = (619 X 10-S) (9119~ 116 bull (67) ( I 5) (26) (1) (1)

14 7-1 16)

Thus for the two larger tanks LB= 472 lbday

For the third tank 0=57 and = 178 lbdayL8 The total emissions become 23724 lbyear ~ince he plant is in the process

of shutdown it is not clear what the liquid levels currently are Note that

if the material in the three tanks were combined into one totamiddot1 -~1nissions

would be reduced markedly Exposure to a population from this site was

determined separately since it is lbullgtcated over six miles fron tl1e plant

142

110

ASSESSMENT OF PUPULATlUN EXPUSUR~

111 OVEKVIEW

A major program goal wa~ to compare emissions from the various

sources and identify and rank any 11 tlot spots in California where the general

population was exposed to elevated concentrations of carcinogens A simple

Gaussian dispersion model was therefore used to obtain order-of-magnitude estimates of exposure of the genermiddotal population surrounding each source

Since this was essentially a screening study use of more sophisticated models

was not appropriate

It cannot be emphcsized too stronyly tnat most California residents

are exposed to emissions of hazardous substances from a variety of natural and

rnanmade sources Urban dwellers dre typically exposed to greater concentrashy

tions than rural residents however all are subjected to so called 11 background 11 levels from multiple sources In order to place stationary

source exposures in perspective the typical ambient levels of each substance

were identified from the literature and compdred with the concentrations due

to the emissions from each ~lant Exposures were thus expressed both as

absolute quantities and as increments above background

Comparison of plants presents a further difficulty in that various

substances are being consict~red No attempt was made to evaluate the relative

importance of exposure to two different substances such as chloroform versus carbon tetrachloride other than by ambient concentration

112 DATA SOURCES

1121 Meteorologicdl Llata

The dispersion model to be descritgted in Section 113 required input

of annual average wind speed and frequency of occurrence of wind from each

compass direction These data were obtained for most of the sites from the

South Codst Air wuality Management District (SCAQMD) In other cases (Kaiser

Jotms-Manvi l le l)ow and OuPlt)nt) only clai ly cJVerage wind speed and frequency

data were available In al I cases we used data from the meteorological

station nearest the modeled e11issrnn source Table 112-1 summarizes the

141

Table 112-1 METEOROLOGICAL DATA BASE

Observation Site Data Period of Pldnt (Distdnce Source Observation Remarks

from Plant)

A11 i ed Che11 i Ca 1 Lennox SCA~MD 1Y71-1974 Averaged hourly readings available ~ 1 Segundo (3 111iles)

S ff~r-- Uemical Long l3each SCA()MD 1962-1974 Averaged hourmiddoty leadir19s avai1able offsite Carson (3 miles) storage tanks appruxi1ntL~ly 5 miies from

rneteoro middot1 ogy s I te

I--

-~ Dow Pittsburg Pittsbury lJOVJ 1955-1974 Daily and averacied Ninnd rrJse on1y--no hourly-~

Uu~ont Antiocn (lt5 miles to data uVdilable Antioch)

r~d i ser Stee 1 Fontdnct SCAQMD 1978-1979 Two years of hourlydaily data printout made r - i -- lt3 1niles available (33000 entries) daily averages

estimated as only practical alternative -middot bullA

Jon 1 s 1middot1 an I i 1 l e Stockton NOAA 1941-1970 No hourly data estiamtes based on monthly S-cJc~ton (1 mile) prevalence of directions and speed

R~R City uf Whittier SCAQMD 1969-1973 Averaged hourly readings available lnuustry (4 miles)

1neteoroloyical data 1iase for eden site Wind speed and 1tJind direction

freljuency dcttd us~lti 1r1 Ull~ model are provided in Apmicroendix Li

11LL Populdtion Ddtct

In oroer to assess popu Idt i 011 exposure in tile same way for a 11 the

sources we defined d lLJ-s4u1re mi le impact area arround each ~ant This

size WdS cnosen since it was found in most cdses to include all distances at

whicn incremental ground lev~I concentrations due to plant emissions would

exceed genera I urban ambient Ieve 1s for the microo 11 utant in question In most

cases the plant was placed ltlt the center of the impact area Where winds

were predominantly from one sector or a few adjacent sectors or where an

unpopulated area (ey the Pacific Ucectn) adjoined one side of a plant the

impact drea was defined to li~ imm~didtely downwind of the site

Once the impact areas were defined we obtained Thomas 8rothers maps

of all census tracts within them These maps are provided in Appendix H

Census tract populations were obtained from the 1980 US Census The

population of each tract was assumed to lie dt tne centroid except when tne

tract was large and most of the population was concentrated away from the

centroid in the latter case the best-defined population center was used

fadial and angular distances from the sources to the population centers were

tr1en determined

11~3 OISPERSION MODELING APPROACH

ln urder tu estiindt~ pomicroulcttiun expusures in the census tracts

surrounding each source a simple Gaussian dispersion model was used Use of

a 1nore sopnisticdtect 111odel ~-1as inamicroproJridte yiven the uncertainty in our

emission rate estir~1ates It is questionable v1hether any real gain in accuracy

would nave resulted

ne -Jell-knovm Gaussian oispersion formula is (Porter 1976)

C = 106 Q

iTUOCT z y

(113-1)

C 9rounct leve1 concentration in (i-igm3)

emission rate (gs)

u average vJi nd speed at the microl1ys i ca I stack nei ght (ms) 0 standard deviation of the vertical concentration distribution z

145

--------

a standard deviatinn d the noriz1intal conentration _y

distribution

H effective stack height (m)

y is the crosswind distance froin the plume centerline to the

receptor point (m)

This e4udtion was assumed to microrov1de no 11r1y average ground level concentrashy

tions (Kanzie1~i 1982) Tle vaiues for the standard demiddot-riations o and 0 are y z functions of the downwind distance x

b iJ dX (113-2)

l d

0 ex (113-3)y

where a b c and d dre constants that fit the function to the empirical

curves presented in Turner (t97C) he v~middotJnd -peed at physical stack height is

given by the equation

u (113-4)

where

u wind speed at physical stack height (ms)

llJ = measured wind speed (ms)0

h physical stack neight (m)s h the hei gtlt at which tile known wind speed was measured (m) and

0

p an empirical constant which varit~s with stability class

Lacking data on the heights at which the all known wind speeds were measured

we followed common practice and assumed a value of 10 m for h bull0

Tridl calculations showed the valu~ of tne plume rise for all the

sources except RSR Johns-Manville and the Kdser final cooler cooling tower

to be negligible (ie less than one meter) Plumbull rise formulas developed by

Christiansen (1975) and cited by Porter (1971gt) wer- used for the exceptions

The rise WdS assumed to be momentum-dominated for ltSK Johns-Manville and

Kaiser cooliny tower and bouyrncy-dominated for tne Kaiser coke ovens

As was discussed above the radial and angular distances from a

source to each surroundinlJ census tract were dett~r1ined Wind direction and

See llu s se 1ltJ 7J

146

smicroe~d data meanwt1ile ~ere obtained for eacr1 of the 16 111ajor compass points

To cr1lculdte tile conu~ntrat1on at a giv n microoirt it was first necessary to

d(termine the co1JpdSS s~ctor in Jhic11 Li1e microui11t ldy Fiyure 113-1 gives an

LXdinmicro le for a census tract at r k1~ from a source and at JU degrees from a

reference ctngle ~~t1ich we defined as north (U deyrees) As seen in the

fi9ure this ~oint lil~S in a sector bouided b the NNE (ZL5 deqrees) and NE

( 1i~ de~JretS) rn111microdss dir1~ct1ons lt1e Cttlculc11ion WdS microertormed once t1)r every

11our of the day since annually averctged values of hourly wind smicroeed were

available The followiny schedule of hourly stability clas was determined to

De cons i stent w i th t ne re I d ti on sh i micros s u mrna r i z ( d by Turner ( 1 7) y i v en the

observed distribution of wind speeds at our meteorologoical measurement

stations The schedule vas m1Jdified slightly at the suggestion of the ARl3

(Ranzieri 198l)

Hour Class Hour Class

0 F 13 B 1 F ltl 8 2 F 11) 8 J F l ~ 13 4 F 1 8 5 F 1 ~ IJN 6 F l J UN 7 DU 20 F 8 d 21 F 9 13 22 8 10 13 iJ 8 11 t1 12 13

Using the aoove equations and adj us tnents the concentration at the

point of interest was then cal cu I at~d as the sum of the concentrations

resulting from plumes naviny middot~ne boimdiny compass directions as centerlines

if the anyular distance to th~ point was conuruent with a compass direction

then only one calculcttion was necessary Let C(Y 1ti) and C(U ti) be the2

concentrations calculctted at 1our i for co111pibull)s directions Y and Y 1 2

resmicroectively 1s d1cusseltJ in Section 11L 1ur 111eteoroloyicdl ddtd in most

cases inc I uded t tie f reyuency ot wind direction for each hour of the day Let

f(Y 1

ti) and f(Q t) oe tne microrobctDilities ot occurrence of wind in the2

147

N

~-------------------------E

Figure 113-1 Determination of Compass Position of Census Tract Centroid

118

directions Y dnd w2

resp1bullctive1y at hour i rt1en the expected value of the1 cunce11Lrit1rgtrI it UH point iri questiun tt iiour I is

C(t) = f(G t )C(G t) + f(IJ t )C(q1 t) (113-5)

l 1 l 1 1 2 1 l 1

nw average dnnua I exposure wa then ca I cul ated as the average exposure on

this composite day

23 C = (124) E C(ti)

(113-6)i=U

The model was programmed in Applesoft BASIC on an Apple II

microcomputer having 48 K bytes of random access memory and a disk storage

capability The program which is included as Appendix I was compiled with

an Un-Line Systems Inc Expediter II BASIC compiler in order to decrease

running time

114 POPULATION EXPOSURE FRUM SURVEYED SOURCES

1141 Modeling Kesults

Using the modeling parameters listed in Table 114-1 tne

incremental pop~lation exposure due to each of the stationary sources was

computed Tables 114-2 through 114-12 show the modeled annual average

incremental exposure for each census tract Jround each plant Census tract

numbers appear on tne maps in Appendix H) The cumulative population column

specifies the total population exposed to all concentrations equal to or

greater than the corresponding source weighted concentration entry of the

table Figures 114-1 through 114-11 illustrate the cumulative population

exposure versus incremental concentration above ambient background concentrashy

tions Table 114-lJ lists rangemiddot of typicJI urban ambient concentrations for

eacn substance fhese gtJere used middoto assess tne incremental contribution of the

plant emissions Table 114-14 s 1 1mmarizes the incremental population exposure

due to each source These were bctsed upon annuc1l averaye source strengths and

do not reflect transients in emisions or worst case meteorolgoical

conditions Note thdt no attempt was made to assess the potential health

effects or risks to the public dut to the resultant combined exposure

149

E storaye tankworkiny loss 27x 0-u 5F fugitiv~ emission bxll -

-JLi ildSteJdter l gtX] U

Em~ssion factors are si~~lJ fer oxychlorination processes

Enlission factor vau-5 (~1rbulle Tdilt2~ frcn tne literature (JR13 198U) and

assume 9b iciency for incineration (A-E) It is also assumed above that

EIJC emissions to water are Lf~ uf ~rihsion to citir ard that in wastewater

treatment 100 of EOC discharged tc1 vater is rel easl~d to a i 1middot ~ These emission

factors 1bull1e1middote used to prioritize r-2ieaies iJut were cldrly uude approximations

to tre planL l)ecause of thlt ncine-xt~on system it was anticipated that

fdctors A-E would be ri2dtKcdft 1-middotactor - 1nii~t1t be reduced since a monitoring

program had been in force al20st o~e y2ar Factor G and the off site storage

tanks~ vni~cn are not tied into lt1n ircinerator system rniyht -iominate emissions

Emissmiddotion limits and concomitant regulations embodied in Rule l00S of

the SC-l)MU have necessitctted tlw incinera-cion system Ninety gas chronatograph

probes are located throughout t11e plant including the stack of the primary

incinerator Con cent rations of vc1v1 are reported es sent id I I y be I ow tt1e r~yu la -

tory limit of 10 ppm and in fact below the limit of detection somewhat less

tl1an 01 micropra

lUl MEASUK~MENT APPROACH

Tne objective of tt1e proyram is to determine a pl ant emissions for

EOC It is not accemicrotabl~ to dtmiddotvelop sucll informatmiddot1on basect upon published

indlllstry Jide esti11dtes of plant control efficiencivs Fortunately it was

possible to desiyn a monitoring and calculational progra1n to determine EUC

emissions for tne site and not ctbsorb a disproportionate share of program

resources

Hie re are tnre~ modes ot re I ease from tt12 micro Iant dnd d fourt11

offsite

bull Post-Process 1nci11er-dtion

Processes A-E of ecc ion lll 1 dre a 11 1ented into the pl ant

incinerdtio11 syste11 rt1e conceritrcttion of VCfmiddotl contir1uously iectsured in till=

stctck cts SJdS output remicroresents d rectsondble UjJper li11it to diJply tor EDC

con cent rat i ons s i nce i ts ef f i c i ency of i nc i nera t i on 1s d t 1 east c l1 u a I to tr1 at

134

ot VCibulll Tlerefore knowledge of the system dirflow rnd VCM concentrcttion are

tt1e necessary and sufficient conditions for determining the bounds of EDC

reledse fhe pldnt expn~sse I wi 11 inyne~)s to provide tnese data in order to

support the calculations

bull Fugitive Emisions - Valves Flanges Seals and Other Sources

Fugitive emission from the valves pumps and flanges can be

determined more precisely tnan was anticipated since Stauffer has completed a

comprehensive leak inventory of all such components in compliance with Rule

466 466 1 and 1005 The Inventory delineates all leaks uncovered by their

three man crew throughout tne year and indicates the screening level in ppmv

and component identity In consultation with Stauffer we will be able to

identify the total number of components associated with EDC handling systems

( 700) the distribution ot substance composition streams the distribution

and incidence of leaks by hardware component type and leak rate We will

utilize this information to provide historical data to compliment our Foxboro

OVA sampling at the site dased upon this monitoring we predict an EDC mass

emission rate for the fugitive releases The plant has agreed to provide the

necessary information We propose to meet witt plant personnel prior to the

start of monitoring and finlize the sdmpliny strategy to accomplish 100

coverage of the lines of hiuhest EDC composition The sampling approach and

calculation of mass emission are described in Section 72

bull ~~as tewater

From examination of the emission factors derived from published

literature (see Section 10~) it is clear that EDC release from wastewater to

air is potentially several 11rders of magnitude higher than any other plant

source l1ased upon microl ant ml~a sured concentrcJ ti ons of 8 ppm EDC in water a more

realistic emission factor wgtuld b 24xlo-4 1-iassunit mass EOC produced

Clearly this could still be the dominant source Furthermore the release

point would be expected to middotgte locctted between the plant and the sanitary

district treatment site whi1 his~ kmfrom the plant at 24501 S Figueroa We

believe it is necessary to ndemicroendently confirm the average 24 hour EDC

concentration in the dischar-qe valer by obt1ininy the refrigerated composite

sample We t1ave identified ttie I 1nes of int~rest and received agrt~ement to

sample and analyze for EDC in the stream ~e propose to draw a duplicate

sample from the compositor nd ani1lyze for its rnc content This will be

135

compared with Stauffers para11eI tnalysis It is not reliable to obtain

direct readings of EDC by survey instrument in the air above the water flow since concentrations at crny single point will be i 1 the near dmbient range

Emissions will be cc1lcJLtecl 1_sing the plants vulumetric daily flm

and ass~m~ng that complete degass40~ of the effluent will eventually occur

This assumption is based umicroon the relative volatility uf EDC and tne distances

involved However it should be noted that no direct experimental or

monitoring data is available with which to confirn this Th2 composition of

the discharge stream is unknomiddotwn since it merges into a larqe multisource flow

Conversation with the LA County Sanitation l)istrict (J f1ilne) reveals the

stream to be both exposed and covered Vents exist where me measurements

could De made to assess gross leaka~e at key points

We will obtain samp1es of several plant process discharge streams in

40 ml bottles witn no head space Transit time fr0m sample collection to

analysis will be minimized and wi l I not exceed 24 nours This time frame is

conservat i v e al though ED C i s a v o l ct ti l e mater i al ar1 ct vJi 11 undergo

concentration degradation and outgdssing lnalysi) will b(~ performed in tt1e

SAI Trace Environmental Chemistry Laborcttory utilizing pur 1Je and trap analysis

FID gas chromatography A trial a11alysmiddotis was conducted and the EOC

characteristic peak was distinctive down to the ppb level Therefore the

analysis should easily confirm concentrations in t1e 8 ppm range

Offsite Storage ranks

Three EOC leased storage tanks are located offsite at the Port

of Los Angeles and are owned by annther firm Annual emissions from the tanks

were very roughly estimated based un AP-42 JRB 1980) as 44 kkgyear

based on preliminary estimates of stored quantites

Considering tne magnitude of this source these storage tanks must

be investigated We will gather information about their configuration and

control i n order to ca I cu l ate their e111 i s s i on s bull

136

lU3 UEfERM I NAmiddot 10~~ OF MISS IONS

1031 Incinerator

Vinyl chl 11ride co11centration is monitored dt approximately 90

locations throughoul the pl trlt (Lanyner 19Bl) including the incinerator

output Conc~ntrations are reported by Stauffer dS less than 0l ppm at a

flow rate between 10000 ant 12000 SCFM Although no direct monitoring of

EDC is conducted it is posible to conservatively bound the concentration of

EDC at U 1 pp111 Tl1at conce 1ltrat ion is a suitable ci10i ce s i nee it represents a

conservative bound on tile VCM levels measured near the stack output

Furthermore the molar volume will be taken as 224 L rather than the higher

value it would nave because of the sl i~thtly elevated temperature at the

detector location

Using 12000 SCFM one has thE annual emission of EDC as 12 x 103

SCFM x 283 LSCF x 526x105 minyr x lmole x 97gmole x 10-7vv 224L

Therefore incinerator emissions of EOG are thought to be bounded by 773 kg or

170 1byear

10J2 Fugitive Emissions

Seven hundred (7UlJ) sources were surveyed comprising nearly 100 of

EDC service All accessibl~ plant areas with streams containing greater than

1 EUC were screened except for a small number of devices located in areas

11lere active maintenance was being condJCted It is believed unnecessary to

p2rf-gtrm any emission factor adjustment since it is estimated that gredter than

95 of the requisite components were screened

Three leaks above an arbitrary OVA reading threshold of 20 ppm were

detected Table 103-1 sumnarizes thei~ screening valves calculation

parameters and mass emission numbers fhe upper bound leak rate was

determined to be close to twice the nominal leak rate which accounted for the

response fltlctor of amicroproximamiddotely 05 by Radirn for TLV detection of EDC

( 13 rown 1980) bull

Stauffer found and reported approximately 10 leaks in the EDC

service during 9 months previous to the pLrnt testing Assuming no111inal leak

137

) )

Table 103-1

STAUFFER CHEMICAL MASS EMISSIONS PREDICTION FROM OVA SCREENING VALUES

OVA SAI Actual Upper Device Response Response Cone Source Leak Rate Bound Location (ppm) Rae tor (ppm) a b Se I Re Function lbyr lbyr

Gate Valve Unit P1559A

Valve Heavy Ends Column Unit 14439

Compressor Seal EDC Storage Unit Tl062

I-- w co

300 11 273 -035 096 0 17 153 D 6

11 I500 114 439 II 241 C 1

r III II II o-is lUI bull2000 vl 1 1818 A 9~

All emissions 100 EDC (12-Dichloroethane)

11

values in the nei~hborhood of 2000 lbs~ total could be emitted annually

Therefore it appears ttiat major reductions in the mass emissions were achieved

by the company run inspection prugram and the fugitive emission source has now

become of seconqary importance as a fraction of total plant emissions

lUJJ Wastewater Discharge

Wastewater samples were collected in duplicate from four sites

within the plant for determination of ethylene dichloride (EDC) concentration

The samples were collected in EPP standard 40 ml VOA vials on July 1 1981

Analyses were performed using standard purge and trap techniques coupled with flame ionization detection gas chromatography The results are given in the

table below

Sample description EDC conc~ntration range (microJml)

EDC concentration average ( 1-19ml)

Approximate fl ow ( gpm)

PVC Interceptor Box 82 - 108 95 200

EDC Stripper number Chlorination area C1404

2

011 - 014 012 25

Final collection site for pH adjustmentP663 349 - 3B2 366 350

Final discharge site sanitary sewer 6 - 254 158 500

The water in the PVC lnterc1~ptor ~ox was warm and represents washings from the

PVC reactor which travels t 1J the Interceptor Box in a concrete drainage ditch

Water from the EDC ctripper was vPry t10t and because of its heat was difficult to collect The ~1eat ot trie wat11 r may in pdrt account for tne

relatively low concentratio11 of ElC here as the EDC would out-gas from the hot

~vater more readily The fir1ctl collection site tor pH adjustment is a large

concrete container middotvith mixers in it located just prior to the final discharge

site Water at the final dischdrg~ site is being constantly aerated due to

the speed that it flows thru~yn the concrete drainage trough This aeration

could account for tne relatively large range in the rnc conotent found here

The average EDC concentraticmiddot1 at tne final discharge site is middotbelow the daily

discharye requirement of 25 19ml EDC

139

Plant oersonnel indicdted monthly iverage readings are typically on

the order of 8 ppm but daily averages can re 1ch greater than 20 microgml In

addition LA County Sanitation Uistrict staff (J Milne 1981) indicated that

an on-site impoundment pond can become laden with EOC during certain abnormal

operating periods and discharge Vdriances arl requested by Stauffer This

mi 9t1t Jccur on the order of O1c0 aCii ecH c1 1d t1erefore is no-c expected to

signifcantly impact averag~ cnnJJI discharg ~ values lt s not expected at

this t1ime tEit evaporative eriss uns from this pond are significant except

infrequently during upset conditions and spi 11 control operations Program

staff ~~ 1er2 una111Jare of any periods ~h 1~n EDC cimtent in the ponds could be

appreciable and therefore no sampling was performed

Utilizing the average uf the two final discharge concentration

1readinJs (158 micrograrnsm~ 1 and 1G 1Ji gpm fl)W one has 34600 lbyear of EDC

released into the sanitary sEwer For the pJrposes of calculating population

exposures from plant releases it will be assJmed that the EDC is locally

emitted from the wastewater streams Tl1ere ire no monitoring data available

with which to develop a more accurate release profile

However emissions of me fron the plant wastewater discharge can be

calculated according to the method of Mcmiddotckay (197S) as modified by Dilling

(1977) It should be noted that this mLst be considered an estimate since

conditions of fl ow and the presence of other substances will influence

emission rates

Using Uilling (1977) for non~erated flow first recalculate

Henry 1 s law constant (dimensionless-my of chmiddot~mical per liter of air divided by

mg of chemical per liter of water) dS

1604 P M 1 Wl

TS l

)

wnere

H Henrys law constant dimensionlessl

Pi tile compounds pur1~ componen- vapor pressure in mm Hg at T

Mwi the 11olecular weiglt

T tile absol ~te temperature of he wastewater in K

Si tt1e compounds ~uluhility in myliter at T

Then for UJC

(l6u4)(7~)(9Y)H = = U0447 with 1

(2lJ4)(8690)

LIii bullul1Jl)i I 1 1y 1 middot UL i11 wt1Ler yiven -1L L0degC is 8u4JU UHI tile vapl)r pressure

1 4 psi ( 7 2mm) at 7 0 degF bull

Tl1e overall liquid mass-transfer coefficient Kil is given by Dilling (1975)

as

(2211)(06) in mhr

-f-0-4~ (M _) 12+ 100 Wl-H-

1

=U108111tH

Then from Mackay the percent desorption is given by

c 1 = exp (-Kil tL) where

c 0

c = the concentrd ti on at time t of EOC 1

co = tile initial concentration

L = the Ii quid depth (Ill)

t = tt1e retention time (in hr) of the liquid in the wastewater

system

Retention time is bc1s~d 11pon a flow velocity range of 3 ftsec as estimated by

the Los Angeles County Sdnitatii)middot1 Dibulltrict (1J Milne) over a middotdistance of

ctmicroprox i 1nate I y S km to the Joi n_t Wdter Po 11 ut ion Contra l Pl ant Sewer fl ow

depth throughout the entire route have been estimated by J Milne as typically

1 foot 030m) to the Davison Pump Plant and 3 feet (09m) to the Joint

Tredtment Plant distances assumed to bt~ 1 mile and 2 miles respectively

Trdnsit tiine l)eco111es between 05 tnd LU hours sequentially Then computing

the net emission reduction as the product of each leg

Therefore 26 of the EOC i erni t ted bdween the pl ant and the Joint Water

Pollution Control Plant Rtgtsfdence ti111e and conditions at the plant account

141

for additional releases and residudl content is forther emitted between the

plant cnd the ocean discharge poiiit ~-Herever the flow is r2tained aerated

or shallow emissions will be accentuated

Offsite Storage Tank Emissions

There are three storage tanks leased by Stauffer for EDC storage

1ocated at 22nd and Gaffey Sts ~ bull San Pedro These tan ks a ~e used for long

term storage rather than providing feed on a routine basis Therefore

breathing loss rather than v1orkir 1J lass is )f concern An tanks are white

two being 67 ft in diameter while t~2 third is 57 ft All are 40 ft 3 inches

high and have capacities of e~thi l0~10000 (2) or 840000 gallons The

tanks are cone roof type and are not tied into vapor recovery systems The

South Coast Air Quality r1dna~J~ment u~ str4 ct has based their estimate of

emissions on the specification of 12 foot vapor level Using the AP-42

formula for breathing loss with the vapor pressure of EDC at 20degc and an

average diurnal temperature variation of 26degF one hds for each of the two

larger tanks

l) 173 H bull51 T bull 5 LlI = (619 x 10-S) M( _P_ tH F CKcp 147-P

68 173 51 5 = (619 X 10-S) (9119~ 116 bull (67) ( I 5) (26) (1) (1)

14 7-1 16)

Thus for the two larger tanks LB= 472 lbday

For the third tank 0=57 and = 178 lbdayL8 The total emissions become 23724 lbyear ~ince he plant is in the process

of shutdown it is not clear what the liquid levels currently are Note that

if the material in the three tanks were combined into one totamiddot1 -~1nissions

would be reduced markedly Exposure to a population from this site was

determined separately since it is lbullgtcated over six miles fron tl1e plant

142

110

ASSESSMENT OF PUPULATlUN EXPUSUR~

111 OVEKVIEW

A major program goal wa~ to compare emissions from the various

sources and identify and rank any 11 tlot spots in California where the general

population was exposed to elevated concentrations of carcinogens A simple

Gaussian dispersion model was therefore used to obtain order-of-magnitude estimates of exposure of the genermiddotal population surrounding each source

Since this was essentially a screening study use of more sophisticated models

was not appropriate

It cannot be emphcsized too stronyly tnat most California residents

are exposed to emissions of hazardous substances from a variety of natural and

rnanmade sources Urban dwellers dre typically exposed to greater concentrashy

tions than rural residents however all are subjected to so called 11 background 11 levels from multiple sources In order to place stationary

source exposures in perspective the typical ambient levels of each substance

were identified from the literature and compdred with the concentrations due

to the emissions from each ~lant Exposures were thus expressed both as

absolute quantities and as increments above background

Comparison of plants presents a further difficulty in that various

substances are being consict~red No attempt was made to evaluate the relative

importance of exposure to two different substances such as chloroform versus carbon tetrachloride other than by ambient concentration

112 DATA SOURCES

1121 Meteorologicdl Llata

The dispersion model to be descritgted in Section 113 required input

of annual average wind speed and frequency of occurrence of wind from each

compass direction These data were obtained for most of the sites from the

South Codst Air wuality Management District (SCAQMD) In other cases (Kaiser

Jotms-Manvi l le l)ow and OuPlt)nt) only clai ly cJVerage wind speed and frequency

data were available In al I cases we used data from the meteorological

station nearest the modeled e11issrnn source Table 112-1 summarizes the

141

Table 112-1 METEOROLOGICAL DATA BASE

Observation Site Data Period of Pldnt (Distdnce Source Observation Remarks

from Plant)

A11 i ed Che11 i Ca 1 Lennox SCA~MD 1Y71-1974 Averaged hourly readings available ~ 1 Segundo (3 111iles)

S ff~r-- Uemical Long l3each SCA()MD 1962-1974 Averaged hourmiddoty leadir19s avai1able offsite Carson (3 miles) storage tanks appruxi1ntL~ly 5 miies from

rneteoro middot1 ogy s I te

I--

-~ Dow Pittsburg Pittsbury lJOVJ 1955-1974 Daily and averacied Ninnd rrJse on1y--no hourly-~

Uu~ont Antiocn (lt5 miles to data uVdilable Antioch)

r~d i ser Stee 1 Fontdnct SCAQMD 1978-1979 Two years of hourlydaily data printout made r - i -- lt3 1niles available (33000 entries) daily averages

estimated as only practical alternative -middot bullA

Jon 1 s 1middot1 an I i 1 l e Stockton NOAA 1941-1970 No hourly data estiamtes based on monthly S-cJc~ton (1 mile) prevalence of directions and speed

R~R City uf Whittier SCAQMD 1969-1973 Averaged hourly readings available lnuustry (4 miles)

1neteoroloyical data 1iase for eden site Wind speed and 1tJind direction

freljuency dcttd us~lti 1r1 Ull~ model are provided in Apmicroendix Li

11LL Populdtion Ddtct

In oroer to assess popu Idt i 011 exposure in tile same way for a 11 the

sources we defined d lLJ-s4u1re mi le impact area arround each ~ant This

size WdS cnosen since it was found in most cdses to include all distances at

whicn incremental ground lev~I concentrations due to plant emissions would

exceed genera I urban ambient Ieve 1s for the microo 11 utant in question In most

cases the plant was placed ltlt the center of the impact area Where winds

were predominantly from one sector or a few adjacent sectors or where an

unpopulated area (ey the Pacific Ucectn) adjoined one side of a plant the

impact drea was defined to li~ imm~didtely downwind of the site

Once the impact areas were defined we obtained Thomas 8rothers maps

of all census tracts within them These maps are provided in Appendix H

Census tract populations were obtained from the 1980 US Census The

population of each tract was assumed to lie dt tne centroid except when tne

tract was large and most of the population was concentrated away from the

centroid in the latter case the best-defined population center was used

fadial and angular distances from the sources to the population centers were

tr1en determined

11~3 OISPERSION MODELING APPROACH

ln urder tu estiindt~ pomicroulcttiun expusures in the census tracts

surrounding each source a simple Gaussian dispersion model was used Use of

a 1nore sopnisticdtect 111odel ~-1as inamicroproJridte yiven the uncertainty in our

emission rate estir~1ates It is questionable v1hether any real gain in accuracy

would nave resulted

ne -Jell-knovm Gaussian oispersion formula is (Porter 1976)

C = 106 Q

iTUOCT z y

(113-1)

C 9rounct leve1 concentration in (i-igm3)

emission rate (gs)

u average vJi nd speed at the microl1ys i ca I stack nei ght (ms) 0 standard deviation of the vertical concentration distribution z

145

--------

a standard deviatinn d the noriz1intal conentration _y

distribution

H effective stack height (m)

y is the crosswind distance froin the plume centerline to the

receptor point (m)

This e4udtion was assumed to microrov1de no 11r1y average ground level concentrashy

tions (Kanzie1~i 1982) Tle vaiues for the standard demiddot-riations o and 0 are y z functions of the downwind distance x

b iJ dX (113-2)

l d

0 ex (113-3)y

where a b c and d dre constants that fit the function to the empirical

curves presented in Turner (t97C) he v~middotJnd -peed at physical stack height is

given by the equation

u (113-4)

where

u wind speed at physical stack height (ms)

llJ = measured wind speed (ms)0

h physical stack neight (m)s h the hei gtlt at which tile known wind speed was measured (m) and

0

p an empirical constant which varit~s with stability class

Lacking data on the heights at which the all known wind speeds were measured

we followed common practice and assumed a value of 10 m for h bull0

Tridl calculations showed the valu~ of tne plume rise for all the

sources except RSR Johns-Manville and the Kdser final cooler cooling tower

to be negligible (ie less than one meter) Plumbull rise formulas developed by

Christiansen (1975) and cited by Porter (1971gt) wer- used for the exceptions

The rise WdS assumed to be momentum-dominated for ltSK Johns-Manville and

Kaiser cooliny tower and bouyrncy-dominated for tne Kaiser coke ovens

As was discussed above the radial and angular distances from a

source to each surroundinlJ census tract were dett~r1ined Wind direction and

See llu s se 1ltJ 7J

146

smicroe~d data meanwt1ile ~ere obtained for eacr1 of the 16 111ajor compass points

To cr1lculdte tile conu~ntrat1on at a giv n microoirt it was first necessary to

d(termine the co1JpdSS s~ctor in Jhic11 Li1e microui11t ldy Fiyure 113-1 gives an

LXdinmicro le for a census tract at r k1~ from a source and at JU degrees from a

reference ctngle ~~t1ich we defined as north (U deyrees) As seen in the

fi9ure this ~oint lil~S in a sector bouided b the NNE (ZL5 deqrees) and NE

( 1i~ de~JretS) rn111microdss dir1~ct1ons lt1e Cttlculc11ion WdS microertormed once t1)r every

11our of the day since annually averctged values of hourly wind smicroeed were

available The followiny schedule of hourly stability clas was determined to

De cons i stent w i th t ne re I d ti on sh i micros s u mrna r i z ( d by Turner ( 1 7) y i v en the

observed distribution of wind speeds at our meteorologoical measurement

stations The schedule vas m1Jdified slightly at the suggestion of the ARl3

(Ranzieri 198l)

Hour Class Hour Class

0 F 13 B 1 F ltl 8 2 F 11) 8 J F l ~ 13 4 F 1 8 5 F 1 ~ IJN 6 F l J UN 7 DU 20 F 8 d 21 F 9 13 22 8 10 13 iJ 8 11 t1 12 13

Using the aoove equations and adj us tnents the concentration at the

point of interest was then cal cu I at~d as the sum of the concentrations

resulting from plumes naviny middot~ne boimdiny compass directions as centerlines

if the anyular distance to th~ point was conuruent with a compass direction

then only one calculcttion was necessary Let C(Y 1ti) and C(U ti) be the2

concentrations calculctted at 1our i for co111pibull)s directions Y and Y 1 2

resmicroectively 1s d1cusseltJ in Section 11L 1ur 111eteoroloyicdl ddtd in most

cases inc I uded t tie f reyuency ot wind direction for each hour of the day Let

f(Y 1

ti) and f(Q t) oe tne microrobctDilities ot occurrence of wind in the2

147

N

~-------------------------E

Figure 113-1 Determination of Compass Position of Census Tract Centroid

118

directions Y dnd w2

resp1bullctive1y at hour i rt1en the expected value of the1 cunce11Lrit1rgtrI it UH point iri questiun tt iiour I is

C(t) = f(G t )C(G t) + f(IJ t )C(q1 t) (113-5)

l 1 l 1 1 2 1 l 1

nw average dnnua I exposure wa then ca I cul ated as the average exposure on

this composite day

23 C = (124) E C(ti)

(113-6)i=U

The model was programmed in Applesoft BASIC on an Apple II

microcomputer having 48 K bytes of random access memory and a disk storage

capability The program which is included as Appendix I was compiled with

an Un-Line Systems Inc Expediter II BASIC compiler in order to decrease

running time

114 POPULATION EXPOSURE FRUM SURVEYED SOURCES

1141 Modeling Kesults

Using the modeling parameters listed in Table 114-1 tne

incremental pop~lation exposure due to each of the stationary sources was

computed Tables 114-2 through 114-12 show the modeled annual average

incremental exposure for each census tract Jround each plant Census tract

numbers appear on tne maps in Appendix H) The cumulative population column

specifies the total population exposed to all concentrations equal to or

greater than the corresponding source weighted concentration entry of the

table Figures 114-1 through 114-11 illustrate the cumulative population

exposure versus incremental concentration above ambient background concentrashy

tions Table 114-lJ lists rangemiddot of typicJI urban ambient concentrations for

eacn substance fhese gtJere used middoto assess tne incremental contribution of the

plant emissions Table 114-14 s 1 1mmarizes the incremental population exposure

due to each source These were bctsed upon annuc1l averaye source strengths and

do not reflect transients in emisions or worst case meteorolgoical

conditions Note thdt no attempt was made to assess the potential health

effects or risks to the public dut to the resultant combined exposure

149