Chemical demulsification of aged, crude oil emulsions - Environmental

Ralph C. Little. Environ. Sci. Technol. , 1981, 15 (10), pp 1184–1190. DOI: 10.1021/es00092a005. Publication Date: October 1981. ACS Legacy Archive...
2 downloads 0 Views 763KB Size
Chemical Demulsification of Aged, Crude Oil Emulsions Ralph C. Little Naval Research Laboratory, Washington, D.C. 20375 ~~~

The viscosities and the rates of evaporation for six tanker crudes of various geographical origins were studied to ultimately determine the resultant effect on the chemical demulsification of their seawater-in-oil emulsions. A 5- to nearly 15-fold increase in viscosity was observed upon loss of their more volatile fractions. In spite of initial differences in viscosity, five of the six crudes had similar evaporation rates. It also was found that the evaporation rate of volatiles from a 50% seawater-in-crude-oil emulsion was not greatly different from that of the raw crude. Aerosol OT was found to be an acceptable chemical demulsifier of the emulsions generated from these crudes. As expected, low temperatures decreased the rate of demulsification for a given crude. Moreover, the demulsification rate was significantly influenced by the demulsifier concentration. In general, emulsions generated from the less viscous crudes were more easily broken probably as a result of lower concentrations of naturally occurring emulsifier and lower viscosity of the continuous phase.

The artificial seawater used was made up by dispersing and dissolving the recommended amount of ‘kea salt” (Lake Products Co.) in distilled water. Methods. A Virtis homogenizer was used to form the emulsions. Equal volumes of oil and artificial seawater were added to the special flask, and emulsification was complete after a 5-min period of high-speed mixing. All rheological measurements were made with both a Brookfield Model LVT rotary viscometer and a series of Cannon-Fenske Capillary viscometers. Sample temperature was controlled by means of a constant-temperature bath maintained at f O . l “C. The processes used in emulsion production and destabilization may be represented as follows: emulsion making homogenizer

water/oil (w/o) system ------+

w/o emulsion time

Introduction

The successful resolution of oil-spill problems requires that the available cleanup equipment be quickly transported to the site of the spill, weather permitting. The Navy is involved in an effort to develop a worldwide oil-spill response capability through the use of strategically located pollution response centers (1).Unfortunately, as a result of weather conditions or equipment limitations, significant quantities of seawater can be entrained or emulsified with the collected crude oil. This report deals with a chemical method to assist the separation of oil and seawater under ambient conditions of temperature. In order to provide as broad a base as possible, tanker crudes of varying viscosity and volatility from different geographical origins have been selected for the purpose of providing a realistic test of the chemical demulsification technique. Experimental Section

Materials. Six 1-gal samples of crude oils were kindly supplied by the Exxon Corp. at Baytown, T X , in response to a request for crudes expected to be representative of “light and heavy types”. They were labeled as Prudhoe Bay, BCF44, Ivory Coast, English, Arabian Heavy, and Nigerian Mix. The sodium dioctyl sulfosuccinate was obtained from American Cyanamid as a 75% active “Aerosol OT” solution. The remainder was water and -5% of a low molecular weight alcohol to provide fluidity. 1184

Environmental Science & Technology

emulsion breaking aged w/o emulsion mixing

+ demulsifier

---+destabilized emulsion

aged w/o emulsion

time

water/oil system

A previous report ( 2 ) suggested that aging the emulsion for 1or 2 days greatly reduced the experimental variations found when fresh emulsions were demulsified. Moreover, it is expected that aged emulsions more closely simulate those emulsions expected in oil spills. Accordingly, all emulsions were aged before chemical demulsifier was applied. An ultrasonic probe (Branson Sonic Power Model S-75)was used to mix demulsifier with 25-mL samples in 50-mL plastic vials followed by separation in 25-mL test tubes. Ultrasonic mixing was 30 s, and the samples were precooled to avoid temperatures in excess of that desired. In each case, samples were placed in a constant-temperature bath maintained at f O . l “C, and volume readings of the separated water layer were taken at selected time intervals and recorded. Evaporation Studies. The evaporation experiments were of two types: (1)preparatiue, to obtain the required sample for viscosity and chemical demulsification measurements, and (2) analytic, to obtain information on the evaporation process. In the first case, 400 mL of each crude oil was placed in a 1000-mL beaker and allowed to evaporate for 2 weeks in a

This article not subject to US. Copyright. Published 1981 American Chemical Society

laboratory hood. Within the context of this report, a sample thus prepared was termed “an evaporated crude”. The total loss of material from the sample (directly weighed by a 2-kg capacity Torbal balance) during this period of time was designated as the percent “volatiles” lost. I t is admitted that the process of evaporation is a continuing one and that, had the evaporation period been increased, additional volatiles would have been lost. The choice of 2 weeks is arbitrary but, as shown below, reasonable. In the second case, samples of crude oil were placed in Petri dishes (7.25-cm radius and 1.9-cm wall height) to a thickness of 0.4 cm. The ducted airflow over the sample was adjusted to a velocity of 5 mph with an air velocity meter (Taylor Instruments, Rochester, NY). The loss in sampie weight was monitored by a Statham transducer cell to which was attached a load cell accessory with its output feeding a recorder. It .was found that the experimental data could be plotted as time/ percent weight lost vs. time (equation discussed in later section). From such plots, an estimate of percent volatiles lost a t infinite time could be made. Such estimates tended to be only 1-3% higher in absolute value than the 2-week beaker results, Le., English crude 25.6% vs. 24.3%, Ivory Coast crude 19.2% vs. 18.1%,etc.

Results and Discussion Basic Considerations. The chemical demulsification of tanker crude emulsions requires that the naturally occurring emulsifiers present in the seawater-oil interface be effectively displaced by the demulsifier. Since successful demulsifiers may be considered to be very poor emulsifiers, effective chemical demulsification will be dependent upon the collision rate between droplet particles. That is, after the chemical demulsifier is thoroughly mixed with the emulsion and steady-state conditions are reached, the flocculation rate, a , may be approximated by (assuming each collision is beneficial) a = 2kT/(3puD)

where Nie = evaporative flux of component i (mol cmw2s-l), = evaporative mass-transfer coefficient (cm s-l), X, = mole fraction of component i, P, = vapor pressure of pure component i (torr), R = gas constant (cm3 torr/(mol K)), and T = air temperature above the oil (K). This relation assumes ideal solutions wherein the heat of mixing is zero. In practice, crude oils are nonideal mixtures of components having widely varying solubility parameters (6). In such cases, the evaporative flux equation might perhaps best be modified to

K,

(3)

where yi is the activity coefficient defined by the relation

and V, = molar volume of compone_nt i (cm3 mol-I), 6, = solubility parameter of component i,6 = 2,=ln&a,, and 4 = volume fraction. However, the problems inherent in accounting for the loss of volatile components by using the above equation led to laboratory experiments with selected natural crudes and ultimately the use of descriptive empirical equations to describe the evaporative fluxes observed. The interplay of raw-crude viscosity, evaporated-crude viscosity, and temperature on the chemical demulsification process is developed in the subsequent sections. Viscosity of Crudes. The viscosities of the crudes varied greatly with their origins and were strongly affected by the temperature and the volatiles lost through evaporative processes (see Table I). Substantial changes in viscosity of well over 1 order in magnitude may significantly affect the resultant viscosity of water-in-oil emulsions and, subsequently, the ease with which these emulsions may be broken through use of chemical demulsifiers, as will be shown later. The crudes were essentially Newtonian in their flow properties, and their viscosities could be described by means of the Walther equation (7): log log (v

(1)

where k = the Boltzmann constant (erg K-l), T = absolute temperature (K), v = viscosity of continuous phase (cSt), D = diameter of drop (cm), and p = density of continuous phase (g cmP3).I t is clear that the flocculation process is a necessary precursor to coalescence (3)and that large values of viscosity will greatly inhibit flocculation even if the emulsifier present is a poor one. The process of evaporation (with subsequent loss of volatiles from the crude) will result in a 2-fold effect: (a) the viscosity of the crude will be increased and (b) the concentration of naturally occurring emulsifiers, such as asphaltenes ( 4 ) ,in the crude will increase. Significant increases in emulsion stability or increased resistance to chemical demulsification should occur as a result of the evaporative process. The evaporative flux of volatile crude oil components has been estimated through use of Raoult’s law ( 5 ) ,the resultant equation having the form

+ k ) = A + B log T

(4)

where k = 0.6 for oils having kinematic viscosities greater than 1.5 cSt, u = kinematic viscosity (cSt), T = absolute temperature (K), and A and B = constants to be determined for each oil. Table I1 lists data for the six crudes studied. Walther equation constants for both the raw crudes and the evaporated crudes are given in the table together with limited data for the intermediate, partially evaporated crudes. Unfortunately, the constants A and B did not vary in a predictable fashion with the percent volatiles lost so that a Walther equation expressing viscosity-temperature relations as a function of volatiles lost could not easily be constructed. However, the Walther equation of the raw and evaporated crudes permits a reasonable estimation of viscosity in the 0-50 “C range for these crudes. Figure 1 features ASTM-derived viscosity-temperature plots for the six oils tested. In the figure, both the raw crudes and the evaporated crudes are plotted for purposes of comparison. The Ivory Coast crude showed the greatest viscos-

Table 1. Viscosity (cSt) of Raw and Evaporated Crudes raw crude source

no.

u5OpC

no.

YOOC

Q59C

Y50°C

6.86

3.53

2.21

1’

84.0

21.2

8.53

9.59

4.61

2.74

2’

359

28.8

7.51

6.18

2.95

3‘ 4‘

168 1430

32.5 55.5

11.2 9.91

5’

3000

384

92.6

6‘

1960

176

37.6

BCF44

1

Nigerian Mix

2

English

3

19.5

Ivory Coast Arabian Heavy Prudhoe Bay

4 5

81.3 85.7

6

evaporated crude Y250C

vooc

104

12.6 28.3

4.26 12.9

26.5

10.5

Volume 15, Number 10, October 1981

1185

Table II. Walther Equation Constants as Function of Percent Volatiles Lost % los13

A

BCF44 Nigerian Mix

0 0

9.60 9.51

-3.96 -3.90

31.9 15.7

8.89 12.06

-3.57 -4.86

36.2 19.1

9.95 14.60

-3.98 -5.86

38.4 22.9

10.42 16.00

-4.16 -6.40

English

0

12.63

-5.14

19.7

10.42

-4.16

22.4

10.74

-4.28

24.3

10.97

-4.36

Ivory Coast

0

15.16

-6.11

11.7

15.17

-6.07

15.0

15.92

-6.35

18.1

16.92

-6.74

Arabian Heavy Prudhoe Bay

0 0

8.09 9.85

-3.20 -3.92

17.6 11.9

8.55 9.25

-3.30 -3.63

20.6 14.3

8.74 10.04

-3.37 -3.93

22.5 17.8

8.85 11.14

-3.41 -4.36

a

B

% lost2

*

% losll a

crude oll

A

B

A

log V I V O = 2.704

loo0

(5)

-

600-

A

B

where u = kinematic viscosity of emulsion, ug = kinematic viscosity of oil, and 4 = volume fraction of seawater. When this equation was used as a tool, Table I11 was generated to determine the approximate range of emulsion viscosities which might be expected from the crudes. Highly viscous emulsions may present additional problems in collection, handling, and chemical treatment. However, a warming of such emulsions to 25 "C greatly reduces their viscosity, and they should become less intractable as a result. Emulsions generated with the lighter crudes, Le., BCF44, English, and Nigerian Mix, remain quite fluid even at 0 "C and therefore should be easily collected and treated. Evaporation of Crudes. As already pointed out, it was felt that the precise calculation of evaporation rates of nonideal crude oil systems might be difficult to treat in a semitheoretical fashion. Therefore, evaporation experiments were performed in the laboratory hood to obtain the necessary data (see Experimental Section for details). I t was found that the experimental data conformed rather well to an emiprical equation of the type

400-

W/Wm = t/(tl/z

200

-

100

-

I

-

v)

2 40> 30

V

20

+t)

(6)

where W = weight percent volatiles lost at time t , W, = initial weight percent volatiles of oil, and t 1/2 = time when 50% of volatiles have evaporated. Figure 2 illustrates the plots obtained for two different crudes. The reciprocal of the slope yields the percent volatiles. The half-life is determined from the reciprocal of the slope times the intercept a t t = 0. Table IV lists the data obtained from the evaporation experiments. The initial evaporation rates, R e , were estimated by differentiating this expression and incorporating the area of the Petri dish, i.e.

8 0 -60-

z

% lost4

Raw crude. * After 72 h. After 168 h. Evaporated crude (2 weeks in laboratory hood).

ity-temperature slope (highest B constant in Table 11)within each separate grouping of the raw and evaporated crudes. The Nigerian Mix crude also showed the second highest slope in the set of evaporated crudes, being fully two slope units higher than the other four crudes. These two crudes also show evidence of substantial wax content because on evaporation a surface crust was observed. On the average, the crudes increase 6-fold in viscosity through evaporation of volatiles. The viscosity of seawater-in-crude-oil emulsions may be estimated by use of the Richardson equation (8).A previous report ( 2 ) suggested that the viscosity data of emulsions of seawater in residual-type oils approximately followed the expression

&0 y u t 5z

B

-

Re

= t1/2Wrn/[A(tl/z + tI21

(7)

where R e = evaporation rate (g cm-2 s-l), Wm = original

10-

-

8.0

9-

6.9.-

8-

-

74.0

-

6-

3.0 -

1

I

I

I

-1 1

0

25

50

2.0

TEMPERATURE ("C)

Flgure 1. ASTM-type viscosity-temperature plots for raw and evaporated tanker crudes studied: (1) BCF44; (2) Nigerian Mix; (3) English; (4) Ivory Coast; (5) Arabian Heavy; (6) Prudhoe Bay. Primes refer to evaporated samples of same. 1186

Environmental Science & Technology

00

I

50

I

1

100 150 TIME (MINUTES)

I

200

2 0

Figure 2. Test of empirical equation W/ W,,, = t , / * / ( t f t , / 2 ) ( A ) for Nigerian Mix and (0)for BCF44; 25 O C , 5-mph air velocity.

Table 111. Calculated Water-in-Oil Emulsion Viscosities (cSt) from Raw and Evaporated Crudes yooc

V25OC

7 5 % emulslon

5 0 % emulsion

25% emulsion

vlscosity of 011

vooc

vooc

V25-C

vooc

V25OC

V25'C

raw crudes 79.0

374

BCF44

6.86

3.53

32.4

16.7

154

Nigerian Mix

9.59

4.61

45.4

21.8

215

103

1020

727

488

English

19.50

6.18

92.3

29.2

437

138

2060

655

Ivory Coast Arabian Heavy Prudhoe Bay

81.30

12.61

385

59.7

1820

282

8620

1340

85.70

28.31

405

134

1920

634

9080

3000

104.3

26.46

493

125

2335

592

11000

2800

100

evaporated crudes BCF44

Nigerian Mix English Ivory Coast Arabian Heavy Prudhoe Bay

1900

475

8900

2200

3050

425

14400

2000

154

3800

730

17800

3400

6800

263

32000

1240

152000

5900

14400

1800

68000

8600

350000

51000

9290

835

4400

4000

208000

19000

84.0

21.2

397

136.3

19.0

645

168

32.5

794

3000

55.5 384

1960

176

1430

'

89.9

Table IV. Evaporation of Crudes a in a Ducted 5-mph Airflow at 25 "C

crude oil

% volatiles

half-llfe, mln

Initial evaporation rate, g/(cm2-s)

x x 5.71 x 5.72 x 1.33 x 5.54 x 1.15 x

BCF44

35.8

16.5

9.40

10-5

Nigerian Mix

24.6

10.3

1.09

10-4

English Ivory Coast

25.6

19.2

19.2

15.4

Arabian Heavy Prudhoe Bay toluene

23.6

8.7

water a

17.8

152

10-5 10-5 10-4 10-6 10-4

1.85 x 10-5

Ca. 4-mm thickness in a low-wall Petri dish of 14.5-cm diameter.

1 MINUTE 10-7

10

I

I

102

10 MINUTES

1 HOUR

I

I

1

103 TIME (SEC)

Figure 3. Evaporation of crudes from 4-mm dishes (25 "C, 5-mph air velocity).

1 DA'I I

104

I

1(

layers in laboratory Petri

weight of volatiles in sample (g), tl/z = half-life (s), t = time (s), and A = area of dish (cm2) and, therefore,

In general, the initial evaporation rates of the crudes were quite similar and varied only by a factor of 2 with the exception of the Prudhoe Bay sample where the evaporation rate was 1 order of magnitude lower. Figure 3 presents evaporation-rate curves for the six crudes, over a 3-h period. In spite of differences in origin and viscosity, the five samples form a fairly narrow envelope of curves with the exception of the Prudhoe Bay sample. After 3 h, a t 25 "C (-0.5 day a t 0 "C) essentially 95% of the volatiles have evaporated from the 4-mm layers of the five crude oil samples a t the 5-mph wind condition. Evaporation rates at lower temperatures, for example, 5 "C, may be estimated by application of the earlier expression relating evaporative flux and the other physical factors together with the Clausius-Clapeyron equation as follows:

Since AH,,, z 2 1 T b (where T b = boiling point in (K)),an approximate change in total vapor pressure for a hypothetical mixture containing a volatile component whose boiling point was 100 "C would be

Hence Pz z 0.36P1 for a component where Tb P 373 K. Correcting for temperature leads to a decrease in the initial evaporation rate of -0.38. The half-life, tllz, would therefore be 2.6 times longer a t 5 "C than a t 25 OC. The Petri-dish data might also conceivably be extrapolated to larger pool sizes representative of oil spills and to higher wind velocities through use of Sutton's equation (9) for the mass-transfer coefficient, K,, i.e. K , = CUO 780-0.11 (10) Where C is a constant, U is the velocity, and D is the pool diameter. Evaporation of Volatile Hydrocarbons from an Emulsion. Experimental Data. In order to determine whether the presence of the dispersed phase had an effect on the evaporative process, experiments were devised simulating a 5-mph wind blowing over a 4-mm thickness of (a) a raw crude oil, BCF44, and (b) an emulsion of 50%ethylene glycol in BCF44. Ethylene glycol was chosen because of its low vapor pressure and the ease with which reasonably stable glycolin-oil emulsions could be made, thus avoiding additional complicating evaporative weight losses had water been used. The mean droplet size of the dispersed glycol was observed Volume 15, Number 10, October 1981 1187

in the microscope to be roughly 10 microns and was thus comparable to a seawater-in-oil dispersion. The crude oil and the emulsion were placed in Petri dishes whose weights were monitored by Statham load cells. Figure 4 reports the evaporative-loss data as a function of time for both the raw crude and the emulsion. It is clear that the evaporation losses in the emulsion (based on the oil content) are nominally the same as with the raw crude. The incorporation of microdroplets of ethylene glycol into the fluid medium apparently does not sufficiently lower the diffusivity of the volatile components to the surface layer to the point that transport across the interface into the gas phase is affected. Figure 5 shows the computed growth of emulsion viscosity with time for a raw Arabian crude emulsion a t two temperatures, 0 and 25 "C. The solid line represents a 75% emulsion wherein evaporation of volatiles increases the viscosity of the base crude a t constant phase volume. The dashed curve more realistically accounts for the additional changes in water and oil phase volumes as the volatile components are lost. The following equations were used for the estimates: log V . = K & + VO

Crude oil emulsions were easier to break and required less agent as the viscosity of the continuous phase decreased within the six-sample set. In the case of the raw English and Nigerian crudes, it was found that their 50% emulsions of seawater were not stable. Separation took place within minutes and was complete within 1 h or less. Raw BCF44 emulsions were, however, reasonably stable. Figure 6 is a smoothed curve for 1 and 34 "C conditions computed from the statistical equations. Experimentally, 0.04% concentration of demulsifier and higher did not break the emulsion. The statistical equation

Table V. Typical Data Sample for Chemical Demulsification of 5 0 % Emulsions of Seawater in Arabian Heavy Crude

tu2

+t

(12)

Where k was experimentally found to be v0/30 for the six oils studied. In general, the major increase in emulsion viscosity takes place within the hour for the 25 "C case and within 3 for the 0 "C case. Again, the data for our laboratory conditions, Le., 4-mm thickness of emulsion, 14.5-cm diameter Petri dish; Sutton's equation must be used to relate this information to oil-spill dimensions. Chemical Demulsification. In order to gain the most information on the effect of temperature and agent concentration on the course of demulsification, it was decided to use a two-variable, second-order, central composite design. Table V shows a typical set of data, the developed equation (in coded form) and a comparison of the experimental vs. calculated demulsification times. The demulsification data are useful for showing generalized trends reflecting the effect of concentration and temperature.

log (exptl demuls the), mln

log (calcd a demuls tlme), min

resldual

0.067

run no.

temp,

1

18.4

0.40

2.550

2.48

2

23

0.20

1.550

2.498

3

23

0.60

1.740

1.871

-0.131

4

34

0.117

2.041

2.190

-0.149

5

34

0.40

1.919

1.949

-0.030

6 7 8 9 10

34

0.40

1.978

1.949

0.029

34

0.683

1.204

1.100

0.104

45

0.20

2.000

1.821

0.179

45

0.60

0.903

49.6

0.40

1.204

0.903 1.320

ktW, tll2 + t

ktW,

agent concn %

OC

'log t = 1 9 4 9 - 0.386Xi - 0.411X2 0.0718XiX2. ' S = R ' / ( N - KJ= 0.1046/(10

0.052

0 -0.116

- 0.152X1' - 0.024.3X22 - 6) = 0.02615. S = 0.1617.

'- I 1

I I

B

HOUR

I

DAY

WEEK

25 "C (see text for equations used).

,

*

Z z3 F 6I

/

I

YEAR MONTH WEEK DAY

I

50

I

100 TIME (MINUTES)

1

150

Figure 4. Comparative evaporation of volatiles from a raw crude (BCF44) and its 50 % emulsions containing glycol (evaporation from emulsions based on its crude content): (e)crude; (9) emulsion: 25 OC, 5 mph. 1188

Environmental Science & Technology

- 10 MIN 1

I

I

I

1 MIN

Table VI. Coefficients for Statistical Equation: log f = bo bo

Oil

Prudhoe Bay, raw

2.106

bi .

b2

-0.177

-0.438

4- b l x l 4- b 2 x 2 4-

b l l x ~ 4-~b 2 2 X 2 2

-0.118

-0.128

-0.0152

BCF44, raw

1.792

-0.221

-0.271

0.0474

-0.0484

-0.0222

Ivory Coast, raw

0.328

-0.128

-0.483

0.00582

-0.209

-0.160

1.949

-0.386

-0.411

-0.151

-0.0236

-0.0718

comments

1.774

-0.148

-0.152

-0.0526

-0.0486

0.111

-0.453

-0.0440

-0.336

emulsions unstable

English, raw Arabian Heavy, raw Nigerian Mix, raw Prudhoe Bay, evaporated BCF44, evaporated Ivory Coast, evaporated English, evaporated Arabian Heavy evaporated Nigerian Mix, evaporated

0.517

2.488

0.0166

NSFO

0.256

0.246

No. 6 fuel oil

1.302

a

4- b 1 2 X 1 X 2

biz

622

b ii

emulsions unstable 0.358

-0.260

emulsions rapidly unmixa emulsions rapidly unmixa 0.0321

-0.0692

-0.0486'

0.111

emulsions rapidly unmixa -0.504

-0.825

1.014

0.172

0.204

-0.812

0.273

0.379

0.460

Rapid unmixing prevents determination of coefficients with experimental setup used.

(see Table VI for the descriptive equation) predicts that concentrations above 0.02% (at 34 "C) and 0.017% (at 1 "C) will be ineffective in breaking these emulsions. Therefore, care must be taken in the use of chemical demulsifiers since, if excess amounts are used, not only will economic waste result but even more intractable emulsions may probably be produced by the added demulsifier itself. With respect to the other raw less viscous crudes, Nigerian Mix and English, it was found that their emulsions were unstable at room temperature. In the case ofthe emulsions generated from evaporated less viscous crudes, it was found that both the Nigerian and English samples were rapidly demulsified; 0.02% concentration sufficed for the English crude emulsion while 0.08%was necessary for the Nigerian Mix. It was difficult to separate demulsifier concentration effects with the equipment used, and the statistical design, therefore, was not pursued further. The evaporated BCF44 emulsion was quite stable, and the effect of added demulsifier is shown in Figure 7. Small amounts of demulsifier of the order of 0.1%work best; larger amounts are tolerated but at lower emulsion breaking efficiency. While the raw crude was easily demulsified in several hours, even a t 1 "C, the evaporated crude requires -1 day a t the 0.1% demulsifier level. While the curves suggest that lower concentrations would work better, it was found that slightly lower concentrations were quite ineffective. It is concluded that the less viscous crudes require minimal amounts of demulsifier

and that small amounts in 0.02% increments be cautiously added to determine optimum working concentrations. Ivory Coast was somewhat unusual in that its raw crude emulsions appeared slightly more difficult to break than in its evaporated form (see Figures 8 and 9 and Table VI). Nevertheless, very minimal amounts of demulsifier-in the range of 0.04-O.l%-are effective in breaking its emulsions over a wide range of temperatures. The Prudhoe Bay and Arabian raw crude emulsions offer more substantial resistance to demulsification at 34 "C (Figure 8). The demulsification characteristics of both crudes are quite similar with respect to 6

1 YEAR

iz 5

1 MIN

PO

s-20.1

0.2

0.3

0.4

0.5

0.6

0.7

PERCENT DEMULSIFIER

Figure 8. Demulsification of raw crude emulsions at 34 Coast; (5) Arabian Heavy; (6) Prudhoe Bay.

z

E

Y F

OC: (4)

Ivory

- 1 YEAR

3

1 DAY

10MIN

W

1 YEAR 1 MONTH 1 WEEK

1 HR

------6

- 1 MONTH 4-

- 1 WEEK

'2

:

2t

4

I! 10 MIN 1 MIN

-2 0.1

u-

Y3

t

410 MIN

In 0

0.2

0.4 0.5 PERCENT DEMULSIFIER

0.3

0.6

0.7

Flgure 7. Demulsification of evaporated BCF44 emulsions at 1 and 34 "C.

Volume 15, Number IO, October 1981 1189

0.2

0.4

I\

0.6

0.8

1.0

I

1.2 I

1.4 1 1

5

0.4

0.6

0.8

1.0

I

1.2

1.4

I

1'A1

I

z

0

0.2

61

I

1 MONTH 1 WEEK

4,

4 1 DAY I

YEAR MONTH 1 WEEK

4

4 t

t 2

1

I , /

#6]10 MIN

1 MIN

-2O0.1

0.2

0.3

0.4

0.5

0.6

0.7 Y

-2 1 0.1

I

0.2

PERCENT DEMULSlFlER

Figure 10. Demulsification of evaporated crude emulsions at 1 "C: (5')

Arabian heavy (evaporated);(6') Prudhoe Bay (evaporated);(NSFO)Navy Special Fuel Oil; (#6 F.O.) No. 6 fuel oil emulsions. concentration. Separation of oil and water at this temperature is achieved in 1 h or less a t demulsifier concentrations as low as 0.1%. Some mild acceleration in demulsification rate occurs above the 0.5% level but not enough, perhaps, to justify the economics of such an increase. A t 1 "C (Figure 9), the Prudhoe Bay emulsion sample appears most resistant to demulsification. Periods of several weeks are required to separate the oil and seawater a t concentrations as high as 0.7% demulsifier. The Arabian Heavy raw crude emulsion sample may still be broken a t this temperature well within a 24-h period. The evaporated more viscous emulsions-with the exception of the Ivory Coast sample-represent the most difficult case for the chemical demulsifier technique. Figure 10 reports the demulsification curves a t 1"C for emulsions of Prudhoe Bay and Arabian Heavy together with the residual-type oils, No. 6 fuel oil and Navy Special Fuel Oil (NSFO). When one works with such evaporated crudes and residual oils, it is extremely difficult to mix the demulsifier with the emulsion since viscosities of several hundred thousand or more centistokes are involved. For the evaporated highly viscous crudes and the residual oil data, it was necessary to double the mixing time to properly disperse the demulsifier through the thick chocolate-mousse-like mass; otherwise, separation would not take place. At such low temperatures, however, emulsions of residual-type oils and probably some aged, weathered, highly viscous crude emulsions will give responses similar to the upper two curves of Figure 10. Such emulsions are thoroughly intractable a t such low temperatures and cannot be broken a t ambient conditions. Both heat and mixing energy are required in substantial amounts, in addition to demulsifier, to bring about phase separation. Interestingly, the Arabian Heavy and Prudhoe Bay emulsion samples could be demulsified if the 30-s ultrasonic mixing period was doubled to 1min; otherwise, phase separation would not take place. At 34 "C (Figure 11), all of the emulsion samples could be easily broken in several hours by using 0.3-0.4% demulsifier. Both the NSFO and No. 6 fuel oil have reversed their position with respect to the evaporated crudes a t this higher temperature. In short, the chemical demulsification of a large variety of raw or evaporated crudes is achievable under ambient conditions provided temperature conditions are not too severe. Chemical demulsification will proceed easily under ambient tropic conditions. Under Arctic conditions, however, chemical demulsification will become much more difficult with additional heat and mixing energy sometimes being required to bring about phase separation.

1190

Environmental Science & Technology

I

1

I

0.3 0.4 0.5 PERCENT DEMULSIFIER

1

I

0.6

0.7

Figure 11. Demulsification of evaporated crude emulsions at 34 "C: (5') Arabian heavy (evaporated);(6') Prudhoe Bay (evaporated);(NSFO) Navy Special Fuel Oil; (#6) No. 6 fuel oil.

Conclusions (1)Crude oils exhibit a 5- to nearly 15-fold increase in viscosity (at room temperature) upon evaporative loss of their volatile content. (2) In spite of differences in viscosity and origin, five of six crudes studied had similar evaporation rates. (3) After 3 h essentially 95% of the volatiles have evaporated from 4-mm layers of oil (25 "C and 5-mph air velocity). (4) The evaporation of volatiles from crude oil emulsions is not significantly different from their evaporation from the raw crudes. (5) Emulsions generated from less viscous crudes are less stable and more easily broken than those from more viscous crude emulsions. (6) Chemical demulsification a t low temperature. Le., near 0 "C, may be arrested by the high viscosity of the oil phase. (7) Aerosol OT is an effective demulsifier for seawater-in-crude oil emulsions when used within the appropriate concentration range for a given crude oil emulsion. (8) Temperature and mixing energy (to disperse demulsifier and promote droplet collisions) greatly aid the demulsification process. (9) Demulsifier concentrations in the range of 0.01-0.1% are useful for less viscous crude emulsions; concentrations of 0.1-1% may be needed for stubborn highly viscous crude emulsions.

Acknowledgment

I gratefully acknowledge constructive aid and support given to me by Joseph Sholander of the Naval Surface Weapons Laboratory, Dahlgren, VA. Literature Cited (1) Sholander, J. E., Proposal for Oil/Water Separator Program submitted to Naval Sea Systems Command, Arlington, VA 20360, 1977. (2) Bolster, R. N.; Little, R. C. Enuiron. Int. 1980,3, 163. (3) Lawrence, A. S. C. Chem. Ind. (London) 1948,615. (4) Lawrence, A. S. C.; Killner, W. J. J . Inst. Pet. 1948,34, 821. (5) Mackay, D.; Leinoinen, P. J. Technical Review Report EPS-3EC-77-19; Minister of Supply and Services, Canada, 1977. (6) Hildebrand. 3. H.: Scott. R. L. "The Solubilitv of Nonelectrolytes", 3rd ed.; Reinhold: New York, 1950; Chapter 12, p 202. (7) Walther, C., Pet. 2. 1930,26, 755. (8) Richardson. E. G. Kolloid-2. 1933.65. 32. (9) Sutton, 0. G. "Micrometeorology;'; McGraw-Hill: New York, 1953.

Received January 15,1980. Accepted June 8,1981. This work was supported by the Naual Sea Systems Command.