Experimental xylene and cumene spill at sea

(1)

Simo Salo

216 Experimental xylene and cumene spill at sea

. . . ■ • . .

ⁱ

• . . . s . . .

(2)

Simo Salo

Experimental xylene and cumene spill at sea

Helsinki 2000

SUOMEN YMIPÄRISTÖKESKUS

(3)

ISBN 952- I I -0860-6 ISSN 1455-0792 Painopaikka: Oy Edita Ab

Helsinki 2001

(4)

The author thanks the Nordic Council of Ministers and its Steering Group for Research and Developement on Combatting Chemical Accidents for financial assistance. The drift model was operated by Jorma Koponen, who also commented on the manuscript, and Hannu Ylinen from the Environmental Impact Assesment Center of Finland. Furthermore the author is also thankful to the crew of RIV Muikku and to the scientists and experts participating in the experiment. Michael Bailey revised the English.

(6)

1 Introduction

A wide rar- ?e of chemical compounds are carried in tankers by sea. Despite all efforts to ensure secure transport, «.e risk of an accident still exists. If an accident occurs in which chemical compounds are spilled in the sea, it is very important to know how the spill will behave. The behaviour of spilled chemicals has been widely studied in laboratory conditions and by theoretical means. On the basis of these studies, empirical equations describing evaporation, dissolution, dispersion, diffusion and transportation of chemicals in marine conditions have been developed and reported in the international scientific literature. However, only a few studies have been carried out in which the behaviour of chemicals has been monitored under observed marine conditions (e.g. Kantin et al., 1990 and Merlin 1991). In order to obtain more information on chemical spill behaviour and to generate experience for responding to, monitoring and understanding chemical spills an experiment was organised on August 17, 1997 in the Gulf of Finland.

The results of the experiment were also used to test an operative chemical model that has been under development in the Finnish Environment Institute (FEI, Vepsä, et al. 1993). The equations in the model describing physical and chemical processes were collected from the literature (Salo 1992).

2 Materials and methods

The experiment area is located about 3 km to the south of Kalbådan off Porkkala (Fig. 2). The coordinates of the spill release sites were 59°50.651' and 24017.648' (xylene) and 59° 50.530' and 24°17.953 (cumene).

The chemical compounds xylene and cumene were used in the experiment. Their molecular structure is aromatic and therefore they can be detected by fluorescence spectrometry. They are also both transported in widely in the Baltic Sea. Both chemicals are floaters and evaporators. They are only slightly water soluble (Table 1).

Table 1. Physical properties of substances used in the chemical spreading experiment.

Property Xylene (dimethyl benzene, Cumene (isopropyl benzene) mixture of isomers)

CAS-nr 1330-20-7 98-82-8

Molecular weight (g mol-`) 106 120

Boiling point (°C) 137 145

Melting point (°C) -49 -97

Vapour pressure (Pa) 900 (20 °C) 613 (25 °C)

Water solubility (g dm 3⁾ 0.2 0.78 (25 °C)

Density (g cm 3) 0.87 0.86

Both chemical compounds are flammable. They are harmful when inhaled or when contacted with skin or eyes, and therefore full protective clothing and a face respirator equipped with an organic gas

(7)

0 cartridge were used. Both substances are biodegradable.

Both chemical was packed in a polyethylene sack of their own. From the sack the chemical was then spilled in a single action on the water surface. First 45 dm 3 of xylene was spilled into the sea. In the second phase 20 dm3 of cumene was spilled. Xylene was dyed by 1.7 g dm 3 fuchsine (CAS-nr 632-99-5) and cumene was undyed.

The location of the slick was determined with a differential GPS from a working boat following the slick. The xylene slick was also detected by side looking airborne radar (SLAR), ultraviolet- (UV) and infrared-scanners (IR) mounted in the DORNIER 228 aircraft of The Finnish Frontier Guard flying at a height of approximately 200m. Some test measurements were also made with a portable Personal Spectrometer II. This 512 channel spectrometer operates in the wavelength range of 325 - 1052 nm.

The spectrometer was produced by Analytical Spectral Devices (ASD), Boulder, CO, USA.

The concentration of chemical in water was measured in situ by flow-through spectrofluorimetry (Fluo-Imager, produced by laser Diagnostic Instruments, Tallinn Estonia). The spectral range of exication was 240 - 360 nm and the spectral range of registration was 250 - 570 nm. The spectrometer was assembled in the working boat. Three water samples were taken for gas chromatography (GC) analysis.

The experiment was simulated with a chemical model developed in FEI in order to obtain a rough approximation of the slick persistence time and its width. Forecasted weather condotions were used for the simulations. During the experiment the drifting was simulated by 3-D flow and a transport model.

After the experiment evaporation and dissolution were modelled using a chemical model and ambient conditions (Vepsä et al. 1993).

A new concentrating sampling method and a combatting method based on polymer fibre were tested (Kuuselal999).

The wind velocity and direction and air temperature were measured using the weather station of RIV Muikku, which was also used as a support ship. The water flow and temperature were determined using the vessel's equipment.

3 ^Results

3.1 Environmental conditions

On August 16 there was very strong wind from the north. On the experiment day August 17 the wind calmed down. Despite the calm weather, there was very strong flow of about 30 cm s-' to the north.

During the first phase of the experiment the wind was about 0.1-2.7 m

s'

from the southwest. During the second phase it was again stronger at 0.8-5.0 m s', still from the southwest. Air temperature was 14.5 °C during the first phase and later 16.2 and the air pressure was 101.9 kPa. The surface water temperature was 16.2 °C during the xylene experiment and 16.5 °C during the cumene experiment.

(8)

å

3.2 Detection of the surface slicks

Both of the chemical slicks were visible when observed from above. From the working boat the detection distance from the slick was less than some ten meters, depending on the mutual location of sun, slick and the detection point, otherwise the slick was invisible. It was obvious that the dye separated very rapidly from the xylene solution and crystallized on the water surface. The reason was that Fuchsine was not soluble in xylene but had first been dissolved into ethanol and then mixed with xylene. However, the xylene slick was visible all the time.

The slick of xylene was not detectable by side SCAN radar, but in the IR-image it was seen as a dark blot (Figs. 1 A and 1C) and in the UV- image as a light blot (Fig.1B).

The odour of both chemicals could still be recognised even after the slick had disappeared. This was due to evaporation of dissolved chemical.

va

Ve ~:

..a .x.:r.^.A .:..

.... .,ter ....:...:.::.°.: :,. .:~. _° :7 =

na

`6s1 n F.

F r etY

ail åq= '<E Fi:

Y:~vei:å VFe>

F,e e~i ~;4;, ^x'e.~'°

p°: °1a

iå:

sy ^vä~~

r.f '

n#}q

2

.>

L

„ba: å~'

i.

i r

::e zii<•

" _

;.: ..v',.= s..'~.F iii'.

sR~ <iac _'_iPi' ~£F =t.,` tl ~~i'xzsx:iå's•5,p ,°.~.s

Ea<

Figure 1. IR-scanner image of the xylene slick 10 minutes after the spill (A), UV-scanner image of the xylene slick 13 minutes after the spill (B) and IR-scanner image from xylene slick 13 minutes after the spill.

The aim of measurements by the Personal Spectrometer II was to test whether an imaging spectrometer (e.g. AISA, Airborne Imaging Spectrometer for Applications) could be an acceptable tool to observe chemical spills. The total reflectances of both slicks in the wavelengths of visible light were 1.5-2 times greater than the reflectance of water. The examination of the ratio of reflectance in 400 nm and 580 nm can also be a possible means of detection. The measurement results showed that cumene and xylene slicks could be observed e.g. by AISA. The relative thicknesses of the slicks could also be

(9)

13

detected but the identification of chemical compounds in field conditions was impossible (Pyhälahti 1997).

The locations of the working boat following the surface slicks of the chemical compounds during the experiment are presented in Figure 2.The spots in the figure describe one side of the slick and a slight deviation from the straight line in the beginning of the experiment may arise from the fact that the boat was located on different sides of the slick when the location was determined.

3.3 Spreading of chemicals

Both of the chemicals spread very rapidly after release. The slicks were at first uniform but seen later became dispersed. The slicks of both chemicals also divided later into several separate slicks, which drifted slightly differently. The slick of cumene had a somewhat greater tendency to this non- uniformity than the slick of xylene, which may also have been due to the slightly stronger wind. The driftings of the surface slicks are presented in Figure 2. The scattering of the location points in the cumene drifting figure towards the end of the experiment could be due to the existence of separate cumene slicks.

y~

EGs Slick at 15:47

Wind direction Xylene spill at 12:24

Cumene spill at 14:48

0 Xylene observed A Xylene modelled o Cumene observed

Cumene modelled

0 250 500 Meters

9

Figure 2. Location and drifting of the chemical surface slicks

The slick area was detected only for xylene 10 and 13 minutes after the release. The detection was made using UV- and IR-scanners assembled in the aircraft (Fig. 1). The slick areas were after 10 and 13 minutes about 600 m2 and 700 m2, respectively.

(10)

3.4 Physical and chemical analyses

Xylene dissolved and evaporated so that after 35 minutes the slick had disappeared completely. The cumene slick disappeared within 1 hour.

The concentration was measured using a flow-through spectrofluorometer. In the device the 2D- fluorescence spectrum was shown on the PC screen after the sample measurement. The concentration was calculated using pre-defined calibration.

Xylene fluorescence was detected only in the 20 cm top layer; in the deeper layers the detection limit was not exceeded. The detected concentrations were between 3 and 4 mg 1-' throughout the period when the surface slick could be detected (Fig 3A). When the surface slick disappeared the concentration rapidly decreased below the detection limit. The detection limit in ambient conditions using predefined calibration was not identified, but it was probaply close to 3 mg 1-`. The same concentration range was detected using GC-analysis.

The cumene spectrum could also be detected in the 30 cm and 50 cm water layers. However, the cumene fluorescence intensities in all water layers were very low when compared the signal to noise level. Therefore the concentration calculations were uncertain and fluorescence is presented only in arbitrary units (Fig. 3B). The measured fluorescence corresponded to about 2 - 3 mg 1-'. 10 Minutes after the release, concentrations of about 2 mg 1-` were measured in the 20 cm top layer with GC. 55 Minutes after the release on a site where the cumene surface slick had already disappeared the concentration was about 0.2 mg 1"1 using GC.

A B

37 120

—6 0-10cm

X 3.6 _X100 ■-10-20 cm

E 3.5 ^I ⁱ—~ 30 - 50 cm

34 80 ^{i i}Spill

i ----Newslick

... End f

0 3.3 60 ⁱ ^i_

ö 32 . 40

V 3.1 X 10 - 20 cm 20

—Spill i

3.0 0

12:14 12:28 12:43 12:57 13:12 14:38 14:52 15:07 15:21 15:36 15:50 16:04

Time 'lime

Figure 3. Measured xylene concentrations in the 10 - 20 cm water top layer (A). Measured cumene fluorescence in different water layers. Vertical lines labeled "new slick" mean that the original slick disappeared and a new, separately drifting slick was taken under observation (B).

(11)

10

3.5 Modelling physical processes

Evaporation and dissolution of xylene and cumene were modelled using the chemical model developed in the Finnish Environment Institute. The chemical process module was used by itself and the drifting of chemical slick was modelled separately.

It is assumed in the model that the chemical spreads on the water surface immediately after release and that the thickness of the slick remains constant thereafter. The thickness of surface slick is an input parameter. For the experiment it was determined in the laboratory by dropping 2 ml xylene and cumene on a sea water surface. The thickness of the slick was calculated on the basis of its known volume and observed surface area. It is obvious that the thickness calculated in this way is only a crude approximation of the true thickness in field conditions.

When the chemical module was used by itself without drifting and spreading calculation the water volume in which the chemical would be dissolved was considered to have the surface area of the original spill area and a depth of 0.4 m. These dimensions were assumed to remain constant during the simulation period.

In the laboratory the measured surface slick thicknesses for xylene and cumene were 0.06 cm and 0.1 cm, respectively. The average surface slick thickness was 0.008 cm 10 minutes after chemical release when determined from an airborne IR-image of the xylene slick. The weakness of this method was the dispersed nature of the slick because an undefined portion of the total slick area consisted of pure water surface. Thus the true slick area remained unknown and the defined thickness was too thin.

However, the model was run with both thicknesses.

Using a slick thickness for xylene of 0.06 cm the slick area was 75 m2, whereas a thickness of 0.008 cm gave an area of 600 m2. In the first case the model underestimated the disappearence of the slick and in the latter case the model overestimated it (Fig. 4A).

When cumene was modelled a slick thickness of 0.1 cm gave 20 m2 slick area and the thickness of 0.008 cm gave 250 m2. The model calculated a somewhat longer disappearance time for cumene than for xylene. This was in agreement with observations. As in the case of xylene the thinner slick disappeared faster than observed and the thicker slower (Fig. 4B).

Evaporation was the main process responsible for diminishing the surface slick. On the basis of the model calculations 99.97 % of the xylene slick evaporated and 0.03 % dissolved. 99.82 % of the cumene slick evaporated and 0.18 % dissolved.

(12)

11

A Residual Mass in the Surface Slick

40

35 Slick thickness 0.06 cm

_30

.'25 ... Slisk thickness 0.008 cm

W 20 A. Observed time of slick

15 disappearence

10 5 0

0 20 40 60 80 100 120 140

Time (rnin)

Area of the Surface Slick

600 Slick thickness 0.06 cm

c

⁴⁰⁰ --- Slick thickness 0.008cm

! Observed time of slick disappearence

.200 0

0 20 40 60 80 100 120 140 Time (min)

B Residual Massin the Surface Slick

Slick thickness

15 ^'. 0.1 cm

Y10 , Slick thickness

cn 0.008 cm

cn

A Observed time

5 of slick

s

disappearence

0

0 20 40 60 80 100 120 140 Time (min)

Area of the Surface Slick

250 Slick thickness

200 ^, ^{0.1 cm}

E 150 ' ... •... Slick thickmess 0.008 cm

Q 100 Observed time

50 ^{of slick}

disappearence 0

0 20 40 60 80 100 120 140 Time (min)

Figure 4. Modelled shrinking of the chemical surface slick assuming two different slick thicknesses (see text) and the observed disappearence of the slick: A) xylene and B) cumene.

Dissolution of chemical from the surface slick was modelled by calculating the concentration in the 0.4 m water layer below the slick. For xylene 0.06 cm and 0.008 cm slick thicknesses were again assumed, corresponding to 75 m2 and 560 m2 surface area of the slick, respectively (Fig 5A). For cumene the applied slick thicknesses were 0.1 cm and 0.008 cm, corresponding to slick areas of 20 m2 and 247 m2, respectively (Fig. 5B). Mixing of chemical with water layers deeper than 0.4 m was not applied but the evaporation of dissolved chemical from the water phase was included in the model.

0.20 3.5

`? -a B 0.16 0.18• A ,r _,. •_._,, '- 3.0 B 0.14

0.12 •

r''

o 2.5

E 2.0 ^'• ---Slick thickness

ö ^0.10 ... Slick thicness r _r _0.1cm

0.08 0.06 cm 1.5

c

^0.06 Slick thickness c 1.0 • Slick thickness

004 0.008 cm 0.008 cm

V 0.02 ö ^0.5

0.00

TTTT11

^V ^0.0

0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140

Time (min) Time (min)

Figure 5. Modelled xylene (A) and cumene (B) concentrations under the surface slick in the 0.4 m water layer using alternative slick thicknesses.

(13)

12

3.6 Drift model

Strong north winds during the night and morning preceeding the experiment caused a strong flow to the southwest in the surface layer of the model. When the wind calmed down the flow became slower and the direction turned to the east. The simulated drifting direction for the xylene experiment corresponded reasonably well with the detected directions but the flow speed was only about half that actually detected. During the cumene experiment the model overestimated the turning of the flow and also underestimated the speed in the same manner as in the case of the xylene experiment (Fig. 2).

One reason for the errors in flow speed could be the earlier strong wind from the north, which transferred warm surface water away from the shore and caused the thermocline to tilt. The returning thermocline could magnify the return flow towards the shore. The model included only the water surface tilting, and because of this underestimated the currents (Ylinen 1997).

4 Discussion

The experiment was organised in order to obtain knowledge of the behaviour of chemical substances in the sea and thus to facilitate chemical combatting. To avoid harmful environmental impacts the volume of the spilled chemical was minimised. The experiment showed that in calm conditions 20 dm3 of chemicals was sufficient when a work boat followed close to the slicks throughout the experiment.

The slicks were visible from the boat all the time. However, separate portions of the divided slick were difficult detect because the angle of view to the water surface became so narrow that reflection from the water and from the slick were almost identical. For better monitoring of the dimensions and location of slick an aircraft could be used. In addition to visible wavelenghts infrared and ultraviolet ranges can also be used for xylene and cumene detection.

The initial aim was to measure the concentration profiles of the dissolved chemical in the water layers under the surface slick. However, the detection limit of the spectrofluorimeter was low enough only for detecting the concentration immediately under the slick. Possibly a better optimization of the spectrometer could decrease the detection limit so that the concentration in the deeper water layers could also be measured. On the basis of the results obtained, the highest concentration in trial conditions existed in a rather thin layer under the surface. In the case of xylene the layer was about 20 cm and in the case of cumene about 40 cm.

The advantage of the spectrofluorimeter which was used in the trial was that the analyses could be made in situ. An advantage was also that the instrument was movable and it could be mounted in the working boat. The use of a small-sized boat allowed manovering around the spill area without disturbing the spill too much. A disadvantage in addition to the too high detection limit for experimental purposes was the limited range of analysable chemical compounds. If fluorescence is used, a conjugated double bond system must exist in the molecular structure of the chemical.

In modelling of evaporation and dissolution, the relation of slick thickness to slick surface area was very critical. In the model the slick thickness was a given parameter and the surface area was then calculated based on the spill volume and slick thickness. The crude estimate for slick thickness had been estimated from laboratory experiments but it was not very realistic because of the scale differences. The slick surface area was determined from airborne images 10 minutes after release.

These images were used to calculate the average slick thickness. The slick thickness estimated from airborne images is underestimated because of the diffuse structure of the slick, whereas the laboratory experiment can overestimate it. This uncertainty could also be seen in the simulation results. When the

(14)

approximation for slick thickness was made from the airborne image the modelled disappearence time of the slick was shorter than the observed time, whereas the time was longer than the observed time when the laboratory experiment was used to estimate the slick thickness.

The transportation of dissolved chemical was not simulated in the model, but the dissolution was assumed to occur in a water volume with a constant depth of 0.4 m and the area of the slick surface area in the beginning of the spill. Due to the lack of diffusion of dissolved chemical to the deeper water layers, the simulated concentration did not decrease after the slick disappeared. The dissolution water volume determined above is obviously unrealistic and explains the wide difference between the simulated and observed xylene concentrations under the surface slick.

5 References

Kantin, R., Merlin, F. and Peigne, G. 1990. Experimental Study of Chemical Behavior at Sea: Pollutmar I and Pollutmar II Trials. Proceedings of the seventh Technical Seminar on Chemical Spills. June 4-5, Edmonton, Albertta: 47-54

Kuusela, T. 1999. Personal communication.

Merlin, F. 1991. Experimental Study of Chemical Behavior at Sea: Pollutmar III Sea Trials. Proceedings of the Eighth Technical Seminar on Chemical Spills. June 10-11,1991, Vancouver: 47-60

Salo, S. 1992. The Fate of Chemicals Spilled on Water. A Literature Review of Physical and Chemical Processes. Publications of Water and the Environment, Administration - series A 91. National Board of Waters and the Environment, Helsinki 1992, 117 pages.

Pyhälahti, T. 1997. Personal communication.

Vepsä, H., Koponen, J. and Salo, S 1993. Finnish operational model system for oil- and chemical accident and sea rescue. Aqua Fennica 23, 251-258

Ylinen, H. 1997. Personal communication.

(15)

14

DOCUMENTATION PAGE

Published by

Finnish Environment Institute Author(s)

Simo Salo

Date of publication February 2001

Title of publication

Experimental xylene and cumene spill at sea

Type of publication Commissioned by

Report

Parts of publication

Abstract

An experimental chemical spill was organised in the Gulf of Finland in Baltic Sea. In experiment 45 dm3 xylene and 20 dm 3 cumene were released separately to the water surface. The drifting and disappearing of the chemical slick was monitored by visual observations, side looking airborne radar (SLAR), ultraviolet- and infrared-scanners and flow-through spectrofluorimetry. The results of the experiment were also used to verify a chemical fate model.

The xylene slick totally disappeared in 35 minutes and the cumene slick in one hour. Both of the slicks were detectable by visual observation and with IR- and UV-scanners. The chemicals were only slightly soluble in water and about 3 - 4 mg 1-1 concentrations were measured for xylene and 2 - 3 mg 1"' for cumene in the upper 20 cm layer.

Keywords

chemical spill, cumene, xylene, sea, slick, evaporation, dissolution, transportation, drift model

Other information

Series (key title and no.) ISBN ISSN

Finnish Environment Institute Mimeograph 216 952-11-0860-6 1455-0792

Pages Language Price Confidentiality

15 English Public

Distributed by

Finnish Environment Institute Custom Services

Tel.+358 9 4030 0100, Fax + 358 9 4030 0190 E-mail: neuvonta.syke@vyh.fi

Publisher

Finnish Environment Institute Impacts Research Division PO Box 140

FIN-00251 Helsinki

(16)

KUVAILULEHTI

Julkaisun päivämäärä Julkaisija

Suomen Ympäristökeskus Helmikuu 2001

Tekijät) (toimielimestä: nimi, puheenjohtaja, sihteeri)

Simo Salo

Julkaisun nimi (myös ruotsinkielinen)

Ksyleeni- ja kumeenivuotokoe merellä

Julkaisun laji Toimeksiantaja Toimielimen asettamispvm

Moniste

Julkaisun osat

Tiivistelmä

Suomenlahdella Porkkalan edustallajärjestettiin kemikaalivuotokoe. Kokeessa valutettiin 45 dm3 ksyleeniä ja 20 dm 3 kumeenia veden pinnlle. Kemikaalikalvojen kulkeutumistaja häviämistä seurattiin visuaalisesti, valvontalentokoneen sivukulmatutkalla (SLAB) sekä ultravioletti- ja infrapuna keilaimilla ja työveneestä käytetyllä läpivirtaus spektroflourometrillä. Kokeen tuloksia käytettiin myös kemikaalien haihtumista ja liukenimista kuvaavan mallin varmentamiseen.

Ksyleenikalvo katosi kokonaisuudessaan 35 minuutin kuluessa. Kumeenikalvon häviämiseen kului yksi tunti. Molempien kemikaalien kalvot veden pinnalla olivat silmin havaittavissa. Läikät näkyivät myös infrapuna- ja ultraviolettikeilaimien kuvissa. Kemikaalit olivat veteen niukkaliukoisiaja n. 20 cm pintakerroksessa mitattiin 3 - 4 mgl-1 ksyleenipitoisuus ja 2 - 3 mgl-' kumeenipitoisuus.

Asiasanat (avainsanat)

kemikaalivuoto, kumeeni, ksyleeni, meri, kemikaalikalvo, haihtuminen, liukeneminen, kulkeutuminen, kulkeutumismalli

Muut tiedot

Sarjan nimija numero ISBN ISSN

Suomen Ympäristökeskuksen Moniste 216 952-11-0860-6 1455-0792

Kokonaissivumäärä Kieli Hinta Luottamuksellisuus

15 Englanti julkinen

Jakaja

Suomen ympäristökeskus, asiakaspalvelu,

puh.(09)4030 0100, faksi (09)4030 0190 sähköpostiosoite neuvonta.syke@vyh.fi

Kustantaja

Suomen Ympäristökeskus PL 140

00251 Helsinki

(17)

(18)

r r • • ■ • • ■ • • • • • • • • • • • • • • • • ■ • ■ ■ • • r •

Experimental xylene and cumene spill at sea

216

Experimental xylene and cumene spill at sea

. . . ■ • . .

• . . . s . . .

Experimental xylene and cumene spill at sea

Contents

1 Introduction

2 Materials and methods

3 Results

s'

2

i r

c

s

c

TTTT11

4 Discussion

5 References

3 ^Results