The Potential for Dispersing Bunker C (IFO-380) Fu

T. Lunel, A. Crosbie, L. Davies and R.P.J. Swannell

AEA Technology

National Environmental Technology Centre

Culham, Abingdon Oxfordshire, OX14 3DB, United Kingdom

e-mail: Tim.Lunel@aeat.co.uk

tel +44 (0)1235 463083

Abstract

The 1997 sea trials carried out by the National Environmental Technology Centre (NETCEN) for the UK’s Maritime and Coastguard Agency (MCA), Exxon Research and Engineering and Alaska Department of Environmental Conservation (ADEC) were a turning point in demonstrating that it may be possible to disperse some types of Bunker C (IFO-180, heavy fuel oils) using dispersants.

The results of follow up laboratory studies suggest that dispersants may be a potential response option for spills of Bunker C (IFO-380, heavy fuel oil) at water temperatures of 15°C or greater. The paper describes how the results from the WSL laboratory dispersant test are correlated with field data to estimate likely dispersibility. However, laboratory tests are not definitive and the results of these experiments need to be verified at sea in controlled experimental trials.

Among the dispersants tested Corexit 9500 proved to be the dispersant of choice for treating a Bunker C from the Milford Haven refinery at the likely operational dosages. Some other dispersants such as Dasic LTSW or Superdispersant 25 proved to be effective on the emulsified Bunker C at high dose rates, while other dispersants such as Agma DR379 and Dasic NS do not appear to be suitable products for treating emulsified IFO-380 spills on the basis of these laboratory tests on the Milford Haven HFO.

1.0Introduction

There is a higher frequency of Heavy Fuel Oil (HFO) spills than spills of crude oil around the world. Typically, around 90-95% of the HFO spilt at sea comes ashore. Currently, the response options to HFO spills are limited. One option is to do nothing and allow the oil to strand and subsequently to clean up the shoreline. However, this could result in ecological damage to resources and a costly shoreline clean-up operation. An alternative option is to mechanically recover the oil using booms and skimmers, but these can only be deployed in low sea states since they are only effective in fairly calm weather conditions. The use of booms and skimmers typically only recovers a small amount of the oil that has been spilled. Heavy Fuel Oil (or Bunker C as it is sometimes known) has previously been viewed as not dispersible.

The 1997 sea trials (Lewis et al, 1998, Lunel & Lewis 1999) carried out for MCA by the National Environmental Technology Centre (NETCEN) demonstrated that it may be possible to use a new dispersant, Corexit 9500, on some types of heavy fuel oils.The laboratory testing following the sea trial aimed to determine the capability of five dispersants (Agma Superconcentrate DR379, Corexit 9500, Dasic Slickgone LTSW, Dasic Slickgone NS and Superdispersant 25) to disperse fresh and emulsified Bunker C (IFO-380 heavy fuel oil). All the laboratory dispersibility tests were carried out using the Warren Spring Laboratory LR448 protocol (Morris and Martinelli 1983) at a temperature of 15°C.

2.0Correlating the WSL laboratory test to field data

It is important to be clear from the outset that the results from laboratory dispersant tests can not be extrapolated directly to give information on the dispersibility of an oil in the field. However, we propose that it is possible to use laboratory tests which have been correlated against field data to give an indication of which oils could be dispersible in the field. In any discussion using these numbers there is the implicit statement “it is essential to validate these results in the field”.

In principle many of the commonly used dispersant tests could be correlated against field data. The steps that should be followed in such a correlation process are as follows:

1.Field data: Generate or identify field data giving quantitative values for the

percentage of oil dispersed;

2.Define “successful dispersion”: For the field data set identify the quantitative

threshold indicating “successful dispersion”;

3.Comparison of test methods with the field data: Identify a laboratory test method

which provides dispersibility numbers which correlate with the field data.

In this paper, we outline the way in which this process has been carried through to propose a dispersibility threshold for the WSL test.

2.1Field data

Most field data on dispersant application provides only semi-quantitative information on how effective dispersants are relative to an untreated control slick. Dispersion is an on-going process, which is not uniform over time because of the random nature of wave breaking and turbulence. Therefore, it is not possible to determine the percentage of oil that is dispersed from measurements made after applying dispersants to a single patch of oil. As soon as the patch of oil is treated, the surface slick and the dispersed oil plume will rapidly evolve with time, making it impossible to make the replicate measurements needed to quantify the percentage of oil which has successfully dispersed.

The approach adopted by NETCEN to overcome this problem is to produce a continuous release of oil to generate a steady-state oil slick. This is illustrated in figure 1. The oil is released at a known and constant rate and treated with dispersant at a known ratio. Replicate measurements can be made at a fixed distance downstream. In a constant tidal stream the fixed distance represents a fixed time after release of the oil. Thus, for example figure 1 shows how repeated transects can be made to produce replicate measurements of the amount of oil dispersing 15 minutes after release. The experiment can be repeated for 15 minutes after treatment with and without dispersant and for different dispersant types.The amount of oil released on the sea surface per metre (e.g 900 g/m) is known from the steady state flow rate and the tidal flow rate. The mass of oil dispersed in the water 15 minutes after treatment can be determined by using fluorometry to measure the total mass of oil in a transect (figure 2). For the 4 contour plots illustrated in figure 2, the percentage of Medium Fuel Oil in each transect is:•OSR-5(303/900) x 100 = 34 %

•Slickgone NS(203/900) x 100 = 23 %

•1100X( 74/ 900) x 100 = 8 %

•Control( 20/ 900) x 100 = 2 %

By carrying out many replicate measurements for different oil-dispersant combinations over field trials carried out in 1993 and 1994 we were able to produce the following table of quantitative oil dispersion efficiencies under wind conditions of 6-10 m/s, 15 minutes after treatment.

T able 1 Summary of the field dispersant efficiency data 15 minutes after treatment

W ind speed (m/s)D ate O il-Dispersant% Dispersed

(mean)

S tandard

deviation

N umber of

replicates

107/9/93M FO20.7

723/8/94M FO4315 725/8/94F orties4236

623/8/94M FO-LA18348530 107/9/93M FO-1100X104

107/9/93M FO-Slickgone NS176

623/8/94M FO-Slickgone NS14820 625/8/94F orties-Slickgone NS17527 722/8/94M FO-Corexit 952720922 107/9/93M FO-OSR5307

I t is important to note that there is a large standard deviation in the results which reflects the random nature of wave breaking and turbulence. This large standard deviation emphasises the need to test conclusions drawn from laboratory tests in controlled field experiments.

2.2.Defining “successful dispersion” for the field data

Dispersants enhance the rate of a naturally occurring process and therefore a judgement must be made at what point the degree of enhancement constitutes a “success”. The aim of dispersant response is to remove oil from the sea surface, hence “success” in the field should be related to this measure. For the steady state slicks the quantitative measure of the percentage of oil dispersed was accompanied by a noticeable change in the persistence of the surface slick.For example, figure 3 shows the marked difference in the surface slick of Forties Blend crude oil (spilt at the Sea Empress) which is visible to the Infra Red (IR) sensor with and without dispersant treatment. The untreated slick persists for the full length of the IR trace. A significant volume of oil is still visible to the IR camera at the point where it is possible to see the sampling vessel at a point 15 minutes after release. By contrast, for the Forties Blend treated with the dispersant Dasic Slickgone NS as it is discharged from the ship with a “t-boom” arrangement (Lunel & Davies, 1996), the surface slick is no longer visible by IR 15 minutes after treatment. Therefore, the value of 17% ± 5% of the oil dispersed in the water column after 15 minutes can be related to the removal of all but the sheen and can therefore be defined as a successful dispersion.

Figure 4 illustrates that without treatment, the slick of MFO will be persistent on the sea surface and visible to IR some 90 minutes after release (the reason for the break in the slick in the first 10 minutes after release is because the discharge of oil has been halted in preparation for the next release). By contrast when treated with Corexit 9527, (bottom image in figure 4) the surface slick was barely visible to IR 15 minutes after treatment. Similarly after treatment with Slickgone NS (second image from the top of figure 4) the surface slick has been broken up into patches of thick sheen 15 minutes after treatment. Both of these treatments were successful at reducing the persistence of the surface slick. These “successful” dispersions can be related to dispersant efficiencies of 20% ± 9% and 14% ± 8 % respectively at 15 minutes after treatment.

These results lead to the conclusion that for these field experiments natural dispersion is associated with measured efficiencies of 4% ± 3%. By contrast, successful chemical dispersions which reduce the persistence of the surface slick are associated with measured efficiencies of 14% ± 8% or greater.

When the MFO was treated with the demulsifier (LA1834/Surdyne) the persistence of the surface slick was reduced but not to the same extent as for Corexit 9527 and Slickgone NS. At 15 minutes after treatment there was no significant visual difference between the MFO slick visible to IR for the MFO treated with

LA1834/Surdyne compared to the control MFO slick. There are indications after a period of 30-40 minutes that the persistence of the surface slick is reduced. For

LA1834/Surdyne and the dispersant 1100X we define the treatment as partially successful. The values of dispersibility measured in the water column 15 minutes after treatment associated with the partial success were 8% ± 5% for the demulsifier and 10% ± 4% for 1100X.

We acknowledge that this field data set does not represent the values which can be applied universally to all oil-dispersant combinations or to all weather conditions. The results do, however, represent the only available quantified data on dispersant efficiencies based on replicate measurements. Taking into account the statistically derived standard deviation from the mean, the measurements for “natural dispersion”, “reduced dispersibility” and “dispersible” fall into 3 different groupings even though there is some overlap in the dispersibility ranges. The field data grouped into these 3 categories are illustrated in figure 5.2.3. Correlation of WSL test method with field data

An international programme of laboratory dispersant testing was carried out for the oil dispersant combinations given in table 1. NETCEN selected the WSL test as providing the closest match to the dispersibility values for this field data set. However, it is equally valid to use the other laboratory test methods which have been calibrated against field data. As described in Lunel and Wood (1996) the attributes of the WSL test were:

•The WSL test generates values of dispersant efficiency which are closer in numerical value to field values with fresh oils than any

other existing test.

•The test has been shown to recreate the observed dispersant

efficiencies for both the oil types used in the field with very

different density MFO = 0.904 kg m-3; Forties Blend = 0.843 kg

m-3.

•The droplet size distribution generated by the WSL test closely

matches that observed in the field (figure 6).

•The inter-laboratory reproducibility of the WSL test has been

shown to be good (Nordvik et al. 1993).

•The test is simple to use and is cost effective to run because of the low investment needed for the equipment and the relatively short time needed

to perform the test.

The field and WSL laboratory test data are listed below in Table 2 and illustrated in figure 7.

T able 2 Statistical Correlation between field and WSL laboratory test values.

O il-Dispersant F ield

(average value)

W SL test (average value)

M FO-Corexit 95272039

D ispersible F orties-Slickgone NS1721

M FO-Slickgone NS1625

R educed

Dispersibility

M FO-LA1834811

N atural F orties41

D ispersion M FO32

C orrelation coefficient = 0.96

If two data sets are correlated then when the values in data set A increase in value they a mirrored by an increase in data set B. When they decrease in numerical value in data set A, they decrease in data set B. The sizes of these increases/ decreases are not necessarily the same numerical value. For example, data set A (1, 2, 3, 4) and data set B (9, 18, 27, 36) has a correlation factor of 1.0.A statistical correlation coefficient of 0.96 between the WSL laboratory test values and the field values is highly significant and implies there is only a 1 in 25 chance that these two datasets are not correlated (the correlation factor is not a factor that can be used to relate the numerical values obtained in the WSL test to the numerical values obtained in the field).

The dispersibility values in the WSL test associated with the oil-dispersant combinations which represent “Dispersible”, “Reduced Dispersibility” and “Natural Dispersion” are given in table 2. By inspection, on the basis of the only available field data set of replicated and quantified dispersant efficiency, we propose the following dispersibility thresholds for the WSL laboratory test:

•Natural dispersion: ≤ 5%

•Reduced dispersibility: > 5% and < 15%

•Dispersible: ≥ 15 %

These proposed thresholds have been derived on the basis of two oil types, with different densities (MFO = 0.904 kg m-3; Forties Blend = 0.843 kg m-3). For oil types outside the range of viscosity and density, field validation is still required. However, the thresholds for the WSL test provide what we believe is the best indication of likely dispersibility in the field currently available.

3.0Use of the Field Correlated WSL Laboratory test to assess the potential dispersibility of Bunker C

The test results quoted in this section have been performed using the WSL test.

3.1The effect of Bunker C composition on dispersant effectiveness

Three IFO-380 heavy fuel oils were selected for initial dispersibility testing. They were produced from different crude oils by different processing routes and are representative of some of the types of IFO-380 that could be encountered at spills. The crude oils tested were from the refineries at Milford Haven, (UK), Slagen, (Norway) and Sriracha, (Thailand).

All three IFO-380s were emulsified to 30% water content and the viscosities of the oils and emulsions were measured at 5°C, 10°C, 15°C and 22°C. Table 3 shows the viscosity (cP) of the oils and emulsions at a shear rate of 10s-1.

T able 3. Comparison of Milford Haven, Slagen and Thailand IFO-380 viscosity results

S ource of IFO-380W ater

C ontent

V iscosity of IFO-380 (cP) at 10s-1

T emperature

5°C10°C15°C22°C

M ilford Haven0% water49,11227,49910,1175,071 S lagen0% water53,42725,86810,13,011 T hailand0% water25,109-7,3493,758 M ilford Haven30% water144,47095,42034,07214,693 S lagen30% water107,55050,00625,3799,769 T hailand30% water66,281-21,4849,934

From Table 3 it may be seen that the three IFO-380s tested had different viscosities at a given temperature. Following emulsification to 30% water content,

the viscosity of all three oils increased considerably and once again there was quite a variation in the observed viscosity values with temperature. As expected, an

increase in temperature caused a decrease in the viscosity of both the oils and their emulsions.

A ll three oils met the specification for an IFO-380 but they were dispersible to differing degrees both before and after emulsification due to variation in the oil properties. An example of the effect on dispersant effectiveness of using different

IFO-380s in the WSL test is shown below. Table 4 gives the results of the tests performed at 15°C using Corexit 9500.

T able 4. Milford Haven, Slagen and Thailand emulsion dispersibilities at 15°C

D isp:

C orexit 9500

D ispersant Efficiency on IFO-380 with WSL test

T emperature 15°C

M ilford Haven

V iscosity 34,100 cP

S lagen

V iscosity 25,400 cP

T hailand

V iscosity 21,500 cP 30% Water30% Water30% Water

1:1025%--1:2523%40%24% 1:5017%34%21% 1:10021%23%10%Table 4 shows that the three different IFO-380 emulsions did not all disperse to the same extent at 15°C when treated with the same dispersant. The Slagen emulsion was more dispersible than either the Milford Haven or Thailand emulsions at 15°C even though it was not the lowest viscosity emulsion. The complete results of the test programme (detailed in Crosbie et al. 1999) details how the most dispersible out of the Milford haven, Slagen and Thai IFO-380s varied, depending upon the temperature at which the tests were carried out.

Whilst the three heavy fuel oils used in these tests all came within the required specification for IFO-380 heavy fuel oils, they did not exhibit the same level of dispersibility at different temperatures. Further those three IFO-380 fuel oils did not cover the full possible range of properties. Therefore, the results of this programme of work can not be extrapolated over the whole range of IFO-380 fuel oils.

3.2Effect of Temperature

As shown in Table 3, the viscosities of the IFO-380 heavy fuel oils and emulsions varied considerably with temperature. Decreasing the temperature caused an increase in viscosity for a given heavy fuel oil and emulsion which made it more difficult to disperse.

The very significant effects of temperature on the Milford Haven dispersibilities both before and after emulsification are shown graphically in Figures 8 and 9 respectively.

Based on the proposed 15% threshold, the potential to disperse the IFO-380s from the Milford Haven and Slagen refineries was significantly effected by temperature. The dispersibility values were above the 15% threshold using Corexit 9500 at both 22°C and 15°C. However, for the emulsified oil at 10°C, and the oil and emulsion at 5°C, the dispersibility value were below the proposed 15% threshold.

Unlike the other two oils, the unemulsified Thailand IFO-380 remained dispersible even at 5°C since the viscosity of the oil was only 25,000cP even at 5°C. However, it was not dispersible at 5°C following emulsification to 30% water content and a viscosity of 66,000cP at 5°C.

Through the effect on the viscosity of the IFO-380 oils and emulsions, the temperature has a significant effect on dispersant effectiveness. Thus, it is not possible to extrapolate from the results obtained at 15°C in this study to lower sea temperatures, where the dispersibility of IFO-380 will be reduced significantly.

3.3Dispersant effectiveness of 5 dispersants on Bunker C from Milford Haven Refinery

The five dispersants were tested on the Milford Haven IFO-380 oil and emulsion at 15°C. The results obtained at 15°C for the dispersants from the MCA stockpile have been taken as an indicator to assess dispersant effectiveness. However, as discussed in the previous sections, the results can not be directly applied to other IFO-380s or to temperatures different from 15°C.Tables 5 & 6 show the results of the tests at 15°C on the unemulsified and

30% water content emulsion Milford Haven IFO-380 respectively. Those tests

where the mean dispersant efficiency values are above the proposed 15% threshold

are shaded (the reasoning behind the 15% threshold is detailed in section 2):

T able 5. Dispersant effectiveness on unemulsified Milford Haven IFO-380 at 15°C

D ispersant efficiency on unemulsified Milford Haven IFO-380

V iscosity 10,117 cP @ 10-1 , Temperature 15°C

D ER C orexit 9500D asic LTSW D asic NS S uperdispers

ant 25 1:10----

1:2526%51%53%22%63% 1:5012%48%42%24%52% 1:1009%45%33%31%50% 1:200-41%-22%-

T able 6. Dispersant effectiveness on emulsified Milford Haven IFO-380 at 15°C.

D ispersant efficiency on 30% water content Milford Haven IFO-380

V iscosity 33,372 cP @ 10s-1 , Temperature 15°C

D ER A gma DR379C orexit 9500D asic LTSW D asic NS S uperdispers

ant 25 1:105%25%18%11%24% 1:251%23%10%5%16% 1:50-17%-2%4% 1:100-21%-3%-In common with other oil types, IFO-380 will rapidly emulsify when released

on the sea surface under most sea conditions. Even with a standby dispersant spray capability, it is likely to be at least 2 hours before a dispersant spray operation can commence. Therefore, it is the dispersibility results on the 30% water content IFO-

380 emulsion that provide the most appropriate indication of the likely success of a dispersant spray operation.

Following emulsification, the Milford Haven viscosity increased from

10,117cP to 33,372cP which substantially reduced the ability of the dispersants to

break the emulsion and disperse the oil. From a comparison of Tables 5 and 6 it may

be seen that emulsification of the Milford Haven IFO-380 to 30% water content

caused a significant reduction in dispersant effectiveness for all of the dispersants

tested in this study.There is an uncertainty surrounding oil thicknesses and the dosages that have previously been quoted from actual incidents. However, for illustration, it has been estimated in the Sea Empress that oil:dispersant ratios were in the range of 1:40 to 1:80 (SEEEC 1998; Lunel & Lewis 1999):

•Over a DER ratio of 1:25 to 1:50 , Corexit 9500 is the only dispersant which produces dispersibility values which exceed the proposed 15% threshold.

Reducing the DER to 1:50 or 1:100 did not appear to have a significant effect on the effectiveness of Corexit 9500 on this emulsion.

•Superdispersant 25 and Dasic LTSW produce dispersions which are at or just below the 15% threshold at ratios of 1:25. However, higher dispersibility values are obtained when the dosage is increased to 1:10.

•Dasic NS and Agma DR379 do not appear to be effective on an emulsion of IFO-380 from the Milford Haven refinery. Dasic NS may be effective on the

unemulsified HFO (table 5). However, the time window for such an application will be extremely short.

In summary, at the likely operational dosages it appears from these laboratory tests that Corexit 9500 may be effective at treating a Heavy Fuel Oil spill. It could be worth considering the use of Dasic LTSW or Superdispersant 25 on a Bunker C spill although it is likely that high dose rates would be required. The results suggest that there would be little merit in using Agma DR379 or Dasic NS on a spill of IFO-380 fuel oil with similar properties to the Milford Haven sample used in this study.

4Recommendations

The laboratory testing detailed in this report indicates that emulsions of IFO-380 may be dispersible at sea temperatures of 15°C or above. These results have also been confirmed in a mesoscale flume test on the dispersibility of a 24,000 cP emulsion of IFO-380 from the Slagen refinery. This larger scale laboratory test recently carried out by SINTEF also indicated that at 15°C an HFO slick may be dispersible using Corexit 9500 (Fiocco et al, AMOP 1999).

All of the results presented in this report and the larger mesoscale test were generated in laboratory studies. Therefore, whilst it is believed that the studies provide an indication of the possibility of dispersing heavy fuel oils under certain conditions, it must be remembered that the laboratory studies only provide an approximate simulation of the dispersion processes that occur at sea.

To confirm the preliminary findings in this report that dispersants might provide a potential response to IFO-380 spills, a field trial is required.

Using the steady state release method described in section 2, would be an ideal method of validating the findings of this study in the field. Firstly, the volumes used would be less than 5 tonnes, considerably smaller than the alternative 20 tonne releases. Secondly, the data set would provide correlation data for the laboratory tests with HFO. This would then give confidence that once we have generated the initial validation in the field, the laboratory tests could be used to extend the testing to the wide range of IFO-380 heavy fuel oils transported in UK waters.In preparing for a sea trial, additional information on the likely behaviour of the selected IFO-380 fuel oil at sea under a range of weather conditions would be essential. Weathering studies should be performed on the candidate IFO-380 fuel oils and the resultant oil residues tested against all five of the dispersants considered in this report. The data from these weathering studies could then be used as input to an oil weathering model such as OSIS which could then be used to predict the time taken for a given oil to reach a certain viscosity under a given set of weather conditions. This information could be used to establish the likely time windows for dispersant use on the IFO-380 under different weather conditions which can be validated in the field programme.

5Conclusions

•Dispersant treatment may be an appropriate response to a spill of IFO-380 (Bunker C) at sea temperatures of 15°C or more. The results to date are a

significant advance but are not definitive:

•The results must be verified in a controlled field trial.

•The results are based on one type of IFO-380 derived from the Milford Haven refinery and can not necessarily be extrapolated to other IFO-

380s.

•On the basis of the laboratory testing carried out for this study, Corexit 9500 appears to be the most likely to produce an effective dispersion when treating an HFO spill at operational dosages.

•Dasic Slickgone LTSW and Superdispersant 25 may also be effective on the IFO-380 emulsion, particularly at 15°C at a DER of 1:10.

•Agma DR379 and Dasic NS do not appear to be suitable products for treating emulsified IFO-380 spills on the basis of these laboratory tests on the Milford

Haven HFO.

•In treating the IFO-380 emulsion, reducing the dispersant to oil ratio caused a noticeable reduction in dispersant effectiveness for all of the dispersants, except Corexit 9500.

•The trends observed in these experiments on the Milford Haven IFO-380 heavy fuel oil will not be the same for all other heavy fuel oils. Other heavy fuel oils may be dispersible to different extents to those tested in this programme of work.•Some of the five dispersants compared in this study may well be effective over a broader range of oil types than others. It should not be assumed that a dispersant that is effective on a fuel oil will necessarily be effective on all crude oils and emulsions, as a dispersant could be fuel oil specific.6References

Crosbie, A., Davies L. & Lunel T. The scope for dispersing heavy fuel oils. AEA Technology report AEAT-4475, February 1999.

Davies, L., Lewis, A., Lunel T and Crosbie A. Dispersion of emulsified oils at sea -Laboratory study. AEA Technology Report AEAT-4347, October 1998.

Lewis A., Crosbie A., Davies L. & Lunel T. Dispersion of emulsified oil at sea. AEA Technology report. AEAT- 3475, June 1998.

Lunel T. & Davies L. Dispersant effectiveness in the field on fresh oils and emulsions. Proceedings of the 19th Arctic and Marine Oil Spill Programme (AMOP) Technical Seminar, Environment Canada, Calgary, Alberta, 1355-1394, 1996.

Lunel T. & Wood P. A Laboratory dispersant effectiveness test which reflects dispersant efficiency in the field. Proceedings of the 19th Arctic and Marine Oil Spill Programme (AMOP), Technical Seminar, Environment Canada, Calgary, Alberta, 461-480, 1996.

Lunel T. & Lewis A. Optimisation of oil spill dispersant use. International Oil Spill Conference Proceedings, API, 1999.

Morris, P. & Martinelli F. A Specification for oil spill dispersants, LR448, 1983. Nordvik A. et al Inter-laboratory calibration testing of dispersant effectiveness – Phase 2 appendix. MSRC Technical Report 93-003.

SEEEC. 1998 The Environmental Impact of the Sea Empress Oil Spill. The Stationery Office, London, 135 pp.