Ecotoxocicological assessment of antifouling biocides and non-biocidal antifouling paints

Appendix 1: Model for calculation of exposure concentrations (PEC)

1. Introduction
2. Establishing af calculation model
2.1 Scenarios
2.2 Emissions of antifoulants
2.3 PEC model
2.4 Data on active substances
2.4.1 DCOI
2.4.2 Zine pyrithione (ZPT)
3. Calculation results
4. Sensitivity materials
4.1 Harbour scenarios
4.2 Sensitivity analysis of other parameters

1. Introduction

This appendix describes the exposure model which was used for calculating the exposure concentrations PEC (Predicted Environmental Concentration) for DCOI and zinc pyrithione (ZPT) and their main metabolites.

The established model applies modelling principles that are generally used at the determination of PEC. Numerous models for calculating exposure are described in literature, i.a.:

	SimpleBox, which is a "multi-compartment" model based on the fugacity principle (Fredenslund et al. 1995). This model is used in the Technical Guidance Document (TGD) for generic risk assessments of individual substances (EC 1996) and is thus incorporated in the edp model EUSES, which is an "electronic" edition of the TGD (European Chemicals Bureau 1997).
	Furthermore, EUSES includes a module for estimation of PEC for antifoulants.
	The CHARM model, which is used for risk assessments of offshore chemicals (Karman et al. 1996).
	EXAMS, which is an interactive computer program for simulation of the fate of environmentally hazardous substances in aquatic ecosystems (Burns et al. 1981).

2. Establishing a calculation model

In general, exposure assessments are composed of the following items:

	Setting up of scenarios describing the environmental parameters of importance to the emission and the fate of the substances
	Determination of the emission of chemicals
	Calculation of PEC in relevant sub-environments
	Sensitivity analysis in which the relative dependency of PEC of the parameters forming part of the standard scenarios and the relative dependency of PEC of the parameters of the substances are estimated.

These elements were also applied in the present exposure calculations.

2.1 Scenarios

Two standard scenarios have been set up for the calculation of exposure concentrations (PEC):

Both standard scenarios are thus placed in Roskilde Fjord.

Each scenario was characterized as regards:

Water exchange The net water exchange between Roskilde Fjord (inlet) and its mouth at Isefjord is assumed to correspond to the net water supply to the inlet, which is stated to be approx. 1.25 × 10-4 m3 × s-1 per metre of the length of the inlet (Harremoës and Malmgren-Hansen 1989). As Roskilde Fjord is approx. 38 km long (Harremoës and Malmgren-Hansen 1989), a total of approx. 410,400 m3 water is supplied a day. With a surface area of approx. 125 km2, this corresponds to a net water exchange of 0.003 m3/m2/day. The net water exchange is thus very low. DHI (1994) thus also indicates that the water level in Roskilde Fjord is primarily determined by wind conditions and tidal variations. The water level variations determine the currents and thus the water exchange of the inlet. The largest variations in water level are induced by the wind but the tide determines the regular minimum variations and thus which minimum water exchange occurs in a short view. At Hundested in the north, the normal range of the tide is approx. 20 cm, which corresponds to the daily variation in calm weather observed by the bridge guard at Kronprins Frederiks Bro near Frederikssund. At the bottom of the inlet near Roskilde, the tidal variations are as low as 6-7 cm (DHI 1994). Due to the wind, normally occurring variations over longer periods of time are, however, much larger. For the pleasure craft harbour of Jyllinge, it is primarily north-westerly winds that may cause up to 1 metre high tides and southeast and southerly winds that may cause 0.5-1.0 m low tides. Data on the water level (for the years 1996, 1997 and 1998) for a station at Værebro Å (stream), which falls into Roskilde Fjord a few kilometres north of the pleasure craft harbour, state a mean daily total change in the water level of approx. 0.6 m/day (0.6 m3/m2/day). The data on water levels were obtained from Ivar Thorstein Hansen, the County of Roskilde. This water level figure is considered to reflect the water exchange in the pleasure craft harbour of Jyllinge and is used in the model calculations. For the narrows at Frederikssund, the difference between daily minimum and maximum water depths (at times without high winds when the differences may be much more considerable) is stated by the bridge guard to be approx. 0.2 m. Furthermore, wind conditions will contribute to the water exchange. We have not succeeded in obtaining exact data on the water depth under the bridge, Kronprins Frederiks Bro, but the water exchange will probably as a minimum be at the same level as the water exchange in the pleasure craft harbour of Jyllinge and at Værebro. Therefore, a water exchange of approx. 0.6 m3/m2/day was assumed for the narrows at Frederikssund.

For both scenarios, the concentration of substance in the water transported into the waters in question is considered insignificant.

	Water depth.
	Composition and characterization of suspended matter. The suspended matter was characterized with respect to the contents of organic matter. Furthermore, the suspended matter was considered negatively charged.
	Salinity.
	Temperature. In the present study, the temperature is put at 12.5°C, corresponding to the mean air temperature from April till September (the sailing season is typically from the end of April to the beginning of October).
	pH.
	The number of m2 of bottom areas of ships that are in the waters per time unit.
	The percentage of the ships having bottom paint with the examined antifoulant (P).
	Water-holding capacity of the sediment. Apart from a minor content of organic carbon in the sediment, the composition of the upper sediment layer was assumed to be identical to the composition of the suspended matter. The sediment was assumed to be anaerobic.
	Number of boats in the waters in question. Danish Sailing Association (personal correspondence with Steen Wintlev, Danish Sailing Association) has passed on information on the capacity (number of sailing and motor boats) of the pleasure craft harbour of Jyllinge. Furthermore, Danish Sailing Association has estimated the average wet surface of the boats to be approx. 18 m2. Statistics on the passing-through of pleasure craft for 1995, 1996 and 1998 (Kronprins Frederiks Bro 1996, 1998, 1999) were used for determining the number of sailing and motor boats passing the bridge. Based on these statistics, the average daily number of passing-throughs in the peak season (May - October) was fixed at approx. 70 pleasure craft a day. The distribution of pleasure craft on sailing boats and motor boats and thus the average wet surface of the boats is assumed to be the same as that for the pleasure craft harbour of Jyllinge.

Average time which the centre of gravity of the boat stay in the waters: For the pleasure craft harbour, it was assumed that all the boats are in the harbour. Danish Sailing Association states that the berths are occupied from approx. mid-May till end-September. During summer holidays (1 July - 15 August), approx. a third of the boats are gone, and there are almost no visitors in the harbour as it is situated inconveniently to tourists on their way through Roskilde Fjord (personal correspondence with Steen Wintlev, Danish Sailing Association). Assuming that all berths are always occupied will thus overestimate the total leaching of antifoulants to the harbour. Furthermore, Danish Sailing Association (personal correspondence with Steen Wintlev, Danish Sailing Association) states that the passing-through at Frederikssund takes place at relatively low speed because of the narrow fairway and the large number of boats. When passing through, the sea speed is estimated to be approx. 3-4 knots, corresponding to 5.5-7.4 km/h. The time that the centre of gravity of a boat stays in the water column in question is thus approx. 5.6 × 10-6 – 7.5 × 10-6 days.

Table B1.1 summarizes the parameters characterizing the two scenarios. These parameters are applied in the basic calculations.

Table B1.1    Look here...
Standard scenario.

2.2 Emission of antifoulants

The rate, at which the antifoulant leaches into the aquatic environment, is expressed as follows:

U = [leached substance per area per time unit].

The measuring of realistic leaching rates of antifoulants is causing great problems as the leaching rate depends on various factors such as:

The time after painting. The leaching rate has often been demonstrated to decrease as a function of the time as a result of the falling concentration of the substances in the paint.

Thickness of the coat of paint

Liberation of other substances in the paint

Outward circumstances, i.e. whether the boat is sailing or not, currents/water exchange, temperature, etc. The leaching rate is typically higher when the boat is sailing than when it is in port.

A draft standard (ISO 1999) is available from which the leaching rate can be calculated. The leaching rate is determined on the basis of an estimate of the life of the paint, in which it is simply assumed that all of the antifoulant will be released throughout the life of the paint. The first two weeks after the boat has been painted, a higher leaching rate is anticipated. After two weeks, the leaching rate is considered to be constant. The standard does not take into account that the leaching rate is probably higher while sailing than when the boat is in port, and the standard will thus be likely to overestimate the leaching rate when the boat is in port and underestimate the leaching rate while sailing. Furthermore, the draft standard proposes typical thicknesses of coating and lives of different paints (ISO 1999). On the basis of the proposals of the draft standard on coat thicknesses and lives, the thickness of the coat worn down in six months (corresponding to a sailing season) can be calculated to be 42 mm (soluble matrix), 38 mm (insoluble matrix), 45 mm (tin-based self-polishing paint) and 50 mm (tin-free self-polishing paint). These coat thicknesses are in good agreement with the estimates that Hempel has made, stating an average worn down coat of paint of 42 mm per sailing season for pleasure craft in Denmark (Hempel 1999c). Hempel has based their calculation on the amount of bottom paint sold in the Danish market a year and the number of sailing/motor boats of more than 6 m (corresponding to those painted) and their average bottom area.

In order to simulate the increased leaching rate while sailing, it is assumed that 60 mm of the coat of paint is worn off at constant sailing for 6 months. In order to simulate the lower leaching rate when the boat is in port, it is assumed that 30 mm of the coat of paint is worn off when the boat is constantly in port for 6 months. The above corresponds to the assumption that the boats are sailing for approx. 2 months of a sailing season and are in port for the remaining 4 months and that the leaching rate while sailing is twice the rate when not sailing.

On the basis of confidential information from Hempel on the content of antifoulants, dry matter and density (Hempel 1999b), the average leaching rates for the two types of antifoulants can be calculated. Table B1.2 gives the results of these calculations.

In the model, the total leaching of the active substance to the water per time unit is expressed as:

2.3 PEC model

The model is divided into the following parts:

1. Mass balance in the water column
2. Mass balance in the sediment

Transformation in the water column
The following conditions in the water column were taken into consideration:

The degradation rate of the parent compound and the subsequent formation of metabolites. Aerobic conditions in the water column were assumed. The effect of the temperature on the degradation rate was included by assuming that the degradation rate is halved when the temperature falls by 10°C (or vice versa). During the modelling period, temperatures are not so low that degradation stops.

Abiotic transformation. Degradation of ZPT by photolysis is included. For the two investigated substances, hydrolysis is not considered a significant reaction (cf. Chapters 3 and 4). Like other reactions, degradation by photolysis is temperature dependent but the effect of the temperature is less than for other reactions. Schwarzenbach et al. (1993) thus states that a change in temperature of 10°C only changes the reaction rate by a factor of between 1.15 and 1.5. Therefore, the effect of the temperature on the degradation by photolysis is ignored in the present study.

A first order degradation kinetics was assumed for all degradation reactions.

No discrimination was made between dissolved substance and substance sorbed to dissolved organic matter (DOC).

Sorption to the suspended matter was expressed as a linear adsorption.

Linear sedimentation of the suspended matter was assumed.

Resuspension of sedimentated matter. In the calculations, the resuspension rate was assumed to be constant.

The following conditions in the sediment were taken into consideration:

Sedimentation of suspended matter from the upper water layer.

Resuspension of sediment to the upper water layer.

A first order anaerobic degradation. The sediment was assumed to be anaerobic for which reason only anaerobic degradation was included.

Only the sediment formed during the simulation period was examined. This sediment layer was considered to be a completely homogeneous mixture.

For both the parent compound and its main metabolites, a mass balance was established for the water column and the sediment.

The following three PECs (Predicted Environmental Concentration) were calculated:

	PEC(water column)
	PEC(sediment)
	PEC(sediment, pore water)

These three concentrations were put equal to the steady-state concentration, i.e. the concentration that the calculated concentrations eventually approach when simulating a continuous leaching of the substance to the aquatic environment.

For all substances, the background level is assumed to be 0.

2.4 Data on active substances

2.4.1 DCOI

Aerobic degradation
Figure B1.1 shows the simplified degradation pattern which was assumed for DCOI. At first, DCOI was assumed to degrade into N-(n-octyl) malonamic acid which, while releasing CO₂, was transformed into N-(n-octyl) acetamide and N-(n-octyl) b hydroxy propionamide. These two metabolites were assumed to be transformed into a large number of different organic compounds, which were comprised under "Other metabolites". The half-life of this pseudo-reaction was assumed to be same as those of N-(n-octyl) acetamide and N-(n-octyl) b hydroxy propionamide. To a certain degree, these metabolites will be mineralized while forming CO₂ (half-life of this transformation is assumed to be the same for all "other metabolites").

The half-life of the transformation of DCOI into N-(n-octyl) malonamic acid was determined on the basis of a test in which the removal of DCOI was measured in seawater from the pleasure craft harbour of Jyllinge for a period of 72 hours at 12°C (Jacobson and Kramer 1999). By minimizing the total of the areas of the relative residues of DCOI (RRSQ) stated by Jacobson and Kramer (1999), using the following equation:

Half-lives can be calculated to be 12.8 hours (for replicate 1) and 15.3 hours (for replicate 2) with an average half-life of 14.1 hours (at 12°C). The other half-lives were estimated on the basis of the quantities which were considered to be present after 30 days’ aerobic degradation (see Figure B1.1) in experiments carried out by Mazza (1993).

The estimated half-lives of the aerobic transformation of DCOI at 25°C are given in Table B1.3.

Table B1.3    Look here...
Estimated half-lives (days) at 25°C of aerobic degradation of DCOI.

Anaerobic degradation
The anaerobic degradation of DCOI was assumed to follow the same reaction pattern as the aerobic degradation. The degradation rates were, however, assumed to be slower for the anaerobic degradation.

The transformation of DCOI at aerobic and anaerobic test conditions is shown as a function of the time in Figure B1.2. Data from Table 3.2 of the main part of this report (aerobic conditions) and from Table 3.4 (anaerobic conditions; concentrations set at 100%). It should be noted that the time axis depicting the anaerobic tests is 4.5 times longer than the time axis of the aerobic tests. Figure B1.2 shows that there is thus a fair correlation between the measured concentrations of the aerobic and anaerobic tests, respectively. In the calculations, it was thus assumed that, under anaerobic conditions, the half-lives of the reactions are 4.5 times longer than under aerobic conditions.

Figure B1.2    Look here...
Degradation of DCOI under aerobic and anaerobic conditions.

Properties of the substance
Table B1.4 gives selected properties of DCOI and its metabolites. DCOI and the three metabolites were not assumed to be present in ionized form at pH = 7.

* Calculated by means of K_owWin (Syracuse Research Corporation 1996).
** Measured (data stated in the main report). Calculated by means of K_owWin (Syracuse Research Corporation 1996).
** Measured (data stated in the main report). * Calculated by means of K_owWin (Syracuse Research Corporation 1996).
** Measured (data stated in the main report).

2.4.2 Zinc pyrithione (ZPT)

Biodegradation
Figure B1.3 shows the simplified biological degradation paths of ZPT which were simulated in the exposure calculations. It is a very simplified model compared to the very complicated degradation pathways of ZPT.

The following abbreviations are used:

Zinc pyrithione Pyrithione Omadine disulfide Omadine sulfonic acid 2-Pyridine sulfonic acid

ZPT PT OMDS OMSo PSoA

Other heterocyclic metabolites with one ring are given as NP1-NP5 (cf. Chapter 4 of the main report). The identity of NP1-NP5 is known to VKI.

Two main degradation paths are assumed:

It was assumed that OMDS, NP3, PSoA and OMSo were further transformed into other compounds, which, to a minor degree, are mineralized.

The half-life of the primary reaction (ZPT ® PT^-® OMDS + NP3) was set at 0.5 days. The other half-lives were estimated on the basis of the quantities found in the aerobic and anaerobic degradation tests in which the concentrations of substance were measured as a function of the time (these tests are discussed in the main report).

The estimated half-lives of the aerobic and anaerobic degradation are given in Tables B1.5 and B1.6. Measured and calculated concentrations are depicted in Figure B1.4.

Photolysis
As mentioned above, the degradation of ZPT by photolysis is included in the model simulations. ZPT is assumed to be transformed into NP3 in the photolytical degradation.

In daylight about noon, the photolytical half-life of ZPT was estimated at:

	1.78 min without cloud cover
	3.74 min with cloud cover

The tests were made in September on the parallel approx. 42°N, where the degradation rate of ZPT in seawater was followed (Fenn 1999). The tests were made in curved tubes. A first order rate constant k_P can be estimated to be 0.18 min^-1 (without cloud cover) and 0.08 min^-1 (with cloud cover). A factor of 2.2 was used in order to correct for the curving of the tubes.

At a cloudless sky, it is assumed that the measured photolytical rate constant may be described by (cf. Schwarzenbach et al. 1993):

where
F	is the so-called quantum yield, which is here assumed to be independent of the wavelength [-]. F states the amount of molecules exited by the light that are transformed into another compound.
	is the specific rate of light absorption [time-1]. This rate was calculated by (Schwarzenbach et al. 1993):
l	is wavelength [nm].
W(l)	is the light intensity at various wavelengths [milliEinstein/cm2/time/nm]. The light intensity in the autumn on the parallel 40°N was borrowed from Zepp and Cline (1977).
D(l)	is the ratio of the average length of the trajectory of the light to the depth of a water element, which can be assumed to be completely mixed [-]. D(l ) is here as- sumed to be equal to 1 for all wavelengths.
e(l)	is the so-called molar extinction coefficient [(mol/l- 1/cm]. For ZPT, these values are borrowed from Fenn (1999).

The light deflection of the cuvette has been taken into consideration.

F was thus determined to be 0.07.

Then the American program GCSOLAR (U.S. EPA 1999) was used for calculating the photolytical half-life. This program uses the so-called attenuation coefficients, a¹, which are applied in order to calculate to which degree the water absorbs the light as a function of the depth.

The attenuation coefficient for the water at Kronprins Frederiks Bro is determined on the basis of measured Secchi disk transparencies at two stations in the vicinity of the bridge (Counties of Roskilde and Frederiksborg 1997). In the summer, the shortest Secchi disk transparency is approx. 2.5 m. By using data from Calkins (1977), the following correlation between the Secchi disk transparency and the attenuation coefficient a was found at Secchi disk transparencies of less than 4 m:

a [m^-1] = 3.05 - 0.57 × Secchi disk transparency [m].

The attenuation coefficient for the pleasure craft harbour of Jyllinge was found on the basis of literature data on coastal areas (Zafirioiu 1977). The values are here stated as a function of wavelength and with a minimum and a maximum value. The highest values are used in the present calculations.

GCSOLAR does not take the effect of the cloud cover on the photolysis rate into account. The American program EXAMS (Burns et al. 1981), which can also be used for simulating the photolysis of substance, does, however, take the effect of the cloud cover on the photolysis rate into account. By using EXAMS, it was estimated that, compared to the half-life at a blue sky, the half-life will be approx. 50% higher at a cloud cover of 60% which is the average cloud cover in Denmark from April to September (Danish Statistical Office 1996).

GCSOLAR can calculate the average photolytical half-life for each of the seasons: Spring, summer, autumn and winter and on various lattitudes (however, only lattitudes divisible by 10). The average photolytical half-life of ZPT for the seasons spring, summer and autumn and on the lattitudes 50° and 60° was found to be 9.8 hours for the pleasure craft harbour of Jyllinge (6.5 hours at blue sky) and 6.6 hours for the narrows at Kronprins Frederiks Bro (4.4 hours at blue sky).

The degradation by photolysis is slightly dependent on the temperature. A dependency of the photolytical degradation on temperature is, however, not included in the present calculations.

The photolytical half-life was determined for open waters where cloud cover and the falling light intensity down through the water column are taken into consideration but where the boats in the harbour and the shadow effects of the pier are not taken into account. Therefore, two types of calculations have been made, one in which the degradation by photolysis is taken into account and another in which the photolytical degradation is ignored. The actual conditions will probably be somewhere between these two reflections but it is not possible right away to quantify the importance of the shadow effect of the boats and the pier on the amount of light actually falling on the water surface. For the busy navigation route under the bridge (Kronprins Frederiks Bro), there will be limited admittance of sunlight right under the bridge while there will be no important shadow effects in the other part of the navigation route.

Properties of the substance
The properties of zinc pyrithione and its metabolites are summarized in Table B1.7.

The different heterocyclic metabolites with one ring have a low calculated log K_OW, which indicates a high water solubility. The sulfonic acids are also expected to be very strong acids for which reason they are probably fully ionized at the prevailing pH in the two waters. For these two compounds, the sorption to sediment and suspended matter is thus expected to be low. The calculated K_OC values are, however, used for estimating the sorption to suspended matter and to the sediment.

Figure B1.4a    Look here...
Calculated and measured concentrations of zinc pyrithione
and its metabolites. Aerobic experimental conditions.

Figure B1.4b   

Figure B1.4b    Look here...
Calculated and measured concentrations of zinc pyrithione
and its metabolites. Anaerobic experimental conditions.

3. Calculation results

Table B1.8 gives the calculated PEC values for the two scenarios and the various substances. As mentioned above, the concentrations are steady-state concentrations.

It appears from Table B1.8 that the highest calculated concentrations were found for the pleasure craft harbour. Here, the calculated concentrations are approx. 100 times higher than the concentrations calculated for the busy navigation route.

For the parent compounds, the following steady-state concentrations for the water phase, PEC(water), were estimated:

Table B1.8a    Look here...
Calculation results of DCOI.

Table B1.8b    Look here...
Calculation results of ZPT.

4. Sensitivity analysis

The calculation results are conditional on i.a. the values assigned to the different parameters.

A sensitivity analysis of the effect of the following parameters on the calculated concentrations of the parent compounds (DCOI and ZPT) was made:

Harbour scenarios. Supplementary PEC calculations were made for five pleasure craft harbours (cf. Section 4.1 below).

Temperature. The temperature was varied between 5°C and 15°C as this was considered a typical variation in the temperatures for the months from May till September (cf. Section 4.2 below).

	Water exchange. The water exchange was varied between 0 m3/m2/d corresponding to no water exchange and 1 m3/m2/d which would occur under specific conditions (cf. Section 4.2 below).
	Sedimentation rate. The sedimentation rate was varied between 0.7 m/d, corresponding to the net sedimentation being almost 0, and up to 1.5 m/d (cf. Section 4.2 below).

Leaching rate. The total leaching rate was varied between 50% and 200% of the total leaching rate used in the basic calculations (cf. Section 4.2 below).

4.1 Harbour scenarios

Supplementary calculations were made for five other pleasure craft harbours. A characterization of these harbours is given in Table B1.9. These data were obtained by Hempel and passed on to VKI.

The water exchange in the harbours was set at 0.6 m³/m²/day for all harbours with the exception of the harbours of Svendborg and Horsens. For the pleasure craft harbour of Horsens, the water exchange was set at 0.8 m³/m²/day. The pleasure craft harbour of Svendborg is a pile-built harbour in Svendborg Sund. Therefore, the flow conditions are assumed to correspond to those of Svendborg Sund. An average flow rate of 0.5 m/s was assumed for the pleasure craft harbour of Svendborg corresponding to the amplitude of the drastic periodic velocity variation caused by the tides in Svendborg Sund (Harremoës and Malmgren-Hansen 1989).

Table B1.9    Look here...
Harbour scenario (prepared by Hempel).

Note: Difference of height is used as a parameter in stead of tides. As the tidal variation in Denmark is very small (10-40 cm), the primary water exchange is made by wind and currents. Therefore, the difference in height is an overall assessment of the different parameters.

Information was obtained from (conversations with):
   Jyllinge: Steen Wintlev DS + map received
   Grenå: Benny Andersen
   Svendborg: Kurt Hansen
   Rungsted: Finn Rosdahl
   Horsens: Hilmer Christoffersen
   Egå Marina: Dan Nilsson (harbour master Børge Heilbach)

The calculated steady-state concentrations for these harbour scenarios are given in Table 1.10.

Table B1.10 indicates that the pleasure craft harbour of Jyllinge results in the highest calculated concentrations. The main reason for this is that there are relatively more boats in the pleasure craft harbour of Jyllinge compared to the volume of water needed in order to dilute the leached chemical (Table B1.10). The scenario used is thus a conservative scenario as assumed at the selection of the pleasure craft harbour of Jyllinge.

4.2 Sensitivity analysis of other parameters

Tables B1.11a and B1.11b show the relation between the calculated concentration of the parent compound in the water phase after changing the parameter and the calculated concentration of the standard scenario.

Tables B1.11a and B1.11b indicate that, within the variation assigned to the individual parameters, it is the total leaching rate that causes the largest variations in the calculated concentrations. The sedimentation rate only slightly influences the calculated concentrations of the parent compounds.

_______________
1) If the light intensity at the surface is expressed as I₀ and the light intensity in the depth z as I, the correlation between I and I₀ is described as log₁₀(I/I₀) = -a × z
Back...