BIOPROTA 2012 Determination of probabilistic Kd values for freshwater Laura Marang and Philippe Ciffroy 1 EDF Recherche et Développement
Over 100 people dealing with Water and Environment issues EDF runs More than 500 dams 19 nuclear power plants and some fossil power plants, which all need water for cooling and develops new hydropower techniques using renewable marine energy 70% of French open-air water meet EDF works 2 - Document name - Chapter - 00 Mois 2009
National Hydraulic and Environment Laboratory One of the 15 departments of EDF research and development Our mission : The National Hydraulic and Environment Laboratory (LNHE) is entirely focused on water and environment with two main objectives: Consolidate the environmental acceptability of EDF production facilities in relation to the aquatic environment (and health via the food chain): thermal, chemical, radiological and biological discharges and the impact on morphodynamics, continuity of water courses, etc. Protect the production facilities (dams, nuclear sites) against environmental hazards: rizing water levels, floods, storms, low water levels, heat waves, etc. 50% of activity devoted to nuclear energy production, 15% to hydraulic production, 7% to marine renewable energies and 28% to cross-disciplinary environmental themes in the various production sectors. Our main activities : numerical simulation laboratory testing of scale models, in situ monitoring (on field measurements) 3 BIOPROTA May 2012
Realistic environmental and health risk assessment (15 staff) Thematic : Environnemental and health risk assessment due to the aquatic transfert of pollutants originally from chronic releases or waste of production facilities. Technical skill : hydrogeology geochemistry ecotoxicology radioecology numerical modelling Help to increase reliability and realism of our impact assessments for sites in operation Propose appropriate and performing tools for monitoring and improving our knowledge of the fate of pollutants in the environment Evaluate the safety to optimize the storage of radioactive waste Develop and support the use of engineering tools to perform risk assessment study 4 - BIOPROTA May 2012
Methods for assessing envirronmental and heath risks Identification and quantification of emission Dispersion in air, aquatic system, terrestrial system Specific site information required to perform a realistic risk assessment Biota Exposition Human Exposition Effets on ecosystem Effets on human Comparison / addition - of risk 5 - BIOPROTA May 2012
Kd value : Key parameter Environment Mobility and Bioavailability are related to the speciation H + Solutes Na, Mg, Ca Al, Fe, Mn Complexing ligands Natural organic matter-humic Substances Oxides of Al, Si, Mn and Fe Clay cations exchangers Nanoparticles Conditions ph Ionic strength Temperature Redox potential Kd : solid-liquid partition coefficient Best estimate Probability density function (PDF) 6 - BIOPROTA May 2012
Use of Kd water/suspended matter values in the model Assessment tool predictions are sensitive to the value of Kd and therefore this parameter value is important. ERICA (dose to biota) developped in the framework of a European project ERICA uses Kd values to predict unknown water or sediment concentrations Water concentration is used with CR to predict wholebody concentration and internal dose Sediment concentration is used for estimation of external dose OURSON (dose to human) developped by EDF R&D OURSON uses Kd values to predict concentrations in the dissolved phase. The dissolved phase is used to predict concentration in fish and in drinking water. Do we have a problem with freshwater Kd value? The following presentation is applied to the radionuclides of interest for EDF in case of chronic releases but the methodology could easily be applied to others radionuclides. 7 - BIOPROTA May 2012
Comparison between different database Ag Am Co Cr Cs Eu Fe Mn Ni Zn Kd L/kg 8 - BIOPROTA May 2012 TRS 364 (1994) TRS 472 (2009) ERICA ERICA / TRS 364 ERICA / TRS 472 GM 85000 120000 1,4 GSD 2.3 GM 5000 850000 530000 106,0 0,6 GSD 3.7 GM 5000 43000 106000 21,2 2,5 GSD 9.5 GM GSD GM 1000 8500 137000 137,0 16,1 GSD 6.7 GM 500 500 500 1,0 1,0 GSD GM 5000 5000 GSD GM 1000 1300000 90800 90,8 0,1 GSD 12 GM 20000 GSD GM 500 500 GSD In the TRS 472, a database containing Kd values has been referenced and a weighted bootstrapping procedure was set up in order to built PDF for environmental conditions. The difference between ERICA values and TRS 472 values are due to the fact that in ERICA, values are the AM of the reported GM. Important disparity between the values proposed by each database High uncertainty on the Kd values referenced in the TRS 472 (several orders of magnitude) Some ERICA values are based on sea water (Ni...) because values for freshwaters are not available. BUT: Kd will also vary considerably within freshwaters with factors such as ph and organic matter content. When multiple values are available, what is the value to be considered in a risk assessment study? Kd values proposed in these databases are generic values, that is to say they are not specific to French rivers. Are these values adapted to our environmental conditions? Is the uncertainty associated with Kd values due to uncertainty about the parameters or due to natural variability?
Development of a methodology to predict Kd values 0.5 Probability density function ph DOC POC Montecarlo sampling 10 000 water compositions + Radionuclides activity in water calculed from the release and water flow ECOSAT 10 000 Kd values accounting for the speciation in each water composition 0.4 0.3 0.2 0.1 0 2.5 3.5 4.5 5.5 logkd (L/kg) Equilibrium Calculation Of Speciation And Transport Chemical speciation software Détermination of Kd specific to french rivers 9 - BIOPROTA May 2012
Probability density function of factors that influence speciation in freshwaters A bibliographic study has been performed to identify factors which potentially influence metal speciation in French freshwaters : ph, major cations, carbonate system, ionic strength, dissolved and particulate organic carbon, oxides Long term-chronic (10 years) have been statislically treated. It is the case for ph, major cations, ionic strength, dissolved organic carbon Data are easily available from the Agence de l Eau for each watershed considered in the study. The total concentrations of Al, Fe and Mn were collected from the monitoring of french freshwaters organised by EDF up-stream and down-stream each site of production. The concentrations of oxides were calculated with a chemical speciation sofware from the total concentrations. 10 - BIOPROTA May 2012
Determination of Particulate organic carbon Only few data are available in littérature for French rivers Relationship between concentration of suspended matter and particulate organic carbon for different rivers in the world : all rivers follow the same trend. 16 %POC 14 12 10 8 6 %POC moyen Calcul %POC moyen %POC minimum Calcul %POC minimum %POC maximum Calcul %POC maximum 4 2 0 1 10 100 1000 10000 MES (mg/l) The probability density function of POC was derived combing the relation between suspended matter and water flow (data easily available) and between suspended matter and percentage of particulate organic carbon. 11 - BIOPROTA May 2012
Probability density function of factors that influence nuclide s speciation Fleuve Loi statistique POC (mg/l) μ σ 5% 95% Geom Mean GSD 5% 95% Nombre de valeurs Garonne log normal -0.25 0.29-0.73 0.23 7.77E-01 1.34 0.48 1.26 115 Loire log normal -0.11 0.17-0.39 0.17 8.97E-01 1.19 0.68 1.19 2557 Meuse log normal -0.39 0.19-0.71-0.07 6.76E-01 1.21 0.49 0.93 97 Moselle log normal -0.27 0.29-0.75 0.20 7.59E-01 1.33 0.47 1.22 108 Rhin log normal 0 Rhône log normal -0.11 0.43-0.81 0.59 8.93E-01 1.53 0.44 1.80 202 Seine log normal -0.11 0.30-0.61 0.39 8.99E-01 1.36 0.54 1.48 309 Vienne log normal -0.32 0.08-0.46-0.18 7.28E-01 1.90 0.63 0.84 716 Tous log normal -0.25 0.29-0.73 0.23 7.77E-01 1.34 0.48 1.26 115 Loi statistique ph μ σ 5% 95% Geom Mean GSD 5% 95% Nombre de valeurs Garonne log normal 2.07 0.04 2.0 2.15 7.95 1.05 7.38 8.56 1256 Loire log normal 2.13 0.08 2.00 2.27 8.45 1.08 7.40 9.66 3058 Meuse log normal 2.12 0.04 2.05 2.18 8.32 1.04 7.80 8.86 527 Moselle log normal 2.07 0.05 1.99 2.14 7.92 1.05 7.35 8.54 867 Rhin log normal 2.09 0.05 2.01 2.17 8.05 1.05 7.44 8.72 382 Rhône log normal 2.08 0.03 2.03 2.12 7.98 1.03 7.62 8.37 4174 Seine log normal 2.07 0.04 2.00 2.13 7.91 1.04 7.42 8.44 1378 Vienne log normal 2.01 0.05 1.92 2.09 7.44 1.05 6.84 8.11 758 Tous les fleuves Fleuve log normal 2.09 0.06 1.99 2.19 8.06 1.06 7.28 8.92 12400 Loi statistique Ca 2+ (mol/l) μ σ 5% 95% Geom Mean GSD 5% 95% Nombre de valeurs Garonne log normal -6.84 0.18-7.14-6.54 1.07E-03 1.20 7.92E-04 1.44E-03 329 Loire log normal -7.06 0.26-7.50-6.62 8.52E-04 1.30 5.50E-04 1.32E-03 2611 Meuse log normal -6.29 0.17-6.58-6.00 1.84E-03 1.19 1.38E-03 2.46E-03 758 Moselle log normal -5.72 0.51-6.56-4.89 3.26E-03 1.66 1.40E-03 7.52E-03 788 Rhin log normal -6.56 0.17-6.84-6.28 1.41E-03 1.18 1.06E-03 1.86E-03 566 Rhône log normal -6.48 0.17-6.77-6.20 1.52E-03 1.18 1.14E-03 2.03E-03 350 Seine log normal -6.14 0.15-6.39-5.89 2.14E-03 1.16 1.67E-03 2.74E-03 181 Vienne log normal -7.90 0.65-8.97-6.84 3.67E-04 1.91 1.26E-04 1.07E-03 533 Tous log normal -6.46 0.27-6.91-6.02 1.56E-03 1.309 9.97E-04 2.42E-03 4795 10 000 water compositions representing the annual environmental variability 12 - BIOPROTA May 2012
Ion-binding models simple solution chemistry Na, Cl, OH, CO 3, SO 4 Clay cation exchanger ECOSAT Humic Ion- Binding Models (NICA-Donnan) Oxide model (CD-Music) AlOx SiOx MnOx FeOx Calculations take into account : competition between the element of interest and major ions (H+, Mg, Ca, Al, Fe etc), complexation by inorganic ligands and natural organic matter (dissolved and particulate) and oxide Calculations assume that : DOC can be represented by average isolated fulvic acid, OM in particulate matter (SPM) can be represented by average isolated humic acid All the oxide have the same properties than FeOX 13 - BIOPROTA May 2012
Advantage of the models included in ECOSAT Set of generic parameters have been derived wich enables the use of these models for a wide range of different aquatic systems (Milne, 2003,ES&T; Ponthieu, 2006, Geo. Chim.Acta) In situ Validation : Good agreement between predicted and measured speciation in aquatic system, soil and sediments Ni Solid-solution partioning in the Humber rivers (Lofts, 2000, The science of the Total environment) Cu concentration in soil solution (Netherlands soil) (Bonten, 2008,Geoderma) 14 - BIOPROTA May 2012
Results Good agrement between calculated Kd and Kd of the littérature when a large number of Kd is available (case of cesium) Calculated values are lower than the values proposed in the TRS 472 and in ERICA Diminution of the uncertainties Determination of Kd for radionuclides such as Ni and Cr (no data available in littérature) 15 - BIOPROTA May 2012
Does it matter? Example of ERICA (dose to biota) : Too High Kd values as presented in the tool : Will give lower water concentration lower whole body concentration: it is therefore not conservative Will give higher sediment concentration higher external exposure - conservative as >90% of most metals are in sediment Example of OURSON (dose to human) : Too high Kd values are not conservative since this leads to lower activity in the dissolved phase and therefore in the drinking water. This methodology could be applied when there is a limited amount of data available in the literature for a radionuclide or when a risk has been identified (needs of site-specific informations). 16 - BIOPROTA May 2012
Limitations and perpectives Limitation : Lack of data to derive generic parameters for some cations (but relation between hydrolysis constant and binding constant with organic matter and oxide are given in the litterature) Hypothesis of reversible and rapid equilibrium is not always true Incorporation of all relevant processes (in particularly knowledge of all the precipitates present and the reliable thermodynamic constant) Application to the soil, marine and estuarine ecoystem 17 - BIOPROTA May 2012