1. Proposal for Task 9: Modelling of LTDE and REPRO
Task Force Meeting 29, November 2012

2. Outline of this presentation
Overview of Task 9 proposal
Summary of LTDE presentation at TF#28
Available data
Examples of results
Possible modelling cases

3. Proposal for Task 9: Modeling of LTDE and REPRO
LTDE is performed at Äspö HRL
Long-Term Diffusion Experiment (sorption)
Lots of data but the results need to be analyzed further
REPRO is on-going and conducted by Posiva at ONKALO
“A combination of LTDE and TRUE”
Gives possibilities to predictive modeling

4. Motivation and features of proposal
LTDE: Large data set, deserves to be analyzed more
Some odd results need to be explained
Educational, training and demonstration
Demonstrate what the Task Force can do e.g. in terms of predictive modeling
Long time since we modeled solute transport
Could fit into a PhD program
Could be a benefit to show that the solute transport tools work
Input to SA/PA modeling
Could address scale issues e.g. transport in the matrix, channels and fractures i.e. give input to the near field transport
Could give input to reactive transport modelling

5. Motivation and features of proposal
Other suggestions on topics that could be addressed?
Would it be good if a pre-study is made first?
Think about this until the parallel sessions on Thursday
Summary LTDE
REPRO
How to start up a task successfully
Task Force challenges

6. Summary of LTDE Sorption Diffusion Experiment (LTDE-SD) Heterogeneously distributed diffusion and sorption in cm scale, a possible modelling case?
# 28 Int. Äspö Task Force Meeting, January 2012
Erik Gustafsson, Geosigma

7. LTDE-SD aims at:
Increase knowledge of sorption and diffusion under in-situ conditions, in crystalline bedrock at depth representative for the final repository
Obtain data on sorption properties and processes of individual radionuclides on natural fracture surfaces and internal surfaces in the matrix rock
Compare laboratory derived diffusion constants and sorption coefficients for the investigated rock system with in-situ determined parameters
Evaluate if laboratory scale/conditions sorption results are representative also for larger scales

8. LTDE-SD Main project tasks
In-situ experiments in isolated test sections with known surface areas and without advection or dispersion effects
Functionality test with 8 short lived radionuclide tracers
Main test with 22 tracers (c. 6 ½ months)
Laboratory program using LTDE-SD site specific rock material
Supporting laboratory program at AECL
Laboratory experiments at CTH using same tracer cocktail as in situ experiment (c. 6 months)

9. Experimental set up at Äspö HRL
LTDE-SD experimental site and borehole layout in niche NASA 3067A (-410 m) at Äspö HRL. Target fracture surface and matrix rock at about 10 m borehole length in experimental borehole KA3065A03. Pilot borehole KA3065A02 is shown beside.

10. Borehole instrumentation and test section design in KA3065A03
Experimental concept
In-situ sorption and diffusion, no advection or dispersion
Natural fracture surface and matrix rock
Ambient hydraulic pressure, rock stress and hydrochemistry at – 410 m level in Äspö crystalline rock
Extraction and analysis of the rock at termination of experiment
Principle of borehole instrumentation and test section design. A about 3 mm wide and 177 mm diameter stub section in front of the natural fracture and a 300 mm long and 36 mm diameter slim hole section in matrix rock beyond the fracture. The rock volume enclosed by the red dotted line was extracted by over core drilling at termination of the experiment.
Isolated test sections with known surface areas and without advection or dispersion effects. Experiment lasted for about 6 ½ months.

11. Experimental rock volume retrieved by over core drilling
One piece, 278 mm diameter, – m core retrieved from over coring of borehole KA3065A03. Target fracture with the 177 mm diameter core stub inside the cylindrical rubber seal where radionuclide tracers were circulated

12. Sample cores extracted from large over core
Cutaway drawing showing core sample scheme of the 278 mm diameter over core from the experimental borehole KA3065A03. The test section at the target fracture (diameter of 177 mm) is confined by a polyurethane cylinder (yellow). The test section (diameter of 36 mm) in the matrix rock away from the fracture is confined using a specially designed packer. The system also contains a Peek dummy in the test section, but not shown in the figure. Sample cores are 150 – 180 mm in length and have a diameter of 24 mm.

13. Example on sample core (A6, from the fracture surface in stub section) Handling procedure of all sample cores before slicing: 1) scanning with scintillation detector 2) detailed geological mapping 3) stereo photography 4) rough measurement of gamma emitting tracers with HPGe
Sealed fractures at about 0.5 and 1.0 cm from fracture surface. 2 fractures c. 45º to borehole axis between the ”0.5” and ”1.0” fractures, not visible around the whole core. Very thin microfractures at c. 8 cm distance from the fracture surface.

14. Available data from LTDE-SD in situ experiment
Tracer concentrations in the groundwater, i.e. input function
Hydrochemical data and speciation of tracers
Diffusion profiles (Na-22, Cl-36, Co-57, Ni-63, Ba-133, Cs-137)
From natural fracture into altered/unaltered rock (stub surface section, 10 profiles)
From groundwater directly into matrix rock (slim hole section, 8 profiles)
Penetration profiles ≤ 3 mm (Cd-109, Ag-110m, Gd-153, Ra-226, Np-237)
Radionuclide distribution in each slice by autoradiograph measurements
Geological characterisation of whole sample cores and the slices

15. Available data and results from the LTDE-SD laboratory program with site specific rock material
Batch tests on crushed rock material and through-diffusion and sorption on intact drill core samples
Porosity measurements by water saturation and PMMA
Specific surface area and electrical resistivity measurements
Parameters determined; Kd, Rd, De, Ff, ε,  α, BET-area
Chemical analysis of rock samples
Rock material used in the laboratory tests
Core samples from the slim hole test section in experimental borehole KA3065A03
Counter part of the stub fracture surface in KA3065A03
Core samples from the exploration borehole KA3065A02
Diffusion and through diffusion tests on 278 mm diameter core from experimental borehole KA3065A03 (De, ε)
Permeability test on 22 mm diameter core sample from KA3065A03 (K)

16. Example of penetration profiles in matrix rock Sample core D13 with relatively deep penetration
The Y-axis refers to the measured tracer concentration in the sample (Bq/kg) divided by the tracer concentration in the water phase (Bq/m3) at termination of experiment. Values where the tracer concentration in the rock samples were below detection limit have been omitted in the figure.
Make a comment on the shape of the curves and penetration depth.

17. Example of migration paths and mineral specific sorption in matrix rock
Upper; sorption on biotite, chlorite, titanite in sample D7.1 (first slice in direct contact with radionuclide labelled groundwater)
Lower; sorption mainly on biotite, but also in pore space at mineral grain boundaries in sample D5.2 (second slice, 1.5 mm from rock surface)
To the left Autoradiography image. To the right photo of the same core sample (16x16 mm).
D.7.1 sorption also associated with a thin micro fracture.

18. Some results of the in-situ experiment
Out of 22 tracers injected, it was possible to follow 21 in the aques phase
11 tracers; Na-22, Cl-36, Co-57, Ni-63, Cd-109, Ag-110m, Ba-133, Cs-137, Gd-157, Ra-226, Np-237 were possible to follow in the rock fabric
Species mainly sorbed by surface complexation (e.g. Gd-153, Ag-110m) is found in the first few millimetres
The moderately sorbing tracers (Cs-137, Ni-63, Ba-133) are present in quite high concentrations in the first slices but have decreased to a level 3 – 4 orders of magnitude at 3 – 6 mm depth
Non sorbing Cl-36 and weakly sorbing Na-22 has penetrated typically up to 30 mm during the 200 days of in situ experiment

19. Single-rate matrix diffusion model Three modelling concepts to fit the experimental data (A16 core sample)
Case 2
Case 3
Case 4
Case 2 Blue line: fixed diffusivity (from laboratory through diffusion experiments with LTDE-SD rock material) and Kd varied.
Case 3 Red line: fixed Kd (from laboratory batch sorption experiments with LTDE-SD rock material) and diffusivity varied.
Case 4 Green line: both Kd and diffusivity varied.
Modelling has focused on the slow penetration process, which constitutes the vast majority of the sorbed tracer, therefore the data representing the fast penetration process have been excluded in this modelling.
Modelling of penetration profile data can give unique values of Kd and Diffusivity. This in contrary to the technique using only the time dependence of the loss of tracer in the aqueous phase which only is sensitive to combined product of Kd and diffusivity.

20. Some conclusions from the in-situ experiment
Autoradiograph analyses of the sliced rock samples indicate that the radionuclides diffuse in a heterogeneous pattern. The migration paths can visually be associated with microfractures and with the biotite part of the rock.
The general shapes of the penetration profiles (and forward tailing) indicates significant influence of heterogeneous distribution of porosity in micro scale creating migration paths were fast diffusion can take place.
Difficult to fit a single-rate based homogeneous diffusion-sorption model to the penetration profiles.
Further interpretation and modelling capable of handling heterogeneous porosity in micro scale is needed.

21. Possible modelling exercises based on data from the LTDE-SD experiment
Connection of the porosity distribution studies to the possible existence of heterogeneous diffusivity
Can the adsorption of the sorbing tracers be predicted using process based modelling (e.g., surface complexation/cation exchange) with literature data?
Testing the possibility of application of the results of the batch sorption experiment to predict and explain the behaviour of the sorbing tracers in the LTDE-SD experiment

22. Thank you for your attention!

23. Outline of this presentation
Overview of project and experimental work
Available data
Examples of results
Possible modelling cases

24. Cutting and slicing of sample cores
Sawing procedure of a core sample (in this example A.1) into subsamples and thereafter into cuboid shape (c. 16×16 mm) and finally thin slices (c. 3×1, 3×3, 3×5, 3×10 and 3×20 mm, see text above). The shaded areas represent surfaces exposed to autoradiographic imaging plates.

25. Comparison penetration profiles Natural fracture surface “stub”, core sample A6 Tracers; Cs-137, Ni-63, Co-57, Cd-109, Gd-153 Slices A6.1, A6.2 and A6.3 are 1 mm, slice A6.4 is 3mm thick
Gadolinium (Gd-153) is very strongly sorbed and penetration is stopped in the first rock slice, i.e. 1 mm from rock – water interface. Only Cesium (Cs-137) and Nickel (Ni-63) reaches the third slice at c. 4 mm distance from fracture surface during the course of the experiment of c. 6 ½ months.

26. Example of migration paths continued
Sample core D13 with relatively deep penetration
← D13.2
← D13.3
Autoradiography of core slices D13.1, D13.2 and D13.3
16 x 16 mm
Cs-137 activity (Bq/g) versus penetration depth in D-cores, 0-15 mm. No representation of the measurements where the tracer concentration was found to be below the detection limit has been included in the figure. The measurements corresponding to the autoradiographs below are marked in the figure.
Autoradiography analyses of selected D-core samples. The size of the samples is 16x16 mm and the corresponding measurements of Cs-137 in the same samples are shown in the Figure above
D13.1
D13.2
D13.3

27. Modelling Case 5, all retention parameters (Kd, Ff, ε) varied simultaneously to fit penetration profile and tracer loss in aqueous phase.
Ni-63
Core A1
Fracture surface
Ni-63
Core D12
Matrix rock
This modelling attempt was considered meaningful only for the tracers where the sorption was high enough to enable an observable loss in the aqueous phase, but still low enough to allow diffusion reaching some of the inner slices of the penetration profiles. For this reason, the 137Cs and the 63Ni data were exclusively selected for performing this Case 5 exercise. As for Case 4, no unique values of  and Kd could be identified from the modelling; this since they both influence the diffusion calculation only by their contribution to the (+Kd ). However, particularly in the present case, when only using the relatively strongly sorbing tracers 63Ni and 137Cs, the neglect of the  versus the Kd   can be very well motivated and Kd for this case is therefore calculated using the assumption that << Kd .

28. Modelling Case 5, all retention parameters (Kd, Ff, ε) varied simultaneously to fit penetration profile and tracer loss in aqueous phase.
Cs-137
Core A1
Fracture surface
Cs-137
Core D12
Matrix rock
This modelling attempt was considered meaningful only for the tracers where the sorption was high enough to enable an observable loss in the aqueous phase, but still low enough to allow diffusion reaching some of the inner slices of the penetration profiles. For this reason, the 137Cs and the 63Ni data were exclusively selected for performing this Case 5 exercise. As for Case 4, no unique values of  and Kd could be identified from the modelling; this since they both influence the diffusion calculation only by their contribution to the (+Kd ). However, particularly in the present case, when only using the relatively strongly sorbing tracers 63Ni and 137Cs, the neglect of the  versus the Kd   can be very well motivated and Kd for this case is therefore calculated using the assumption that << Kd .

29. Comparison of Kd ∙ Ff intervals, determined by different techniques
Comparison of the intervals Kd •Ff (m3/kg) values determined by the different techniques applied within the LTDE-SD experiment. Besides the evaluation performed for the results of the in situ experiments, a comparison is also made to laboratory experiments with sorption on entire drill cores
Using data only from penetration profiles and fitting to a homogeneous single rate model gives the lowest Kd values (dark green lines).
Using only data from the loss in aqueous phase gives the highest Kd values, except for Cs (blue lines).

30. Comparison of Kd value intervals, determined by different techniques
Comparison of the Kd value intervals determined by the different techniques applied within the LTDE-SD experiment. Besides the evaluation performed for the results of the in situ experiments, a comparison is also made to batch laboratory experiments and laboratory experiments with sorption on entire drill cores.
In-situ experiment using data only from penetration profiles and fitting to a homogeneous single rate model gives the lowest Kd values (green lines).
Laboratory experiments and in-situ experiment using only data from tracer loss in aqueous phase gives the highest Kd values (blue, blue dotted line).

31. Comparison of sorption coefficients (Kd, m3/kg) recommended for SR-Site Forsmark and results from LTDE-SD LTDE-SD results from modelling using data both from penetration profiles and tracer loss in aqueous phase.
Sorption coefficients determined from LTDE-SD in-situ data. Compared to recommended values for SR-Site Forsmark. LTDE-SD results are within recommended values for SR-Site
The threevalent lantanoid Gd and the quadrovalent Hf (analogues to three- and quadrovalent actinides, e.g. Pu and Np) shows typical high Kd.
Also the mono- and divalent cations Cs, Ni, Cd and Ra have Kd within expected range.

32. Possible modelling exercises based on data from the LTDE-SD experiment
Tentative modelling using a two pathway model for Cs-137
Matrix Fracture surface (A6 core sample from stub section)
A premature modelling attempt addressing a heterogeneous approach. The modelling is performed in order to investigate to what extent the tailing observed can be explained with realistic diffusion rates using a heterogeneous multi-rate model.
The diffusivity obtained for the tail part of the penetration profile is in agreement with what has been observed in laboratory experiments.
However it must be emphasized that this modelling uses 6 fitting parameters (two of them shown to be insensitive) and there is no independent data determining that the diffusion should be described as a two pathway process.

33. LTDE-SD Participating organisations
SKB Geosigma AB
Project management, In-situ experiments, sample preparation and analysis, evaluation, modelling
University of Helsinki
Rock core sample preparation and analysis
Swedish Defence Research Agency
Sample preparation and analysis
Technical Research Institute of Sweden Sample preparation and analysis
Chalmers University of Technology Supporting laboratory experiments
OPG/NWMO AECL Whiteshell Supporting laboratory experiments, modelling

34. Tracers injected in main sorption diffusion experiment

35. Illustration of the dependence of the formation factor (Ff) on the Kd-value in the curve fitting procedure for the loss of Cs due to sorption on intact rock
Illustration of the dependence of the formation factor (Ff) on the Kd-value in the curve fitting procedure for the loss of Cs due to sorption on intact rock. The major figure shows that, within at least a range of 6 orders of magnitudes, the best fit can be obtained independently of the individually employed Ff and Kd, as long as their product remains constant (in this particular case Ff•Kd=1E-5). However, as expected the penetration profiles differ strongly between the different cases which illustrate that penetration profile measurements are necessary in order to identify the individual values of Ff and Kd, respectively.

36. Experimental set- up at Äspö HRL
This modelling attempt was considered meaningful only for the tracers where the sorption was high enough to enable an observable loss in the aqueous phase, but still low enough to allow diffusion reaching some of the inner slices of the penetration profiles. For this reason, the 137Cs and the 63Ni data were exclusively selected for performing this Case 5 exercise. As for Case 4, no unique values of  and Kd could be identified from the modelling; this since they both influence the diffusion calculation only by their contribution to the (+Kd ). However, particularly in the present case, when only using the relatively strongly sorbing tracers 63Ni and 137Cs, the neglect of the  versus the Kd   can be very well motivated and Kd for this case is therefore calculated using the assumption that << Kd .

37. Some conclusions from the in-situ experiment
Autoradiograph analyses of the sliced rock samples indicate that the radionuclides diffuse in a heterogeneous pattern. The migration paths can visually be associated with microfractures and with the biotite part of the rock.
The general shapes of the penetration profiles (and forward tailing) indicates significant influence of heterogeneous distribution of porosity in micro scale creating migration paths were fast diffusion can take place.
The interaction observed in this experiment (occurring mainly in the rock material less than 5 mm from the water-rock interface) seems to be influenced by a somewhat decreased Kd compared to batch sorption experiments and with an increased diffusivity compared to the results obtained by laboratory through diffusion experiments.
Difficult to fit a single-rate based homogeneous diffusion-sorption model to the penetration profiles.
Further interpretation and modelling capable of handling heterogeneous porosity in micro scale is needed.

38. Possible modelling exercises based on data from the LTDE-SD experiment
Connection of the porosity distribution studies to the possible existence of heterogeneous diffusivity
The penetration profile measurement of the tracers indicates that there are significant deviations from a homogeneous diffusivity pattern and also a significant variation in diffusivity rates of the different measured samples. An interesting task would be to investigate if the porosity distribution measurement could be used to explain and predict the shape of the penetration profiles, e.g., influence of microfractures
Can the adsorption of the sorbing tracers be predicted using process modelling (e.g., surface complexation/cation exchange) with literature data?
Using surface complexation models combined with the mineralogical data to make prediction of the adsorption and other processes (e.g., precipitation and co-precipitation) for the different sorbing tracers.

39. Possible modelling exercises based on data from the LTDE-SD experiment continued
Testing the possibility of application of the results of the batch sorption experiment to predict and explain the behaviour of the sorbing tracers in the LTDE-SD experiment
How do the results of the batch sorption experiment relate to the losses in the aqueous phase and/or the penetration profile of the sorbing tracer of the experiment?

40. LTDE-SD project performance and results are presented in three reports:
Widestrand H, Byegård J, Selnert E, Skålberg M, Höglund S, Gustafsson E, Äspö Hard Rock Laboratory. Long Term Sorption Diffusion Experiment (LTDE-SD). Supporting laboratory program - Sorption diffusion experiments and rock material characterisation. With supplement of adsorption studies on intact rock samples from the Forsmark and Laxemar site investigations. SKB R Svensk Kärnbränslehantering AB.
Widestrand H, Byegård J, Kronberg M, Nilsson K, Höglund S, Gustafsson E, Äspö Hard Rock Laboratory. Long Term Sorption Diffusion Experiment (LTDE-SD), Performance of main in situ experiment and results from water phase measurements. SKB R Svensk Kärnbränslehantering AB.
Nilsson K, Byegård J, Selnert E, Widestrand H, Höglund S, Gustafsson E, Äspö Hard Rock Laboratory. Long Term Sorption Diffusion Experiment (LTDE-SD). Results from rock sample analyses and modelling. SKB R Svensk Kärnbränslehantering AB.