Hydrogeological conceptual model development and numerical modelling using CONNECTFLOW, Forsmark modelling stage 2.3 Forsmark Hydrogeological conceptual model development and numerical modelling using CONNECTFLOW, R-08-23 Hydrogeological conceptual model development and numerical modelling using CONNECTFLOW, Forsmark modelling stage 2.3 Sven Follin, SF GeoLogic AB Lee Hartley, Peter Jackson, David Roberts Serco TAP Niko Marsic, Kemakta Konsult AB May 2008 Svensk Kärnbränslehantering AB Swedish Nuclear Fuel and Waste Management Co Box 250, SE-101 24 Stockholm Tel +46 8 459 84 00 R-08-23 CM Gruppen AB, Bromma, 2008 ISSN 1402-3091 Tänd ett lager: SKB Rapport R-08-23 P, R eller TR. Hydrogeological conceptual model development and numerical modelling using CONNECTFLOW, Forsmark modelling stage 2.3 Sven Follin, SF GeoLogic AB Lee Hartley, Peter Jackson, David Roberts Serco TAP Niko Marsic, Kemakta Konsult AB May 2008 Keywords: Forsmark, Hydrogeology, Modelling, Calibration, Sensitivity. This report concerns a study which was conducted for SKB. The conclusions and viewpoints presented in the report are those of the authors and do not necessarily coincide with those of the client. A pdf version of this document can be downloaded from www.skb.se. Abstract Three versions of a site descriptive model (SDM) have been completed for the Forsmark area. Version 0 established the state of knowledge prior to the start of the site investigation programme. Version 1.1 was essentially a training exercise and was completed during 2004. Version 1.2 was a preliminary site description and concluded the initial site investigation work (ISI) in June 2005. Three modelling stages are planned for the complete site investigation work (CSI). These are labelled stage 2.1, 2.2 and 2.3, respectively. An important component of each of these stages is to address and continuously try to resolve discipline-specific uncertainties of importance for repository engineering and safety assessment. Stage 2.1 included an updated geological model for Forsmark and aimed to provide a feedback from the modelling working group to the site investigation team to enable completion of the site investigation work. Stage 2.2 described the conceptual understanding and the numerical modelling of the bedrock hydrogeology in the Forsmark area based on data freeze 2.2. The present report describes the modelling based on data freeze 2.3, which is the final data freeze in Forsmark. In comparison, data freeze 2.3 is considerably smaller than data freeze 2.2. Therefore, stage 2.3 deals primarily with model con- firmation and uncertainty analysis, e.g. verification of important hypotheses made in stage 2.2 and the role of parameter uncertainty in the numerical modelling. On the whole, the work reported here constitutes an addendum to the work reported in stage 2.2. Two changes were made to the CONNECTFLOW code in stage 2.3. These serve to: 1) improve the representation of the hydraulic properties of the regolith, and 2) improve the conditioning of transmissivity of the deformation zones against single-hole hydraulic tests. The changes to the modelling of the regolith were made to improve the consistency with models made with the MIKE SHE code, which involved the introduction of spatial variability of the hydraulic properties within soil layers and horizontal versus vertical anisotropy in the hydraulic conductivity of soils. For the deformation zones, the same prescription for assigning transmissivities was followed as for stage 2.2, but a new method for automating the local conditioning of the deformation zone transmissivity in the vicinity of a measurement interval was used. The numerical simulations carried out in stage 2.2 demonstrated that the three geological units: deformation zones, fracture domains and regolith, can be parameterised by means of single-hole hydraulic tests and satisfactorily transformed into heterogeneous hydraulic conductor domains (HCD), hydraulic rock mass domains (HRD) and hydraulic soil domains (HSD). This means that the conceptual model developed from the interpretation of Forsmark data in stage 2.2 can be used to predict a wide range of different types of data and processes such as 1) large-scale cross-hole test responses, 2) natural point-water heads in the bedrock and the regolith, and 3) hydrochemistry profiles along the many cored boreholes drilled in close proximity to the so- called target area. It is noted that a primary idea of the confirmatory testing applied in stage 2.2 is that the same groundwater flow and solute transport model is used for each type of simulation to make it transparent that a single implementation of the conceptual model could be calibrated against all three types of field observations, although it may have been possible to improve the modelling of a particular data type by refining the model around a relevant observation borehole, for example. The conceptual modelling in stage 2.2 invoked a number of hypotheses, three of which that were addressed in stage 2.3 by means of complementary field investigations (hydraulic tests). The results from these investigations do not falsify (contradict) any of the three hypotheses, hence none of them should be rejected. In fact, the three hypotheses are supported by new evidence, which strengthens the overall credibility in the conceptual model presented in stage 2.2. 3 The sensitivity of the calibrated groundwater flow and solute transport model developed in stage 2.2 to parameter uncertainty was addressed in stage 2.3 by means of numerical modelling. A comprehensive set of uncertainties have been quantified to each of the model elements: HCD, HRD and HSD as well as boundary conditions both in terms of their effects on the model calibration processes and in predictions of discharge areas for groundwater flow through the repository candidate volume. The results from the sensitivity tests carried out confirm that high degrees of anisotropy in the regolith and/or the uppermost bedrock seem to characterise the near-surface hydrogeological conditions of the site. Furthermore, the envelope of realisations used to simulate the structural-hydraulic heterogeneity of the deeper parts of bedrock captures much of the sampled hydrochemical data and shows that the few shortcomings in the predic- tions of using a calibrated single realisation (here called the base model simulation) may be explained by the heterogeneity in the structural-hydraulic properties. Sensitivities studies considering the role of structural-hydraulic heterogeneity on discharge locations for the repository target area by means of particle tracking corroborate that the base model simulation gives a consistent prediction of the main discharge areas, confirming that the geometry of deterministically modelled structures (larger deformation zones and sheet joints) are the dominant control on groundwater pathways, and that hydraulic heterogeneity does not disperse the exit locations to any radical degree. Exploratory simulations considering the Singö deformation zone as potential barrier to flow emphasise the importance of this zone for control- ling the ultimate fate of any release. 4 Sammanfattning Tre versioner av den platsbeskrivande modellen för Forsmark har färdigställts. Version 0 beskrev kunskapsläget innan platsundersökningarna påbörjades. Version 1.1, som var en övningsversion, färdigställdes år 2004 och version 1.2 färdigställdes i juni år 2005. Version 1.2 utgör den preliminära platsbeskrivningen för Forsmark och beskriver kunskapsläget efter det inledande platsundersökningsskedet. Det avslutande platsundersökningsskedet består av tre steg, vilka betecknas 2.1, 2.2 och 2.3. En viktig uppgift för arbetet inom var och ett av dessa steg är att tydligt redovisa kunskapsläget samt osäkerheter av betydelse för projektering och säkerhetsanalys. Steg 2.1 syftade till att ge feedback till genomförandet av de återstående platsundersökningarna och innehåller dessutom en uppdaterad geologisk modell över Forsmark. Steg 2.2 och 2.3 karaktäriseras av ett stort antal ämnesspecifika underlagsrapporter. Den föreliggande rapporten ingår i steg 2.3 och beskriver det hydrogeologiska kunskapsläget i Forsmark och den numeriska modellering som utförts baserat på datafrys 2.3, som är den sista datafrysen inom ramen för platsundersökningarna i Forsmark. I jämförelse med datafrys 2.2 är datafrys 2.3 betydligt mindre rik på data. Syftet med steg 2.2 var att redovisa en konceptuell modell baserat på datafrys 2.2 och pröva densamma medelst numerisk modellering. Syftet med steg 2.3 är att verifiera de olika hypoteser som ställdes i steg 2.2 och demonstera beräkningsresultatens känslighet för olika typer av osäkerheter inkl parameterheterogenitet. På det hela taget är arbetet i steg 2.3 att betrakta som ett komplement till arbetet i steg 2.2, där tonvikten ligger på verifiering och osäkerhets analys. De numeriska simuleringarna i steg 2.2 visar att hydrauliska egenskaper som tilldelats jordlager, sprickor och deformationszoner kan användas för att prediktera olika datatyper och processer som 1) responser vid storskaliga mellanhålstester, 2) naturliga grundvattennivåer i berggrunden och i det kvartära jordtäcket och 3) hydrokemiska profiler längs med ett stort antal kärnborrhål borrade i närheten av det potentiella förvarsområdet. Det bör påpekas att en huvudtanke i steg 2.2 har varit att använda en och samma grundvattenflödes- och transporttmodell för de olika simuleringarna, dvs skapa transparens och konsistens i modelleringen. Två mindre ändringar har utförts