By Gordon P.L.Scott
Newcastle University, School of Civil Engineering and Geosciences

We’re pleased to be able to publish out first guest blog this week! Gordon P.L.Scott, from Newcastle University, School of Civil Engineering and Geosciences, has kindly let us use his current research paper for our Leapfrog blog. Have a read of this fantastic article and learn how Leapfrog software helped Gordon with his research.

 

Summary: Evaluating tritium plume

The evaluation of a developing tritium plume using a new approach through the 3D characterisation of the hydrogeology in a complex glacial environment is presented through the use of new geological modelling software and spatially varying Kh and Kv within a lithofacies method.

 

Why monitor?

The main tritium waste repository in the UK is at the Drigg Low Level Waste repository in West Cumbria (Figure 1), where the waste is stored in a series of vaults excavated into the shallow drift sediments. Over the years, it has been proven (Henderson and Smith 2011) that tritium has been leaking from the depository, and found its way into the groundwater system. The presence of ‘tritiated’ water in the groundwater system is undesirable as it is a radioactive compound that is easily absorbed into the body and poses a risk to our DNA.

The tritium concentration in groundwater can be used to check the degree of confinement of an aquifer and the rate of flow of groundwater and provide data with which to validate the hydraulic parameters of the numeric model used to monitor the tritium plume.

Drigg LLWR location map, West Cumbria.

Drigg LLWR location map, West Cumbria.

Geology and hydrology

The storage vaults within the Drigg LLWR have been built in Quaternary glacial to post glacial sands, gravels and tills, which have a complicated history of deposition, erosion and reworking as the glaciers have advanced and retreated across the area. This has resulted in a complicated distribution of the sediments both horizontally and vertically and results in the groundwater flow paths being complex and hard to determine.

Numerous boreholes (over 700) have been drilled around the Drigg area to understand the lithological and hydrological variations across the site area as well as to monitor the tritium pollution. This data has been used as part of the input for a detailed numerical model to explain the current groundwater flow patterns and to show where the evolving tritium pollution plume will migrate.

Hydrogeological modelling– how it’s done.

Traditionally, a hydrogeological model is built as a series of layers of bodies that have been defined by their lithologies and hydraulic conductivities. Lithologies logged in the borehole are correlated on the basis of chronostratigraphy, using a series of cross sections to define their structure and variograms to define their properties. The hydraulic conductivities (Kh and Kv) obtained for each formation are then combined with their respective formations to create the hydrogeological model.

Such an approach works well in areas of good borehole control and relatively laterally continuous layer, with slowly varying conductivities.

However complex glaciated terrains such as the study area do not need these criteria. Lithologies vary rapidly and therefore so do the hydraulic conductivities. Simple layer- based modelling reaches its limits, and a step change towards object based modelling is required.

 

Problems with the traditional approach

The traditional approach to hydrogeological modelling fails for the following reasons:

  • The process is time consuming and manually intensive; as new data is added significant time is spent updating the cross sections and variograms needed for the correlation purposes,
  • The interpretation is inherent in the modelling stage rather than as a separate stage as the selection of boreholes for the cross sections affects the geological model output.
  • Gridding algorithms struggle with the issue of partial thicknesses of formations when boreholes stops midway through a formation.
  • In addition, the previous hydraulic modelling has taken a single set of hydraulic conductivities for each lithofacies and stochastically modelled the variation of the K values which conflicted with borehole values.

A different approach was needed at the Drigg LLWR for an accurate and structurally consistent 3D hydrogeological model which incorporates the geology and hydrology in a fashion which could be easily updated, and able also to display the results displayed to a wider audience who were concerned about the tritium pollution.

 

The new approach?

Lithofacies

Eyles et al (1983) applied the lithofacies approach to glacial terrains and defined a lithofacies in terms of the dominant lithology within an area with a similar style of deposition. This was applied to West Cumbria by Mcmillan et al (2000) where 7 distinct lithofacies were identified. This approach has been used at the Drigg LLWR site for the recent geological evaluation.

 

Geological modelling

The development of fast radial basis functions (RFBs) by Hardy (1971) has allowed geological models to be constructed accurately and very quickly in 3D and so overcome the limitations stated earlier. The ‘Leapfrog Hydro’ software which uses these RFBs was used to construct the hydrogeological model used in this study. Figure 2 below is a representative cross section through the Drigg LLWR site showing boreholes, lithofacies and the sub surface structure.

Cross section through part of the Drigg 3D geological model showing the lithofacies, boreholes and the regional aquifer.

Cross section through part of the Drigg 3D geological model showing the lithofacies, boreholes and the regional aquifer.

A 3D geological model can be sliced to review and present the geological results not only conventional maps and cross sections (Figure 2) but also at any desired surface. Figure 3 shows the sub crop of the different lithofacies and lithologies against the regional aquifer.

Surface slice through the Drigg 3D geological model showing the sub crop of the various lithofacies against the regional aquifer.

Surface slice through the Drigg 3D geological model showing the sub crop of the various lithofacies against the regional aquifer.

 

From geology to hydrogeology

With the new 3D geological model comes a different way to model the hydraulic conductivities within this area. Mcmillan et al (2000) stated that that the average hydraulic conductivity (K*)within a lithofacies was a product of a heterogeneity factor (Ch) and the geometric mean of the individual layers thicknesses and hydraulic conductivities.

Table 1 illustrates the methodology for a single well. The average hydraulic conductivities for each lithofacies were calculated in Excel and gridded in ArcGIS. The grids were imported into the hydrogeological model to define K for each cell and finally the hydrogeological model was available as input to the numerical modelling.

Effective hydraulic conductivities for the lithofacies within borehole 6125 using the Mcmillan et al (2000) proposed method.

Effective hydraulic conductivities for the lithofacies within borehole 6125 using the Mcmillan et al (2000) proposed method.

 

A way forward?

Previous work on the Drigg model failed to capture the detailed hydraulic layering within the B2 and B3 lithofacies and relied instead on stochastic modelling. The detailed layering was not accurately captured by this method and requires more work to explain the relationship between water levels and the 2 aquifers in certain parts of the Drigg site.

The new method outlined here can be refined to create a hydraulic lithology log for each borehole using Excel. These logs can be upscaled using McMillan et al’s (2000) method and incorporated into the hydrogeological model.

This technique can create a 3D hydrogeological model which separates geological interpretation from model building which capture the spatial variation across the Drigg LLWR site within the individual lithofacies. This will lead to a better understanding and modelling of the movement of the tritium plume.

 

Further reading:
Read more about 3D implicit modelling.