Presentation I gave in Valencia, Spain.

1. International Groundwater Symposium
Valencia, Spain
September 22-24, 2010
Greg Ruskauff1
Nicole DeNovio2
Edward Kwicklis3
1 NNES LLC
2Golder Associates
3 Los Alamos National Laboratory
1

2. Background
The Nevada Test Site was used from
1951 to 1992 for nuclear weapons
testing
Great Basin area of western United
States
Facility area is 3,500 km2
(> Luxembourg)
The tests left behind radioactive residue,
which may be migrating in groundwater
NNSA/NSO established the
Underground Test Area (UGTA)
Subproject to ensure the protection of
the public
Site characterization and modeling used
to negotiate compliance boundaries with
Nevada Department of Environmental
Protection
2

3. Frenchman Flat
Ten underground shaft tests detonated
between 1965 and 1971
All less than 20 kt
Informally divided into Northern and
Central areas
3

4. What happens near an underground nuclear
test?
Rock and water vaporized immediately
around device, forming a cavity
Formation of a chimney as
pressure subsides
Alteration of rock properties to some
distance away from the test
Cavity size related to yield and depth of
burial
The exchange volume, where
radionuclides can be found, forms out to ~
2 cavity radii
Preferential distribution of radionuclides
Higher boiling point radionuclides become
incorporated in the nuclear melt glass
Lighter radionuclides condense from vapor
and remain in the water and sorbed to
cavity/chimney rubble
4

5. Underground Nuclear Tests in Frenchman Flat
Ten tests
Nine conducted in alluvium
One conducted in vitric tuff
The CAMBRIC event (entirely in alluvium)
has been studied extensively by
Lawrence Livermore National Laboratory,
who developed the following conceptual
model:
In alluvium, shock-compressed zones of lower
permeability, porosity limit groundwater flow
through the cavity
Radionuclides assumed to be evenly distributed
within exchange volume (2Rc), and nuclear melt
glass zone.
Supported by several numerical models
5

6. The need for other conceptual models
Within 2 cavity radii rocks vary,
especially at the water table
IAEA (1998) reviewed French
underground nuclear tests in basalt
lava in the Pacific
“The Radiological Situation at the
Atolls of Mururoa and Fangataufa”
Conceptualized high permeability
zone out to 2.5 cavity radii
Cavity infilling studies broadly support
spherical zone of enhanced
permeability from test effects
Adopt another conceptual model for
tests with the exchange volume
extending to lava and welded tuff
6

7. Consequences of near-test conceptual models
Evaluated the effect of different near-field conceptual models on
radionuclide release
Same inventory uncertainty
Different conceptual models received different parameter assignments
Alluvium/vitric tuff – permeability reduction factor from 1.5 to 100
IAEA – permeability enhancement factor of 5 to 100
Different conceptual model results in higher peak concentrations
Alluvium Concept IAEA Concept
7

8. Basin-Scale Conceptual Model
Top of the Model Domain
Deep, complicated geology Alluvial
Aquifers
Volcanic
Initial model didn’t have data Aquifers
on the depth of the alluvium,
thickness of confining units
Aquitards
Cross Section through Center of Model Domain
HLCA < HAA Regional
Carbonate
Aquifer
8

9. Hydrogeologic Conceptual Model
Easterly component from
leakage across Cane Spring fault
from CP basin
Based on age and head
0.43 relationships groundwater flows
0.25 from the perimeter of the basin
0.85 1.1 toward the basin center and out to
the south-south east
0.20
For the regional carbonate flow
system, Rock Valley fault system
provides regional drainage that
will focus flow to the southwest
out of Frenchman Flat
Groundwater velocity in
alluvium from 0.1 to 1 m/yr
9

10. Approach to Uncertainty Analysis
External project review during 1999 suggested
the need to consider conceptual uncertainty
reflected in basin conceptual model
Revised approach incorporated 5 interpretations HFM Alternative Description
of geology DETA – detachment This alternative is a no detachment fault
fault alternative model.
Tried to focus alternative interpretation in portions DISP – This alternative is concerned with the
of the model likely to have transport Displacement fault locations and displacement of basin-
alternative forming faults.
Several sets of boundary conditions CPBA – CP basin The CP basin alternative extends the
alternative UCCU beneath all of CP basin.
One different hydrologic conceptual model
The BLFA HSU is modeled as a single
BLFA – Basalt Lava-
Three permeability parameterization Flow Aquifer
continuous flow, rather than three
separate zones
approaches
Several discrete flow model cases of different
geologic models, parameterization approaches
Performed Monte Carlo transport analysis on
each case
Flow model calibration constrained Monte Carlo
analysis
Monte Carlo transport analysis on ensemble
10

11. Pooled Uncertainty Analysis
All discrete cases
considered
Alternative Hydrologic
Conceptual Model
“Null-space Monte
Carlo” of Tonkin and BASE Hydrologic
Conceptual Model
Doherty (2009) (100
realizations)
Calibration
constrained
uncertainty of flow
model parameters
Specified plausible
parameter ranges
Heads and flows always
calibrate acceptably
11

12. Which models matter?
Try to identify similar behaving
models that can be used as Alternative Hydrologic
Conceptual Model BASE Hydrologic
surrogates for others
Conceptual Model
Used cavity flow rate as
transport surrogate
BASE geology
Evaluate range to see if relevant variable
uncertainty is being sampled parameterization
Cannot discern that alternative
geologic models have any
importance
Alternate geology
Scatter among models using interpretations
variable parameterization similar
to alternative geology scatter
Two (NHA and BASE-USGSD)
really are different
12

13. Remaining Geological Uncertainty
Welded tuff under PIN STRIPE modeled
as continuous
North-south normal faulting appears to die
out, not certain
Is the welded tuff really continuous?
Missed conceptual uncertainty
13

14. Things We Learned
In one instance, exchange volume altered properties, conceptual model uncertainty
was clearly important
We need to be alert to conceptual uncertainty when we are extrapolating outside the
information we have
CAMBRIC conceptual model vs. other tests
Refsgaard, J.C., J.P. van der Sluijs, J. Brown, and P. van der Keur. 2006. “A Framework for
Dealing with Uncertainty due to Model Structure Error.” In Advances in Water Resources,
Vol. 29, 1586-1597.
Can exist at all scales
Having an initial idea of what might be uncertain is useful, but it can’t override the
ability to make adjustments
Two of the alternative geologic models were not insightful
It may take some time to understand the base interpretation enough to identify sensitive
assumptions (e.g. the geologic uncertainty at PIN STRIPE)
“If the confidence intervals associated with the models predictions are greater than
the difference between the predictions, this difference cannot be considered
meaningful, at least in a statistical sense.” National Research Council, Ground Water Models Scientific and
Regulatory Applications, 1990.
14

15. International Groundwater Symposium
Valencia, Spain
September 22-24, 2010
Greg Ruskauff1
Nicole DeNovio2
Edward Kwicklis3
1 NNES LLC
2Golder Associates
3 Los Alamos National Laboratory
15