ML20035A260
| ML20035A260 | |
| Person / Time | |
|---|---|
| Issue date: | 06/30/1992 |
| From: | Codell R NRC OFFICE OF NUCLEAR MATERIAL SAFETY & SAFEGUARDS (NMSS) |
| To: | Federline M NRC OFFICE OF NUCLEAR MATERIAL SAFETY & SAFEGUARDS (NMSS) |
| References | |
| REF-WM-1 NUDOCS 9303250063 | |
| Download: ML20035A260 (7) | |
Text
n.
89001832 JUN 3 0 NE MEMORANDUM FOR: Margaret Federline, Branch Chief Hydrology and Systems Performance Branch, DHLWM FROM:
Richard Codell, Senior Systems Analyst Repository Performance Assessment Section I
Hydrology and Systems Performance Branch, DHLWM
SUBJECT:
ATTENDANCE AT INTRAVAL MEETING, SYDNEY, AUSTRALIA, FEBRUARY 10-14, 1992 Enclosed is a detailed trip report which contains a summary of my observations that were made as a result of my attendance at the INTRAVAL meeting in Sydney, Australia, from February 10 to 14,1992.
Richard Codell, Senior Systems Analyst Repository Performance Assessment Section 4
Hydrology and Systems Performance Branch, DHLWM d
Enclosure:
As stated cc: R. M. Bernero, NMSS B.J.Youngblood, Director, DHLWM i
J. Linehan, DHLW J. Taylor. EDO Division of Security GPA DISTRIBUTION Central File BJYoungblood, HLWM JJLinehan, HLWM HLHP r/f PD E-MVFederline, HLHP RLBallard, HLEN JHolonich, HLPD NMSS r/f g pp/T_
RCodell, HLHP NMSS Off.Dir r/f gg
)? C HLHPAh HLHP OFC HLHP NAME RCodell/cj NEisenberg 0
DATE 6/j//92 6/30/92 0FFICIAL RECORD COPY
'f f
Ahb?b[a G:\\ SYDNEY b
Gy'
= 223asa emesse o.@
gg t
t
OFFICIAL TRAVELLER:
TRAVEL T0:
Richard B. Codell Sydney, Australia BEGINNING ON: 02/08/92 UNTIL: 02/16/92 l
OFFICE: Office of Nuclear Material Safety and Safeguards e
DIVISION: Division of High-Level Waste Management 6
BRANCH: Hydrology and Systems Performance Branch MEETING TITLE AND/0R AFFILIATION:
February 10-14, 1992, Attendance at the INTRAVAL, Phase 2 Workshop, Sydney, Australia
~
ORGANIZED BY: Australian National Science Technology Organization (ANSTO) and Swedish Nuclear Power Inspectorate (SKI)
)
i
SUMMARY
OF MEETING RESULTS:
INTRAVAL is an international exercise whose aim is to increase understanding of how natural phenomena can be described by models that will be used to license radioactive waste repositories, and to explore the feasibility of validating these models.
Phase 2 of INTRAVAL consists of ten experiments in a range of media types, flow and transport processes, spatial and time scales and various environmental conditions.
Phase 2 of INTRAVAL emphasized increased interaction between experimentalists and modelers so that the results of the models could be coupled back into the experimental program.
l INTRAVAL participants come from 13 countries. The United States is represented by NRC, DOE and EPA, the State of Nevada, and the State of New Mexico. George 1
l Birchard of RES also represented NRC. Gordon Wittmeyer from the CNWRA, Richard Hills from New Mexico State University, and Schlomo Neuman from University of Arizona represented the NRC as contractors to RES.
I The INTRAVAL meeting was hosted by the Australian National Science Technology I
Organization (ANSTO). ANSTO participates in INTRAVAL mostly in regard to the Alligator Rivers Analog Project (ARAP), of which NRC is a sponsor.
As most of the sessions ran concurrently, it was impossible for me to attend all 1
meetings.
I have a complete set of all overhead slides presented at the meeting, however. While there were many interesting presentations concerning validation of models in a variety of different media, I will summarize the presentations and I
conversations of most interest to the NRC waste management programs.
)
[
t Unsaturated sites Las Cruces i
Richard Hills is the principal investigator of the Las Cruces trench experiment sponsored by NRC. The Las Cruces experiment consists of a deep trench in an unsaturated desert soil. The site is artificially irrigated, and the movement of I
water and tracers observed with a dense network of sampling ports drilled from the side of the trench and from the top. The experiment has beer going on for several years and was the subject of an INTRAVAL Phase I test case. The current phase 2 test case has moved to the opposite side of the trench.
It is being conducted as a blind experiment; i.e., the modelers are being given measurements of physical parameters and the results of hydraulic and tracer tests from the earlier experiments on the original side of the trench, but not the results of the current 4
migration and tracer tests. There is more emphasis on performance assessment in this phase; one of the objectives of the experiment is to calculate the flux of water and tracer past a horizontal plane to simulate cumulative release of radionuclides as they would be expressed in 10CFR60.112.
Gordon Wittmeyer presented results of modeling of the Las Cruces experiments 4
using the PORFLOW computer code developed by Budhi Sagar and currently in use at the CNWRA. Wittmeyer attempted to simulate the experimental observations using 5 levels of detail ranging from the assumption of spatially uniform properties to i
spatially varying properties based on a large quantity of data collected at the site. A comparison of the model and prototype results seemed to indicate that the simpler representation of the site gave better agreement than the more complicated representation. Wittmeyer offered the explanation for this apparent contradiction:
- There is limited spatial resolution of the concentration sampling, which may not represent what is actually happening.
l
- It was difficult to determine the initial value of water content, a key parameter of the flow model.
He concluded that none of the results were resoundingly successful, and that one i
should not on the basis of these results determine that only simple models and sparse site characterization are adequate for determining performance.
Richard Hills presented an initial demonstration of a validation methodology to a simplified subset of the experimental data, total water content. The objective of this exercise was to compare predictions of total water content to observations using statistical inference, and to accept or reject models on that basis. He discovered that models can be rejected even though the maximum predictive error is small.
Furthermore, the statistical tests do not specify the reason for rejection; e.g., error in conceptual or mathematical model, parameter estimation error, or invalid statistical assumptions. Therefore the use of standard statistical inference on complex problems such as model validation for HLW repositories may not be appropriate.
In a presentation on another test case, Natalie 01 ague demonstrated the rejection and acceptance of models (See Twin Lakes study below).
)
4 Apache Lean Schlomo Neuman of the University of Arizona and an NRC contractor described current experimental and modeling at the Apache leap unsaturated rock site in Arizona. The Apache Leap experiment is in an unsaturated tuff having characteristics in common with Yucca Mountain, and has been the subject of many NRC reports. The most current experimentation and modeling at the site deals with pneumatic measurements for cross-borehole tests using air. Measurements of water flow at the site under unsaturated conditions lead only to estimates of hydraulic parameters close to the sampling location, usually a distance of a meter or less.
To get more extensive measurements over a larger scale would either mean saturating the site with liquid water or using air.
Of course, the former practice would alter the unsaturated nature of the site. Air measurements have the advantage of causing no long-term changes to the site, and can be performed quickly and cheaply. The interpretation of the information in terms of parameters meaningfui to nydroiogy ana contaminant transport is problematic however. The focus of the current research is to generate a sufficient data base to allow meaningful statistical analysis of rock heterogeneity, and to use the statistics on pneumatic properties in conjunction with limited hydraulic measurements to characterize site hydrology and transport.
In addition to characterizing the hydraulicsofthesite,Ipointedoutthatthepneumaticpropertiesofthes{te would be necessary for the succes:ful modeling of radionuclide transport of C in the gas phase.
Yucca Mountain Although Yucca Mountain was not originally a Phase 2 test cases, Alan Flint of the U565 ano one of tne principal Investigators of Yucca Mountain site hydrogeology gave an interesting exposition on current findings at the site. USCS is currently installing a new unsaturated zone borehole UZ-7.
This borehole has been drilled at the bottom of a wash, and therefore is probably an area of high infiltration.
Furthermore, the soil in the wash aids infiltration which would otherwise run off.
The main purpose of this borehole is to measure water contents and zones of saturation in oraer to cetermine intiitration at the site. He noted that this has been a wet winter. A recent storm dropped 3 inches of rain, yet he observed no runoff, indicating that the water must have infiltrated some of the porous media at the site (although this is not necessarily an indication that the recharge reached great depths.) Most of the washes have had no runoff for 6 years until last year. Most of the fractures near the top of the site are filled with sediment and solidified minerals and could not carry large quantities of water, but there could be some fracture flow above the Tiva Canyon nonwelded unit.
Downhole cameras did not show any fracture flow at depth however. The borehole shows that some of the rock is near saturation near the top of Tiva Canyon nonwelded zone. Saturation measurements indicate nearly constant water content, but saturation varies because of varying porosity in the rock.
Isotopic measurements at depth have shown an apparent pulse of water that had infiltrated from 1984. Alan indicated that some of the results of tritium measurements in sample cores may have been the result of accidental contamination.
Experimenters noted an apparent correlation between tritium values and relative humidity. Apparently, luminous safety lights in the core storage laboratory contained 20 curies of tritium, some of which escaped into the atmosphere and 1
e i
i contaminated atmospheric water. Tritium measured in the cores declined significantly after removal of the offending signs.
Some of the samples of core from the new unsaturated borehole indicated the possible presence of zeolites, indicating rock that may have been wet for a time long enough to allow the alteration of volcanic glasses. He indicated that these samples had low permeability which could lead to conditions of perched water and flow in fractures.
One intriguing piece of new information was the possibility of a continuous low permeability layer at the base of Tiva Canyon. Hespeculatedthatthepresenceof this layer could serve as a barrier to gas flow and 'C transport at the site. He also noted that underground tunneling could disturb the unsaturated zone, compromising data on water contents, infiltration and pressure variations. He recommended that any underground tunneling for data collection be delayed until all surface based testing in the unsaturated zone is completed.
2 I
Alan also offered data from UZ-7 as a possible INTRAVAL experiment for unsaturated i
flow, in lieu of the canceled Apache Leap heater experiment. The objective of the INTRAVAL exercise would be to predict moisture content below the base of the Tiva Canyon member from available information in the unsaturated zone boreholes.
i The modelers would be provided matrix hydrogeological properties on rock core, and i
i would have to predict moisture contents. The key issues of the test case would
[
I be:
l 1.
What is the model scale necessary to accurately predict saturation?
I 2.
Can existing computer codes model infiltration over long distances vertically? and 1
i 3.
How reliable are the relationships for permeability and water retention properties?
Linda Lehman, representing the State of Nevada, presented an alternate conceptual model of groundwater flow at Yucca Mountain. Her contention is that there might 2
l be focused recharge along the flanks of the mountain where runoff from intense i
1 storms is diverted. Rainfall would saturate the thin soil and talus in Solitario I
l Canyon, and would flow along fractures hidden beneath the loose cover and soil.
j She wants DOE to measure water content and isotopes in the possible infiltration routes. She suggested that the Apache leap research site could serve as a useful j
analog to study this phenomenon of focused recharge because of the presence of tunnels and an effluent stream near that site.
I made a presentation on recent efforts at NRC to model the migration of "C and other volatile radionuclides in the gas phase at the succa Mountain site. My 1
presentation covered (1) releases from the waste form, (2) transport in the geosphere under the influence of thermal buoyancy caused by repository heat, (3) 1 the interaction between released "C and the non-radioactive carbon in the rock i
and water, and (4) release of volatile radionuclides such as Iodine and oxides of j
Tc and Se.
I noted the possible usefulness of pneumatic tests being conducted at the Apache leap site in determining crucial model parameters and validating the model s.
1 S
i
a 1
1 l
Alliaator Rivers Analoc Pro.iect (AP,AP)
The location of the meeting in Australia was an opportunity to showcase the Alligator Rivers Analog Project. The Alligator Rivers analog site is of interest to the United States HLW program because it represents the release and transport of uranium, thorium, plutonium and other minerals from a high-grade uranium are under oxidizing conditions; a situation similar in some respects to the processes that might occur at the Yucca Mauntain site. The ARAP site however is in a saturated medium while YMP would be predominantly unsaturated. Oxygenated water apparently infiltrates the site through fractures, and oxidizes the reduced ore as well as weathering the rock. The oxidized uranium in the ore dissolves along with other minerals held in the matrix, and is carried downgradient in a large plume away from the ore body. There 11 some evidence for migration of radionuclides in colloidal form, particularly insoluble actinides. Most radionuclides other than uranium seem to be associated with iron mineral phases. It may be possible that some radionuclides are transported as pseudo-colloids, attached to iron mineral particles, but this is not clear from the data.
This test case is one of the few in INTRAVAL where there has been an application of geochemical models and an attenpt to validate them with field data.
In this case, the alteration of the uranium ore by oxidation of the U0, wcathering of the 2
rocks and sorption play key roles in the release and transport of dissolved uranium and other minerals. George Birchard of NRC presented results from geochemical modeling being performed mainly at Johns Hopkins University. There were also about a half dozen presentations on geochemistry by several other national groups. Modelers included a number of processes in their geochemical l
models such as rock alteration, secondary mineralization, complexation, nucleus recoil and sorption onto altered rock.
In addition to the geochemical modeling, there is an extensive ongoing program of field data collection and modeling to l
describe the velocity and flux of ground water at +he site.
l l
Twin Lakes Studv l
I l
The Twin Lakes experiment in Canada is set up to measure the dispersion of tracers l
introduced to a phreatic aquifer. The medium is a relatively uniform sand and is heavily instrumented with monitoring devices to measure the concentration of radioactive and chemical tracers in the spreading plume. This site bears little l
resemblance to current U.S. sites for either high or low level waste, but l
nevertheless provides the modeler with a well-instrumented, relatively simple site on which to test their models.
Natalie Olague of Sandia National Laboratories presented a demonstration of model validation process for the case of the Twin Lakes study. Olague stated that one of the purposes of validation is to assure that models used in licensing a repository will be capable of accepting " good" sites and rejecting " bad" sites.
Such a goal does not necessarily imply that the models simulate exactly the performance of the site; e.g.,
in some cases, simple bounding estimates that nevertheless predict compliance would be just as good as more complicated models, i
so long as they did not wrongly accept a bad site.
She used the Twin Lakes data l
in a hypothetical probabilistic analysis of site performance, using the l
methodology likely to be employed in licensing of a high-level waste repository.
From the available hydrogeologic data, she estimated the distribution of model 5
g 4
L.
parameters for hydraulic conductivity, porosity, dispersivity and gradient and generated sample vectors using Latin Hypercube Sampling. These sample vectors were then used with a flow and transport model to predict maximum concentration at downgradient locations (which were taken as the measure of repository performance), and presented as cumulative distributions. The predicted results were then compared to the highest concentration actually measured at that location. The distribution of predicted concentration was then compared to the actual measurements. On the basis of this albeit simple demonstration, Olegue concluded that the procedures would have a very low probability of accepting a t
" bad" site, providing that conservative models are used with complete sampling ranges of the parameters. However, the same procedure has a high probability of I
rejecting a " good" site.
i Near-Field Modelino Forum Stig Wingefors of SKI announced the formation of a forum for near-field modeling.
The forum would study phenomena associated with the near field of high-level nuclear waste repositories, particularly coupled effects of heat on rock mechanics, chemistry and release of radionuclides from the waste form. The t
objective would be to determine the extent that performance assessment models can rely on simplified models while still ensuring that the key processes determining radionuclide releases were included. SKI has agreed to support one third of the funding for a pre-study which would establish topics to be covered in the near field forum.
I expressed interest because of my involvement with the source term modeling for Iterative Performance Assessment.
I do not know the schedule for t
this forum, but could pursue it if there is interest in NRC participation.
Conclusions i
INTRAVAL continues to be a useful forum for modelers to compare their results to experimental data and to discuss different approaches to modeling and validation.
It is part of an ongoing process whereby modelers build confidence and credibility in their approaches for the purpose of demonstrating the safety of nuclear waste repositories.
I was an active participant in INTRAVAL Phase I, for which I had primary responsibility for one of the validation experiments.
I have been unable to actively participate in modeling activities in Phase 2 because of other comitments at work. Nevertheless, taking part in INTRAVAL has, I believe, been worthwhile because it allows me to come into contact with some of the most advance modelers involved in problems of waste management.
The next meeting will be held in San Antonio Texas, November 9-13, 1992.
I have recently learned that the test case suggested by Alan Flint based on new data from Yucca Mountain has been approved, and will be modeled by 8 groups, including one from the Office of Research.
I believe we should follow this test case closely, and perhaps take advantage of the location to send several staff to San Antonio in i
November.
E h
,