risk-based performance assessments for long-term … · risk-based performance assessments for...

1

Sandia National LaboratoriesAlbuquerque, New Mexico

(505) [email protected]

Spectrum 2002 ConferenceAugust 4-8, Reno, NV

Clifford K. Ho, Bill W. Arnold, John R. Cochran, and Randal Y. Taira*

Risk-Based Performance Assessments for Long-Term Cover Systems:

Sensitivity and Uncertainty Analyses

*Pacific Northwest National Laboratory

Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under Contract DE-AC04-94AL85000.

2

Overview

• Introduction

• Approach

• Demonstration

• Impact and Summary

3

Introduction

• Engineered covers are needed to assist on-site isolation of subsurface contaminants in landfills, waste tanks, and other disposal sites

• Seven DOE Operations Offices have indicated major needs for long-term cover systems

– Albuquerque, Idaho, Nevada, Ohio, Rocky Flats, Richland, and Savannah River

• DOE 2006 Accelerated Cleanup Plan identifies use of surface barrier systems as a vital remedial option

4

Problem Statement

• Need rigorous method to evaluate long-term performance of covers with quantification of risk and uncertainty

• Need way to identify parameters and processes for performance verification and monitoring

• Need more rigorous methods to compare alternative designs and approaches

– Reduce Costs– Meet regulatory performance metrics and schedules

Develop risk-based performance-assessment approach for selection, design, modeling, and monitoring of long-term covers

5

Overview

• Introduction

• Approach

• Demonstration


6

Approach

Performance Assessment

Engineering Design/Deployment

Performance Monitoring and

Stewardship

Environmental Setting/Site

Characterization

Public and Stakeholder OutreachPublic and Stakeholder Outreach

Regulatory RequirementsRegulatory Requirements

7

Performance Assessment

Process

Scenario 2Scenario 1

Select Select Reject

Scenario 3

Develop and Screen Scenarios

KsatClimate Change Defects

Estimate Parameter Ranges and Uncertainty

ClimateEvapotranspirationSource TermVadose ZoneSaturated ZoneHuman Exposure

Develop Models

PA_process.ai

Perform Calculations00E000E000E000E000E000E000E000E000E000E000E000E0000D63768118>I<FFF8FFF8FFF80038003800380038003800380038003800380038003800380038003800380038003003800380038003800380038003800380038003800380038003800380038003800380038003800380038003800380038003800380038003800380038003800380038003800380038003800380038003800380038003800380

Uncertainty AnalysisSensitivity AnalysisAlternative Designs

Risk/Perfor manceCost/ScheduleRegulatory Compliance

Interpret Results

Monitoring Requirements

8

Overview

• Introduction

• Approach

• Demonstration


9

Monticello Disposal Site

• Permanent repository for 2.5 million cubic yards of low-level radioactive mill tailings and contaminated soils

• Sub-humid environment– Average annual temperature ~ 46 F

– Average annual precipitation ~ 15 in

• Repository design consists of– composite cover system

– double-liner system beneath tailings

– leachate collection and removal system and leak-detection system

10

2.6 m

12.5 m Source Term

1.5 cmComposite Liner

Vadose Zone4.6 m

Receptor Well

Confining Layer

Groundwater Flow

Cover

Aquifer

not to scale

Conceptual Model Development

Vertical Percolation/

Evapotranspiration Layer (soil)

Lateral Drainage Layer (sand)

Geomembrane Liner

(60 Mil HDPE)

Barrier Layer(clay)

Uranium Mill Tailings

66”

24”




Geomembrane Liner

(60 Mil HDPE)

Barrier Layer(clay)





Geomembrane Liner

(60 Mil HDPE)

Barrier Layer(clay)





Geomembrane Liner

(60 Mil HDPE)

Barrier Layer(clay)


66”

24”

PrecipitationPrecipitation

Lateral DrainageLateral Drainage

RunoffEvapotranspiration

RunoffEvapotranspiration

InfiltrationInfiltration

Percolation

Radon-Gas Flux

Groundwater Concentration Peak Dose

(HELP, RAECOM codes)Cover

(FRAMES/MEPAS codes)Total System

11

Integrated Performance-Assessment Model• Used FRAMES (PNNL) to integrate models for total-

system performance assessment– Drag-and-drop software platform

– Stochastic (Monte Carlo) analyses

• Source Term

• Vadose Zone

• Saturated Zone

• Human Exposure

http://mepas.pnl.gov:2080/earth/earth.htm

• Radon-Gas Flux (RAECOM)

• Percolation (HELP)

12

Performance-Assessment AnalysesExamples from Monticello Study

• Risk/Uncertainty Analysis

• Sensitivity Analysis

• Alternative Design Comparison

• Meet regulatory performance metrics

• Identify important parameters

• Reduce costs

13

Risk/Uncertainty Analysis• Multiple computer “realizations” are simulated

using stochastic inputs

0

2

4

6

8

10

12

-7 -6 -5 -4 -3 -2

Ksat_1.qpc

Co

un

t

log10

(Ksat

) (cm/s) for Layer 1

Stochastic Inputs(Latin Hypercube Sampling)

Multiple Computer Simulations(HELP, FRAMES)

• Ensemble of realizations yields probability distribution for “performance metric”

Distribution of Results(Present and Future Conditions)

10-6 10-5 10-4 10-3 10-2 10-1 100 1010

20

40

60

80

100present_future_cdf.qpc

Average Annual Percolation through Layer 4 (inches)

Cu

mu

lativ

e P

rob

abili

ty

sing

le d

eter

min

istic

resu

lt

present

future

14

Uncertainty Analysis of Percolation through the Cover

• 95% of all simulated water percolation through cover are less than 10-7 cm/s

• Future conditions yield poorer performance– Increased defects in

liner

– Cooler, wetter climate10-13 10-12 10-11 10-10 10-9 10-8 10-7 10-60

20

40

60

80

100present_future_cdf_cm_per_s.qpc

Percolation Flux through Cover (cm/s)

Cu

mu

lati

ve P

rob

abili

ty

present

future

40 C

FR

Par

t 264

.301

15

Uncertainty Analysis of Peak Dose

• Uncertainties can be quantified

• Can use risk-based evaluation of performance

0.0

0.2

0.4

0.6

0.8

1.0

1.0E-23 1.0E-18 1.0E-13 1.0E-08 1.0E-03 1.0E+02

Peak Cumulative Dose (mrem/yr)

Cu

mu

lati

ve P

rob

abili

ty

present

future

100

mre

m/y

r (D

OE

Ord

er 5

400.

5 II

1.a)

• Single deterministic result cannot quantify uncertainty or risk

16

Sensitivity Analysis

place

ment qu

ality

Ksat-

layer

1

evap

orative

zone

depth

precip

itation

multi

plier

tempe

rature

adjus

tmen

t

maxim

um lea

f-area

index

0

0.1

0.2

0.3

0.4

0.5

Sensitivity Analysis for Percolation through Cover

Present Climate

Future Climate

• Sensitivity analysis can be used to identify parameters and processes most important to performance

• Stepwise linear-regression analysis quantifies relative importance of parameters

∆R2

17

Aquifer

Porosity

Aquifer

Velocity

Distanc

e to w

ell

Aquifer

Radium

Kd

Source

Width

Source

Radium

Kd

Percola

tion Rate

Liner

Hydrau

lic Con

ductiv

ity

Vadose

Zone

Radium

Kd

0

0.2

0.4

0.6

0.8

1

Sensitivity Analysis

∆R2

Sensitivity Analysis for Peak Cumulative Dose

Shallow Aquifer (present climate)

Shallow Aquifer (future climate)

Deep Aquifer (future climate)

18

Alternative Design Comparison• Alternative designs can be compared using a probabilistic

risk-based approach

• “ET” cover design performed adequately for percolation, but it greatly exceeded radon-gas-flux metric

0

0.2

0.4

0.6

0.8

1

0.001 0.01 0.1 1 10 100 1000

Surface Radon-222 Gas Flux (pCi/m2-s)

Cu

mu

lativ

e P

rob

abili

ty Existing Cover Design

Alternative "ET" Cover

40 C

FR

192

.02(

a)

Present Climate

10-13 10-12 10-11 10-10 10-9 10-8 10-7 10-60

20

40

60

80

100ET_cover_comparison.qpc

Percolation Flux through Cover (cm/s)

Cum

ulat

ive

Pro

babi

lity

40 C

FR

Par

t 264

.301

Alternative "ET"Cover

Existing CoverDesign

Present Climate

Cum

ulat

ive

Pro

babi

lity

40 C

FR P

art 2

64.3

01

19

Overview

• Introduction

• Approach

• Demonstration


20

Impact and Summary

• Probabilistic PA approach provides rigorous regulatory-based evaluation of long-term cover systems

– Quantifies risk and uncertainty

• Sensitivity analyses identify processes and parameters most important to performance

– Site characterization– Performance verification and monitoring

• Provides risk-based comparison between alternative approaches and designs

– Reduce costs and meet schedule– Meet regulatory performance metrics

• Software and methods are ready for use– www.sandia.gov/eesector/gs/gh/SAND2001-3032.pdf

21

Backup Slides

22

Monticello Validation

0 20 40 60 80 100

Time (years)

0

200

400

600

Ann

ual P

reci

pita

tion

(mm

)

0.001

0.01

0.1

1

10

100

1000

10000

Ann

ual D

rain

age

Laye

r 2

(mm

)

HELP Simulation Results, Monticello Landfill Cover, Expected-Value Run

2001 Measured Precipitation (278 mm)

2001 Measured Drainage (0.17 mm)

23

Uncertainty Distributions for Input Parameters

• Future Climate Conditions– Precipitation

– Temperature

• Percolation through Cover– Evapotranspiration

– Hydrologic Properties

– Geomembrane Quality

• Transport Parameters– Sorption Coefficient

– Dispersivity

– Distance to Well

• Radon-Gas Flux Parameters– Diffusion Coefficient

– Moisture Content

0

2

4

6

8

10

12

-7 -6 -5 -4 -3 -2

Ksat_1.qpc

Cou

nt

log10

(Ksat

) (cm/s) for Layer 1

risk-based performance assessments for long-term … · risk-based performance assessments for...

Documents