Managing Risk at Northern Contaminated Sites: Differentiating - - PowerPoint PPT Presentation

RPIC Montreal 2016

Presented by: Francois Lauzon Prepared by: David Wilson Stantec Consulting Ltd. April 27, 2016

Managing Risk at Northern Contaminated Sites: Differentiating Assessment-derived Uncertainty from Risk Assessment Conservatism in Remedial Action Planning

Agenda

1 Uncertainty in Site and Risk Assessment 2 Role of Risk Assessment in Remedial Action Planning 3 The Uncertainty (Risk) Budget 4 Working with Risk 5 Case Example

Uncertainty in Site and Risk Assessment

“Risk is like fire: If controlled it will help you; if uncontrolled it will rise up and destroy you”

• Theodore Roosevelt

1

Definitions

Risk

• Treasury Board (2010b): “the effect of uncertainty on
• bjectives” or “the expression of likelihood and

impact of an event” [4]

Uncertainty

• Treasury Board (2010a): “the state, even partial, of

deficiency of information related to understanding

• r knowledge of an event, its consequence, or

likelihood” [3]

• Environment Canada (2012): the “state of having

limited knowledge where it is impossible to exactly describe an existing state or future outcome” [1]

• Types of uncertainty:
• Aleatory or Exogenous Uncertainty - statistical

variability and heterogeneity of the system (e.g., standard deviation of sample results)

• Epistemic Uncertainty - model and parameter

uncertainty (e.g., infiltration rate) “Deep Uncertainty”:

• uncertainty

about fundamental processes or assumptions [2] Often forgotten: also includes scenario and decision-rule uncertainty

Building the Conceptual Site Model

Phase CSM COCs* Terrestrial/ Land Use Climate/ Hydrology Hydro- geology Aquatic ESA I I(G) Historical land use/SARA/ terrestrial species Identify surface water (SW) bodies Water wells SARA/aquatic species ESA II C(G) Characterize soils (impacted and background) Characterize SW (impacted and background) Stratigraphy, groundwater (GW) Characterize SW and sediment ESA III D(G) Delineate soil impacts/ background Characterize SW transport Porosity/ Gradients/ GW Transport Delineate SW and sediment impacts HHERA C(SS) Define human health and eco exposures Consider seasonality and trends Characterize GW exposure Characterize SW/sediment exposure RAP D(SS) Risk mitigation (remedial options) / management measures * I(G) = Identify (generic guideline) C(G) = Characterize (generic guideline) D(G) = Delineate (generic guideline) C(SS) = Characterize (site-specific guideline) D(SS) = Delineate (site-specific guideline)

Uncertainty and the CSM

How should it be developed? [5] Dealing with uncertainty:

• Understand the data by doing exploratory analysis
• Identify and quantify uncertainty – revisions to the CSM are

expected

• Question assumptions
• Supplement data where needed, re-analyze, update the

CSM

[6]

SAP = sampling and analysis plan

Sources of Uncertainty

Uncertainty is generally identified within a phase but is often not carried to a subsequent phase (e.g., from ESA to ROA/RAP; from RA to ROA/RAP; from ROA to RAP; from ROA/RAP to Cost) [7]

SOURCE e.g.s: NORTHERN SITE ‘MAGNIFIERS’ Quantification Background conditions poorly defined Limited time/samples Parameter inclusion as COC: Y/N/Unk. CSM not fully developed Limited site-specific physical system data Some COCs not identified Limited historical data Parameter inclusion as COC: Y/N/Unk. Impacts not delineated Risk assessment exposure scenarios not ‘typical’ Impacted soil volume range: x m3 ± y m3 Decision criteria not defined Identifying/engaging stakeholders challenging Identify reliance of decisions on site

• bjectives

Characterizing Uncertainty

Aleatory or Exogenous Uncertainty

• Measures of statistical variability:
• standard deviation (parametric and non-parametric)/

standard error; variogram

• Bayesian probabilities

Epistemic Uncertainty

• Model sensitivity analysis
• Monte Carlo simulation

Deep Uncertainty

• Expert judgment
• Pairwise comparison of significance

Role of Risk Assessment in Remedial Action Planning

"Reality is that which, when you stop believing in it, doesn’t go away“

• Philip K. Dick

2

Remedial Action Planning

Objective:

• Reduce risk to acceptable

levels as effectively and efficiently as possible

Inputs:

• Site hazards posing

significant human health or ecological risks

Outputs:

• Recommended remedial/risk

management options that eliminate risk or reduce to acceptable levels

ESA RAP

Risk Assessment

Reduce Uncertainty of Using Generic Criteria:

• Screen out/in COCs
• Eliminate non-existing

pathways and receptors Define Site-specific Criteria:

• Based upon site COC

concentrations

• For site receptors

Dependencies:

• Exposure scenarios: land

use assumptions

• Pathways: CSM

COPCs & Media Pathways Receptors Risk resides here

Remedial Options Analysis

Hazard Toxicity Location Size/volume Stability Attributes: Uncertainties: ESA-CSM or RA: TRVs; Exposure duration Pathways/receptors + / - Trend RA Both ESA-CSM ESA-CSM

How do uncertainties affect the ROA? Significant or not: elimination, reduction or management RO Technology RO Cost/Duration RO Technology/Cost/Duration

The Uncertainty (Risk) Budget

“There are those who are so scrupulously afraid of doing wrong that they seldom venture to do anything”

• Vauvenargues

3

The ESA Risk Budget

Component Phase of Definition Nature of Uncertainty Management of Uncertainty Background Conditions II, III

• Nat. Var. – mean, S.D.

Epistemic - 95% UCLM; High natural Additional sampling CSM: Land use I, II, III Epistemic - generic criteria Deep - future intentions Conservatism (lowest criteria) CSM: Terrestrial Environment I, II (Preliminary) II, III (Mature)

• Nat. Var. - soil depth, grain

size, FOC, groundwater Epistemic - porosity Conservatism (e.g., coarse grained); additional sampling CSM: Aquatic Environment I, II (Preliminary) II, III (Mature)

• Nat. Var. – seasonality, TDS/

TSS, pH, species Epistemic – flows (e.g., 7Q10) Conservatism (e.g., max values); additional sampling COCS: Identification I, II Epistemic - COCs > Tier 1 criteria not identified Gap analysis; additional sampling COCs: Characteri- zation II

• Nat. Var. - mean, S.D.

Epistemic - generic vs. site- specific criteria Gap analysis; additional sampling COCs: Delineation III

• Nat. Var. - extent in x, y, z,

duration in t Epistemic - qualifiers (L, M, H) Gap analysis; additional sampling; contingency volume

The RA Risk Budget

Component Nature of Uncertainty Management of Uncertainty Inputs Background Conditions As per ESA Reference sites 95% UCLM CSM: Land use Human health exposure scenario; Applicable pathways Conservatism (chronic

• vs. acute exposure)

CSM: Terrestrial and aquatic environments Ecological exposure scenario; Applicable pathways Modeled vs. measured concentrations ESA COCs

• Nat. Var. – mean, S.D.,

extent in x, y, z, duration in t 95% UCLM = EPC OR

• max. conc.

Models Human health dose model TRV Conservatism (e.g. published uncertainty factors for TRVs) Ecological dose model Species; Exposure area and duration Conservatism (e.g. most sensitive species; max. concentrations)

ESA and RA Risk Budgets Compared

How much of the uncertainty is due to natural variability, to ESA-CSM models used, to RA models used, or to deeper issues? Find out: predict expected case, best case, and worst case values E.g. area of soil impacted with arsenic

COC – arsenic Volume – 100 m3 +/- 20 m3 Soil texture – undefined Sample results: # samples: 7 95% UCLM = N/a Log-norm. mean = 136 mg/kg

• Max. concentration =

367 mg/kg HH chronic TRVs: Ingestion 1.8 (mg/kg-d)-1 Inhalation 27 (mg/kg-d)-1 HH HQ – 1.6 (FN toddler) Eco subchronic TRVs: Dog 0.55 (mg/kg-d) UF: 3 Eco HQ – 10 (Masked shrew) ESA RA 20%

• 10%
• 10%
• 10%
• 20%
• 5%
• 10%
• 50%
• 10%

The Conservatism Cascade

ESA RA to Error Type*

Component Background Conditions CSM: Land use CSM: Terrestrial Environment CSM: Aquatic Environment COCS: Identifica- tion COCs: Characteri- zation COCs: Delineation Not enough samples Future use unknown GW pathway undefined Aquatic species undefined COCs not identified COCs not character

• ized

COCs not delineated Max value

• vs. mean

Most sensitive use GW pathway assumed Species assumed present COC list assumed complete Charac- terization assumed Delineation assumed

Impact

• n Risk

* Assuming a site is contaminated, Type 1 = false negative (i.e., incorrectly assuming site is clean, and Type 2 = false positive (i.e., incorrectly assuming site is contaminated)

1 2 2 2 1 1 1

Working with Risk

“Fail to plan, plan to fail”

4

Revisiting the CSM Post-RA

The Dynamic CSM

• Before beginning the ROA, update the CSM based upon the

RA results

Investments in Uncertainty Reduction

• Best approach for a given hazard still unclear (i.e.,

remediate vs. risk manage)?

• Is the information needed for ROA (application of selection

criteria) available?

Managing Residual Uncertainty (Risk)

• Define triggers and thresholds in the LTMP

Mitigating Uncertainty

Exogenous Uncertainty

• Variability is due to spatial, temporal, or individual randomness and

cannot be decreased by further data collection: it’s impact can

• nly be (and should be) managed

Epistemic and Deep Uncertainties

• Rank first by risk significance (e.g., hazardous vs. non-hazardous,

COC HQs), then by magnitude

• Work down ranking, and answer the questions:
• Does the uncertainty span a decision threshold? [e.g.,

remediate or risk manage; on-site or off-site disposal]

• Worth investing in reduction of uncertainty (mitigate), or better

to manage? [Cost-benefit analysis]

• What is the source of uncertainty? [Where should the

investment in uncertainty reduction be made?]

Reducing and Managing Uncertainty based upon Source

ESA (COCs):

• Additional sampling
• Advanced analysis

CSM:

• Additional characterization (e.g., pump tests)
• Additional analysis (e.g., time series analysis)
• Monitor inputs to remedial option (RO) or risk management

approach (RMA) and compare to design criteria RA:

• Development of short-term TRVs
• Use of measured vs. modeled concentrations in vegetation

and animal tissues Uncertainty in Technology Performance:

• Monitor outputs of RO or RMA and compare to

performance prediction

Case Study

“Everything that can be counted does not necessarily count; everything that counts cannot necessarily be counted”

• Albert Einstein

5

Case Study Example

Scenario: gold mining site decommissioned to the standard of the day in the late 1980’s: arsenic impacts in sediments and surface water adjacent to a tailings containment area (TCA) and flooded underground

• workings. Best option for the

TCA?

Assessment History: Phase Scope Ph I/II ESA

• historical

2010

• identification of APECs
• limited test pits/boreholes
• soil. tailings and WR samples

Ph III ESA

• characterization of AECs

2013

• additional boreholes
• SW, GW and sediment

samples

• background soil samples

HHERA

• regional background soil

2014 analysis; additional samples

• COPC screening-to-COCs
• calculation of soil SSTLs
• Suppl. ESA
• additional soil, SW, GW,

2015 sediment samples

• benthic organism toxicity

sampling and analysis 1988

Case Study Example cont’d

ESA Uncertainty Budget: Background [As] mg/kg:

• Geo. Mean/S.D. = 5.1±2.5
• 95% UCLM = 22.4

TCA [As] mg/kg:

• ‘09 Mean/S.D. = 150±109
• ’09+’12 Mean/S.D. = 131±101
• All Mean/S.D. = 136±124
• 95% UCLM = 340

CSM:

• Tier 1 crit. (Agric.) = 12 mg/kg
• Tier 1 crit. (Res.)* = 31 mg/kg
• SSRT = 69 mg/kg
• SW (ug/L) and sediment:
• SW Mean/S.D. (#sm 3) =

16.1±3.6 mg/kg

• Sed. Mean/S.D. (#sm 14) =

38±33 mg/kg

• Groundwater regime unknown
• Impacted area volume:
• ‘12 (#sm 84) = 42,000±15,000 m3
• ‘15 (#sm 106) = 44,800±9,400 m3

2012

* Carcinogenic

Case Study Example cont’d

ESA Uncertainty Analysis:

• is a decision threshold spanned?
• background exceeded? Y
• volume estimate:
• Tier 1 = 66,000±20,000 m3
• SSRTs = 44,800±9,400 m3
• remedial action required
• can options be evaluated? N

Cont’d:

• need to know if the aquatic

environment is being impacted

• additional assessment of

aquatic environment required RA Uncertainty Budget: undertake a similar process

• Uncertainty in inputs
• Uncertainty in models

Conclude options analysis:

• assessment shows impacts

above Tier 1 but RA shows non- toxic to benthic invertebrates

• Class A cost estimate

achievable within uncertainty? Y

2012

References

References 1. Environment Canada. 2012. Federal Contaminated Sites Action Plan (FCSAP) Environmental Risk Assessment Guidance. ISBN no. 978-1-100-22282-0. Cat. no. En14-19/1-2013E-PDF. 2. Committee on Decision Making Under Uncertainty (CDMU). 2013. Environmental Decisions in the Face of Uncertainty. Board on Population Health and Public Health Practice. National Academy of Sciences. ISBN 978-0-309-13034-9. 3.

• TBS. 2010a. Framework for the Management of Risk.

4.

• TBS. 2010b. Guide to Integrated Risk Management.

5. Maheux, P., Lauzon, F., Wilson, D., Sundaram, S., Bouchard, M. 2012. Developing a Good Conceptual Model for Federal Contaminated Sites – Common Shortfalls and Data Needs. Pres. at the RPIC Federal Contaminated Sites National Workshop, Toronto, Ontario. 6. Evolving Conceptual Site Models (CSMs) in Real-time for Cost Effective Projects, Kira P. Lynch, US Army Corps Seattle District. 7. Wilson, D. 2015. Advancements in Managing Uncertainty in Remedial Options Analysis and Remedial Action Plan Development for Northern

• Sites. Pres. at the RPIC Federal Contaminated Sites Regional Workshop,

Edmonton, Alberta.

Questions?

David Wilson, M.A.Sc., P.Eng. Senior Associate Stantec Ottawa (613) 738-6091 david.wilson@stantec.com