© 2017 Chevron

Evaluating Oil Spill Risks through Stochastic and Deterministic

Modeling

Don Danmeier

Environmental Risk Scientist

San Ramon, March 2 2017

2© 2017 Chevron

Outline

• How modeling fits into an oil spill risk assessment

• Define stochastic and deterministic modeling

• Modeling considerations when assessing likelihood & consequence

• Comments on the common practice

3© 2017 Chevron

Part of a bigger picture

consultation with stakeholders

establish the context

risk assessment

risk treatment

monitor and review

Ris

k M

an

ag

em

en

t Pro

ce

ss

4© 2017 Chevron

Elements of the risk assessment

Hazard identification

HAZOP, PHA, What-If-Analysis

Likelihood analysis

historical data and extreme value theory

Consequence analysis

Predicted exposure & receptor sensitivity

Evaluate risk

Compare severity/prob. to tolerance criteria

5© 2017 Chevron

Hazard identification

event

blowout

penetration

depth

entering

formation

full

penetration

flow path

annulus

open hole

drill pipe

…

BOP status

fully open

partially

closed

…

release

location

sea surface

mudline

…

…

Hypothetical Event Tree Analysis (ETA) for exploratory drilling

6© 2017 Chevron

Likelihood analysis

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

10 100 1,000 10,000 100,000 1,000,000

pro

ba

bil

ity

tons

0.0E+00

2.0E-05

4.0E-05

6.0E-05

8.0E-05

1.0E-04

1.2E-04

pro

ba

bil

ity

flow rate (m3/d)low

high

Characterize volume & probability

For blowout example:

• Model flow for each ETA pathway

• Vol. based assumed intervention time

• Assign prob. based on historical data

Representative spill scenarios

• Evaluating all hazards is intractable

• Select representative scenarios for

consequence analysis

Consider complete risk profile

• Production & transportation risk

7© 2017 Chevron

Consequence analysis

Severity of spill determined by:

• Modeling fate & trajectory of spilled oil

• Comparing exposure to environmental sensitives

Importance of metocean inputs

• Drive trajectory…but also affect weathering

• Transient and seasonally variable

Day 8 Day 14 Day 20

spill scenario

oil properties

volume, duration, location

metocean conditions

8© 2017 Chevron

Establish & evaluate risk

Establishing risk

• Combine likelihood and consequence

– Establish risk level from one specific set of metocean parameters

– Do different metocean inputs affect severity?

Evaluating risk

• Compare risk level to tolerance criteria

• Are safeguards needed?

• Are risks as low as responsibly practical (ALARP)?

9© 2017 Chevron

Deterministic modeling

Simulate fate & transport under specific metocean forcing

10© 2017 Chevron

Stochastic modeling

Combine multiple runs under various metocean forcing

• Spatial variation of conditional probability

• Does not show extent of oiling

Histograms describe distribution of low/high exposure

• Rank runs by surface area oiled, shoreline oiled, …

• Shape of distribution indicates importance of metocean variabilty

+⋯ →+

trajectory #1 trajectory #2 probability plot

11© 2017 Chevron

A ‘representative’ or ‘worst-case’ run?

Some exposure metrics may not vary much….

but simplification may mask severity

12© 2017 Chevron

A ‘representative’ or ‘worst-case’ run?

…but some may vary considerably

13© 2017 Chevron

Conflicting guidance & practice

IPIECIA Good Practice Guide

The stochastic analyses should be paired with a most probable deterministic

case that can be utilized to support response planning

Australian Maritime Safety Authority

Stochastic modelling is the recommended method for determining the Zone of

Potential Impact…assess the likely effect of the spill scenarios for each resource

type identified within the ZPI .

BOEM risk assessments

• Simplified consequence analysis (no weathering, 30-day trajectory, no ecology)

• Summation of prob. x extent of oiling over stochastic set

Typical practice

Use P100 run as a (overly?) conservative assumption

14© 2017 Chevron

closing thoughts

• Don’t blindly adopt P100 run for consequence analysis

• Consider how severity varies with conditional probability

– Is comparison to tolerance criteria robust?

• When practical, establish risk by weighting consequence by conditional

probability

• If establishing risk only from central portion of stochastic set, confirm ability to

scale-up response

P50P95 P95or

15© 2017 Chevron

Questions?