using flexible loss modelling platforms to develop unique ... arango.pdf · ©aon limited trading...
TRANSCRIPT
Prepared by Cristina Arango and Martin Kadlec, Impact Forecasting
RAA Cat Risk Management 2020
Using Flexible Loss Modelling Platforms to Develop Unique Terrorism Risk Management Tools
Impact Forecasting | Proprietary & Confidential 2
Agenda
Section 1 IF’s Hi-Res Terrorism Blast Modelling using Computational Fluid Dynamics (CFD)
Section 2 CFD Blast model multiplatform implementation using Oasis format
Section 3 Key takeaways
Impact Forecasting | Proprietary & Confidential 3
IF’s Hi-Res Terrorism Blast Modelling Using Computational Fluid Dynamics (CFD)
Impact Forecasting | Proprietary & Confidential 4
Real Blast Testing
Karagozian & Case, Inc. https://www.youtube.com/watch?v=QRx1_nuP2Jw
Impact Forecasting | Proprietary & Confidential 5
IF’s Terror Blast Modelling Using CFD
Property damage and workers’
comp injuries & fatalities
.
Façade, structural damage
and collapse evaluation
.
CFD blast simulationsDB with individual building
information.
VulnerabilityHazardExposure Loss
Impact Forecasting | Proprietary & Confidential 6
Blast Description
3D Computational Fluid Dynamics Simulation timeline
(overpressure)
Pressure-time history can be effectively represented by two parameters:
▪ Over-Pressure (OP)
▪ Pressure Impulse (PI)
Impact Forecasting | Proprietary & Confidential 7
CFD Blast Modelling Approach
3D urban environment Blast simulations Outputs
▪ WALAIR software is used to perform full
3D blast simulations
▪ Uses CFD techniques – able to capture
channelling, shielding and reflection of
incident shockwave
▪ Simulation inputs:
o domain geometry
o blast location
o mass of explosive
o type of explosive
▪ Runs on Graphical Processing Units
making it faster than other comparable
software
▪ Model output includes blast metrics in 3D at
predefined model resolution (2m across the
surface of building and domain):
o Peak overpressure
o Impulse
o Pressure time history
▪ 3D urban environment is generated from
buildings’ footprints and heights datasets
▪ Structure and façade info are defined at
individual building level (IF’s own
exposure DB)
WALAIR has been validated using experimental data. Results have been published in research papers such as Nicholson A., Stirling C. G., Misselbrook N. K. – Fast running CFD code
using GPGPUs for high fidelity airblast calculations.
Impact Forecasting | Proprietary & Confidential 8
CFD Blast Modelling Framework
PI
OP
DR
▪ Pressure Impulse (PI) curves describe the damage caused by the blast
▪ Point-wise damage ratio (DR) from PI curves allows to evaluate:
(1) severity of façade damage and
(2) structural damage level reached and possible building collapse
PI curves for specific structural and façade types and
damage levels
Impact Forecasting | Proprietary & Confidential 9
▪ Extend of collapse depends on:
(1) Point-wise damage: blast (external forces)
(2) Column size/strength: building (internal forces)
▪ Workers in collapsed zones suffer mostly fatal injuries
▪ Façade damage: casualty distribution as a function of DR
Example: Workers Comp. Damage & Collapse Framework
Compute damage ratio DR @ 2m
Estimate # of failed columns
Estimate total collapse extent
Combine collapse and façade damage
Colla
pse
exte
nt
eva
lua
tio
n
> 5 columns
Localised
substantial
damage
Destruction
Perimeter above column collapse threshold
Extended collapsed area
Façade only damage
Transition zone/collapse buffer
Up to 2 columns
Casualty
Dis
trib
ution
Damage Ratio
Fatal
PPtotal
PPmajor
PPminor
TempTotal
MedOnly
Façade
Damage only
range
Impact Forecasting | Proprietary & Confidential 10
Case Study: CFD Terrorism Blast Model for New York
IF’s own exposure DB defines 3D cityscape & individual building data
Property module: Buildings, Contents and BI
Worker’s comp module: Worker’s comp. injuries/fatalities
Blast location uncertainty: densely placed targets at every
street’s intersection and midpoint
Explosive size uncertainty: 2, 5 and 10 ton blast
Workers location within a building are randomized
CFD modelling to capture complex blast wave-urban
environment interaction
Blast metrics (pressure and impulse) at 2m resolution on
building’s surface
Considers buildings’ structural resistance to blast and collapse
potential:
▪ Façade only and structural damage
▪ Different collapse criteria for property and workers’
compensation modelling
Damage &
Collapse
CFD blast
simulations
Uncertainties
Exposure DB
and model
scope
Impact Forecasting | Proprietary & Confidential 11
Model Implementation Using Oasis Format
Impact Forecasting | Proprietary & Confidential 12
Oasis Model Format in ELEMENTS
ELEMENTS allows Oasis formatted models to be hosted:
▪ Hazard and vulnerability in Oasis format
▪ Exposure in ELEMENTS format
▪ ELEMENTS loss calculation engine
Enhanced Oasis model format in ELEMENTS for:
▪ High resolution flood models
▪ Secondary perils
Greater model interoperability:
▪ 3rd party models in Oasis format can be implemented in
ELEMENTS
▪ IF models in Oasis format can be deployed in Oasis LMF,
ModEx or client’s Oasis environment
Implementation in ELEMENTS
ELEMENTS
production ready
platform
Oasis hazard /
vuln files ELEMENTS Oasis
database
ELEMENTS
Client
Model
Development
Model is ready for
operation on production
ELEMENTS platform
Development
following Oasis
model format
Files transferred
into Oasis DB
Impact Forecasting | Proprietary & Confidential 13
Oasis Model Format in ELEMENTS
Hazard
Information
Vulnerability
Information
Each model has its own database
▪ All hazard and vulnerability information are stored in tables
▪ Table design is the same for both probabilistic and deterministic/scenario models
EventInfoEventID
Frequency
EventFootprintCount
DemandSurge
Region
Intensity
….
EventFootprintEventID
MastertableID
IntensityBinIndex
Probability
….
HazardIntensityBinIntensityBinIndex
BinFrom
BinTo
….
VulnerabilityDamageFunctionID
IntensityBinIndex
DamageBinIndex
Probability
….
DamageBinDamageBinIndex
BinFrom
BinTo
….
Impact Forecasting | Proprietary & Confidential 14
Oasis Model Format Implementation – Model Developer’s View
Key questions to answer
Can the format handle the model complexity?
Hazard
How to measure the hazard?
Is the measure relevant for all
coverages?
How to discretize it?
Vulnerability
How to discretise damage ratio
into bins?
What options/modifiers should be
considered?
Loss
What is the definition of the risk –
SI vs. # workers, pay-outs
Impact Forecasting | Proprietary & Confidential 15
Property
Building
▪ Uses a single damage ratio
per building and blast
scenario
▪ Straightforward
implementation in Oasis
format
Property BI
▪ Uses building damage ratio
and distance from the blast
▪ Moderate complexity of
implementation in Oasis
format
Workers
Comp.
▪ A function of building
damage, workers position
within building and injury
class
▪ More complex
implementation utilising
Oasis format flexibility
Oasis Model Format Flexibility – 3 Examples
Impact Forecasting | Proprietary & Confidential 16
Example 1: Property Building Damage – Coupled Hazard & Vulnerability
▪ Hazard quantity is building level damage ratio (BDR):
– Construction and façade type modifiers embedded in BDR – taken
from model building DB
▪ Vulnerability defined as 1:1 mapping with hazard (customisable)
▪ No building properties are required in portfolio for modelling
▪ Oasis format flexibility supports this scheme
▪ Deterministic event set
▪ Thousands of blasts locations
with different explosive sizes
▪ 3D fields for 2 hazard metrics:
overpressure OP and pressure
impulse PI
▪ Vulnerability/PI curves in terms
of 2 hazard metrics: OP and PI
▪ Construction and façade type
specific
Event set Hazard Vulnerability
Implementation:
Coupled hazard & vulnerability
▪ Risk lat/lons mapped to model
building ID
▪ Embedded building properties
DB (structure and façade)
Exposure
Impact Forecasting | Proprietary & Confidential 17
Evacuated zone
▪ Measure: distance from blast center
▪ Discretized into M classes:
– <100m
– <200m
…
– >1km
Damaged offices, shops, etc.
▪ Measure: building level damage ratio BDR
▪ Discretized into N classes:
– no damage
– broken glazing
…
– total collapse
Example 2: Property Business Interruption BI – Hazard
All combinations are possible (almost), 2-dimensional hazard (matrix N x M)
Oasis format requires “flat” hazard table. Implementation solution: matrix de-pivoted into vector of hazard
bins representing each combination
Property
BI
Direct
Property
BI
Indirect
Impact Forecasting | Proprietary & Confidential 18
Example 2: Property Business Interruption – Vulnerability
Damage bins
Damage uncertainty
Vulnerability matrix
− Example: Building with broken glazing 300 m from blast center: closed for 3 months, DR 0.25
− Building, content and business interruption are implemented as one model with multiple
coverages to allow proper application of financial conditions
▪ Each hazard case (bin) may be assigned to more bins with certain probability
▪ Currently implemented as deterministic
▪ Each bin may have a range of damage ratio to sample from
▪ Implemented with fixed damage ratio 0/12, 1/12, … 12/12
▪ 13 damage bins representing BI for 0, 1, … 12 months
▪ Sum insured assumed to represent payout for a year lasting BI
Impact Forecasting | Proprietary & Confidential 19
Risk value defined by
sum insured (SI)
vs
Risk value defined by
number of workers
Loss (GU) = DR x SI
vs
Loss (GU) = ∑payouts
Damage ratio (DR)
relative to SI
vs
Payout per worker -
depending on injury class
Example 3: Workers Compensation
Oasis does not directly support workers compensation exposure and payout calculations.
Differences between property and workers compensation modelling
Exposure Vulnerability Loss
Impact Forecasting | Proprietary & Confidential 20
Risk value defined by
sum insured (SI)
vs
Worker defined by
“virtual” value (VV)
Loss (GU) = DR x SI
vs
Loss (GU) = PR x VV
Damage ratio (DR)
relative to SI
vs
Payout ratio (PR)
relative to VV
Example 3: Workers Compensation – Redefining the Approach
We can now use standard Oasis model format to calculate workers comp. losses
Redefinition of workers compensation modelling approach
Exposure Vulnerability Loss
Impact Forecasting | Proprietary & Confidential 21
▪ Collapsed zone
▪ Façade damage
▪ Measure: pointwise damage
ratio
▪ Defines injury profile for each
damage ratio
▪ Applied on all building points
provides distribution of injury
classes for whole building
▪ Represents probabilistic
distribution of injury classes for
single worker in given building
and scenario
Example 3: Workers Compensation – Implementation
Building hazard profile Casualty matrix Oasis hazard table
Casualty D
istr
ibutio
n
Damage Ratio
Fatal
PPtotal
PPmajor
PPminor
TempTotal
MedOnly
+ Final injury distribution includes location uncertainty
− Exposure (total workers per building) has to be disaggregated - else workers’ location
is random but correlated
Hazard (building hazard profile) and vulnerability (casualty matrix) are again coupled to create Oasis hazard table
Impact Forecasting | Proprietary & Confidential 22
Example 3: Workers Comp. Model in Oasis – Key Implementation Points
▪ Probability to fall into one of
the injury classes for a
worker randomly positioned
within building
▪ Exposure has to be
disaggregated
▪ Worker represented by virtual
value (VV)
▪ Damage ratio represented by
payout ratio (payout relative
to worker virtual value VV)
▪ Sum of payouts – each injury
class represented by payout
ratio multiplied by worker
virtual value (VV)
Implementation not straightforward, but successful!
Workers comp. model in
Oasis model format
Impact Forecasting | Proprietary & Confidential 23
IF CFD Terrorism Blast Model in Oasis format – Inputs and Outputs
▪ Risk location (latitude & longitude)
▪ Sums insured
▪ Policy conditions
Property Workers Compensation
Model
built-in
features
Exposure
data
Loss results
▪ Risk location (latitude & longitude)
▪ Head count/number of workers per building/floor
▪ Policy conditions
▪ Risk latitude & longitude mapped to closest building ID using IF’s exposure DB
▪ Building properties/modifiers assigned from IF’s exposure DB
▪ If workers/floor are unknown, total workers/building are dis-aggregated based on floor footprint area
▪ Occupancy rate (default is 100%)
▪ Pay-outs per injury class (can be customised)
▪ 2, 5 and 10 ton blast scenarios
▪ Property loss per blast
▪ Property loss per blast & building ID
▪ Other breakouts (e.g. coverage)
▪ Workers’ comp loss per blast
▪ Workers’ comp loss per blast & building ID
▪ Other breakouts (e.g. per building ID & injury
class)
Impact Forecasting | Proprietary & Confidential 24
Key Takeaways
Impact Forecasting | Proprietary & Confidential 25
Key Takeaways
Can “non-standard” models be implemented in Oasis format?
Implementation not always straightforward, but possible
Oasis files
structure
Exposure needs
Hazard and
vulnerability
definitions
Solution
Sta
rt
Fin
ish
Making a decision on
what to implement as
“hazard” and
“vulnerability”
Assessing if Oasis
format/file structure
can accommodate
modelling needs
Assessing if exposure
characteristics can be
represented
Develop a technical
solution for model
implementation
Testing
Testing results
throughout model
development cycle
Impact Forecasting | Proprietary & Confidential 26
Contacts
Martin Kadlec
Impact Forecasting Prague
Cristina Arango
Impact Forecasting London
Impact Forecasting | Proprietary & Confidential 27
Legal Disclaimer
©Aon Limited trading as Aon Benfield (for itself and on behalf of each subsidiary company of Aon Corporation) (“Aon Benfield”) reserves all rights to the content of this report (“Report”). This Report is for distribution to Aon Benfield
and the organisation to which it was originally delivered only. Copies may be made by that organisation for its own internal purposes but this Report may not be distributed in whole or in part to any third party without both (i) the
prior written consent of Aon Benfield. and (ii) the third party having first signed a “recipient of report” letter in a form acceptable to Aon Benfield. Aon Benfield cannot accept any liability to any third party to whom this Report is
disclosed, whether disclosed in compliance with the preceding sentence of otherwise.
To the extent this Report expresses any recommendation or assessment on any aspect of risk, the recipient acknowledges that any such recommendation or assessment is an expression of Aon Benfield’s opinion only, and is not a
statement of fact. Any decision to rely on any such recommendation or assessment of risk is entirely the responsibility of the recipient. Aon Benfield will not in any event be responsible for any losses that may be incurred by any
party as a result of any reliance placed on any such opinion. The recipient acknowledges that this Report does not replace the need for the recipient to undertake its own assessment.
The recipient acknowledges that in preparing this Report Aon Benfield may have based analysis on data provided by the recipient and/or from third party sources. This data may have been subjected to mathematical and/or
empirical analysis and modelling. Aon Benfield has not verified, and accepts no responsibility for, the accuracy or completeness of any such data. In addition, the recipient acknowledges that any form of mathematical and/or
empirical analysis and modelling (including that used in the preparation of this Report) may produce results which differ from actual events or losses.
The Aon Benfield analysis has been undertaken from the perspective of a reinsurance broker. Consequently this Report does not constitute an opinion of reserving levels or accounting treatment. This Report does not constitute
any form of legal, accounting, taxation, regulatory or actuarial advice.
Limitations of Catastrophe Models
This report includes information that is output from catastrophe models of Impact Forecasting, LLC (IF). The information from the models is provided by Aon Benfield Services, Inc. (Aon Benfield) under the terms of its license
agreements with IF. The results in this report from IF are the products of the exposures modelled, the financial assumptions made concerning deductibles and limits, and the risk models that project the pounds of damage that may
be caused by defined catastrophe perils. Aon Benfield recommends that the results from these models in this report not be relied upon in isolation when making decisions that may affect the underwriting appetite, rate adequacy or
solvency of the company. The IF models are based on scientific data, mathematical and empirical models, and the experience of engineering, geological and meteorological experts. Calibration of the models using actual loss
experience is based on very sparse data, and material inaccuracies in these models are possible. The loss probabilities generated by the models are not predictive of future hurricanes, other windstorms, or earthquakes or other
natural catastrophes, but provide estimates of the magnitude of losses that may occur in the event of such natural catastrophes. Aon Benfield makes no warranty about the accuracy of the IF models and has made no attempt to
independently verify them. Aon Benfield will not be liable for any special, indirect or consequential damages, including, without limitation, losses or damages arising from or related to any use of or decisions based upon data
developed using the models of IF.
Additional Limitations of Impact Forecasting, LLC
The results listed in this report are based on engineering / scientific analysis and data, information provided by the client, and mathematical and empirical models. The accuracy of the results depends on the uncertainty associated
with each of these areas. In particular, as with any model, actual losses may differ from the results of simulations. It is only possible to provide plausible results based on complete and accurate information provided by the client and
other reputable data sources. Furthermore, this information may only be used for the business application specified by Impact Forecasting, LLC and for no other purpose. It may not be used to support development of or calibration
of a product or service offering that competes with Impact Forecasting, LLC. The information in this report may not be used as a part of or as a source for any insurance rate filing documentation.
THIS INFORMATION IS PROVIDED “AS IS” AND IMPACT FORECASTING, LLC HAS NOT MADE AND DOES NOT MAKE ANY WARRANTY OF ANY KIND WHATSOEVER, EXPRESS OR IMPLIED, WITH RESPECT TO THIS
REPORT; AND ALL WARRANTIES INCLUDING WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE HEREBY DISCLAIMED BY IMPACT FORECASTING, LLC. IMPACT
FORECASTING, LLC WILL NOT BE LIABLE TO ANYONE WITH RESPECT TO ANY DAMAGES, LOSS OR CLAIM WHATSOEVER, NO MATTER HOW OCCASIONED, IN CONNECTION WITH THE PREPARATION OR USE
OF THIS REPORT.
Disclaimer