PerilX Physical Risk Assessment Methodology

How to read this document

This document explains PerilX's technical foundation without turning it into a detailed engineering manual. Reading the “Earthquake assessment approach”, “Interpreting scores” and “Appropriate use” sections first is sufficient.

PerilX assessment framework

PerilX turns location-based hazard data and declared building information into an explainable preliminary assessment output within the same model. The framework considers each hazard according to its own data scale and uncertainty; it does not produce a definitive strength or damage result for an individual building.

How earthquake hazard is considered

Earthquake hazard is represented by ground-motion indicators from probabilistic earthquake-hazard models such as EFEHR/ESHM20. PGA is used as an explanatory indicator in the location-hazard view, while earthquake assessment related to structural behaviour is based on a spectral-demand approach associated with building height and ground effects.

Ground effects and indicators used

Ground effects are interpreted through VS30 and the local-ground approach in TBDY 2018. This indicator is not a parcel-specific ground investigation; local fill, liquefaction, foundation systems and site-specific geotechnical conditions require separate review.

Structural typology and construction period

Structural type, construction period and building height are the main differentiating variables for earthquake vulnerability. Reinforced-concrete frame, reinforced-concrete shear wall, tunnel formwork, precast reinforced concrete, steel, masonry and timber classes are interpreted within the scope supported by the source model.

Turkey-specific vulnerability approach

The PerilX earthquake approach references the ELER methodology and earthquake-engineering literature on Turkey's building stock. It does not silently create a new fragility curve for structure-type and height combinations that have no direct equivalent in the source taxonomy; this is stated as a data and model limitation.

Damage-state probabilities

In earthquake analysis, damage states are considered as a probability distribution ranging from no or negligible effect to progressively more severe damage levels. The distribution explains expected relative vulnerability for the selected inputs, rather than the exact damage level that will occur in the field.

Vulnerability Class

Vulnerability Class is a relative decision-support indicator derived by interpreting the hazard signal and structural characteristics together. It is not a statement confirming regulatory performance, an official damage determination, an insurance price or market value.

Design Demand Ratio

The Design Demand Ratio is an explanatory indicator describing conceptually how earthquake demand increases or decreases together with structural class and ground effects. It is provided for decision support and does not replace a formal code check or engineering performance analysis.

Portfolio comparison approach

Unlike an individual-building report, portfolio analytics compares groups of assets using limited CSV inputs. Its purpose is to make damage-state distributions, expected economic loss and the need for prioritisation visible at portfolio scale.

Uncertainties and data limitations

Data resolution, user declarations, the accuracy of permit and project information, current interventions and local environmental conditions affect results. Earthquake, flood and landslide layers do not have the same resolution or analytical depth.

Correct interpretation of results

PerilX outputs indicate which building and location factors merit closer review. A lower result does not mean that a building is safe, and a higher result does not mean that damage will certainly occur.

Work the assessment does not replace

PerilX does not replace an on-site engineering inspection, ground investigation, structural-project review, performance analysis, official appraisal, damage assessment or insurance pricing.

Methodology version and last update

This public methodology summary explains the perilx-methodology-v0.2.0 approach. The last documentation update is 20 August 2026; content responsibility is assigned to the PerilX technical methodology team.

Abbreviations and key terms

Abbreviations are written out on first use. The short form is used thereafter; hover over an abbreviation to view its explanation again.

EFEHR

European Facilities for Earthquake Hazard and Risk: a network of institutional and research resources developing earthquake-hazard and risk models for the Euro-Mediterranean region. Official description

ESHM20

European Seismic Hazard Model 2020: an updated earthquake-hazard model for the Euro-Mediterranean region. Model overview

SA

Spectral Acceleration: represents the approximate acceleration demand of a building with a particular vibration period under earthquake motion.

PGA

Peak Ground Acceleration: the highest expected acceleration level at the ground during an earthquake.

TBDY 2018

Turkish Building Earthquake Code 2018: Turkey's code framework for building design and assessment under earthquake effects. AFAD page

VS30

The average shear-wave velocity in the upper 30 metres of ground; an indicator used to interpret ground stiffness and local earthquake effects. USGS explanation

JRC

European Commission Joint Research Centre: the source institution for flood and depth-damage studies used in PerilX's flood assessment. JRC flood studies

NASA

National Aeronautics and Space Administration: provider of global landslide-susceptibility and landslide-monitoring data products. NASA Earthdata

FEMA / Hazus

Hazus, developed by the Federal Emergency Management Agency, is a standard risk-assessment approach for estimating the potential physical and economic effects of hazards including earthquakes and floods. Hazus overview

g

The unit of gravitational acceleration. Earthquake-engineering ground-motion values are often expressed as multiples of gravitational acceleration.

Fragility model

A statistical model representing the probabilities of exceeding different damage states at a given ground-motion level.

Return period

The long-term average recurrence interval of a particular hazard level. For example, “476 years” does not mean that an event occurs exactly once every 476 years.

Deterministic calculation

A calculation approach that produces the same result for the same inputs and does not include random sampling.

Monte Carlo sampling

A calculation method that produces many random scenarios to represent uncertainty. The current portfolio-loss output does not use this method.

RC, IM and EDP

Technical abbreviations used in the references: RC denotes reinforced concrete, IM an intensity measure for ground motion, and EDP an engineering demand parameter.

AFAD

Disaster and Emergency Management Presidency: Turkey's competent public institution in the field of earthquake regulations and disaster management.

ELER

Earthquake Loss Estimation Routine: an earthquake-loss assessment methodology and software developed by Boğaziçi University/Kandilli Earthquake Engineering Department. PerilX uses its Level 2 spectral-capacity, performance-point and fragility approach as a scientific reference. Boğaziçi University ELER page

CSM

Capacity Spectrum Method: an approximate analysis method that identifies a performance point by comparing earthquake demand with a building's nonlinear capacity spectrum in the same spectral acceleration–spectral displacement plane.

Sd / Sdp

Sd denotes spectral displacement; Sdp is the spectral-displacement component of the performance point found from the capacity-demand relationship. It is the main demand variable in Engine B fragility calculation.

DS0–DS4

Mutually exclusive damage states used by the PerilX earthquake model, ranging from no/negligible effect to four progressively more severe damage levels. The user score is the product-layer interpretation of these probabilities; it is not an economic loss ratio.

AAL and MDR

MDR (Mean Damage Ratio) converts damage probabilities at a hazard level into an expected loss ratio using economic consequence coefficients. AAL (Average Annual Loss) is the annual average-loss indicator obtained by integrating those loss ratios with annual exceedance rates along the hazard curve.

YYBM

Approximate Building Unit Cost: the 2026 Ministry/High Technical Council cost reference used with building-use/class information and gross construction area when a user has not supplied a replacement cost for the portfolio economic AAL view.

Engine B / Engine A

Engine B is the primary earthquake-vulnerability engine, using an ELER-compatible spectral-capacity and CSM approach for supported typologies. Engine A is a transparent fallback/QA engine based directly on SA(T) fragility and used only within its defined scope.

1. Purpose and scope

In brief: PerilX evaluates the natural-hazard conditions at a location together with basic building information to create an earthquake-focused physical-risk profile.

The analysis considers location-based earthquake-hazard data, local ground conditions, construction period, structural system and number of storeys together. Its purpose is to present a building's relative condition under potential earthquake effects quickly and explainably.

In the current version, the individual-building risk profile is earthquake-focused. Flood and landslide indicators are treated as complementary location information in the natural-hazard summary.

2. Transparency and model limits

In brief: PerilX's scientific basis and principal calculation approach are disclosed; internal calibration coefficients and software-level implementation details are not published.

PerilX does not run the ELER software directly as a “black box”. It draws methodologically on ELER's Level 2 approach, which addresses structural damage through spectral capacity, the performance point and fragility curves.

Current EFEHR/ESHM20 data are used to define earthquake hazard, while TBDY 2018 and VS30 information are used to assess local ground effects. A PerilX result should therefore not be regarded as a direct “ELER output”, but as a PerilX assessment that draws on the ELER methodology and uses current hazard and ground data.

The purpose of this documentation is to present the scientific and technical background of the result clearly and to help users interpret the output in the correct context.

3. Core principles

1. Hazard and structural behaviour are evaluated together: two buildings at the same location may produce different results when their structural systems and construction periods differ.
2. Earthquake demand is not reduced to a single PGA value: spectral-acceleration data representing structural behaviour at different vibration periods are used.
3. Structural-system distinctions are preserved: reinforced-concrete frame, reinforced concrete with shear walls, tunnel formwork, precast, steel, masonry and timber buildings are not treated as one model family.
4. Model limits remain explicit: no new fragility model is silently created for building classes unsupported by scientific sources.
5. Results are presented for preliminary assessment: PerilX supports review and prioritisation rather than replacing detailed engineering analysis.

4. Inputs used

In brief: Results are based on the combined evaluation of earthquake-hazard and ground data read automatically from the location and building information supplied by the user.

Information obtained from the location

• Earthquake-hazard and spectral-acceleration data sourced from EFEHR/ESHM20.
• A PGA-based general hazard view and multi-period SA(T) values.
• TBDY 2018 local-ground coefficients and VS30-based ground information.
• JRC flood and NASA landslide layers used in the natural-hazard summary.

Information supplied by the user

• Year or period of construction.
• Structural system and total number of storeys.
• For reinforced-concrete buildings, the distinction between frame + infill wall, frame + shear wall and tunnel formwork.
• Other structural-system types, including precast reinforced concrete, steel, masonry and timber.
• Complementary building characteristics such as engineering services, maintenance condition, overhangs, building geometry, soft storey and adjacency.

Correct structural-system selection is particularly important for post-2020 reinforced-concrete buildings. Reinforced-concrete frames and reinforced-concrete shear-wall systems may behave differently under earthquake effects, so this distinction is directly reflected in the calculation.

5. Calculation flow

In brief: PerilX first establishes the earthquake demand at the location, then evaluates it together with the capacity and fragility characteristics of the building's structural system.

Primary assessment approach

For supported building classes, the primary earthquake assessment is performed through Engine B, which uses a spectral-capacity approach compatible with the ELER methodology. Engine A, based on a simpler direct spectral-acceleration approach, is used as a fallback model only in explicitly defined, limited cases.

The outputs of the two models are not mixed within the same calculation. This keeps it traceable which method is activated under each building and data condition.

6. Earthquake assessment approach

In brief: The earthquake assessment considers spectral hazard at the location, local ground conditions and the behaviour of the building's structural system together.

Hazard and ground effects

The analysis uses European Seismic Hazard Model 2020 (ESHM20) data provided by EFEHR. The individual-building assessment is based on multi-period spectral demand corresponding to a 476-year return period. This does not mean that an earthquake of the same magnitude occurs exactly once every 476 years; it describes a long-term probabilistic hazard level.

Local ground effects are considered using VS30 information and the ground approach defined in the Turkish Building Earthquake Code 2018. PGA is used to display general location hazard, while structural behaviour is assessed through spectral-acceleration values at different vibration periods.

How is the ELER methodology used?

PerilX's structural-vulnerability approach draws on the ELER — Earthquake Loss Estimation Routine methodology developed by Boğaziçi University/Kandilli Earthquake Engineering Department. In ELER's Level 2 approach, the building is represented by a spectral-capacity curve, and an approximate performance point is identified from the relationship between earthquake demand and that capacity curve.

The performance point is associated with fragility curves for different damage levels to obtain damage-state probabilities. PerilX combines this methodological framework with current EFEHR hazard data and the TBDY 2018 ground approach.

ELER is not run directly as a software component within PerilX. The published ELER methodology, building taxonomy, capacity model and fragility approach are scientific references for the PerilX calculation engine.

Building classification

The structural system, construction period and number of storeys are evaluated together to select the appropriate behaviour family. Reinforced-concrete frames and reinforced-concrete shear-wall systems are treated separately. Tunnel formwork, precast reinforced concrete, steel, masonry and timber buildings are evaluated within their respective building families and the limits supported by the source models.

Post-2020 reinforced-concrete buildings

In the ELER source classification, buildings constructed from 2001 onward are grouped in the same principal period. PerilX treats post-2020 reinforced-concrete buildings through a separate, versioned adaptation representing the more recent construction period in which the Turkish Building Earthquake Code 2018 became established in practice. The aim is to represent modern reinforced-concrete building stock more realistically without departing from the source ELER model.

Because shear walls have become more common in modern reinforced-concrete buildings, “reinforced-concrete frame + shear wall” is highlighted in the interface. Users may nevertheless declare a frame-only system. This product choice is not a regulatory-compliance determination and does not assume that every post-2020 building must contain shear walls.

Effect of building characteristics

Engineering services, maintenance condition, soft storey, overhangs, building geometry and adjacency do not directly change earthquake hazard. They are evaluated through a controlled adjustment layer applied to the model of structural vulnerability. This accounts for structural characteristics while preventing the same factor from being counted more than once.

From damage probabilities to a score

The calculation produces probabilities for different damage states. To make buildings easier to compare, these probabilities are converted into a relative earthquake-vulnerability indicator on a 0–100 scale.

This indicator must not be interpreted as an economic loss ratio, an exact damage percentage or a probability of collapse. A more detailed economic assessment is performed separately in the portfolio module using an AAL approach.

Applicability limits

• The ELER source taxonomy is used for defined building classes between 1 and 19 storeys; buildings of 20 storeys or more require a separate high-rise approach.
• Source-model support for precast reinforced concrete is limited to low-rise buildings.
• No new fragility model is created for building-class and height combinations without a direct source equivalent.
• A PerilX result does not replace detailed structural-project review, a nonlinear building model or performance analysis.

7. Flood hazard context and passive risk module

In brief: Flood is not included in the individual-building risk profile in this version; it may be shown as location context in the natural-hazard summary.

The methodological description below explains the operating framework of the flood-risk module retained in the codebase and available for future reactivation.

The flood module uses raw water depth read from the available JRC-based flood-scenario layer for the selected location. Rather than first reducing depth to a broad class, it relates it to the occupancy-sensitive depth-damage function developed by Huizinga and colleagues. This function represents how expected economic impact changes as water depth increases.

Residential and commercial occupancies are treated as different behaviour families. The floor of the independent unit, basement presence, basement use and ground-floor use alter the expected economic effect of flooding.

A flood score does not mean structural-system collapse. It is a relative physical-risk indicator focused on outcomes such as interruption of use, contents loss and effects at lower elevations. No score is produced when no meaningful flood effect can be read.

8. Landslide hazard context and passive risk module

In brief: Landslide is not included in the individual-building risk profile in this version; it may be shown as location context in the natural-hazard summary.

The methodological description below explains the operating framework of the landslide-risk module retained in the codebase and available for future reactivation.

The landslide module uses the NASA-sourced global landslide-susceptibility layer as its base signal. This layer is intended for broad-scale screening and must not be interpreted as a definitive parcel-scale slope-stability analysis or ground investigation.

The actual value in the data cell intersected by the queried point is used. If the point does not intersect the layer, no automatic fallback value is taken from a neighbouring cell; “no meaningful landslide indicator found” is recorded and no score is produced.

The susceptibility value is evaluated together with retaining walls, ground improvement and observed cracking/settlement. Because slope effects are already represented in the principal susceptibility layer, they are not applied again as a strong secondary hazard multiplier in the individual calculation.

The landslide-susceptibility value is not read directly as a percentage economic loss. It is converted into a relative risk indicator through a bounded impact curve and used to support prioritisation for site investigation.

9. Portfolio analysis approach

In brief: Portfolio analysis considers earthquake levels with different probabilities of occurrence together, rather than one earthquake scenario, and calculates Average Annual Loss (AAL).

The portfolio module uses EFEHR hazard curves; hazard levels at different spectral periods are evaluated on a common annual-exceedance axis. Structural vulnerability is calculated at each hazard level and damage probabilities are converted into an economic damage ratio.

The resulting AAL ratio therefore represents the long-term annual average economic effect of different earthquake levels, not the loss from a single 476-year earthquake scenario.

Economic valuation

Where the user supplies a replacement cost directly, that value is used. Otherwise, when sufficient building-use and area information is available, replacement cost is estimated using the 2026 Approximate Building Unit Costs (YYBM) issued by the Ministry of Environment, Urbanisation and Climate Change.

This value is not market or sale value. Land value, contents loss, business interruption, rental loss and off-building infrastructure are not included in the basic AAL calculation.

10. Interpreting scores

In brief: The 0–100 earthquake score is a relative prioritisation indicator that considers both the hazard environment and modelled structural vulnerability.

• The need for more detailed technical review increases as the score rises.
• The score is derived from calculated damage-state probabilities.
• It is not economic AAL, an exact damage percentage or a probability of structural performance.
• Results close to a class boundary should not be treated as fixed, definitive categories.
• “No data”, “outside model scope” and “score could not be produced” remain distinct from a 0/100 result.

11. Data completeness level

In brief: This indicator describes how much of the required user information has been completed; it does not measure model accuracy or certainty of the result.

Data Completeness Level is an operational indicator calculated from completion of user-supplied fields and unknown values in critical inputs. For example, 100% completeness means that all required fields have been completed; it does not mean that the result is 100% accurate.

12. Appropriate use framework

Appropriate uses

• Performing a rapid preliminary physical-risk assessment of a building.
• Establishing review priorities across multiple assets.
• Identifying areas requiring attention before fieldwork or expert review.
• Establishing a shared and explainable physical-risk language within an organisation.

Decisions for which it should not be used alone

• An official engineering report or structural-project decision.
• A ground investigation or parcel-scale slope-stability assessment.
• A real-estate appraisal or final legal/administrative decision.
• Insurance pricing, policy acceptance or a definitive economic-loss decision.

13. Limitations and uncertainties

• The spatial resolution of earthquake-hazard data and the representativeness of ground information may affect the result.
• Classification uncertainty may increase when building information is based on user declarations.
• Incorrectly distinguishing an RC frame from an RC shear-wall system may materially change the model result.
• The post-2020 reinforced-concrete approach is a PerilX adaptation built on existing source models; it is not an empirical fragility curve derived from direct field observations for every building.
• No artificial model is generated for combinations of building type and height unsupported by the source model.
• PerilX does not replace detailed engineering analysis, structural-performance assessment or regulatory-compliance review.

14. Calibration, validation and versioning

The PerilX methodology is developed through versioned releases. Hazard data, building classifications, fragility models and calculation components are reviewed regularly using scientific publications, source-model comparisons and controlled tests.

ELER's published technical documentation and validation examples are among the principal references for the spectral-capacity approach. PerilX also applies quality controls to ensure that identical inputs produce identical results, relationships between building classes remain consistent and invalid calculations are not interpreted as “no damage”.

Model and data versions are retained in result records so that the method and dataset used to produce an analysis can subsequently be traced.

15. Sources

The publications and institutional sources below are among the principal references underpinning PerilX's approach to earthquake hazard, spectral capacity, structural vulnerability and economic loss.

These references demonstrate the methodological foundations of PerilX. Active PerilX model versions and product calibrations are versioned and validated separately.

Closing note

PerilX is not a “one-click final decision” tool. It is a decision-support layer that accelerates expert assessment, facilitates site and portfolio prioritisation, and enables different users to work with a shared physical-risk language.