Vulnerability#

Vulnerability is a key component in risk assessments, defining the extent to which infrastructure assets are susceptible to damage when exposed to natural hazards. It is influenced by factors such as material strength, structural design, maintenance level, and hazard intensity.

The United Nations Office for Disaster Risk Reduction (UNDRR) defines vulnerability as:

“The characteristics and circumstances of a community, system, or asset that make it susceptible to the damaging effects of a hazard.”

Despite its importance, accurately determining the vulnerability of infrastructure remains a challenge. Many assets lack detailed vulnerability studies, and existing assessments are often based on assumptions, expert judgment, or data from similar regions. The validation of vulnerability relationships requires extensive empirical data, which is often unavailable, making it necessary to rely on literature-based estimates.

To support infrastructure risk assessments, we use the Nirandjan et al. (2024) database as a starting point when local vulnerability or fragility information is unavailable. This dataset provides a collection of standardized vulnerability and fragility curves across different hazard types and infrastructure systems.

Types of Vulnerability Curves#

Two key approaches exist for quantifying infrastructure damage in risk modeling:

1. Vulnerability Curves#

Vulnerability curves describe the expected level of damage to an infrastructure asset based on hazard intensity. These curves express damage as a fraction of the total asset value, ranging from 0 (no damage) to 1 (total destruction).

Vulnerability curves are commonly used for:

  • Flooding (damage as a function of water depth)

  • Windstorms (damage as a function of wind speed)

  • Landslides (damage as a function of ground displacement)

These curves are particularly useful for estimating economic losses, as they provide a direct way to compute expected reconstruction costs based on the level of hazard exposure.

2. Fragility Curves#

Fragility curves describe the probability that an infrastructure asset will exceed a specific damage state at a given hazard intensity. Unlike vulnerability curves, which predict direct damage ratios, fragility curves provide probabilistic estimates for different damage thresholds, such as:

  • Slight Damage (minor repairs needed)

  • Moderate Damage (significant functional impairment)

  • Extensive Damage (structural failure but not complete collapse)

  • Complete Damage (total destruction)

Fragility curves are most commonly used for:

  • Earthquakes (damage probability as a function of peak ground acceleration)

  • Windstorms (damage probability as a function of wind speed)

  • Landslides (probability of collapse under soil movement)

The Nirandjan et al. (2024) Vulnerability Database#

A comprehensive open-access database developed by Nirandjan et al. (2024) provides over 1,250 vulnerability and fragility curves for critical infrastructure across multiple hazard types. The database was compiled through a systematic review of scientific literature and engineering reports, ensuring that the collected curves represent a broad range of hazard intensities and asset types.

The dataset includes:

  • Hazards covered: Flooding, earthquakes, windstorms, and landslides.

  • Infrastructure types: Roads, power plants, telecommunications, buildings, and more.

  • Different structural classifications: Variations based on material type, size, and function.

  • Geographical applicability: The curves originate from multiple regions, with a significant portion of data sourced from high-income countries. However, the dataset also includes estimates for low- and middle-income regions where empirical data is scarce.

One of the challenges addressed in the database is the lack of standardized vulnerability curves for certain asset types, such as telecommunications infrastructure. Where empirical data is missing, the database provides modeled estimates based on engineering assumptions and analogues from similar structures.

The dataset is publicly available and can be accessed via Zenodo:
📥 Download the dataset
The page provides an overview of dataset structure and metadata. New versions will be uploaded on the Zenodo database page.

For more details, refer to the published paper: Nirandjan et al. (2024).

Using the Database in Notebooks#

The section below provides an example on how to read the vulnerablity data. The code below provides an example on reading flood vulnerability information for different road types.

import pandas as pd

vulnerability_path = "https://zenodo.org/records/10203846/files/Table_D2_Multi-Hazard_Fragility_and_Vulnerability_Curves_V1.0.0.xlsx?download=1"
vul_df = pd.read_excel(vulnerability_path, sheet_name='F_Vuln_Depth')

# Display the first two rows with all available options
with pd.option_context('display.max_rows', None):
    display(vul_df.iloc[:2, :].T)

And select a curve to use for each different subtype we are analyzing:

sub_types = ['motorway', 'trunk', 'primary', 'secondary', 'tertiary', 'other']
selected_curves = dict(zip(sub_types, ['F7.5', 'F7.5', 'F7.5', 'F7.7', 'F7.7', 'F7.9']))

damage_curves = vul_df[['ID number'] + list(selected_curves.values())]
damage_curves = damage_curves.iloc[4:125, :]
damage_curves.set_index('ID number', inplace=True)
damage_curves.index = damage_curves.index.rename('Depth')  
damage_curves = damage_curves.astype(float)
damage_curves.columns = sub_types

Make sure we set the index of the damage curves (the inundation depth) in the same metric as the hazard data (e.g., meters or centimeters). In our case, the hazard data is in centimeters, so we have to multiply our inundation values of the damage curves by 100:

damage_curves.index = damage_curves.index * 100

Locally Tailored Vulnerability Curves#

While the Nirandjan et al. (2024) database provides a comprehensive and standardized set of vulnerability and fragility curves, it does not account for local asset characteristics, which can significantly influence infrastructure vulnerability.

In real-world assessments, infrastructure assets differ based on their construction materials, maintenance levels, age, and deterioration state. A recently build highway may have significantly different flood resilience compared to an aged highway in a region with poor maintenance. Similarly, building vulnerability to earthquakes can vary greatly based on local construction codes and retrofitting practices.

Incorporating Local Information#

An ideal vulnerability assessment integrates:

  • Local engineering data (e.g., material specifications, construction techniques)

  • Maintenance records (e.g., frequency of inspections, past damage history)

  • Asset deterioration assessments (e.g., signs of wear, structural weaknesses)

  • Expert validation (consulting engineers, policymakers, and local stakeholders to refine assumptions)

Validating and Adjusting Vulnerability Curves#

  • Empirical Validation: Comparing predicted damages from vulnerability curves with real damage data from past events.

  • Expert Calibration: Adjusting damage ratios based on engineering expert judgment.

  • Field Surveys: Collecting direct information on the structural integrity of critical assets.

  • Scenario Testing: Running sensitivity analyses to see how different vulnerability assumptions impact risk outcomes.

By integrating locally validated vulnerability curves, risk assessments become more representative of real-world conditions, leading to better-informed decision-making for infrastructure planning and disaster resilience.