Damage and Loss Model Library (DLML)

A curated, open-source, version-controlled library of standardized damage and loss models for natural-hazard risk assessment.

About the library

The Damage and Loss Model Library (DLML) is a project from the NHERI SimCenter that addresses a persistent gap in natural-hazards engineering: the lack of a centralized, standardized, and easy-to-use home for damage and loss models. It gathers performance models from established sources — the component library from FEMA P-58 and building- and infrastructure-level models from FEMA Hazus — alongside peer-reviewed models from the research community, and publishes them in one consistent, machine-readable form.

The library is built on three principles:

• Version-controlled. Every model belongs to a released snapshot, so an analysis can be tied permanently to a specific version of the data — supporting reproducible research and defensible professional practice.
• Standardized and machine-readable. Models are converted to a single tabular schema (CSV parameters plus JSON metadata), ready for computational workflows with no manual transcription.
• Documented provenance. Every model carries rich metadata and a citation to its source, so users can trace it to its origin and credit the original developers.

Hazard coverage: seismic, hurricane, and flood — spanning building components, whole buildings, and infrastructure network assets.

Data Organization

The data is a tree of datasets. A dataset is a leaf folder identified by its path (hazard/asset_type/resolution/methodology), for example seismic/building/component/FEMA P-58 2nd Edition. Each dataset provides one or more collections — fragility, consequence_repair, and/or loss_repair — and each collection is a table of models, one per row (typically one per component/building/asset). Parameters are in <collection>.csv; the matching <collection>.json holds the metadata (descriptions, limit and damage state details, citations).

How to use the DLML

Each of these is a different route to the same curated data.

Install the Python package

pip install simcenter-dlml

The data ships inside the package (no runtime download), and you can read it with the dlml API:

import dlml

dlml.list_datasets()
# ['flood/building/portfolio/Hazus v6.1',
#  'seismic/building/component/FEMA P-58 2nd Edition', ...]

fema_p58 = "seismic/building/component/FEMA P-58 2nd Edition"
dlml.available_collections(fema_p58)      # ['consequence_repair', 'fragility']
fragility = dlml.get_fragility(fema_p58)  # a pandas DataFrame of model parameters
metadata = dlml.get_metadata(fema_p58, "fragility")

# Validate asset features against a dataset's input schema:
seismic = "seismic/building/portfolio/Hazus v6.1"
dlml.validate_asset(seismic, {"StructureType": "W1", "DesignLevel": "Pre-Code"})
# []  -> valid
dlml.validate_asset(seismic, {"StructureType": "W1"})
# ["$: 'DesignLevel' is a required property"]

The core package needs only Python 3.9+, pandas, numpy, and jsonschema.

Explore the models with the DLML Explorer

The DLML Explorer is a web app for discovering, visualizing, and assembling models: search and filter the library, inspect interactive fragility curves and consequence functions, add components to a selection, and download a project-ready bundle. The easiest way to try it is the hosted version:

Open the DLML Explorer »

You can also run it locally — for example, to explore alongside your own private models — by installing the explorer extra:

pip install "simcenter-dlml[explorer]"
dlml explorer

(The explorer extra targets Python 3.10+.) And you can inspect the library straight from the command line:

dlml list                 # every dataset and its collections
dlml info <dataset-id>    # a dataset's collections and model counts

Use with Pelicun and the SimCenter tools

The library is the data backbone for Pelicun, SimCenter's open-source damage-and-loss engine, and is available automatically through the PBE and R2D desktop applications that run Pelicun under the hood.

Pelicun 3.10 and later install this library from PyPI and need no further setup. Pelicun 3.9 and earlier read the model data from this repository instead; see Release policy if you are still on one of those versions.

Work with the raw data

You can also read the CSV/JSON files directly from a clone of this repository:

git clone https://github.com/NHERI-SimCenter/DamageAndLossModelLibrary.git

A legacy documentation website with auto-generated model pages also remains available; it will be revised over time to complement the DLML Explorer.

Release policy

Releases are published to PyPI as simcenter-dlml and tagged here, with matching entries in CHANGELOG.md. Version numbers use two segments (3.0, 3.1, ...), and every model in a release is a fixed snapshot, so an analysis can cite the exact data it used.

Migrating from Pelicun 3.9 and earlier

Pelicun 3.9 and earlier downloaded the model data from this repository's latest GitHub release when you imported them. Since v3.0 the data lives under src/dlml/data/, a layout those versions cannot read, so they now fail with a download error.

We recommend upgrading. Pelicun 3.10 and later depend on simcenter-dlml directly and download nothing:

pip install --upgrade pelicun

If you need to stay on Pelicun 3.9 or earlier, pin the model data to the last release those versions can read:

pelicun dlml update v2.1.0

The v2.1.0 release stays available for this purpose.

Contributing

We welcome new and improved models from researchers and practitioners. A formal contributor's guide is in progress; for now, the best way to start is to open a GitHub Issue to discuss the model data you'd like to add. Our goal is a simple pull-request workflow for data that conforms to the schema.

License and acknowledgments

This library is distributed under the BSD 3-Clause license. See LICENSE for details.

We gratefully acknowledge the generous in-kind contribution from Degenkolb Engineers — and in particular Tshajlij Lee and Hannah Thompson — to the development of the DLML Explorer.

This material is based upon work supported by the U.S. National Science Foundation under Grants No. 1612843 and No. 2131111. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the U.S. National Science Foundation.

Contact

For questions or support, please open an issue on the GitHub repository.

Adam Zsarnóczay, NHERI SimCenter, Stanford University — adamzs@stanford.edu