For engineering teamsBespoke engagement

Custom Fragility Engineering.

Asset-specific vulnerability studies, delivered when archetype-level curves are not enough. Structural mechanics, regional calibration, and a parameter lineage you can defend in front of a regulator.

Engagements typically run 4–8 weeks. Scope is agreed in writing after a discovery call: number of archetypes, depth of structural modeling and the local data the client can share. Fully auditable deliverables.

When you need it

The standard Xpectral platform covers 95% of insured portfolios. The remaining 5% needs engineering.

Trophy assets

Tall towers, signature buildings, single-asset exposures large enough that an archetype average — calibrated on a thousand buildings you do not own — is the wrong answer.

Critical infrastructure

Hospitals, ports, refineries, data centers, transmission backbones. Loss of function is not a financial event; it is a continuity event with regulatory implications.

Regional typologies

Caribbean masonry, Soviet-era panel housing, Andean informal construction, MENA precast — building stock that off-the-shelf archetype catalogs were not built to represent.

Post-event repricing

An earthquake just hit your portfolio. You need to recalibrate vulnerability against observed damage before the next renewal cycle, with full documentation.

What's in the engagement

Six deliverables. Fixed scope. No ambiguity about what arrives at the end.

01

Non-linear dynamic analysis

Equivalent single-degree-of-freedom oscillators, calibrated to the structural properties of your asset, exercised against ground motion records selected for the local tectonic regime — subduction, intraplate, near-source — rather than a generic catalog.

02

Capacity-spectrum derivation

An explicit capacity curve per archetype: yield and ultimate displacement, lateral capacity, drawn either from analytical pushover or from your own structural model when you provide one. No black box between mechanics and the curve.

03

Regional θ, β calibration

Lognormal vulnerability parameters specific to country, typology and construction era — calibrated against local post-event observations: DesInventar damage records, USGS PAGER fatality data, your own claims history when you share it.

04

Decomposition of uncertainty

β reported as two components: aleatory (record-to-record variability, material scatter) versus epistemic (classification, model selection, code enforcement). You see exactly which uncertainty more data will reduce — and which will not.

05

Validation against history

Each derived curve is cross-checked against historical earthquakes in the same country: surveyed damage, claims data, official shake intensity maps. Any divergence is documented before delivery.

06

Auditable deliverables

Per-asset fragility curves in NRML XML — the format consumed by the reference hazard engines — plus a methodology paper with full references, an uncertainty dashboard, and a parameter lineage table that traces every number back to its source.

Methodology, in plain English

The same workflow large modeling vendors run internally — the difference is that we hand you the lineage in writing.

Why off-the-shelf archetypes are not the answer

Off-the-shelf vulnerability libraries are useful, peer-reviewed foundations. Most were built around a US construction census from the 1990s, with a finite number of archetypes. A high-rise tower in Manila, a refinery in Algeria, or a 1960s panel block in the Caucasus does not cleanly map to any of them. Use the published θ, β values for those assets and your tail risk is wrong — usually in the direction your reinsurer will not appreciate.

We start from structural mechanics

An equivalent oscillator is calibrated to how your asset actually deforms — from drawings, from a structural model you provide, or from a pushover analysis we run. We exercise it with ground motion records selected for the site's tectonic environment, not a generic catalog. We watch yield, we watch collapse, and we record the spectral acceleration that drives each damage state.

We anchor mechanics to observation

Pure mechanics produces a candidate curve. We then anchor it to what buildings actually did in real earthquakes: DesInventar damage records, USGS PAGER fatality estimates, the client's own claims data when shared. The lognormal parameters — the median capacity θ and dispersion β — stop being textbook estimates and start reflecting observed performance in the same country.

Why this beats proprietary black boxes

Legacy CAT modeling vendors deliver a number for the Caribbean, the Andes, and a hundred other markets. They keep the parameter values proprietary — you accept them on trust. We deliver the same level of regional calibration, and you get the lineage in writing: which structural calculation, which observation set, which modeling decision produced each θ and each β. Auditable to your reinsurer, defensible to your regulator.

“θ and β are not magic numbers. They are the output of structural mechanics plus observed damage. Let us compute them for your asset, in your country, against your loss history.”

Carlos Caramés Molero — Founder & Partner, Dynamis

Deliverables

Four artifacts at the end of the engagement. Designed to land directly in the hands of your cat modeling team, your auditor, and your reinsurer.

NRML XML

Per-asset fragility curves

Lognormal curves for each archetype across damage states, ready to drop into your modeling stack — or directly into the reference hazard engine for portfolio runs.

Methodology paper

Technical documentation

A signed engineering paper documenting record selection, capacity derivation, calibration sources, and every assumption — referenced and reviewable.

Uncertainty dashboard

Aleatory vs epistemic split

An interactive view of where the dispersion comes from, so your portfolio team knows which uncertainty is irreducible and which a future data investment would close.

Parameter lineage

Source-to-curve traceability

Every θ, every β, every assumption traced back to its source: drawing, calculation, observation set, or modeling decision. The artifact your audit trail needs.

DynamisDynamis

Built by

Dynamis

Custom fragility studies are authored by Dynamis — engineering consultants in earthquake engineering and structural dynamics. The same team that engineered Panama Canal bridges, the high-speed rail between Dubai and Abu Dhabi, and nuclear-grade facilities now works on your portfolio.

15+
Years of critical-infrastructure engineering
100+
Engineering projects delivered
20+
Countries across 5 continents
Audit-ready
Every parameter, full lineage

Schedule a discovery call

Tell us what you are insuring. We come back within 48 hours with scoping options and an indicative timeline.