Audit the experiment before fitting the model.
Track specimens, loading sequences, prior damage, repairs, channels, units, processing choices, source documents, and unresolved gaps.
See the NIST data audit ↗
Industries / Seismic engineering research
Turn test evidence into clearer decisions.
Clear Anvil is exploring software for reproducible seismic test data, nonstructural component evidence, and decision-focused test planning.
Exploratory research
This is a research direction, not a deployed engineering product or a claim of validated building performance.
The research direction
A software layer between experiments and design choices.
Existing tools can simulate structural response, quantify uncertainty, and estimate damage or recovery. A remaining practical problem is keeping physical evidence, model assumptions, and the next research decision connected.
Make every result traceable to the raw record.
Compare commanded and measured table motion, document transformations, preserve versioned outputs, and regenerate tables and figures.
Read the test-data workflow ↗Ask which test would change a defined choice.
Prototype a test-selection policy around an engineer-approved component or retrofit decision, then compare it with conventional schedules and simpler baselines.
The research contribution must be demonstrated. Bayesian and adaptive experiment selection already have prior art. A credible advance would need independent physical validation, explicit damage history, strong baselines, and measured decision performance—not a new dashboard around an old idea.
A bounded first program
Begin with one nonstructural component family.
Ceilings, partitions, façades, equipment, and building services can interrupt use after an earthquake even when the main structure remains standing. A focused benchmark can test the data method before anyone makes a building-level recovery claim.
Reconstruct the evidence
Pin a public dataset version and trace records back to specimens, campaigns, source reports, loading order, damage, and repairs.
Define the decision
With an engineer, specify the component family, two approved alternatives, test cost, evidence threshold, and conditions for “insufficient evidence.”
Compare policies
Evaluate conventional schedules, random or space-filling selection, information gain, and decision-focused selection under the same budget.
Validate prospectively
Freeze the method before a partner laboratory runs new specimens, then report calibration, wrong-choice rate, test count, and failures.
Clear Anvil's role
Research software and transparent evidence workflows.
Clear Anvil can build the information and evaluation layer around a public benchmark or collaborator-supplied research dataset. Engineering objectives and physical-test authority stay with qualified partners.
- Dataset and provenance auditSource-linked inventories, missing-field reports, specimen-history reconstruction, and reproducible snapshot counts.
- Analysis pipelineVersioned channel dictionaries, quality checks, processing manifests, calculated metrics, and regenerated figures.
- Research benchmarkGrouped data splits, declared baselines, calibration checks, decision metrics, and an explicit out-of-scope region.
- Review interfaceA tool that shows sources, assumptions, uncertainties, exceptions, and the evidence behind a proposed next experiment.
Start with a public benchmark.
Bring a component family, research question, or public dataset. We can map the evidence and define what a useful software prototype would have to prove.
Architecture & functional recovery
Connect the analysis to how the building must work.
Earthquake-resistant architecture involves early coordination among architectural configuration, the structural system, the envelope, interiors, services, access, and the owner's recovery goals. Software can compare explicit alternatives; it cannot replace the responsible design team.
Questions to raise before detailed structural analysis.
Review site assumptions, plan and section, load-path continuity, façades, interiors, services, and recovery objectives while alternatives are still open.
Read the guide ↗Damage is only one part of reopening.
Building use can depend on nonstructural systems, utilities, access, inspections, repairs, staffing, and occupancy-specific requirements.
See NIST's research program ↗Risk reduction is different from earthquake prediction.
The research concerns test evidence and performance questions. It does not predict the date, location, or magnitude of an earthquake.
Read the USGS explanation ↗
Research boundaries
Keep the accountable experts in the loop.
Any future prototype must be evaluated within a declared domain and under the procedures of the engineering and laboratory teams responsible for the work.
- No structural certificationThe software does not approve a building, connection, retrofit, test plan, or code-compliance decision.
- No autonomous laboratory controlIt does not command shake tables or bypass facility safety, instrumentation, acceptance, or test procedures.
- No unsupported recovery promiseComponent damage evidence alone does not establish when a building can be occupied or return to function.
- No earthquake predictionHazard probabilities, forecasts, early warning, and prediction are different concepts; this work does not claim prediction.
START WITH THE EVIDENCEWhat research decision
What research decision
needs a clearer record?
Tell us about the public benchmark, test-data workflow, or component question. We can define a bounded software and evidence project with the right engineering collaborators.
jamil@clearanvil.com