Building a Platform for Aerospace Test Data

Summary

Testing avionics for the Space Launch System (SLS) Booster Obsolescence Life Extension (BOLE) program generates enormous volumes of timeseries data. That data is only valuable if engineers can find it, analyze it, and report findings effectively.

As part of DMC’s automated test system for Artemis boosters, DMC developed DART (Data Analysis and Reporting Toolkit), an end-to-end platform that streamlines the entire test data workflow. DART ingests test data from the BOLE Electrical Production Support Equipment (the “EPSE”), organizes it in the cloud, and puts powerful analysis tools directly in engineers’ hands.

The platform manages data throughout its lifecycle:

• DART CLI, a cross-site edge ingestion tool, uploads test data from test stands.
• DART Data Store, a cloud data store built on AWS GovCloud organizes it in an export-controls compliant environment.
• DART Desktop, a desktop application, lets engineers visualize, analyze, and report on results.

These components work together to provide an effortless path from test execution to engineering insight, helping teams access critical data and make informed decisions throughout the BOLE program.

Customer Benefits

Solution

Avionics test campaigns produce high-rate timeseries data across many channels, tests, and (in this case) multiple sites. Without a shared system, that data tends to accumulate on separate file systems and often becomes practically inaccessible. Locating a specific run often depends on fleeting institutional knowledge, and analysis is done with one-off scripts that are difficult to review or repeat. For a program with export-controlled data and sometimes unreliable connectivity, off-the-shelf cloud analytics tools weren’t a viable option.

The client needed a system that could:

1. Ingest test data reliably from test stands at multiple sites.
2. Store it in a compliant, well-organized, and cost-efficient way.
3. Give engineers flexible analysis tools that don’t require programming.
4. Produce consistent reports in PDF and XLSX format.
5. Work in both cloud-connected and fully disconnected environments.
6. Give repeatable results.

Cross-Site Edge Ingestion

Data enters DART through the DART CLI, a Python-based edge ingestion tool that runs alongside the EPSE. The CLI parses raw test output, converts it to an efficient columnar format (Parquet), and uploads it to AWS. Ingestion also parses the test sequence that produced the data, automatically generating DART analyses from the sequence’s inline checks (described below). Because ingestion runs at the edge, each test site can push data into the shared data store as soon as a test completes. Some sites cannot guarantee reliable edge connectivity, which is where DART’s offline compatibility shines; data can be uploaded once connectivity is restored, or passed to other machines for upload or offline analysis.

Efficient & Compliant Data Storage

The DART Data Store organizes timeseries data as Parquet files in Amazon S3, with an RDS (PostgreSQL) database serving as a metadata catalog that makes every test, channel, and configuration discoverable. All of it lives in AWS GovCloud to satisfy export controls requirements.

This architecture pays off in both cost and speed. Parquet enables efficient compression and query-in-place, allowing DART users to download only the data they need. For large tests, this often means data can be analyzed in seconds rather than minutes or hours.

DART Desktop: Flexible Analysis Features for Engineers

DART Desktop is a desktop application built with Electron, React, and TypeScript, paired with a sidecar Python service that handles heavy data processing. It acts as the primary user interface to DART.

Rich Plotting & Data Visualization

DART’s plotting system handles upwards of 1 million points at once, allowing engineers to visualize full-resolution test data, overlay channels, and zoom into transients without decimating the data.

No-Code Analysis & Report Builder

Engineers compose analysis scripts and dictate report formatting using a flexible visual editor. Python-based expressions can be used to mathematically combine signals and perform advanced analysis.

Automatic Conversion of EPSE Runtime Checks

The test sequences on the EPSE contain a large number of checks that run at test-time. For example, it might monitor a parameter continuously to verify it stays within numeric limits. A single sequence file can contain 700+ such checks, and these checks need to be reliably replicated during post-processing.

Instead of requiring engineers to manually parse sequence checks into DART evaluations, DART automatically parses the sequence files into evaluations during edge ingestion. Engineers get the full set of checks without any transcription, and because the generated analyses are ordinary DART analyses, they also serve as starting points for more elaborate evaluations in the analysis editor.

Data Trend Plots

Analysis results can be pulled from an archive to create trend plots. These plots can span months or years, and make it possible to spot drift and catch anomalies. They come with a plot creation wizard that allows engineers to precisely pick which analysis results to include.

Data Store Explorer

DART Desktop provides user-friendly screens for browsing tests and signals saved to the cloud. Engineers can search and filter the catalog by test metadata, inspect what a given run contains, and pull down the data they need for local analysis. Downloads can be limited to the relevant channels and time ranges rather than entire test files, which keeps transfers small even when the source data is large.

Automated Reporting

DART generates polished PDF and XLSX reports directly from analysis scripts. Reasonable defaults ensure that most reports look good with minimal configuration, but the appearance of the reports can be heavily customized using the no-code editor described above. Requirements added in the editor flow through to the report, which allows engineers to use DART for formal requirements-based qualification.

Offline Compatible

Network connectivity cannot be assumed in all circumstances, so DART is built to work against local data. As long as tests are downloaded locally, DART can analyze, grade, and report locally using the same tools as it would with a network connection.

Key Technologies

Core Data Tooling

Processing & Storage

Application Layer

Conclusion

DMC is no stranger to managing large volumes of test data, and BOLE was no exception. Large testing initiatives, like the Artemis project, generate terabytes of valuable timeseries data, but if that data is difficult to access and even more difficult for engineers to interpret, the path from data analysis to action can become encumbered.

Value lies in how quickly engineers can access and act on their data. DART provided our client with an end-to-end solution that securely collects, organizes, analyzes, and reports on test data throughout its lifecycle. By creating an effortless path from test stand to engineering insight, DMC helped ensure that critical booster test data remained an asset, not a bottleneck, throughout the Artemis program.