# SAS-300 Software & Analytics Suite

> AI defect recognition & C-scan reporting


The analysis layer of the inspection chain. SAS-300 ingests raw A-scan data from any RAS system and renders B-scan and C-scan maps. The AI model classifies indications at 99.2 % agreement with ASNT Level III review, which means the accept call no longer waits on a second pair of eyes. Report generation falls 73 % against manual workflows, so a full blade report lands in under 5 minutes. Every indication is versioned against the blade serial number, so the audit trail is rebuilt from data rather than memory.

The suite ships as five applications: **RASight** for blade acquisition and review, **RASScan** for in-line plank monitoring, **RASView** for offline analysis, and **RASServer** with **RASAutoLT** for the AI analysis queue and its labelling tool. They share one data model, so a blade scanned on RAS-10 and a plank inspected on RAS-90 land in the same review workflow.


## Overview

The SAS-300 is the analysis layer of the inspection chain. It ingests raw A-scan data from any RAS system, renders B-scan and C-scan maps, and classifies indications with an AI model at 99.2 % agreement with ASNT Level III review, so a full report is ready in under 5 minutes.

## Highlights

- **AI defect recognition**: Delamination, porosity and disbond are typed automatically at 99.2 % agreement with ASNT Level III review, on a model trained across 10,000+ blade inspections in 10+ countries.
- **C-scan analysis & traceable reporting**: Every indication is versioned against the blade serial number, the operator ID and the calibration block, so an audit trail is rebuilt from the data rather than from memory.
- **Open integration**: REST and OPC-UA connectors push each result into MES, ERP and asset-management platforms, and the same feed syncs a digital twin for predictive maintenance.
- **Server-side analysis queue**: A dedicated Ubuntu 22.04 analysis server with GPU acceleration takes scans as they are acquired, queues them by priority and files the finished report, so review does not wait for an operator to open each file.

## Technical data

| Parameter | Value |
|---|---|
| Views | A / B / C-scan |
| Detection accuracy | 99.2 % |
| Report time | -73 % vs manual |
| Integration | REST, OPC-UA, MES / ERP |
| Deployment | On-premise or cloud |
| Analysis server | Ubuntu 22.04, GPU, browser access |
| File formats | Native scanner files; Olympus .odat / .nde |
| Applications | RASight, RASScan, RASView, RASServer, RASAutoLT |
| Audit | 21 CFR-style change log |

## FAQ

### What scan views does the C-scan software support?

A-scan, B-scan and C-scan views are linked, with 3D tomography alongside them, so clicking a C-scan indication opens the raw A-scan at that point.

### How accurate is the AI defect recognition?

The model runs at 99.2 % agreement with ASNT Level III manual review for delamination, porosity and disbond, measured against 10,000+ blade datasets.

### Can the NDT data analysis software integrate with our digital twin?

Yes, over REST and OPC-UA. Each dataset is geo-tagged to the blade 3D model, so a defect is tracked at the same coordinate across inspections.

### How long does it take to generate an inspection report?

Report time falls by 73 % against manual workflows. A full report, with C-scan maps, classification and the acceptance comparison, arrives in under 5 minutes.

### Can we try the software before purchasing a full system?

A 30-day trial is available with sample datasets and a live demo, and we can run the evaluation on your own scan data.

### Does the AI analysis run on the scanner or on a separate server?

On a separate server. Scans queue as they are acquired, priority can be raised per blade, and the analysis returns the defect list, screenshots, grading and a PDF or Excel report without an operator opening the files one by one. Users work through a browser with role-based permissions.