A FLIR GF320 optical gas imaging camera is being used in an innovative robot system developed in Germany for remote detection and localisation of gas leaks.

Detection of gas leaks in industrial facilities is not only dangerous and time-consuming, but also susceptible to human error and interpretation. 

The RoboGasInspector was conceptualised as part of the German technology program AUTONOMIK by a consortium of nine companies and research institutes to provide a safer and more efficient and reliable detection solution using a FLIR GF320 optical gas imaging camera. 

The consortium developed a prototype of an autonomous mobile robot for gas leak detection and localisation in large industrial facilities, with the system able to perform inspection tasks without having to access hazardous areas directly, and without requiring any human presence. 

Ideal for routine inspections of facilities or for targeted inspections of specific system parts, the robot incorporates various navigation sensors for independent mobility along with the option of manual intervention via remote control at any time. The system is also equipped with video and optical gas telemetry, which enables it to inspect system parts that were previously difficult to inspect due to restricted accessibility.

This collaborative research project was headed by Dr Andreas Kroll and Dr Ludger Schmidt at the Mechanical Engineering Department of the University of Kassel and subsidised with € 2.4 million from the Federal Ministry of Economics and Technology. 

Dr Andreas Kroll explains that the objective of this project was the development and testing of an innovative human-machine system with inspection robots equipped with remote gas measuring technology and local intelligence. The mobile robots should not only independently perform the detection and localisation of gas leaks but also analyse the measured data and document the inspections.

Operators of industrial plants give top priority to the safety of their staff and their production facilities. Exercising the utmost care wherever toxic or explosive gases are used is absolutely essential. That’s why rigorous inspection specifications apply to the chemical industry, biogas facilities and gas suppliers. Usually, preventive inspection programs require personnel to perform time-consuming routine inspection procedures on a daily basis. 

During these regular inspections, staff members check the system for proper functionality, usually relying on perceptions and experience without making use of measuring technology. 

For professors Andreas Kroll and Ludger Schmitt, a top requirement for the new system was that it should allow for automated, hazard-free inspection and monitoring and that it should be able to respond independently to problems.

During routine procedures such as repeat inspections, there is always a risk of inadvertently overlooking possible sources of danger due to inattention. 

Therefore, the development of innovative inspection technologies and focusing the flexibility and performance of human operators on managing the technological systems make sense not only for economic reasons, but also with regard to relieving humans from repetitive routine tasks and improving coverage of the wide range of inspection tasks.

First demonstrated in a hall at the University of Kassel, RoboGasInspector independently completed an inspection route, successfully overcoming obstacles and a ramp in the process. At specified inspection sites it inspected various pipelines and found a methane leak. 

In the following months, the robot was tested on several square kilometres of large industrial facilities under laboratory conditions, in which environmental factors such as wind and sun as well as confounding factors resulting from system operation were included. 

The RoboGasInspector consists of three modules: a chain-driven mobile platform, a navigation module and an inspection module, which incorporates the FLIR GF320 optical gas imaging camera. The chain-driven platform is equipped with an electric drive and conventional batteries. 

The navigation module consists of 2D laser scanners as well as a GPS for outdoor orientation. Continuous comparison of the area to be inspected with a digital map enables the chain-driven RoboGasInspector to determine its position at any time; obstacles and blocked areas can be noted on this map. 

The 2D laser scanners enable the RoboGasInspector to avoid unexpected objects such as parked cars, pallets, barrels or even persons, moving around the obstacles or stopping till the path is clear again.

The inspection module combines various metrological instruments on a pan-tilt unit, including a Remote Methane Leak Detector (RMLD), which is based on an active Turnable Diode Absorption Spectroscopy (TDLAS) instrument. It works by means of an infrared laser; when the laser beam hits a surface, it is reflected and its residual intensity is measured. In addition, a FLIR GF320 thermal imaging camera is mounted on the inspection module to visualise the gases.

A built-in gas sensor in the RoboGasInspector shuts down the entire system from 10% of the lower explosion limit (LEL) onwards in order to prevent possible danger to a flammable atmosphere. 

Processing of the measured data and pattern recognition are performed independently by the robot. The RoboGasInspector also carries out the inspection of the specified routes and performs measurements on its own while being continuously in contact with the control room, and can be remotely controlled from there if necessary using a video camera incorporated in the pan-and-tilt measuring module. However, in normal operating mode, the RoboGasInspector works independently and merely transmits all measured data to the control room via WLAN.

Key advantages of the RoboGasInspector include facilitating independent gas detection and leak localisation in difficult to access sites, and avoiding the use of human inspectors in potentially dangerous environments. The FLIR Systems GF320 is a vital part of the RoboGasInspector, helping to detect harmful gas leaks from a safe distance.