Surveying high structures in tanks and holds in many cases requires staging. DNV GL introduced drone-based surveys to avoid this costly preparation. Still, drone pilot and surveyor need to enter the tank. What would be needed to avoid human tank entry?
Through periodic surveys, classification societies’ surveyors verify that appropriate condition of the ship, its components and systems is maintained. A[ds_preview] main portion of surveyor’s activities on board is related to physically accessing tanks and compartments for assessing hull condition, i.e. overall and close-up visual inspections as well as evaluating thickness measurements. Internal examination of ballast tanks and cargo holds requires extensive preparation for physical entry: emptying the tanks, cleaning the structure, gas freeing and maintaining continuous ventilation, and installing sufficient access means such as scaffolds.
In 2016, DNV GL has introduced drone-based production surveys on a case-by-case basis. In that way, often staging can be avoided. E.g. 20m high web frames can be examined through imagery captured by the drone. Significant time and cost savings for shipowners have been achieved through this. Nevertheless, for controlling the drone and for understanding the position of captured images, the surveyor and the drone pilot need to enter the tank. Both from a safety and from a cost perspective it would be attractive to replace human tank entry by alternative techniques. How could that be achieved?
Access to the structure would need to be provided by an autonomous or remotely controlled vehicle. This can be e.g. a drone or a diving or crawling robot. Both orientation of the drone and positioning of captured images need to be enabled through automatic indoor positioning technology. The inspection path would need to be guided through a pre-existing map which can be a 3D model.
Where the above is required, other optional techniques could further significantly enhance the value. A database with historical findings would help focusing on critical areas. Measurement gear could be carried by the drone for measurements of thickness and deformations and/or detection of cracks. Remote connectivity would allow a user to interact with and guide the drone for additional close-up capturing where found necessary. Processing of the resulting footage could be facilitated through Virtual Reality (VR) techniques. Image recognition methods could automatically identify photos for necessary follow-up/human assessment. Finally, captured images could be used to automatically reconstruct a 3D model which refines the existing coarse structural model. All these technologies are under development in the industry today, leading to the question what the future inspection process could look like.
The remote hull survey process
A remote survey process would comprise survey planning, pre-scan on board and remote hull condition assessment.Different scenarios can be imagined depending on the maturity of the technology. For survey planning, the surveyor would explore in Virtual Reality the vessel’s digital twin enriched with historic data, including results of previous class surveys and owners’ inspections, as well as related data for sister vessels. In this way tanks to be inspected and special focus areas can be defined in the office. Hull inspections would be performed with the help of remotely operated or even autonomous inspection means. This can be done in a pre-scan unattended by the class surveyor, e.g. during voyage or unplanned downtimes.
As the drone or crawler would carry an automatic indoor positioning system, inspection photos could automatically be mapped onto a 3D model. If autonomous flight is not yet feasible, pre-scan can be performed by a service technician carrying a camera system applying the same automatic image positioning techniques. After pre-scan, inspection data captured on board (overall and close-up footage, measurement readings) can be explored by the surveyor in virtual space while staying in the office. Thereby, a higher degree of surveyors’ specialisationm (on specific ship types or even individual ship series) would become possible. In an intermediate scenario, the surveyor would remotely interact with the drone in the tank. The reason is the need for additional close-up photos in case of uncertain assessment. Although being on the ship, the surveyor could avoid tank entry. Results could be presented and necessity of possible maintenance and repair measures be explained to the superintendent in a virtual meeting on-board the virtual ship in an efficient and intuitive way.
Can a drone replace tank entry?
A class surveyor, physically present on board during survey, uses more than his visual sense. He can feel: is the surface smooth, dry? Is this water or oil? A hammer or other force can be used to remove scaling or to test the response of structure. And the smell can give an indication of smoke or cargo vapours. How could an autonomous vehicle compensate the lack of these abilities? If an automatic visual drone scan of the structure becomes viable, a more frequent and more extensive examination of the structure would become cost efficient and could therefore replace the infrequent partial checks today. Additionally, image recognition techniques can increase the probability of deficiency detection. Furthermore, techniques such as hyperspectral imaging (able to analyse the full spectrum of the light reflected by the structural surface) could detect corrosion or coating breakdown with higher likelihood than with the naked eye.
Avoiding human entry into enclosed spaces is a valuable ambition. By automating hull pre-scans with the help of drones, the surface coverage and frequency of inspections could be increased. This, and the use of new image recognition technology then has the potential to compensate for the missing human perception of the tank environment. Further conceptual work should be performed to investigate safety equivalence.
It is not yet clear when a full autonomous scan of a ship compartment might become reality. But building on the fast technological progress, we will surely experience a step-by-step move towards less hazardous and more efficient inspections.
Author: Christian Cabos, Head of Information Management Technologies, DNV GL – Maritime
christian.cabos@dnvgl.com
Christian Cabos
