(139c) Enabling Autonomous Operations | AIChE

(139c) Enabling Autonomous Operations

Authors 

Mikhailov, O., Xplorobot
Each use case for autonomous operations is different and there is currently no one-size-fits-all solution available on the market. We will present our experience with multiple robots, drones and sensors at various locations both onshore and offshore, covering installation, data acquisition and processing, as well as the key challenges and results.

There are numerous factors affecting the choice of robot for data acquisition. For example, many areas require equipment which is certified as intrinsically safe due to potentially explosive gases. There are very few such robots, and if additional sensors are required then recertification is necessary, which can be a long process. Wheeled robots typically have a longer range than legged robots, and although they can avoid small obstacles, they are not able to step over broad obstacles such as hoses. In some cases, the terrain may be such that a hand-held sensor package is preferred, or the inspections are infrequent and investing in automated data acquisition does not make economic sense.

The selection of sensors is also very important and depends on the particular use case. Each high-end sensor may represent a significant investment, so we need to ensure that they are fit for purpose, and we do not carry redundant sensors. Note that our “mobile sensor” approach has several advantages over an IoT solutions, particular at large sites where the large number of fixed sensors would be cost prohibitive. It is flexible: if a site is modified, the robots can be given a new route to monitor. IoT sensors require permits and installation, and as a fixed part of the site, themselves need inspecting. The mobile approach is also future proof as sensors can be easily upgraded as technology advances.

The key to a successful deployment however is the integration of all this hardware through well designed software, from data acquisition, though an automated data processing pipeline to visualization of the results that allows engineers to make decisions quickly. Rather than focus on reading gauges, as is commonly the case, we prefer to acquire dense data about the entire site. A typical inspection can thus generate ten to twenty gigabytes of data per hour from multiple different types of sensors and creates significant data management challenges.

After an initial site visit and 3D model building stage, we can process the data “at Edge” and distil the dense data down to easily interpretable heat maps, gas concentration maps, etc, in quasi real time. This process leads to a data compression of up to a thousand, allowing the results to be transmitted over low bandwidth networks, such as 4G, to the remote engineers for analysis. The use of 3D models facilitates time lapse analysis even when the data is acquired from different locations due to obstacles on the inspection route, for example. This allows to generate alerts when an inspection result is outside of normal conditions. With longer historical data, we can use repeat visual data to detect areas with higher corrosion rates, which normally go unnoticed by humans.

Finally, it is obvious that regular inspections followed by this processing build an incredibly rich and organized database that can be used for predictive analytics that will generate insights and solutions that we are only just starting to imagine.

Checkout

This paper has an Extended Abstract file available; you must purchase the conference proceedings to access it.

Checkout

Do you already own this?

Pricing

Individuals

AIChE Pro Members $150.00
AIChE Emeritus Members $105.00
Employees of CCPS Member Companies $150.00
AIChE Graduate Student Members Free
AIChE Undergraduate Student Members Free
AIChE Explorer Members $225.00
Non-Members $225.00