John Barratt discusses how an established tracking system used in TV and movies is being adapted for the process industries
IN THE chemical industry there is often much debate as to whether a process should be batch or continuous. Most processes at low production volumes start off as batch, largely because this is the way they are first developed in the laboratory. Chemical engineers spend time scaling up these processes to improve economic performance and satisfy increasing demand. And when demand reaches a particular level, a process is converted to continuous due to the greater economies of scale and the efficiencies that come from continuous operation. In continuous operation each part of the plant is used all the time, while in a batch plant equipment often stands idle while different steps in the process take place.
Customer demands, driven by end-consumer desires are getting more complex, making the case for batch. So how do we make batch processing more efficient?
The batch or continuous question was often debated in the Process Engineering Group at ICI where I worked in the 1990s. Tasked with designing a laundry detergent plant on an industrial scale, my initial conventional design became untenable as customers’ needs became more complex. An experienced colleague suggested I redesign it as “pipeless” (ie batch) – to solve the problems I was facing with complexity, flexibility and avoiding cross contamination. What resulted was a cheaper plant that was more adaptable and responsive to customer needs.
The only drawback was that it required a significant number of operators, moving intermediate bulk containers between mixing, blending and formulation stages (such plants do exist in the chemical industry, but this type of operation is more common in the food and formulation industries).
In 2009, as Core Technology Officer at BASF, I looked at what technology was available that might improve manufacturing at BASF generally, and for my business area in particular.
Most of BASF’s process plants are highly automated, typically using distributed control systems for process control. Most operators now spend their time in the control room monitoring the automated system and checking the operation of the plant. However, for many plants much of the raw materials are transported to the plant in intermediate bulk containers (IBCs) and flexible intermediate bulk containers (FIBCs). Similarly, the products are often packed in IBCs, FIBCs or 25 kg bags. Even in highly automated specialty chemical plants, labour is a major operational cost. In most plants, further automation of the process is quite expensive and does not yield significant labour saving. However, the transport of raw materials and product to and from the plant uses a large amount of labour and would seem a key area to target.
This got me thinking that if industries such as automotive, warehouse and customer fulfilment can use automatic guided vehicles (AGVs) extensively to move parts and packages, then could we not do the same for raw materials and products in the chemical and process industries, where this is not widely adopted? After consulting with internal experts, Innovate UK, and the manufacturing catapult centres, it became clear that improved technology was required if my dreams were to become a practical reality.
I found that the existing guidance technologies were not well suited to process industry environments (see Table 1). For example, weakened signal due to metallic structures and equipment is a particular issue. Having a sterile environment where only AGVs are allowed to operate works well in automotive and warehouse environments but was less practical in process industries where maintenance activities, people and other vehicles do need to be present. Also, most AGVs do not have the ATEX certification to work in potentially inflammable atmospheres.
To significantly further automate plants, autonomous material handling is required. This requires the transfer of liquids and solids and potentially even gasses fully autonomously. To do this, safe zero-loss couplings exist but the ability to operate these with AGVs was not possible due to the inaccuracies in guidance and other reasons explained earlier.
Fast forward to 2018, where I met Michael Geissler at a trade show. His firm Mo-Sys has developed several world-leading camera tracking technologies used in media, television, film and professional virtual reality. In this field, accurate tracking of a moving camera is vital to special effects and combining virtual and the real picture. It must be done seamlessly in real time – otherwise the effect is destroyed. In our discussions we realised that we could potentially apply the company’s StarTracker technology to AGVs in the process sector to provide real-time accurate positional information.
The system uses a camera and lights to look at a number of retro-reflective spots, typically randomly placed on the ceiling of a studio. It learns the random patterns and using trigonometry can work out, in 3D, precise location and orientation (pose) of the tracker. The calculation is done rapidly, with refresh rates of 100 Hz or greater. It does not depend on external electrical signals, maps or other connections, and can work if several of the dots are obscured. It therefore would work reliably in dense process plants where electrical signals, GPS, may be weakened by steelwork and equipment. It is not dependent on fixed routes but knows where it is and how it is oriented wherever it can see the start map. It provides accuracies of better than 2 mm in real time and pose data of ~0.01° as standard, that would allow it to guide an AGV precisely into a zero-loss coupling and enable fully autonomous material transfer.
Realising the potential to modify this existing tech to use in our industry, while at BASF I sponsored a Mo-Sys application for Government innovation funding to develop the idea further. The grant was successful and the company has over the last two years been working with University College London to build a prototype AGV and integrate it with its StarTracker system.
The requirement to accurately position the IBC or other load has led to the development of a secondary pallet. The secondary pallet sits under the pallet the material is on. The secondary pallet is lifted and lowered by the Mo-Sys AGV. The precise location of the connection which is part of the secondary pallet is known to the AGV, which allows accurate connections with multiple IBC-types and pallet sizes. It also allows one AGV to serve many loading and unloading points.
For example, automated liquid supply would work as follows. A typical liquid containing IBC will have transport caps on the outlet and a sealed lid. These need to be removed before unloading and the sealed lid replaced with a breathing lid. At the same time the IBC is connected by a short hose to the secondary pallet and the manual valve on the IBC is opened. This hose and the coupling heights are designed so that the IBC can be fully drained. The operator then informs the system that the load is ready. These jobs remain manual.
The IBC on the secondary pallet is now ready for transport. The AGV drives to the IBC using the tracking system navigation, picks up the IBC on its secondary pallet and then drives to the required unloading point, wherever that is on the site. The AGV drives into the zero-loss coupling using the precise navigation. The AGV confirms a good coupling has been made and closes the locking device on the coupling. It then signals the plant distributed control system (DCS) that the IBC has been replaced. Unloading from the IBC then takes place automatically through the DCS as it would in a manual operation. Once the IBC is empty, this AGV or another will return, automatically lift the empty IBC and decouple it from the plant.
Mo-Sys has built and run the AGV. It has now further developed the AGV with additional capabilities. For example, the existing system guides with reference to ceilings, walls and structures, but at many plants, forklifts operate both inside and outside of buildings, where fitting tracking spots may be impractical. While accurate position information is needed for loading and unloading operations, GPS with accelerometers provide sufficient accuracy to drive along site roadways and other open spaces, so Mo-Sys has integrated GPS navigation with its existing system to provide a flexible AGV.
Safety in chemical plants is of the highest priority. Forklift trucks are a cause of many accidents from impact, loss of containment and muscular skeletal injuries to operators. The new system aims to greatly reduce these incidents and improve safety, and could also significantly reduce the need for shift work and the health issues associated with disturbed sleep.
It provides accurate navigation around plant equipment and allows the setting of exclusion zones, when required (geofencing). Additionally, the plant-based AGV incorporates four additional cameras which provide 360° vision around the AGV, and has sonar detectors that work like parking sensors. These cameras can detect obstacles, vehicles or people near the AGV. The speed of the AGV is limited to ~1 m/s (or 2 miles/h). The collision avoidance system allows the AVG to stop in a short distance, even if someone steps out in front of it. As the AGV has a full 3D map it can navigate around the obstructions, person, or maintenance activity. This ability to work real-world, complex environments is a major step forward.
Fitting the Mo-Sys AGV to an existing chemical plant is straightforward. The reflective spots are typically stuck on the roof structure or process equipment or any static item. A higher density of stickers can be used where particularly precise navigation is required. The existing plant coupling is changed to a zero-loss coupling with pneumatically-activated closure and the DCS and scheduling and warehouse software systems are adapted to interact with the AGV. The AGV rapidly learns the random pattern of reflective spots, self-guides, self manages and automatically recharges its lithium-ion batteries. As the plant AGV is smaller, lighter and drives more slowly than a typical forklift truck there is less wear on floors and roads. The manoeuvering space required is smaller, and accidental damage, so often seen in areas where forklift trucks operate, can be avoided. The automation of raw material delivery and product movements allows higher levels of assurance and tracking, a particular benefit for highly-regulated industries such as pharma, agrochemicals and food.
This new capability will also allow for the realisation of low-labour, efficient, flexible pipeless plants. In such a plant the product would move from processing station to processing station where each of the process steps would be made. This would provide true customisation and a great variety of products while still achieving the economies of scale. Specialist IBC could be used to further increase automation.
The technology can also be fitted to larger or smaller AGVs and other types of vehicles. For example, it can be used at laboratory scale, moving flasks and small containers or sample racks around. It could also be used at a larger scale, for example where automated vehicles are moving 100 t tanks around a site from a railway yard, it could provide more accurate guidance and key loading and unloading points.
Inspection is another application under development. Automated inspection is already being used but in many cases information as to exactly where the guided vehicle was and its pose when the inspection was made is not known. The system could provide that information and accurate re-inspections could take place, with results compared to give a true history and accurate comparisons.
While StarTracker and its guidance software are fully commercialised in the media industry, its use in industrial and chemical applications is just beginning. Mo-sys has proven the application over the last two years of testing on our own AGV design, with its unique autonomous liquid transfer system. We are in discussions with other robot and AGV producers about fitting our guidance system to their products, from inspection to agile production, and are keen to work with industrial partners.
Further testing is planned on operational chemical sites. We’ll be looking to optimise the system with improvements to operating in all weathers; ATEX certification for operating in flammable atmospheres; and integrating with common control systems and communication protocols. We’re expecting initial industrial operation in around 6 months.
This new technology has potential applications in many sectors of the processing industry both in existing sites and for new plants, and plays a part in achieving better safety, lower costs, and greater flexibility.
Catch up on the latest news, views and jobs from The Chemical Engineer. Below are the four latest issues. View a wider selection of the archive from within the Magazine section of this site.