Making for the future

Manufacturing practice evolves naturally within companies, but to make a significant leap forward you need new knowledge and new ways of thinking.

“You can't always anticipate where the breakthrough or the bottleneck will be,” says Professor Nilay Shah, head of Imperial's Department of Chemical Engineering. “Quite often, it's not directly in the manufacturing research, but in fundamental research. It could be in maths, physics, chemistry, computing, or aspects of engineering.”

This makes Imperial a natural home for advanced manufacturing. “Our strength is the breadth and depth of our fundamental research into science, engineering, and to a certain extent in medicine, which can then make its way into manufacturing innovations.”

For that happen, two kinds of connection are necessary. The first is between fundamental research and its application. “The best channel for that is industrial collaboration, and we have more industrial collaborations than any other university in the UK,” says Professor Shah. “That means we can work with companies to embed these developments into new technologies.”

The second kind of connection is between disciplines, so that people with complementary knowledge can find and inspire each other. This is achieved within networks that cross Imperial's departments and faculties, such as the Manufacturing Futures Lab. “This brings together a community of people who are interested in manufacturing, either because they are already engaged in it, or because they realise their skills and capabilities are highly relevant for manufacturing.”

A digital leap for synthetic chemistry

Digital methods have also been instrumental in advancing synthetic chemistry. “It's about turning chemical synthesis into a data-led subject,” says Professor Mimi Hii, who directs the Centre for Rapid Online Analysis of Reactions (ROAR) at Imperial's White City campus. This is a new national facility that includes systems for chemical synthesis, reaction analysis, laboratory automation and optimisation, and – of course – data analysis.

By drawing together data on all known chemical reactions, and making smart predictions about untried pathways and combinations, these systems should make it possible to digitally 'dial up' any molecule that needs to be manufactured. “If we do it right, this should lead to faster discovery, where we don't waste so much of our time in trial and error. At the same time, through automation and automatic data collection, we can eliminate human error and human biases.”

One early project for ROAR has been a partnership with South Africa, supported by the UK government's Global Challenges Research Fund, looking at ways to make the country's pharmaceutical industry less reliant on imported drugs and chemical components. The first target was a breast cancer drug, which is both expensive and in high demand. “We set out to devise a manufacturing route that was cost effective, environmentally sustainable, and worked with the resources available in South Africa,” Professor Hii explains. “Using the resources in ROAR, we came up with a new synthetic route within three months. Normally that might have taken many more months or even years.”

There are also plans underway to establish a Centre for Doctoral Training in synthetic chemistry alongside ROAR. Instead of learning traditional synthetic skills, PhD students will cover subjects such as data analysis, automated systems, and how to operate the new generation of continuous flow reactors capable of on-demand synthesis. “Over five years, we will train at least 60 people, with more studentships to come, because the industrial interest is still growing.”

Algorithms for optimal chemical processing

Dr Ruth Misener started out as a chemical engineer, then switched to computing in order to work on new optimisation methods in oil refineries and petrochemical plants. “I'm solving problems that come up in chemical engineering, but we are doing it by making fundamental contributions to computer science.”

Her team in the Department of Computing is particularly interested in common problems that remain challenging to solve. “If it is something that only comes up once or twice, it is not worth our time.”

One fruitful subject is scheduling, where tasks need to take place in a set order, within a set time. “That can get very complicated indeed on a petrochemical refinery, for instance. What we do is select the aspects that are theoretically or computationally difficult to handle, and develop mathematical strategies and algorithms that take care of that part of the problem.”

To do this she can draw on advanced techniques that were not available just a decade ago. “We have a lot of new methods in mixed integer nonlinear optimisation, which give us algorithms able to solve problems that are ten times bigger than before,” she explains. Developments in machine learning are also proving useful. “My team collaborates with machine learning researchers at Imperial to integrate the kind of models they create into our optimisation systems.”

Optimising systems for uncertain conditions has also come a long way. “We try to understand, within the algorithm, that things are going to go wrong, and respond actively to that kind of situation.” For example, if a machine breaks down or a reagent fails to arrive at the right time the aim might be to recover quickly, but also to limit disruption by minimising the number of changes that have to be made. “The problem that you have to solve is different. There are extra restrictions that you want to include so that the process has good recovery properties.”

While originating in the chemicals sector, these optimisation solutions can have a much wider application. The work on scheduling, for instance, resulted in a collaboration with Royal Mail. “Delivering mail also requires scheduling jobs onto machines,” Dr Misener says.

She has also benefited from connections made through Imperial's networks, in particular the Centre for Process Systems Engineering. “The Centre's industrial consortium is where I can workshop new ideas, and show industry what I'm working on,” she says.

The same kinds of connection are also possible through initiatives such as the Energy Futures Lab and the Data Science Institute. “These involve different people from different places, so you get to meet lots of like-minded industrial colleagues.”

Novel reactors for fuel cells

Metal 3D printing can also make novel structures, a capability Dr Wu is exploring in connection with fuel cells, for applications ranging from cars to drones. Fuel cells typically operate by combining hydrogen and oxygen from the air to generate electricity, with the only by-product being water. Conventional fuel cell designs use plates with winding, snake-like channels that the reactants flow through to reach the reaction site. However, this design has limited efficiency when operated at high powers, due to difficulties in getting the gas through the channels.

So Dr Wu and his colleagues looked to the human body for inspiration. “We've designed a fractal-based fuel cell, where the serpentine flow channels have been replaced with a structure that takes the efficient gas transfer principles of the human lung to achieve much higher performance,” he says. “Given the complexity of this geometry, it could only be manufactured with 3D printing.”

Meanwhile Dr Wu is also working on a more economic approach to 3D metal printing, based on electroplating. He has replaced the plastic extruder in a conventional 3D printer with a syringe filled with a metal depositing solution. When a current is applied to an electrode immersed in this solution with a bead of fluid at its tip, and brought into contact with a surface, the device lays down a precise layer of metal.

“That allows us to create a metal 3D printer at a very low cost,” he says of the method, which is still in the early stages of development. It also opens up new possibilities, such as the ability to print layers of different metals, and – when the current is reversed – to recover and recycle previously printed metals.

Robots for higher productivity

One of the people at Imperial who is rethinking robotics is Dr Nicolas Rojas, also in the Dyson School of Design Engineering. His work is shaped by fundamental questions about how robots are designed and operate, but also by the challenges faced by industry.

“One of the current needs, particularly in the UK, is to increase the productivity of small and medium sized manufacturers. The problem is that they tend to be producing multiple, high-quality products, but in small batches. To do that you need humans, but if all you have is humans, then it's impossible to improve productivity.”

Robots that work with humans – known as cobots – offer one way forward, although their potential has yet to be fully realised. One problem is that many cobots are over-specified and so take up a lot of space on the factory floor. “The idea is that you need these very big robots, with many motors and degrees of freedom, because you need flexibility. So we are working on robot arms with a minimum number of motors, but the same dexterity.”

Another goal is to produce malleable robots, which a human operator can reshape to perform different tasks. “Robot arms are generally inspired by our arms. But imagine you had a forearm that is not constant, that you can reshape,” Dr Rojas says. “We are developing a variable stiffness limb that becomes very flexible when the user want to reshape it, and then becomes rigid once the system is activated.”

This might be used, for example, to optimise a cobot's speed. “We know that some morphologies of robot arms are faster than others for some operations, so we are designing technology that can be shaped according to the task.”

Another line of thought moves even further from anthropomorphic robot arms to look at parallel robots, with multiple arms that connect to move single gripper or robot hand. “The payload capacity of these robots is very high, and they can move very fast,” says Dr Rojas. However, the design introduces constraints that still need to be overcome for high adaptability. “So we are developing a very dextrous, parallel robot with a large workspace that can be used for pick-up and place operations.”

Ian Mundell is a journalist who specialises in research and higher education. He divides his time between London and Brussels.

This Imperial Story was commissioned by the College's Enterprise Division.

The Division helps businesses to address their challenges by accessing the College's resources, talent and expertise - and helps academics and students find new ways to turn their expertise into benefits for society.