“It’s about process optimisation, not about the techniques that we use to solve the challenges,” Dr Katharina Roettger, Principal Scientist at CPI.
MEPCA are delighted to introduce Dr Katharina Roettger, Principal Scientist at CPI, a pioneering social enterprise dedicated to building innovation ecosystems that connect academia, industry and government. Kat applies her expertise in digital technologies to process improvement, assisting manufacturers, of all sizes, to scale innovative projects.
After completing her at PhD at Kiel University (CAU), Germany, Kat re-located to the UK to continue her postdoctoral research on molecular dynamics, spectroscopy, and time-resolved spectroscopy. Seeking a break from academia and wishing to direct her expertise away from more fundamental research to industrial research, Kat joined CPI in 2016.
“I wanted to see what industry is doing. That’s how I came to work at CPI, because [at the] CPI, we’re a research and technology organisation and we sit between academia and industry,” Kat explained to MEPCA’s editor, Oliver Batt. This positioning enables CPI to help companies navigate through that difficult gulf between ideas and commercialisation.
Her initial project for CPI took her in a refreshing new direction: “I led the development of a digitally enabled pilot plant for liquid mixing, which was very different from my postdoc.”
This project involved developing digital capabilities for this small pilot plant, including process analytic technologies, to gain valuable insights into the manufacturing process, measuring things like particle size and viscosity. They collected this data to provide real time insight into liquid mixing processes.
“That was really interesting and I learned a lot in that time about both digital infrastructure, data collection, analytical instruments, and so on.”
However, it was the challenges that could not be solved that held as much fascination, particularly those related to how prohibitive the cost of measurement technologies is to smaller companies or companies in low margin industries. This led her to looking at alternative solutions; an area she has continued to focus on.
“If you want to measure viscosity, for example, you can either buy an instrument that can do that in line or you could build something which is called [a] soft sensor, where you infer viscosity based on other measurements, like pressure sensors, for example.”
Soft sensors are a non-invasive, digital approach to sensing technology that takes data already held by a manufacturer and uses digital technologies to infer further data from; this forms the basis of predictive modelling, which provides information on what a process will look like over time.
It’s often the case that manufacturers don’t believe that they hold much data, or that the data they hold is unusable, but Kat insists there is nearly always something that can be done with that data, if the right approach is taken. Rather than asking a company what data they have, or offering to build them an AI tool, she wants to understand what their challenges are, as the right approach is “problem-led and not solution-led.”
Referring to research released by the RAND Corporation, Kat explained that among key reasons for digital technology and AI adoption project failure is that these organisations “look at solutions first and don’t define the problem properly. And that is how a lot of these projects fall over because you need to know what you want to know first before you start looking for a solution.”
When asked whether the current hyper-focus on AI as a solve-all technology was perhaps overshadowing other equally important technologies, she agreed that, to an extent, it did. Taking predictive modelling as an example:
“We are not yet at a place where you could just feed all the data into an AI and it will give you the answer […] a lot of it is still, in a way, a very manual process where you need to understand what the process is, you need to understand what the challenges are.
“That’s something that sits with the operators, with the technical people on the plant; they know everything there is to know. No AI has that knowledge, at the moment.”
AI is of course important, and CPI utilise AI and machine learning, but returning to her earlier point on challenges before solutions, she said, “It’s about process optimisation, not about the techniques that we use to solve the challenges.”
At CPI, Kat has dedicated her expertise to helping manufacturers find digital solutions to their challenges, and she finds it most satisfying to work on projects with smaller companies – helping those technical operators to improve processes – and seeing the difference more directly.
An example that stood out for her was Therma-Chem, a family-run business based in Glasgow that makes cleaning products for industrial boilers, such as those found in the heating systems at oil refineries. CPI started working with them circa five years ago, looking at how they analyse and use their data, and upskilling their team.
“What’s really, really nice to see is that they really value what we are doing. They want to change the way they do their business by adding more digital services to their business model. And it’s just a wonderful collaboration. They are very, very innovative. If we had more companies like that in the UK, I think it would be great.”
CPI’s continued efforts were recognised in the Industrial Strategy, which referenced the Advanced Materials Battery Industrialisation Centre (AMBIC) in Sedgefield, aimed at enabling the development of new battery materials and accelerating the clean energy transition. Further details of this and other innovative projects CPI have been involved in can be found on its website.
