Interesting past projects

Kimon Roussopoulos

(principal investigator, Kition Research)

I endeavour to give, below, an introduction to a few interesting product developments, and to indicate the breadth of my experience. This is a limited selection with a general description as, inevitably, much of my professional work remains subject to confidentiality agreements.

Peptide synthesis technology

A Swedish start-up purchased the rights to a potentially disruptive technology for peptide synthesis (solid phase), using continuous instead of batch by batch synthesis. I was invited to review the technology, and proposed a project for building simple prototypes for proving the basic principles followed by more complex systems for testing the commercial performance of the system. I became project leader and specified a new design for the system, overseeing details for all the mechanical aspects, and myself designing the electrical/electronics systems and writing the control software necessary for the technology to be demonstrated. I also designed the experiments, working with chemists to specify the optimal conditions and so on. The technology was build and demonstrated with very limited resources, very successfully: technical performance is comparable with the highly optimised incumbent technology. (I was also very involved in the commercial aspects.) The technology has been presented twice in public at trade conferences, garnering great interest - the commercial exploitation is under consideration.

Grid load modulation

A company owned a technology for controlling power grid loads – in particular, for example, the ability to turn off a large load quickly in response to a fall in grid frequency, by turning off power quickly in refrigerator apparatus. (The on and off timing of refrigerators can be adjusted readily while keeping the contents within a defined temperature band). Their initial implemented algorithms showed an unexpected non-linearity of response. I showed in simulation and analytically how this was a consequence of the algorithm design, and demonstrated how a different algorithm gave liner response (this was patented by the company under my name). (I also implemented their algorithm in a commercial HVAC setting, the first live commercial trial of grid load response of this type, working with a supermarket chain and commercial control company.)

Medical device performance

A start-up developing a novel medical “autoinjector” design, was struggling to consistently develop devices giving the correct delivery performance (injection time, in particular), essentially using trial and error. I showed analytically how the delivered performance depended on factors including the spring parameters, the needle dimensions, drug parameters, friction and other factors, and developed a model that has allowed the required performance to be designed on the “drawing board”. Final performance of designs made using the model have agreed very closely with prediction, drastically reducing the amount of trial and error required during prototype testing.

A more challenging problem came along when a candidate drug turned out to be highly non-Newtonian, rendering delivery performance counter-intuitive and delivery hard to control. A literature search threw up little prior work on this problem; previous consultants had failed to develop an autoinjector for the drug. I developed new mathematical models for this situation and experimentally parametised the drug’s characteristics, allowing the performance to be very accurately predicted. I also led the development of a successful new autoinjector architecture able to deal with this type of drug, to the technical satisfaction of the Pharma company client.

(I also have modelled gas transfer through drug containers, studied extractables and leachables for different polymers, and done much other fundamental science relating to autoinjectors. An image below [taken from Wikipedia and “free to use”] shows components of a typical autoinjector but, for confidentiality reasons, not one have worked on.)

Motorsport radio noise cancelling

At the time of this work (early 2000s) the radios used in motorsport, even Formula 1, were not very advanced, and the high noise levels in a Formula 1 cockpit meant that it was often only possible to understand the driver’s speech when he was braking (and engine noise was lower). However the g-forces generated during braking make this a hard time for the driver to speak. Several misunderstandings occurred during races, which led to me being asked to look into noise reduction for the link for our associated team.

I gathered noise level data from the driver’s cockpit, with speech throughout several laps, during an F1 testing session. Several commercial noise cancellation software products were evaluated using this data, but none was adequate for this extreme case. I developed a special optimised bespoke algorithm for eliminating just enough engine noise from the transmitted signal (without eliminating so much that key speech elements were lost, which can destroy intelligibility). I then determined which commercial noise cancelling software would further enhance the signal. The results were quite startling – a raw recording in which no speech could be detected by ear, on treating, would suddenly reveal the clear tones of Mark Webber’s speech! The relative improvement in signal to noise, in speech bands, was estimated to be over 30 dB.

Novel gas metering technology

The advent of smart metering with electronic data gathering has recently made a strong incentive to develop inherently electrical measurements of gas for domestic metering (as well as for electricity and water metering). Several proposals had been made historically for fiscal gas metering using low frequency acoustic time of flight measurement, but no-one had previously managed to meet the accuracy requirements (around 2% of reading, over a flow range of over 150:1, meaning an accuracy at low flows of around a 0.01% of the maximum flow). (Using low frequencies allows the use of mass produced audio components; ultrasonic gas metering is well established but the electronics and sensors are expensive.)

I took a “back of the envelope” concept for a novel method of achieving this, computed the theoretical performance required from the elements and developed the necessary algorithms for determining gas flow from recorded signals. I led the manufacture of a series of prototype devices, and demonstrated that the best of these was able to meet the performance requirements for fiscal gas metering. The final device is roughly price competitive with incumbent technology, but has not yet gained commercial traction in a conservative industry.

KR testing water purification technology deep below the streets of Stockholm!

Contact

Dr. Kimon Roussopoulos (Director, Principal Investigator)

Cambridge, UK

kimon@kitionresearch.com

phone +44 7909 548421