Dr Corey O’Meara, Chief Quantum Scientist at E.ON recently spoke to Grant Powell MBCS to explain the role of quantum computing in the energy industry.
Growing up in Canada and fascinated by the prospect of becoming a scientist, Corey O’Meara pursued a bachelor’s degree in theoretical physics followed by a master’s degree in quantum information. Next came a relocation to Munich, Germany to undertake an international scholarship, studying for a PhD at the Technical University of Munich. Following four years working for the German Aerospace Centre on applied AI projects and predictive satellite maintenance, Corey joined energy provider, E.ON, as a senior data scientist. Within months, the buzz generated by quantum saw him being headhunted for a new team that would use quantum computers to solve real world energy problems.
How is the energy sector evolving in the digital age?
The energy sector is currently undergoing a massive digital transformation. Traditionally the electrical grid is non-digital. Now we are seeing a shift towards the integration of renewable energy, which is only possible with a digital grid. This unlocks a whole area of opportunity for digital products. Right now, we have power plants that generate electricity and deliver it directly into the grid. The generated energy is transferred from a high voltage grid to a medium voltage grid, and then to a low voltage grid, at which point it can be used by individual consumers and businesses.
What's interesting is that when we move to renewable injection into the grid — so things like roof-based solar PV panels on groups of buildings with the ability to transfer power — we have a situation that previously didn’t exist. Now we have electricity being put back into the grid at the low voltage grid level, rather than always having to be made at high voltage level hundreds or thousands of kilometres away. The effective management of this requires an ability to monitor data relating to electricity generation and consumption, and this introduces a plethora of new computational problems.
Where does quantum come in, and how can it help develop renewable energy solutions?
We're looking at quantum as a tool to enable us to calculate and monitor certain elements that otherwise would not be possible with the limitations of traditional classical computing. It’s not about big data, but more accurately about complex data. Quantum computers are particularly good at solving complex optimisation problems. If we took all of the buildings in a city, for example, and wanted to organise those buildings into groups of 10 based on energy use and energy generation, how many combinations are there? It's an exponential amount, and you cannot brute force the amount of time it would take to check every possible combination. The best way to group them would be in respect of a useable metric or measurable quantity that stipulates what the best grouping will be. And this is actually a real-life example for which we're looking at using quantum computers. They can help us determine the best way to group buildings that are locally generating and using energy. If solar panels are installed, how do you group neighbourhoods together such that they may form a self-sufficient community? Some houses generate electricity and then their neighbours use electricity. If you generate excess electricity, why would you send it across the city to the grid, congesting power lines, when it could be used by your local neighbourhood?
So, you can naively do that by area code, fine. But the area code doesn't fit one to one with the wiring that you see on the streets. So, you need to look at the types of wires, the length of the wires, the year they were installed. These wires may not even be digitally monitored everywhere. How do you check you're not going to blow one from a voltage overload? You can already get an idea of the sheer number of factors to be considered, and quantum has enormous potential to look holistically at many factors simultaneously to help us determine the optimum approach.
Can you tell us about any other specific projects that you’re working on?
We’ve been engaged in a multi-year partnership with IBM Quantum to collaborate on various energy sector problems where quantum computing can be applied. One of the things we're looking at is a concept called vehicle to grid, otherwise known as ‘V2G’, for which E.ON participated in another V2G project with Nissan. The basic concept is that if you have a fleet of electric vehicles plugged in during the day while people are at work, or overnight while they sleep, the regular way to charge those cars is you just plug them in; power is drawn from a power plant somewhere and the battery charges. But, let’s say you have 50,000 car batteries plugged in overnight — why not take a bit of the energy from each of those vehicles, and market that electricity? As long as the car is charged the vehicle will operate and the user is unaffected. Plus, they’ll receive a discount on their energy bill, so essentially make money on the energy their car discharges. And if you take little slices from all of the different car batteries and you add them up to make an amount of energy, then that can be delivered somewhere else in the energy market where it is needed.
But what happens if some people unplug their cars right before you’re supposed to be delivering a set amount of energy? You've already assumed that those cars will have little bits of energy pulled from them. Now, you check with your software and you say, ‘oh, instead of 50,000 cars plugged in, we have 45,568’. You need to recalculate as fast as possible to determine which cars you will take a little bit more from. And that's a very difficult problem to do quickly. A prototype is running now in the UK for a few thousand cars, and energy is traded every 30 minutes, which means that any recalculation has to happen extremely quickly before the next batch of energy is traded. We already hit computational limits with a few thousand cars, so we worked with IBM Quantum and recently released a scientific paper online, which shows that we were able to speed this up using a hybrid classical and quantum computer by 10,000 times the speed. We can currently compute this solution for three million vehicles in 15 minutes.
Do you typically model these problems on classical computers first?
We develop the approach theoretically, thinking about each problem in a way that would make sense when applying quantum optimisation or quantum machine learning. We then run a quantum simulation on our own computers — so typically just a laptop running a quantum computer simulator. We get the core of the idea worked out end to end for the whole solution, and when we're ready we test the experiments on real quantum computers. We have access to IBM Quantum's fleet, which has quantum computers in New York State and in Germany. We regularly run examples on the real hardware. What we're striving to do is run real experiments on hardware in the so-called ‘utility’ scale. Utility scale quantum computing experiments are those which use over 100 qubits, which means you're not able to simulate them even with the world's largest super computers.
As quantum computers get bigger and better, we'll be able to run more complex algorithms, but right now the ones we run on the real machines have to be quite simple without too many operational steps. We'll be able to run much larger, much more robust calculations once we have quantum computers with quantum error correction, which is the next era; the fault tolerance computers which IBM and others have outlined in their development road maps. But it’s a very exciting time. We're already developing algorithms that will work on tomorrow's quantum computers, so it's still useful today to onboard the organisation and train teams to think from a quantum perspective. And in the meantime, we’re generating scientific publications and patents in preparation for when the next generation of production machines are ready.
For you
Be part of something bigger, join BCS, The Chartered Institute for IT.
How will the wider use of quantum computers affect energy consumption?
There are different types of quantum computers. The ones we commonly talk about today are those that use superconducting qubits and rely on liquid helium to cool them down to several millikelvin, thousandths of a degree above absolute zero. But, there are other kinds of quantum computers, such as nitrogen vacancy (NV) based quantum computing, which uses wafers of diamond that operate at room temperature; so essentially room temperature qubits. At the moment, it’s a race between hardware providers to develop the most appropriate style of quantum computer that will allow scaling to millions of cubits. If we do end up going down the superconducting route, we’re not necessarily limited to the use of energy dilution refrigerators to cool down the chips in the classic chandelier picture of a quantum machine that everyone is familiar with, as technology and our approaches to finding cooling solutions are evolving so fast. In fact, if you look at the IBM road map, chips sizes and therefore compute capability is growing year on year, with essentially the same amount of cooling energy required. You haven't changed the energy. And when you consider that the ultimate goal of quantum is to quickly solve problems that classical computers will never be able to solve, even given years of continued supercomputing runtime to run their classical algorithm versions, then quantum suddenly becomes a much more efficient prospect.
Do you have some final thoughts for BCS members?
I would encourage people who are curious about quantum to look at the online resources out there, particularly through IBM Quantum. They have many free quantum related resources including a Python software development kit. For student members, if they’re interested in computer science and physics, then quantum is a fascinating area to work in because it really does represent the future, with game changing technology that operates in a different way to anything we’ve ever experienced. And, if you are a senior business leader or IT leader I would encourage you to start thinking about whether quantum may make a difference to your business. Are there certain organisational problems to which quantum computing may hold the answer? I believe so. 20 years ago people could not have imagined where AI could be applied in a business setting, and now it’s everywhere. Quantum has fantastic potential.
