Quantum Computing and Computational Fluid Dynamics
The Time to Plan Is Now
If you’re wondering why a former banker and a steel industry contributor are writing an article about quantum computing, don’t be too surprised. We, like any other business person, are trying to solve problems. People use tools to solve problems and quantum computing is merely another tool. An incredibly fast processing tool. One that will change the face of so many industries in the future, much like artificial intelligence is doing today.
Today’s manufacturing sector faces a myriad of challenges including tariffs, foreign dumping, skilled labour shortages, cost escalations and more. So, executives are looking for ways to automate repetitive tasks and transition employees on to more value-add activities. In a 2020 Deloitte Insights study, they argued that “manufacturing leaders will likely need to understand the following to differentiate themselves… The pace of disruption that digitization brings, and the investments manufacturers can make to best position themselves to succeed.” In the same study, Deloitte argues that while manufacturers will be heavily disrupted by digitization, many are not prepared for the transition. So, we would argue the time to plan is now.
Computational Fluid Dynamics
In this article, we take on the challenge of a key manufacturing industry research tool, namely computation fluid dynamics or CFD. CFD is used for product development and fluid flow problems, providing solutions for optimizing manufacturing equipment and processes. It is also used for specialized, high-value problems to understand fluid behaviour for certain given conditions.
According to Marcin Rojek, co-founder of byteLAKE, Computational Fluid Dynamics tools combine numerical analysis and algorithms to solve fluid flow problems. They are used to model fluid density, velocity, pressure, temperature, and chemical concentrations in relation to time and space. A range of industries such as automotive, chemical, aerospace, biomedical, power and energy, and construction rely on fast CFD analysis turnaround time. It is a key part of their design workflow to understand and design how liquids and gases flow and interact with surfaces. Typical applications include weather simulations, aerodynamic characteristics modeling and optimization, and petroleum mass flow rate assessment.
A lot of CFD modeling effort goes into asking the right questions and optimizing the computing power to get meaningful results — including mesh size, geometry, and fluid characteristics. These simulations currently take up to a month to run with one set of parameters because of the complexity. Parameters often include fluids, density, velocity, elements, purity, equipment specifics, air mixing, and more.
What is Quantum Computing?
Quantum computing (QC) is a new (think the past decade) way of computing. Rather than using the classical bit paradigm (1’s and 0’s, on or off states), quantum computers use something called qubits. Qubits can have two states, 1 and 0 at the same time. Quantum computers take advantage of this entangling of the states to, in highly specific cases today, achieve algorithms that are exponentially speedier than their classical counterparts.
Several companies have taken the lead in developing quantum computers, namely Honeywell (as of writing, the fastest quantum computer at 64 qubits), as well as DWave, IBM and Righetti. These companies, much like those manufacturers the original personal computers and their chips, are racing for supremacy in the fastest, least error, most cost-effective quantum computers.
How AI & QC are changing/will change the face of CFD?
Most of the leading insights about quantum computing originate from academic and research environments. Take for example, MIT, Brian’s alma mater. Professor Seth Lloyd and team experienced a breakthrough recently applying algebraic equations (used also in solving CFD problems) to bridge the gap between CFD and quantum computing algorithms.
In 2018, Steijl and Barakos published a paper called “Parallel evaluation of quantum algorithms for computational fluid dynamics”. Sounds complex right? Well, while it was complex, what this pair tried to do was better understand the noise and error impact within quantum computer algorithms (as they were in 2018) within a flow solution. They found that “meaningful flow simulations can be performed using a hybrid classical/quantum hardware approach.”
So, work is being done specific to CFD and QC, but what are the benefits and what are the realities?
Increased Number of Viable Designs
Growing the capability within an organization to use CFD in conjunction with QC could have breakthrough benefits in the long term as a research department could make use of several thousand simulations to develop better manufactured parts (lighter, stronger, cheaper, more consistent, better quality) or to solve production problems that required deep study using CFD today. These solutions could be optimized over several thousand scenarios, using more parameters, if the computation time was reduced using QC.
Because of the computational power needed to run CFD simulations, assumptions are made about some model inputs. This would be less relevant when using QC because more models could be run in a much shorter time. So, these inputs could be varied until the model is more accurate when compared to the real world. Then that model could be used to hone in on a better solution before starting trials in the plant.
Reduce Modelling Time
Solving computational fluid dynamics problems using traditional methods requires a lot of time and a lot of effort. For researchers who rely on data, and their interrelationships, running simulations to understand complex dependencies is an everyday task.
Combining AI and quantum computing will offer faster computing (e.g. architectures that are optimized for high volume, rapid, physics-based calculations) and have multiple effects on CFD modeling and use for manufacturing. AI models have been developed to simulate the results of physics-based models with fair results, but a lot of training data is required to get AI models of comparable accuracy for a given problem. QC would not only allow for faster training of AI models, it could allow for multiple variations of the research goal to be simulated and compared in seconds instead of days or weeks.
In one example, a CFD simulation exercise was performed by Germany’s Renumic GmbH and UberCloud with the objective to decrease solving a fluid flow problem by 1000 times with the same accuracy as a traditional CFD solution process. The exercise found, not surprisingly with large amounts of data, they had better accuracy, six times faster than the traditional process, with tens of thousands of samples available in hours rather than days. Add the capabilities of quantum computing and that time span could eventually be minutes.
Realities of Investing in Quantum Computing
Four years ago, when artificial intelligence was still a nascent concept for the majority of people compared to where it is today, few people globally believed in its capabilities. Four years on, artificial intelligence is embedded in our smartphones, our online movie platforms, our research environments, even our car’s GPS. We believe that quantum computing will have a similar trajectory. Yet we also acknowledge business leaders must consider the risks, compliance, rewards and cost considerations for QC & CFD going into the future.
At this point in time, investing in quantum computing is likely best entered into as a partnership between academic institutions, quantum computing firms like Honeywell, D-Wave and IBM, and with research arms of companies that already use CFD to good effectiveness. Having that expertise built into your company and/or your strategic partnerships will be key to leveraging QC in the future.
There are still concerns about the errors in QC computing and comparable results may be as much as 10 years off for some applications, but this is the time horizon of difficult research programs and considering QC as one arm to a comprehensive research program will mean early and quick adoption when technology is proven.
To minimize business risk and build capability, business leaders can start out with pilots and small use cases, like adopting any new digital tool. With low-risk investment in AI/QC at the small scale, leaders could demonstrate competency and build out capability prior to large scale investment. This will increase the trust in results when the time comes for those larger investments.
Conclusion
Quantum computing is considered by the financial services industry as an emerging technology with high potential business impact, second only to artificial intelligence. The manufacturing industry, and CFD participants more specifically, will benefit from advances in technology like quantum computing, however the time is now to start planning for the introduction of such capabilities within your own organization.
Copyright 2020 Aquitaine Innovation Advisors
Brian Lenahan is the author of three Amazon-published books on artificial intelligence including the Bestseller “Artificial Intelligence: Foundations for Business Leaders and Consultants”. He is a former executive in a Top 10 North American bank, a University Instructor, the Author-in-Residence of the AI Geeks community and mentors innovative companies in the Halton and Hamilton areas.
Email: ceo@aquitaineinnovationadvisors.com
LinkedIn: https://www.linkedin.com/in/brian-lenahan-innovation/
Stephanie Holko is Digital Transformation Project Manager at ArcelorMittal Dofasco in Hamilton, Ontario. She has over 15 years of steel industry experience, graduated with a Bachelors in Chemical Engineering, and holds an MBA with specialization in Management of New Innovation & Technology.
Email: stephanie.holko@outlook.com
