Optimization And Collaboration Between Research And Industry. Interview to Álvaro García Sánchez.


Hi Alvaro, thank you for accepting this interview. To start, could you briefly tell us about your
academic and professional journey, and how your interest in renewable energy research
developed?

Sure, I am an industrial engineer and I earned my master’s degree and PhD from the Technical
University of Madrid. My career has always been focused on solving combinatorial problems,
especially in the fields of production and research. In 2011, I founded Baobab Solutions, a
company that allows me to combine my passions for research, teaching, and collaboration
with companies. It was from this that my interest in wind energy optimization was born, a field
with enormous potential for collaboration between academia and industry.
At Baobab, we also work on projects for corporate clients, addressing real problems that allow
us to apply our research. Additionally, we participate in funded projects, which not only
improve our technological capabilities but also allow us to share our solutions with the
scientific community. This combined approach of research and practical application is at the
core of our work.
In your latest study, you focused on optimizing turbine placement in wind farms. Why is this
such a crucial factor for maximizing energy production?

The arrangement of turbines has a significant impact on the efficiency of a wind farm. If
turbines are too close together, they interfere with each other, reducing the amount of wind
available for each. Conversely, placing them in areas with more wind ensures greater
production. The challenge is to find the right balance between these variables, as an optimal
arrangement can significantly increase long-term revenues without incurring additional
operational costs.
The interesting part is that, once a wind farm is built, operational costs generally remain
stable. Therefore, optimizing turbine placement means obtaining more energy over the life of
the farm without further investment. It’s a strategy that increases profitability without raising
costs. Thanks to this approach, companies can reinvest the generated profits into new projects
or improvements, making energy more accessible overall.

It’s fascinating how optimization can increase production without additional costs. In practice,
how does this process work, and how much can it impact the results?
Exactly, this is the crux of the issue. When we optimize turbine placement, we can make the
most of the available wind for decades, without additional operational costs. By simply
improving placement, a wind farm can produce up to 10% more energy, which represents a significant increase in revenue for the company. It’s a strategy that generates immense long-
term value.

In comparing different optimization methods, you found that random biased optimization is
particularly effective. How does this approach differ from traditional techniques like integer
programming or particle swarm optimization (PSO)?

Integer programming works well for small problems, but it becomes ineffective at larger
scales. Random biased optimization, on the other hand, is a method that allows us to explore
many more combinations of solutions in a smarter and more flexible way. We don’t limit
ourselves to selecting the best solution in a single step; instead, we create a variety of
possible solutions, exploring them in a guided manner.
This technique, when combined with PSO, allows us to achieve even better results. We have
discovered that by combining these two techniques, we can improve the efficiency of a wind
farm by 10% compared to traditional solutions already in use in the industry.


This 10% improvement in energy production is significant. Can it also be applied to other
industries?

Yes, absolutely. This approach is applicable to many other industries, such as logistics and
manufacturing. In these sectors, similar problems are faced, where there are many variables to
consider, and the optimal choice is not always obvious. Combinatorial optimization is a very
powerful technique that can be used to solve complex problems in sectors like healthcare,
resource management, and logistics. Ultimately, any industry facing complex decision-making
with many options could benefit from these techniques.


You mentioned the relationship between research and companies. How has the attitude of
companies toward using optimization techniques changed over time?

When we started Baobab, it was very challenging to convey what we did to companies,
especially when we talked about “operations research.” It was an abstract concept for many,
and there was skepticism about whether we could find better solutions than those developed
by business experts over decades. Often, we heard comments like, “I’ve worked in this factory
for 20 years; how can an algorithm do better than me?” This kind of distrust was common.
Now, thanks to machine learning and artificial intelligence, companies are much more open.
The concept of “machines that learn” is easier to understand, and people are accustomed to
seeing practical applications of these technologies, such as sales forecasting models or image
classification. As a result, it is easier for us to explain how optimization can improve their
processes and provide tangible results.
Today, companies trust the mathematics and models we propose much more; the challenge is
no longer to convince them of the validity of the technology but to demonstrate that the
investment is worth the benefits gained. The focus has shifted to proving the economic return
because trust in technology is now established.

Looking to the future, do you see new areas of research or emerging applications?
Certainly, there are several promising lines of research. Historically, machine learning and
operations research have been two separate worlds, but now there is a growing integration
between these approaches. For example, techniques like reinforcement learning are showing
tremendous potential for further improving business decisions.
In the future, I see the possibility of combining these tools to tackle even more complex
problems. The idea is not only to use machine learning for predictive data but to integrate it
directly into operational decision-making processes to generate optimal solutions in real time.
It’s a rapidly growing research field that could provide extraordinary solutions for sectors like
energy, logistics, and manufacturing.


Finally, how do you see the future of decision science and what can its greatest contributions
be?

Decision science has enormous potential in fostering collaboration between
academia and industry. This will enable us to tackle problems in a more realistic and efficient
way, as well as to open new avenues for addressing complex issues such as demographics and
natural disasters. I am convinced that the DSA will play a central role in this evolution,
facilitating collaborations and promoting innovation on a global scale.


Thank you very much, Alvaro, for your time and availability. We look forward to having you
with us again at the next ISC conference.

Thank you, Matteo! I will do my best to be there; I definitely don’t want to miss it!