Daniel Montes López is an engineer in the process systems area, where he uses modeling, simulation, optimization, and control tools to find solutions to industry problems. During his degree, Daniel was particularly interested in modeling and simulation, which led him to develop his undergraduate thesis in this area. He discovered that the same knowledge required for simulating processes could also be used for controlling and optimizing them.
Daniel continued his research and is currently working on his doctoral thesis at the Sustainable Processes Institute at the University of Valladolid. His research is focused on developing decomposition methods to solve large-scale optimization problems involving discrete variables. In other words, optimization seeks to find a series of decisions that minimize a certain objective function and satisfy some constraints. Some of these decisions can be represented by discrete variables, such as turning a machine on or off, feeding a reactant at a certain time instead of another, etc. This greatly complicates the resolution of the general problem, so the aim of his work is to break it down into smaller parts that are easier to solve and allow for shorter resolution times.
The decomposition method he has developed is intended to be applied to scheduling problems, in which some variables may be uncertain, such as weather or material arrival. This can be incorporated into the decision-making process through what is called two-stage programming. The problem is that when applied to real-scale facilities, the computation time becomes prohibitive due to the mathematical problem’s dimension. With the method he has developed throughout his thesis, significant reductions in computation time have been found, making it possible to take into account uncertainty in large-scale industrial installations.
One of the case studies he used in his research was a wood fiber production factory. The main process of this factory uses five solvents to generate the wood fibers, which are diluted and lose quality over time. To avoid disposal and save costs, they are processed in 23 evaporation plants. However, these plants get dirty over time, increasing the facility’s energy consumption, so they must be cleaned after a certain time to recover their nominal efficiency. On the other hand, each plant’s processing capacity depends on the outside temperature, and a one-month stoppage for cleaning and product allocation to plants must be planned. If uncertainty is taken into account, the resulting optimization problem has around 600,000 variables, 90% of which are discrete, requiring a computation time of 10 days. With the decomposition method studied by Daniel, the same solution can be achieved in just 3 hours.
In summary, it could be said that incorporating uncertainty in the planning process of an industrial plant would allow for better decisions by adapting operations to the situations that may arise over time. In this sense, Daniel Montes’s research contributes to achieving that goal by greatly reducing computation times without compromising the quality of the mathematical solutions. For Daniel, his work is fascinating, and he enjoys finding solutions that can improve the efficiency of industrial operations. Furthermore, his experience in this area has taught him that “problems must always be seen from different angles, from a general perspective. Sometimes solutions come from where you least expect them.”
