Abstract
SINTEF is a not-for-profit research institute working with industrial and governmental customers on both small and large projects. In this talk we will give an overview of the ongoing optimization modelling work at SINTEF using JuMP and Julia with a focus on creating flexible and generic optimization frameworks.
Two examples will be highlighted, a location and dimensioning model for hydrogen infrastructure and a model for handling circular logistics in the crushed stone and gravel industry. Through these examples we will show how JuMP can be used to create large scale optimization models in a modular manner. Based on these experiences and our needs, we have been experimenting with several packages that extend or complement JuMP functionality. This includes UnitJuMP for using physical units in the modelling, SparseVariables for efficient and easy handling of optimization variables with a sparse structure and TimeStruct for flexible time structures. Combining their features together with JuMP has reduced development time, increased robustness and paved the way for more code reuse through the use of separate and well tested modules.
Two examples will be highlighted, a location and dimensioning model for hydrogen infrastructure and a model for handling circular logistics in the crushed stone and gravel industry. Through these examples we will show how JuMP can be used to create large scale optimization models in a modular manner. Based on these experiences and our needs, we have been experimenting with several packages that extend or complement JuMP functionality. This includes UnitJuMP for using physical units in the modelling, SparseVariables for efficient and easy handling of optimization variables with a sparse structure and TimeStruct for flexible time structures. Combining their features together with JuMP has reduced development time, increased robustness and paved the way for more code reuse through the use of separate and well tested modules.
