[Paper Review] COMANDO: A Next-Generation Open-Source Framework for Energy Systems Optimization

Marco Langiu, David Yang Shu+8 more|arXiv (Cornell University)|Feb 3, 2021

Integrated Energy Systems Optimization89 references63 citations

TL;DR

COMANDO is an open-source, component-oriented Python framework for nonlinear, dynamic, and stochastic optimization of integrated energy systems. It enables flexible modeling of hybrid mechanistic/data-driven components, supports advanced formulations like stochastic programming and surrogate modeling, and provides extensible tools for global optimization, demonstrating its versatility through four diverse case studies including dynamic and stochastic optimization with neural networks.

ABSTRACT

Existing open-source modeling frameworks dedicated to energy systems optimization typically utilize (mixed-integer) linear programming ((MI)LP) formulations, which lack modeling freedom for technical system design and operation. We present COMANDO, an open-source Python package for component-oriented modeling and optimization for nonlinear design and operation of integrated energy systems. COMANDO allows to assemble system models from component models including nonlinear, dynamic and discrete characteristics. Based on a single system model, different deterministic and stochastic problem formulations can be obtained by varying objective function and underlying data, and by applying automatic or manual reformulations. The flexible open-source implementation allows for the integration of customized routines required to solve challenging problems, e.g., initialization, problem decomposition, or sequential solution strategies. We demonstrate features of COMANDO via case studies, including automated linearization, dynamic optimization, stochastic programming, and the use of nonlinear artificial neural networks as surrogate models in a reduced-space formulation for deterministic global optimization.

Motivation & Objective

• Address the lack of open-source tools for nonlinear, dynamic, and uncertain energy system design and operation that go beyond traditional (M)ILP formulations.
• Enable component-oriented modeling with hybrid mechanistic and data-driven models for greater modeling freedom in technical system design.
• Support simultaneous optimization of design and operation under uncertainty using two-stage stochastic programming.
• Facilitate integration of custom algorithms, including initialization, decomposition, and sequential solution strategies.
• Provide a flexible, extensible framework for algorithm development and advanced optimization techniques such as global optimization with surrogate models.

Proposed method

• Uses SymPy as a core computational engine for symbolic manipulation of mathematical expressions and automatic reformulation (e.g., linearization).
• Models components using nonlinear, dynamic, and discrete characteristics via differential-algebraic equations (DAEs), enabling accurate representation of system dynamics.
• Supports multiple problem formulations—deterministic, stochastic, and reduced-space optimization—by varying objective functions and data inputs.
• Enables integration of artificial neural networks as surrogate models in a reduced-space formulation for deterministic global optimization.
• Provides modular, reusable component models that can be interconnected to form complex system architectures.
• Supports both automatic and manual reformulation of optimization problems, enhancing compatibility with a wide range of solvers.

Experimental results

Research questions

• RQ1 Can a component-oriented, open-source framework support nonlinear and dynamic optimization of integrated energy systems more effectively than existing (M)ILP-based ESMFs?
• RQ2 How can hybrid mechanistic/data-driven models be seamlessly integrated into a single optimization framework for energy systems?
• RQ3 To what extent can stochastic programming formulations under uncertainty be efficiently implemented in a flexible, extensible open-source framework?
• RQ4 Can surrogate models based on neural networks improve the efficiency of deterministic global optimization in complex energy systems?
• RQ5 How does the framework enable advanced algorithm development, such as problem decomposition or sequential solution strategies?

Key findings

• COMANDO successfully enables the optimization of energy systems with nonlinear, dynamic, and discrete characteristics, overcoming limitations of traditional (M)ILP-based frameworks.
• The framework supports automated linearization of nonlinear models, demonstrating its capability to handle complex reformulation tasks efficiently.
• Dynamic optimization of a heat exchanger network was achieved using DAE-based modeling, showing accurate representation of transient behavior.
• Stochastic programming formulations were successfully applied to optimize system design under uncertain demand and price scenarios, improving robustness.
• A case study using a nonlinear artificial neural network as a surrogate model in a reduced-space formulation enabled deterministic global optimization with significant computational savings.
• The framework’s modular and extensible architecture allows for integration of custom solution strategies, initialization routines, and solver interfaces, enhancing its adaptability to complex problems.