First-principles-based dynamic models of industrial-scale chemical processes involve the definition of many equations and variables, which can demand a considerable effort to setup and maintain. Existing tools require the modeler to either write all of the equations or to select them from a repository. The Daedalus Modeling Framework is a novel software library that expedites the creation of nonlinear dynamic lumped parameter models by employing an automatic equation and property selection algorithm. It uses a phenomena-oriented modeling strategy to define a higher-level model description, which can generate different systems of equations, depending on the available model inputs and requested outputs. This paper describes the design decisions and implementation strategy for this framework, such as its basic building blocks, model structure selection algorithm, differentiation index reduction, algorithmic differentiation, and documentation generation. Two examples demonstrate the modeling capabilities of Daedalus and its integration with a nonlinear model predictive control framework.