Greybox modelling of buildings
Author
Harnik, Michael
Term
4. term
Publication year
2019
Abstract
Dette speciale undersøger, hvordan gråboksmodeller kan udvikles og anvendes til at beskrive og styre varmeforbruget i et enfamiliehus i konteksten af smarte energinet. Arbejdet opstiller en metodik fra første principper: et RC-baseret (modstand–kapacitans) varmemodelkoncept formuleres som differentialligninger, parametre estimeres ud fra målinger, og modellen valideres i tilstandsrumrepræsentation. Som led i metodeafklaringen indgår en EnergyPlus-reference samt valg af software og en systematisk modelprocedure. I resultatafsnittet sammenlignes modeller med forskellig kompleksitet, herunder lette og tunge konstruktioner, og der diskuteres centrale udfordringer som parameterestimation og modelvalidering. Afslutningsvis implementeres modelprædiktiv styring (MPC) baseret på en udvalgt gråboksmodel for at undersøge potentialet for at reducere varmeenergiforbruget sammenlignet med traditionel PI-regulering. Kvantitative resultater er ikke angivet i uddraget, men specialet fokuserer på at etablere en robust metode, belyse afvejninger mellem nøjagtighed og kompleksitet og evaluere praktisk anvendelighed for energifleksibel bygningsdrift.
This thesis explores how grey-box models can be developed and applied to represent and control heating demand in a single-family house within the context of smart energy grids. The work establishes a step-by-step methodology: an RC-based (resistance–capacitance) thermal model is formulated as differential equations, parameters are estimated from measurements, and the model is validated using a state-space representation. The methodology is supported by an EnergyPlus reference, software choices, and a systematic modeling procedure. In the results section, models of varying complexity—including light-weight and heavy-weight constructions—are compared, and key challenges such as parameter estimation and model validation are discussed. Finally, model predictive control (MPC) is implemented using a selected grey-box model to assess the potential for reducing heating energy use relative to traditional PI control. While quantitative outcomes are not provided in the excerpt, the thesis emphasizes a robust modeling workflow, trade-offs between accuracy and complexity, and practical applicability for energy-flexible building operation.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents