## Stochastic Optimization of Zero Emission Buildings

##### Master thesis

##### Permanent lenke

http://hdl.handle.net/11250/2561562##### Utgivelsesdato

2018##### Metadata

Vis full innførsel##### Samlinger

- Institutt for elkraftteknikk [1636]

##### Sammendrag

Zero Emission Buildings (ZEBs) are energy efficient buildings that produce on-siterenewable energy to compensate for their consumption. The concept of ZEBs isbased the EU's Energy Performance of Buildings Directive (EPBD) of 2010,demanding that all new buildings constructed after 2020 are to reach "near zeroenergy level. Previous research on energy systems in ZEBs have used deterministiclinear optimization techniques to determine the cost-optimal design of investedtechnologies in such sustainable buildings.
The main contribution of this thesis is the development of a stochastic two-stagemodel, formulated as a Mixed Integer Linear Program (MILP), that determines thecost-optimal investments and operations of a ZEB. The model accounts foruncertainty in the short-term operational patterns; the fluctuations in the outdoortemperature, the spot price of electricity and solar irradiation. The two the mainobjectives are: 1) To compare the optimal technology design of the deterministic andstochastic model counterparts and 2) to investigate the possibilities of the investmentof an electric battery. Emissions constraints are formulated to fit the ambition levelknown as the "ZEB-O" level, only considering emissions caused in buildingoperations.
The model input data is simulated to fit the hourly demand of electricity and heatingin a Norwegian passive house. Time series on simulated demand from 2010 to 2014are used to construct operational scenarios. Realistic investment costs of buildingtechnologies are used based on an extensive survey of Norwegian manufacturers'prices. Clustering analysis is used to reduce the computational effort by selectingseasonally representative hours to imitate a full year of operations.
Results show that a stochastic model can better, than its deterministic counterpart,account for the following: (i) Cover the peak heat demand of periods colder than thedeterministic input data, and (ii) avoid over-dimensioning of the installed base-loadcapacity. The net present value of the total costs can be reduced by 1/6, whichrepresents the quantitative value of using a stochastic model in the place of adeterministic model. Furthermore, the stochastic model is used to analyze the impactof a "power subscription" grid tariff scheme and battery operations in ZEBs. Thebattery is not a cost-optimal technology in ZEBs due to the forced reinvestmentsevery 10th year imposed by the stochastic two-stage formulation. Sensitivity analysisshow that the battery specific investment costs (EUR/kWh of storage capacity) mustbe reduced by 90 \% to become part of the solution.