Date of Award

1-1-2013

Document Type

Masters Thesis

Degree Name

M.S.

Organizational Unit

Daniel Felix Ritchie School of Engineering and Computer Science, Electrical and Computer Engineering

First Advisor

Kimon Valavanis, Ph.D.

Second Advisor

Matthew Rutherford, Ph.D.

Third Advisor

Davor Balzar

Fourth Advisor

Robert Whitman

Keywords

Building model, Electricity markets, Predict and supply, Smart building, Smart grid, Supply and demand

Abstract

A smart grid is an electricity network that accommodates two-way power flows, and utilizes two-way communications and increased measurement, in order to provide more information to customers and aid in the development of a more efficient electricity market. The current electrical network is outdated and has many shortcomings relating to power flows, inefficient electricity markets, generation/supply balance, a lack of information for the consumer and insufficient consumer interaction with electricity markets. Many of these challenges can be addressed with a smart grid, but there remain significant barriers to the implementation of a smart grid.

This paper proposes a novel method for the development of a smart grid utilizing a bottom up approach (starting with smart buildings/campuses) with the goal of providing the framework and infrastructure necessary for a smart grid instead of the more traditional approach (installing many smart meters and hoping a smart grid emerges). This novel approach involves combining deterministic and statistical methods in order to accurately estimate building electricity use down to the device level. It provides model users with a cheaper alternative to energy audits and extensive sensor networks (the current methods of quantifying electrical use at this level) which increases their ability to modify energy consumption and respond to price signals

The results of this method are promising, but they are still preliminary. As a result, there is still room for improvement. On days when there were no missing or inaccurate data, this approach has R2 of about 0.84, sometimes as high as 0.94 when compared to measured results. However, there were many days where missing data brought overall accuracy down significantly. In addition, the development and implementation of the calibration process is still underway and some functional additions must be made in order to maximize accuracy. The calibration process must be completed before a reliable accuracy can be determined.

While this work shows that a combination of a deterministic and statistical methods can accurately forecast building energy usage, the ability to produce accurate results is heavily dependent upon software availability, accurate data and the proper calibration of the model. Creating the software required for a smart building model is time consuming and expensive. Bad or missing data have significant negative impacts on the accuracy of the results and can be caused by a hodgepodge of equipment and communication protocols. Proper calibration of the model is essential to ensure that the device level estimations are sufficiently accurate. Any building model which is to be successful at creating a smart building must be able to overcome these challenges.

Publication Statement

Copyright is held by the author. User is responsible for all copyright compliance.

Rights Holder

Jonathan Falcey

Provenance

Received from ProQuest

File Format

application/pdf

Language

en

File Size

220 p.

Discipline

Electrical engineering, Energy

Download



Share

COinS