Available via license: CC BY 4.0
Content may be subject to copyright.
Energy Efficiency Effectiveness of Smart Thermostat Based BEMS
Koushik Mandlem, Bhaskaran Gopalakrishnan
*
, Ashish Nimbarte, Roseline Mostafa and Rupa Das
Industrial and Management Systems Engineering, West Virginia University, Morgantown, 26505, USA
Corresponding Author: Bhaskaran Gopalakrishnan. Email: Bhaskaran.Gopalakrishnan@mail.wvu.edu
Received: 07 May 2020; Accepted: 18 June 2020
Abstract: The research details the design and development of a spreadsheet based
software system that has been built using the principles of Monte Carlo simula-
tion. The simulation has been applied to a residential building with a certain num-
ber of rooms, each with specific characteristics pertaining to volume, occupancy,
and thermostat set point. The consideration of variables related to the building
envelope and weather, and the HVAC system have provided a realistic view that
enables the accurate estimation of annual energy usage and costs. The main find-
ings of the research are reflected in the sensitivity analysis that estimates energy
use and cost based on variance in the set point of thermostats that operate based
on occupancy levels in various times of the year when the weather pattern could
change drastically with respect to time, under the inherent weatherization levels of
the building. The most important conclusion is that, upon analyzing the results of
the sensitivity analysis, specifically with respect to the operations of the thermo-
stats and occupancy levels under constant building envelope characteristics, deci-
sion makers will be able to incorporate energy efficiency measures and
infrastructure improvements over time, in order to save energy and cost.
Keywords: HVAC; BEMS; smart thermostat; HVAC zoning
1 Introduction
Global energy consumption continues to increase each day. In the 1970s, the global energy usage was
about 3860 Million Tons of Oil Equivalent (MTOE), and within ten years, the numbers are almost doubled
and took the global energy usage to 6630 MTOE [1]. The average US energy consumption in the previous
decade was around 100 quadrillion BTUs, varying from 94 quadrillion BTUs to 101 quadrillions BTUs [2].
According to the US Department of Energy, buildings account for about 40% of the total energy
consumption, of which HVAC systems is 42.5% in the United States [2]. Hence, enhancing the research
of the HVAC operation system can result in opportunities to save energy and therefore reduce the
operating costs for a residential or industrial area.
People install thermostats at fixed places to sense the temperature of rooms and regulate the operation of
the HVAC systems. Room temperature is changed with the occupant’s requirements. A typical residence uses
a single thermostat in a fixed position, although the temperature profile is not uniform across the thermal
space. The insulation and infiltration in individual rooms may not be the same, so they respond
differently to the conditioning systems. For the shortcoming of occupancy detection, a single sensor
This work is licensed under a Creative Commons Attribution 4.0 International License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original
work is properly cited.
Energy Engineering
DOI:10.32604/EE.2020.011406
Article
ech
T
PressScience
control system may unnecessarily cool or heat the thermal environment and thus result in higher energy
consumption. New generation control systems along with zoning, could deploy several sensors or
thermostats in different rooms to regulate the flow in the unoccupied spaces. Then the system will learn
human behavior continuously and control the airflow into the rooms. Though implementing this strategy
helps in enhancing the occupant’s comfort with a reduction in energy consumption, it still has drawbacks.
As these systems have several moving parts, the equipment has a high probability of breaking down. A
survey was done by EIA about human behavior shows that 45% of people use central control, and 31%
of people use individual control air conditioning units. Thus, it is essential to know the estimated energy
savings if the zoning systems with advanced controls are to be installed.
The primary research objectives are:
To develop a model to determine the HVAC energy consumption. The model will have inputs with
varying parameters that are associated with building characteristics and HVAC characteristics.
The research aims to compare the impact of factors occupancy, weather, the volume of the room,
building envelope and thermostat setpoint on energy consumption.
The main findings of the research are reflected in the sensitivity analysis that estimates energy use and
cost based on the variance in the set point of thermostats that operate based on occupancy levels in various
times of the year when the weather pattern could change drastically with respect to time, under the inherent
weatherization levels of the building. Upon analyzing the results of the sensitivity analysis, specifically for
the operations of the thermostats and occupancy levels under constant building envelope characteristics,
decision-makers will be able to incorporate energy efficiency measures and infrastructure improvements
over time, to save energy and cost.
1.1 HVAC Controls
HVAC usually works at part load or is turned off depending on the heating and cooling loads,
occupancy, weather and other factors. Controls usually have a couple of sensors or a single sensor that
senses the current state and compares it with the desired state. Finally, it turns the equipment on or off
based on the desired state. There are several ways to control the temperature. One such method is to
control the air volume supplied into the room and the other is to regulate the load factor of the
equipment. Without proper controls, there might be overheating or overcooling of the space.
1.1.1 Single Thermostat System
In a single thermostat control system, there is only one thermostat in the whole space governing the
operation of the HVAC system. The thermostat senses the temperature of the area around it and regulates
the HVAC system accordingly. A research done by EPA shows, in winter, about 42% of them turn the
heat down and 2% of them turn their heating systems off [3]. This creates an opportunity to employ
programmable thermostats. Programmable thermostats are the thermostats that have the feasibility to
adjust the temperature based on the programs. It has been thought that programmable thermostats should
reduce up to 30% of the HVAC consumption by altering the heating and conditioning of the building as
per the occupant’s daily schedule. But the energy wastage factors of the programmable thermostat are: 1)
The temperatures are not set back if the occupants leave the house before the schedule, 2) The setback
temperatures are higher, which causes the system to run even when the house is vacant, 3) If the
occupants return back to the house earlier than the scheduled time they feel uncomfortable. This resulted
in people reducing their set back schedules or even stopping the usage of the schedules. This explains
why only 12% of the 118 million households program their thermostat though 41% of them own a
programmable thermostat [4].
166 EE, 2020, vol.117, no.4
The updated version of the programmable thermostat is the smart thermostat. The only difference
between a programmable and smart thermostat is the ability to connect to the internet and allowing the
users to adjust the heating or cooling settings from other internet-connected devices like smartphones and
computers. Though the smart thermostats were introduced, the energy wastage could not be minimized
completely. As a result, Smart Learning Thermostats were announced to the market. Smart Learning
Thermostats are those which repeatedly learn the occupancy and the setback schedule of the occupant
and act according to the learned behavior. This helps the thermostat to analyze the lifestyle of the
occupant and then preheat or cool the building while being energy efficient.
The problem with the single thermostat system without zoning is that the rooms which do not need
conditioning are also conditioned. Many people close the vents in the rooms, not in use. The HVAC system
and ductwork are designed considering the area of the entire building. If the vents are closed in the rooms
that are not in use, the air does not have any chance to move out of the system, resulting in an increase of
pressure in the ductwork and thereby developing air leaks in the ductwork. This may also result in
overheating the furnace or freezing the air conditioner. Zone controls are an answer to these problems.
1.1.2 Zoning Systems
HVAC zoning system employs dampers in the ductwork to regulate and redirect the airflow into different
areas (zones) in the building. Zoning helps in gaining more control over the HVAC system by turning the
system off as soon as the desired temperature is met in a particular zone. Zoning does not affect the
efficiency of the HVAC system and eliminates the problem of overheating or overcooling of the furnace or
air conditioner. It reduces energy consumption while addressing the comfort requirements of the occupants
by maintaining the zone at the desired state. The main components required for HVAC zoning include;
Central Control System: It is a control hub for all components such as thermostats, HVAC systems,
dampers, etc. The control system is the only source of communication between the system
components. It is the only processor for the entire zoning system.
Thermostats: In order to regulate the temperatures, each room should be facilitated with a thermostat.
The thermostat in each room sends the signal to the central control system. This helps the control
system to control the dampers to vary the airflow into the room to maintain the desired temperature.
Zone Dampers: Damper is a valve or a plate which can be regulated to vary the airflow. These are
wired to the control system. They open or close based on the signal from the thermostat present
in the room.
Bypass Damper: In a single duct system, if the damper is closed for one of the zones, pressure builds
up in the ductwork. Bypass damper acts as a relief valve, which helps in venting out the air.
1.1.3 Building Energy Management System (BEMS)
The main function of the Building Energy Management System is to regulate, monitor and control
HVAC systems. BEMS will employ computers to manage the energy consumption of the devices and
building technical services. BEMS can integrate the building’s HVAC and lighting. However, some
expensive state of the art BEMS can incorporate additional building systems such as video surveillance,
access control and interfacing with fire and security controls.
This research considers a residential house that has been zoned. It is installed with a BEMS, which takes
feedback from the smart thermostats with occupancy detection capability present in each zone. Thus, it
regulates the HVAC system’s load factor.
2 Literature Review
There are many options to reduce the energy consumption of the buildings sustainably. Because of the
numbers, there is more research done on improving the efficiency of HVAC systems. The factors which
EE, 2020, vol.117, no.4 167
influence the HVAC energy consumption are building envelope, HVAC equipment and its efficiency, level of
the thermostat, outside temperature, occupancy and controls.
There are many studies done on building envelope design and retrofits. Al-Saadi S and Al-Jabri K
believe that thermal characteristics of the building and building envelope such as heat load in the
building, indoor temperature, heat transmission and heat storage should be accounted for in the design
and construction phase of a building [5]. They developed several building envelopes designs and have
done a sensitivity analysis to narrow down several design factors like the amount of insulation for the
walls and doors, length of overhang on the windows, external shading, air infiltration, etc. The results
showed that minimum energy savings of 18% might be obtained by enhancing the design factors.
Another study done by McKay [6] proved that the amount of insulation in the walls and roof would have
a significant effect.
Many studies and surveys proved that there is an increase in satisfaction level for the occupants when
there is a thermostat in their room that they can control [7]. Guo et al. examined a double corridor classroom
building for the setback and set up temperature [8]. Guo et al. in their study created a smart thermostat that
acquires information by X10 sensors deployed in several rooms in 8 houses they experimented on [8]. There
are many ways to detect the occupancy; some of them require customers to wear a tag [9]. Gao et al. in their
study created a smart thermostat that acquires information by X10 sensors deployed in several rooms in 8
houses they experimented on [9]. Gupta et al. proposed an innovative way to estimate the occupancy be
calculating the arrival time of the resident [10]. Qela et al. attempted to study on self-learning thermostats
[11]. A study was done to evaluate the percentage of savings obtained on heating and cooling in
residential houses using Nest thermostats compared to manual and programmable thermostats [12]. Jin
et al. [13] proposed energy savings technique for BEMS which is based on occupancy sensors. Existing
IT infrastructure is used to collect accurate occupancy data inexpensively for non-energy purpose. This
study suggest to incorporate those occupancy information with the current BEMS system where, building
electronic devices such as HVAC and lighting usage can be monitored and controlled more effectively.
The simulation results showed significant energy reduction while they used occupancy information for
controlling building electronic devices.
After realizing the importance of energy savings, which can be achieved by installing programmable
thermostats, US Department of Energy’s Energy Star program started promoting programmable
thermostats as a means of reducing energy usage for residential customers anywhere from 10 to 30%
[14]. After such similar studies were done by US EPA and Energy Star [3], realizing that there were no
savings or sometimes increased usage from programmable thermostats, Energy Star decided to stop the
program in 2009. Two methods were developed by Van der Ham et al. [15] where they trained smart
thermostats to gain information about the house characteristics such as heat loss rate and heat capacity
over the time in order to ensure that homeowners are knowledgeable about the energy efficiency
improvement opportunities present. In another study, Danassis et al. [16] discussed about decision
making mechanisims to guide the temperature regulation of existing HVAC installations in a smart
building environment. Researchers used Artificial Neural Networks and Fuzzy Logic to train the decision
making process for the occupants’thermal comfort at an acceptable level of cost.
A model was developed by Muratori et al. [17] based on the overall thermal resistance theory. The model
performs a control volume analysis based on the fundamentals of thermodynamics. The model then estimates
the energy required for the HVAC system to raise or drop the ambient temperature of the room. White et al.
[18] developed a new model that uses monthly average outside temperature only to calculate total monthly
energy consumption. Some researchers have adopted either multiple linear regression methods or artificial
neural networks to estimate the energy consumption of HVAC systems. For example, Zhou et al. [19]
developed a model for commercial building by using a time series analysis method combined with a
168 EE, 2020, vol.117, no.4
regression model. Olofsson et al. [20] developed a model using neural networks to estimate the energy
demand based on a time interval of two to five weeks. Many engineering and statistical models were
proposed and developed by the authors for validation purposes.
Currently Building Energy Management (BEMS) is an important technology for energy consumption
monitoring and reduction. Cheng et al. [21] studied the published research work about the return on
investment of BEMS through the timeline from 1977 to 2017. Researchers have determined that the
payback periods for the BEMS reduced from 5.4 to 0.7 years for implementation in commercial
buildings. In another study Jones et al. [22] investigated the operational characteristics of energy
management systems for reducing the energy consumption. The designed system studied the efficiency
benefits of low temperature heating and the prosumer concept using a simple and inexpensive hardware
configuration.
3 Methodology
With million residential houses being built every year, it is important to know the effect of various
factors on the HVAC energy consumption [23]. Most of the studies overlooked the combined effect of all
the factors that affect energy consumption. In the residential sector, the lack of knowledge among the
users on the effect of factors such as insulation, thermostat level, zoning, and building envelope, leads to
significant investment in HVAC systems, which may not prove to be economically feasible.
The main objective of this research is to develop a simulation model that will allow the users to estimate
the total energy consumption of an HVAC system with a smart thermostat based BEMS in a residential
house. It provides an opportunity for the user to decide on purchasing the HVAC system and the
associated controls. The information flow diagram shown in Fig. 1 details the parameters used in the
simulation system.
Building Envelope
and Internal
Infrastructure
Characteristics
Smart Thermostat
Settings
HVAC System
Parameters
Energy Type and
Energy Cost
Occupancy
Energy Use
Simulation
Annual Monthly
Energy Usage Cost
for HVAC System
Geographic Location
and Weather
Figure 1: Flow Diagram of the simulation system
EE, 2020, vol.117, no.4 169
3.1 Description
The simulation is done in Microsoft Excel
®
based on the discrete event Monte Carlo Simulation
methodology. The randomness of the variables is generated using the heuristic approach. HVAC units are
usually measured often in terms of BTUH or Tons of HVAC. A ton of air conditioning can remove
12,000 BTUs per hour of heat energy from a space. The energy consumption by the system can be
calculated using the formula;
Energy consumption ¼RLF UF OH
where, R = Rating of the System (Btu/hr); LF = Load Factor (%); UF = Utilization Factor (%); OH =
Operating hours (hours).
The simulation is done with a frequency of 30 minutes for a whole year, specifying the utilization factor
and the annual operating hours. One of the main goals of the model is to determine the load factor of the
HVAC system at various intervals of time. As mentioned earlier, the load factor of the HVAC system is
adjusted based on five factors. The simulation model was created with a user-friendly interface while
maintaining a high degree of detail.
3.2 Monte Carlo Simulation
Monte Carlo Simulation is a method that is used to find solutions to mathematical and statistical
problems through statistical sampling. The results are computed based on random sampling and statistical
analysis. The main purpose of the Monte Carlo simulation is to understand the impact of the risk and
uncertainty of a particular system. The behavior of the systems can be observed by the experimental
process of drawing many samples and then analyzing their behavior. Though we have ample information
about the factors affecting the outcome, it is difficult to predict the future with minimum error. Monte
Carlo Simulation generates a possible outcome of a decision, thereby helping us to make a better decision
under uncertainty. The simulations help to build a model by substituting various values for any factor that
has uncertainty based on probability distribution. It records the results and continues to perform the same
calculations repeatedly. It uses a different set of random numbers from the distribution every time. To
generate different sets of random numbers continuously, it requires programming of a software system.
This study utilizes Microsoft Excel
®
VBA
®
(Visual Basic for Applications) to perform the simulation.
3.3 Heuristic Approach
Heuristics are simple and efficient guidelines that people often use to make conclusions. It is a method of
problem-solving, learning, or discovery that employs a practical method, which may or may not provide
optimal solutions but is sufficient for immediate goals to achieve the necessary levels of satisfaction.
3.4 Simulation Software
Visual Basic Application (VBA)
®
is a programming language for Excel
®
and other Microsoft Office
®
programs has been used for developing the simulation. It enables building user-defined functions and
automating processes. VBA allows users to automate by writing code or by creating macros. Excel
VBA
®
is employed to code the excel model, to perform the calculations and run the simulation model.
The visual basic code was used in this study. The model was developed using Visual Basic Applications
(VBA)
®
in Microsoft Excel
®
. The model was created with a user-friendly interface while maintaining
high degree of detail.
3.4.1 User Inputs and Model Outputs
One of the inputs which the user needs to provide is the configuration of the house, mainly the number of
bedrooms. The simulation considers one kitchen and a living room along with the bedrooms. The model also
requests the user if there are any plans for building envelope improvements and asks for the month in which
170 EE, 2020, vol.117, no.4
improvements are made if any. The model also asks for the number of times the thermostat settings are
changed. The user also has the privilege to input the volume, building envelope and thermostat levels. If
the user fails to input these values, the model generates a random level of all three factors. The model
also checks if there are any power outages in the area. If yes, the user needs to provide the maximum
number of power outages and the maximum time one power outage may last. The model also considers
the location where the user’s house is located as the weather varies all over the United States. The user is
guided with a map and region number for information, which can be helpful to determine the appropriate
region where the house is located.
This is a user interactive simulation. All the probability values for the occupancy, weather and room
volume (from Tabs. 1–5) are predetermined number for running the simulation. Then users can change all
these values according to their requirements. The simulation model allows the user to input the capacity
of the HVAC system and rates associated with the cost of electricity and natural gas. This allows the
model to calculate the energy consumption and the associated costs to run the HVAC systems. As many
residential areas are not subjected to electrical demand costs, this study does not consider any peak
demand or any costs associated with demand. After the user provides all the inputs, the simulation model
is executed. The completed simulation then provides an estimate of the annual energy consumption and
the energy costs that are generated based on the inputs provided by the user.
Table 1: Rules for occupancy of spaces
Bedroom
Start Time End Time Weekday Weekend
11.00 PM 9.00 AM 85% 85%
9.00 AM 5.00 PM 15% 50%
5.00 PM 11.00 PM 50% 50%
Kitchen
Start Time End Time Weekday Weekend
11.00 PM 6.00 AM 10% 10%
6.00 AM 2.00 PM 65% 65%
2.00 PM 5.00 PM 40% 40%
5.00 PM 11.00 PM 75% 75%
Living Room
Start Time End Time Weekday Weekend
11.00 PM 9.00 AM 20% 20%
9.00 AM 5.00 PM 40% 70%
5.00 PM 11.00 PM 60% 60%
Bathroom
Start Time End Time Weekday Weekend
12.00 AM 12.00 AM 15% 15%
EE, 2020, vol.117, no.4 171
3.5 Factors
3.5.1 Occupancy
This study calculates the energy consumption with an assumption that each room has a smart thermostat
with occupancy detecting capability. These can sense if the room is occupied or not and set back the
thermostat when the room is unoccupied. The occupant’s schedule and some heuristic rules are used to
sense the occupancy of a person in the room. If the user of the model feels that the schedule does not
match the base schedule used for the simulation, he or she can change the schedule accordingly. The
model has the capability to differentiate between a weekday and a weekend and change the heuristic rules
accordingly. As the frequency of the model is for 30 minutes, for each step, i.e., for every 30 minutes,
occupancy is sensed for all the rooms and occupancy level is established. The model estimates the time
of sleep and sleeping time of the user and retains the occupancy level until he or she sleeps and then the
occupancy level continues to update. The model is flexible; hence, if the user's activity is varied or if the
user opines to alter the probability matrices, this can easily be done in any of the following Tables,
starting with Tab. 1.
3.5.2 Weather
In this study, the weather is assumed to be at three levels—level 1 as mild, level 2 as moderate and level
3 as extreme weather conditions. The weather factor is updated every eight hours and is assumed to be the
same for the next eight hours. To determine the weather factor, the year is split into four seasons and the rules
are shown below.
It should be noted that the climate conditions vary as per the location. The United States is divided into
nine regions, as shown in Fig. 2.
The model uses heating and cooling degree days to determine if heating or cooling is required. Tab. 2
shows the probability chart for the three different weather levels throughout the year.
3.5.3 Volume of the Room
To lower or increase the room temperature, the HVAC should either add or remove heat from the area.
The amount of heat or cold air added to the room is directly proportional to the volume of the room. Three
levels are considered for the volume of the room—level 1 as small, level 2 as medium and level 3 as large.
The level is randomly generated if the user does not provide input. The randomness is generated, assuming
that all three levels have an equal probability of 33%. Tab. 4 below shows the standard floor area of different
rooms in residential buildings.
3.5.4 Building Envelope
The building envelope is considered as one of the critical factors to determine the load factor. The
building envelope includes walls, fenestration, roofs and doors. Two levels—level 1 as good and level 2
as bad are considered for the building envelope. The level for the building envelope is an input given by
Table 2: Rules to determine weather factor
Months Season level 1 level 2 level 3
March 1st–May 31st Spring 40% 40% 20%
June 1st–August 31st Summer 40% 35% 25%
September 1st–November 1st Fall 40% 40% 20%
December 1st–February 28th Winter 10% 30% 60%
172 EE, 2020, vol.117, no.4
the user. If the user fails to input the level for the building envelope, it is generated randomly. For the study,
two levels are considered for the building envelope and are distributed at 50% each.
3.5.5 Thermostat Setpoint
People prefer their comfort levels, i.e., one might prefer their room at a higher temperature and the others
might prefer their room at lower temperatures. As mentioned, the thermostat setpoint can affect energy
consumption. This is also another input from the user. Like the volume and building envelope, if the user
does not want to enter the level, it is randomly generated. There are two levels of thermostat factor—
level 1 for lower setpoint and level 2 for the higher set point. The tables below help the user and the
model to determine the thermostat level, respectively.
Figure 2: Census region [24]
Table 3: Rules to determine weather factor
Months Season Level 1 Level 2 Level 3
March 1st–May 31st Spring 40% 40% 20%
June 1st–August 31st Summer 40% 35% 25%
September 1st–November 1st Fall 40% 40% 20%
December 1st–February 28th Winter 10% 30% 60%
Table 4: Standard sizes of rooms in residential building
Floor Area (Sq.ft)
Level Bedrooms Bathroom Living Room Kitchen
1 144 54 216 50
2 250 70 320 104
3 384 96 616 120
EE, 2020, vol.117, no.4 173
The model also considers the number of times the thermostat settings are changed yearly. Then it updates
the thermostat level with the updated frequency (365/number of changes per year). If there are no changes to
the thermostat, the initial input from the user will be used for the calculations.
3.6 Load Factor Determination
Load factor is a function of the five factors—occupancy, outside weather, thermostat set point, building
envelope and volume of the room. Thus, assigning levels for all the five factors is a pre-requisite to
determining load factor. The level for the five factors is occupancy-2, weather-3, volume of the room-3,
building envelope-2 and thermostat set point-2. Thus, the maximum number of possible combinations are
considering all these levels and factors are 72 sets (2 × 3 × 3 × 2 × 2). Load factor is assigned to each
possible set (pre-determined). To determine the load factors, worst conditions are considered where
HVAC system is assumed to be running at full load. Then the load factors are assigned to the remaining
71 sets. At 30 minutes interval, new generated sets (5 factors) is compared to the pre-determined set and
a load factor is assigned to that set. An example of the determination of load factor for six sets is shown
below in Tab. 6.
3.7 Energy Costs Calculations
The capacity of the heater, tonnage and COP of the air conditioner, type of fuel and the rate structure is
the inputs given by the user. The heater and the air conditioner capacity are used while calculating the energy
required to heat and cool the building, respectively. kWof the air conditioner is calculated using the COP and
tonnage of the air conditioner.
The model has the capability of identifying if cooling or heating is required. The model uses degree days
of the respective region selected by the user. It then establishes a probability of cooling using the degree
heating days and degree cooling days. Tab. 7 below shows the heating degree days and cooling degree
days of the region 8—South Atlantic.
Table 5: User guidelines to determine the level for the thermostat
Level 1 Level 2
Summer 70–80°F (68%) 60–70°F (32%)
Winter 60–70°F (37%) 70–80°F (63%)
Table 6: Load factor assignment
Category Case 1 Case 2 Case 3 Case 4 Case 5 Case 6
Load Factor 100% 40% 80% 70% 20% 0
Occupancy 1 0 1 1 1 0
Weather 3 3 3 3 1 1
Volume of the Room 3 3 2 2 1 1
Building Envelope 2 2 2 1 1 1
Thermostat Set Point 2 2 2 2 1 1
174 EE, 2020, vol.117, no.4
Energy is calculated for the heater and air conditioner using the formula shown below.
kWh ¼kW 0:5LF
MMBtu ¼MMBtu=hr 0:5LF
kWh Cost ¼kWh $=kWh
MMBtu Cost ¼MMBtu $=MMBtu
where, kW = kiloWatt of Air Conditioner or Heater, MMBtu/hr = Capacity of the heater, LF = Load factor of
the system. The formula contains 0.5, as the energy consumption and energy costs are calculated on a 30-
minute interval. Once the energy is calculated for all the 30-minute intervals, the energy for the heater
and air conditioner are summed up on a monthly basis and on an annual basis.
3.8 Assumptions and Limitations
The following assumptions are made in this study: the frequency of the simulation is 30 minutes,
determined load factor would be the average load factors in that 30 minutes interval, building envelope
improvement stars at the beginning of the month, natural disasters were ignored, a smart thermostat is
used with occupancy sensor, zoning with variable air volume (VAV) control systems are available,
restrooms are tied with the respective bedrooms, ventilation and exhaust air effects are ignored, weather
factor updates only after 8 hours, no demand cost for residential area.
The limitations are that the system is not able to handle more than five bedrooms, only two energy
sources are considered, namely natural gas and electricity.
4 System Execution and Results
A two bedroom house as an example will be discussed here to explain how the model works. All the
inputs will be entered into the program.
Table 7: Probability of air conditioning [24]
Name HDD CDD Cooling
January 569 29 5%
February 460 30 7%
March 322 53 15%
April 129 84 40%
May 39 189 83%
June 2 338 100%
July 1 412 100%
August 1 392 100%
September 11 264 96%
October 118 123 52%
November 315 47 13%
December 445 44 9%
EE, 2020, vol.117, no.4 175
4.1 Example Input
The HVAC systems employ a natural gas furnace with 30,000 Btu/hr and 2 Tons air conditioner with
COP of 3 for heating and cooling, respectively. The user plans to upgrade the building envelope in June.
The user updates the thermostat two times a year. The user expects a maximum of 2 power outages in the
Morgantown, WV area; each power outage may last for 6 hours. It is assumed that the user’s house is in
Morgantown (Region 8—South Atlantic). The cost of natural gas and electricity considered are $7/
MMBtu and $0.1/kWh.
The monthly energy consumption and the energy costs associated with the simulation performed are
shown in Tab. 8 below.
It can be observed that there is no natural gas consumption in the months of “June, July, August and
September”as the user’s location does not have any requirement for heating in these months according to
the past ten years data. Thus, the possibility of heating in these months is very low. It can be observed
that there is a possibility for cooling in months “December and January”although they are winter months.
4.2 Sensitivity Analysis
To study the effect of different levels of each factor on energy consumption, simulation is performed
using varying combinations of pertinent variables. The results from the simulation software executions
are plotted. Though the levels for the factors change for each simulation run, some inputs are considered
to be the same for each run and they are; region, zero power outages, region 8, heater capacity 50,000
Btu/hr, and using a 3 ton air conditioner.
The total energy consumption for a residential house with regular controls are about 50% higher
compared to the house employed with a smart thermostat with occupancy sensor-based BEMS. To
observe the effect of energy consumption with and without occupancy detection, a graph is plotted
between the monthly energy consumption of HVAC system (Electricity and NG) vs. the occupancy
detection capability as shown in Figs. 3 and 4.
Table 8: Energy consumption
Month Electricity Consumption
(kWh/month)
Natural Gas Consumption
(MMBtu/month)
Electric
Costs ($)
Natural Gas
Costs ($)
January 9 12 1 84
February 68 11 7 77
March 35 8 4 56
April 242 5 25 35
May 491 1 50 7
June 470 0 47 0
July 485 0 49 0
August 487 0 49 0
September 483 0 49 0
October 261 2 27 14
November 56 5 6 35
December 54 10 6 70
176 EE, 2020, vol.117, no.4
The simulation results for the different volume level shows that energy consumption, both natural gas
and electricity increase as the volume level increase. The effect of the volume of room on energy
consumption is shown in the Figs. 5 and 6.
To inspect the effect of thermostat level, two simulation runs are performed; case 1 (level 1—low
comfort levels) and case 2 (level 2—high comfort levels). The energy consumed by the HVAC system is
higher in case 2, compared to the energy consumption in case 1. A bar graph is plotted between
thermostat comfort levels vs. the energy consumption as shown in Figs. 7 and 8to study the effect of
thermostat set points on the energy consumption.
The model also considers how many times the thermostat comfort levels are changed, and three different
simulation runs are completed. It can be observed that the energy consumption does not have any relationship
with the number of times the thermostat setpoints are changed. The Figs. 9 and 10 showcases bar graphs
showing the outputs from the three simulation runs.
The HVAC energy consumption depends on the building envelope and observed that the energy
consumed in the case 2 (level 2), is same as the energy consumed in case 3 (level 2 and will be upgraded
in July) for the first six months and case 1 (level 1) in the last six months, as the building envelope is at
level 2 till June 30th and at level 1 for rest of the year. To study the effect of building envelope on
energy consumption a graph is plotted as shown in Figs. 11 and 12.
Figure 3: Electricity and natural gas consumption for regular controls vs. smart controls
Figure 4: Natural gas consumption for regular controls vs. smart controls
EE, 2020, vol.117, no.4 177
Figure 6: Natural gas consumption for varying volume levels
Figure 5: Electricity consumption for varying volume levels
Figure 7: Natural Gas consumption for different thermostat comfort levels
178 EE, 2020, vol.117, no.4
To understand the importance of the user’s location while calculating the energy consumption, eight
simulation runs, one run for each location are performed. It can be observed that the costs associated with
heating (natural gas) is higher compared to costs associated with cooling (Electricity) in all the regions
except for South Atlantic and West South Central. Fig. 13 gives us information regarding energy costs
associated with a residential house with a similar configuration and thermal comfort levels in different
census regions throughout the United States.
4.3 Validation
The model has been applied to simulate the HVAC system for a typical household located in the area of
Morgantown, WV (i.e., South Atlantic region). The user has a heater with a capacity of 85,000 Btu/hr. The
user’s house has five bedrooms, living room and a kitchen. However, there is only one regular thermostat in
the living area. The occupancy of the house is very minimal, and the user prefers energy-efficient thermostat
Figure 8: Electricity consumption for different thermostat comfort levels
Figure 9: Natural gas consumption vs. # of times thermostat set points changed
EE, 2020, vol.117, no.4 179
setpoints (i.e., 60 degrees Fahrenheit in winter and 80 degrees Fahrenheit in summer). The building envelope
is tight with good insulation and very minimal insulation. The natural gas bills are obtained from the user and
heater energy consumption is estimated after considering the energy consumption for cooking and the water
heater. Heater energy consumption was observed to be 32 MMBtu per year. The simulation is performed
considering three bedrooms in order to mimic the configuration, as the user’s house is not zoned and has
only one regular thermostat. The occupancy pattern in the reference sheet is changed to meet the user’s
occupancy pattern. The model is run to simulate the energy consumption of the heater. The total energy
consumed by the heater turned out to be 37 MMBtu per year. The difference between the simulated and
the actual energy consumed is 5 MMBtu per year.
Figure 11: Electric consumption vs. building envelope levels
Figure 10: Electricity consumption vs. # of times thermostat set points changed
180 EE, 2020, vol.117, no.4
5 Conclusion
A simulation tool to estimate the energy consumption of a residential HVAC system with smart
thermostat BEMS was developed. The tool was implemented in Microsoft Excel
®
with Visual Basic
®
.
The technical features of the model include: 1) Compatibility with Windows operating system 2)
Capability of handling up to five bedrooms with a living room and a kitchen 3) Provides the user an
option to select from nine different locations where his residential area is located. The results of the
model include the total monthly and annual energy consumption and associated cost.
Acknowledgement: This research work acknowledges the association of the TransTech Conference held
annually by the National Research Center for Coal and Energy at West Virginia University.
Funding Statement: This research work was funded in part by the Industrial Assessment Center project,
supported by the US Department of Energy and by the West Virginia Development Office through grants
Figure 13: Total energy costs vs. location of user
Figure 12: Natural gas consumption vs. building envelope levels
EE, 2020, vol.117, no.4 181
1 and 2 received by Dr. Bhaskaran Gopalakrishnan and Dr. Ashish Nimbarte, professors in the department of
Industrial and Management Systems Engineering at the Statler College, West Virginia University. The URL’s
for the funding agencies are: 1) https://www.energy.gov/eere/amo/advanced-manufacturing-office and
2) https://www.energywv.org/.
Conflicts of Interest: The authors declare that they have no conflicts of interest to report regarding the
present study.
References
1. Capehart, B. L., Turner, W. C., Kennedy, W. J. (1994). Guide to energy management. 8th edition. Lilburn, GA:
Fairmont Press.
2. Beiter, P., Tian, T. (2016). 2015 renewable energy data book. Golden, CO: National Renewable Energy Lab
(NREL).
3. ENERGY STAR for Programmable Thermostats Definitions (2005). Summary of research findings from the
programmable thermostat market. Washington, DC: Energy Star Information Publication.
4. Lawrence, C. B. C., Woodward, M. (2017). Smart Controls Usage in USA. Washington, DC: U.S. Energy
Information and Administration.
5. Al-Saadi, S. N. J., Al-Jabri, K. S. (2017). Energy-efficient envelope design for residential buildings: a case study in
Oman. IEEE Smart City Symposium Prague,1–8.
6. McKay, J. R. (1987). The effect of wall and roof insulation on energy consumption for environmental control in a
small, 225-square foot outside plant building. INTELEC ’87—IEEE The Ninth International Telecommunications
Energy Conference, pp. 543–550.
7. Moon, J. W., Kim, J. J. (2010). ANN-based thermal control models for residential buildings. Building and
Environment, 45(7), 1612–1625. DOI 10.1016/j.buildenv.2010.01.009.
8. Gao, G., Whitehouse, K. (2009). The self-programming thermostat. Proceedings of the First ACM Workshop on
Embedded Sensing Systems for Energy-Efficiency in Buildings. ACM Press, 67.
9. Smith, J. R., Fishkin, K. P., Jiang, B., Mamishev, A., Philipose, M. et al. (2005). RFID-based techniques for
human-activity detection. Communications of the ACM, 48(9), 39–44. DOI 10.1145/1081992.1082018.
10. Gupta, M., Intille, S. S., Larson, K. (2009). Adding GPS-control to traditional thermostats: an exploration of
potential energy savings and design challenges. International Conference on Pervasive Computing, 95–114.
11. Qela, B., Mouftah, H. T. (2012). Observe, learn, and adapt (OLA)—an algorithm for energy management in smart
homes using wireless sensors and artificial intelligence. IEEE Transactions on Smart Grid, 3(4), 2262–2272. DOI
10.1109/TSG.2012.2209130.
12. Aarish, C., Perussi, M., Rietz, A., Korn, D. (2015). Evaluation of the 2013–2014 programmable and smart
thermostat program. Waltham, MA: The Cadmus Group Information Publication.
13. Jin, X., Wang, G., Song, Y., Sun, C. (2018). Smart building energy management based on network occupancy
sensing. Journal of International Council on Electrical Engineering, 8(1), 30–36. DOI 10.1080/
22348972.2018.1462608.
14. Malinick, T., Wilairat, N., Holmes, J., Perry, L. (2012). Destined to disappoint: programmable thermostat savings
are only as good as the assumptions about their operating characteristics. Washington, DC: ACEEE Summer
Study on Energy Efficiency in Buildings.
15. Van der Ham, W., Klein, M., Tabatabaei, S. A., Thilakarathne, D. J., Treur, J. (2016). Methods for a smart
thermostat to estimate the characteristics of a house based on sensor data. Energy Procedia, 95, 467–474. DOI
10.1016/j.egypro.2016.09.067.
16. Danassis, P., Siozios, K., Korkas, C., Soudris, D., Kosmatopoulos, E. (2017). A low-complexity control mechanism
targeting smart thermostats. Energy and Buildings, 139, 340–350. DOI 10.1016/j.enbuild.2017.01.013.
17. Muratori, M., Marano, V., Sioshansi, R., Rizzoni, G. (2012). Energy consumption of residential HVAC systems: a
simple physically-based model. IEEE Power and Energy Society General Meeting, 1–8. DOI 10.1109/
PESGM.2012.6344950.
182 EE, 2020, vol.117, no.4
18. White, J. A., Reichmuth, R. (1996). Simplified method for predicting building energy consumption using
average monthly temperatures. Proceedings of the 31st Intersociety Energy Conversion Engineering
Conference, 3, 1834–1839.
19. Zhou, R., Pan, Y., Huang, Z., Wang, Q. (2013). Building energy use prediction using time series analysis. IEEE 6th
International Conference on Service-Oriented Computing and Applications, 309–313.
20. Olofsson, T., Andersson, S. (2001). Long-term energy demand predictions based on short-term measured data.
Energy and Buildings, 33(2), 85–91. DOI 10.1016/S0378-7788(00)00068-2.
21. Cheng, C. C., Lee, D. (2018). Return on investment of building energy management system: a review.
International Journal of Energy Research, 42(13), 4034–4053. DOI 10.1002/er.4159.
22. Jones, S. R., Beardmore, J., Gillott, M., Boukhanouf, R., Walker, G. (2018). A control methodology for building
energy management systems (BEMS) in heat networks with distributed generation. Energy Procedia, 153, 295–
302. DOI 10.1016/j.egypro.2018.10.011.
23. Monthly New Residential Construction (2017). Monthly new residential construction, Information Publication, US
census bureau, US Department of Housing and Urban Development.
24. Degree-Days—Energy Explained, Your Guide to Understanding Energy (2017). Your guide to understanding
energy. Washington, DC: Information Publication, US Energy Information Administration.
EE, 2020, vol.117, no.4 183