Renewable and Sustainable Energy Reviews Volume 59 Issue 2016 (Doi 10.1016/j.rser.2015.12.328) Okochi, Godwine Swere Yao, Ye - A Review of Recent Developments and Technological Advancements of Var

Renewable and Sustainable Energy Reviews 59 (2016) 784817
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser
A review of recent developments and technological advancements

of variable-air-volume (VAV) air-conditioning systems
Godwine Swere Okochi, Ye Yao n
Institute of Refrigeration and Cryogenics, School of Mechanical and Power Engineering, Shanghai Jiao Tong University, Dongchuan Road No. 800,
200240 Shanghai, China
art ic l e i nf o
a b s t r a c t
Article history:
Received 27 October 2014
Received in revised form
22 December 2015
Accepted 23 December 2015
This study reviewed VAV systems modeling and simulations, control strategies and optimization tools,
the airow characteristics of VAV systems, some common VAV systems' faults, detection and diagnosis,
energy usage and analysis, and the current applications of variable air volume (VAV) air-conditioning
systems. VAV system modeling is very complex as it involves complex structures and parameters a result
of which has led to lack of models that combine both the AHU and building with all the required
parameters.
The most common controllers used in VAV systems are the PID controllers. We saw that supply air
temperature and the ow rate of supply air are the best parameters that can be optimized in a VAV
system as they greatly minimize energy consumption. Genetic algorithms have good robustness, and can
be easily parallelized. However, they suffer from shortcomings such as slow convergence rates under
some conditions, and have difculty in adjustment of algorithms since there are no rules for determining
the number of individuals in populations. FLCs boost of advantages such as less or minimum overshoot,
oscillation and power consumption compared to conventional PID controllers, can be used in MIMO
systems, and they do not require models as they can control non-linear processes.
Airow control in VAV systems can be achieved through controlling static pressure and position of
the damper. Literature survey shows that balancing and distribution of airow in VAV air-conditioning
systems can be considered to be one of the main challenging areas of research concerning VAV system
control. Most methods used today for detecting and diagnosing faults are hybrid. These are superior to
the conventional methods of FDD.
In conclusion, VAV air-conditioning systems are the most energy efcient systems in use today.
Despite of their current strengths, VAV systems energy saving potential can still be improved.
& 2016 Elsevier Ltd. All rights reserved.
Keywords:
Variable air volume (VAV)
Modeling
Optimal control
Control algorithms
INDOOR air quality (IAQ)
Energy efciency
Contents
1.
2.
3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 785
Modeling and simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 787
2.1.
Model types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 787
2.1.1.
Mathematical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 787
2.1.2.
Empirical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 789
2.1.3.
The hybrid models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 791
2.2.
Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 791
2.3.
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 791
Operations and control of VAV systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 791
3.1.
VAV controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 792
3.1.1.
Classical controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 792
3.1.2.
Optimal, predictive, and adaptive controllers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 794
3.1.3.
A combination of FLC, neural controller, adaptive and PID controller. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 794
Corresponding author. Tel.: 86 21 13641943577; fax: 86 21 34206814.

E-mail address: yeyao10000@sjtu.edu.cn (Y. Yao).
http://dx.doi.org/10.1016/j.rser.2015.12.328
1364-0321/& 2016 Elsevier Ltd. All rights reserved.
G.S. Okochi, Y. Yao / Renewable and Sustainable Energy Reviews 59 (2016) 784817
785
3.2.
Control strategies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 794

3.2.1.
Optimal start/stop control strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795
3.2.2.
Duct static pressure reset control strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795
3.2.3.
Air temperature reset control strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 796
3.2.4.
Demand controlled ventilation (DCV) control strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 797
3.3.
Control algorithms for optimization tools in VAV AC systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798
3.3.1.
Genetic algorithms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798
3.3.2.
Fuzzy-logic control algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
3.3.3.
Fuzzy-Genetic controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 799
3.3.4.
Direct Digital Control (DDC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 800
3.4.
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
4. Air properties. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 802
4.1.
Airow rate measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803
4.1.1.
Airow rate measuring devices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803
4.1.2.
Airow control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803
4.1.3.
Improvement of airow control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803
4.2.
Indoor climate requirements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804
4.2.1.
Indoor climatic problems related to VAV systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804
4.2.2.
Indoor air quality (IAQ) parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 804
4.2.3.
Indoor air quality (IAQ) control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
4.3.
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
5. VAV system faults detection and diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
5.1.
Some common VAV system faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
5.2.
Methods of fault detection and diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805
5.3.
Studies on fault detection and diagnosis (FDD) in VAV AC systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 806
5.4.
Energy penalty due to VAV systems' faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808
5.5.
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808
6. VAV systems energy consumption analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808
6.1.
Energy saving potential (efciency) of VAV AC systems compared to other systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 808
6.2.
Improving the energy performance of VAV systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809
6.3.
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809
7. Current applications of VAV AC systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809
7.1.
Working principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809
7.2.
Examples of VAV systems in practice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 811
8. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 811
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812
Denition of terms used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812
1. Introduction
The term variable air volume (VAV) rst came into existence in
the mid to late 1960s after studies by Urban [1]. After the world
energy crisis of 1970s, these systems gained popularity in the
United Kingdom (UK) and other countries across Europe in early
1980s as part of the efforts by engineers to come up with energy
efcient air conditioning systems [2].This was necessary due to the
cost of energy that was increasing at that time and has continued
to increase up to date. VAV systems by denition are simply air
conditioning (AC) systems that are designed to promote existence
of constant temperature in a conditioned space by varying the
volume of air supplied to the conditioned space instead of varying
the temperature of supplied air [3].Therefore, these systems vary
supply air volume at a constant temperature in order to meet the
demand caused by the changing heat load in the conditioned
space [4]. Generally, VAV systems can be broadly classied into
two categories as chilled water VAV air conditioning systems and
direct expansion (DX) cooling coil VAV air conditioning systems.
The basic components of a VAV air-conditioning system are: a
central air handling unit (AHU) with a variable speed supply fan
(can vary volume of air), coils used for heating or cooling, controls,
lters, mixing box, return or relief fan, air supply duct, VAV
terminal unit (device) connected to thermostats and supply diffusers and return duct or plenum.
VAV systems work on the principle of opening or closing

mechanical dampers or by modulating the airow through mixing
boxes powered by VAV fans as loads in various conditioned spaces
of a building. For instance, if a given conditioned space requires
more cooling, the damper to that space is opened wider to
increase the inow of cold air until the required temperature is
achieved. During the opening of the damper, there is pressure
drop in the supply duct which signals the supply fan to increase air
delivery. On the other hand, if an area is too cool and requires
temperature rise, the damper is gradually closed so as to reduce
the inow volume of cold air. This is usually applied in combination with variable-speed drives (VSDs). The result is decrease in air
ow which results in cutting down fan power needed thus saving
energy [5]. In a further effort to reduce the energy requirements,
most VAV systems utilize the return air in order to cut down the
power requirement and energy use when the outdoor temperature is higher than exhaust air temperature [6].
These systems have several advantages over other HVAC systems [711]. These include, less fan capacity compared to constant
volume systems since in VAV systems only the needed air is used;
greater exibility with respect to varying loads; improved indoor
environment, the system can incorporate an economizer to utilize
the outside air to provide cooling at times when temperature is
appropriate; reduced size of the main ducts since there is no
simultaneous coincidence of the maximum cooling/heating load
demand in all spaces [12].
786
Nomenclature
Symbols
adp , bdp ,dp Damper coefcients
Adp
Area of the damper (m2 )
Awdn
Area of the nth window (m2 )
Awln
Area of the nth wall (m2 )
Az
Is the area of zone oor (m2 )
Azj
Contact area of thermal transmission of jth zone (m2 )
C ahu
Overall

thermal capacitance of the air handling unit
J=kg

C dp
Thermal capacitance of the damper J=kg
cout
Coil outlet
C pa
Specic heat capacity of air at constant pressure (J/
kg K)
C pw
Specic heat capacity of water (J/kg K)
C ref
Refrigerant mass ow coefcient
C sd
Thermal capacitance of supply duct (J/kg)
Cz
Zone thermal capacitance (J/kg)
D
Diameter of the valve (m)
Dsd
Diameter of supply duct (m)
et
Disturbance
f
Fraction of the fan/pump power converted to
thermal energy
K
Static gain
L
Dead time
_ aex
m
Mass ow rate of external air (kg/s)
_ aj
m
Mass ow rate of air in jth zone (kg/s)
_ amix
m
Mass ow rate of mixed air (kg/s)
_ aout=p Is the amount of outside air required per person (kg/s)
m
_ ra
m
Mass ow rate of re-circulated air (kg/s)
_ ref
m
Refrigerant mass ow rate (kg/s)
_ sa
m
Mass ow-rate of supply air (kg/s)
_ samin
m
The minimum volume of supply air when the zone is
not occupied (kg/s)
M sd
Mass of supply duct (kg)
_w
m
Mass ow rate of water (kg/s)
Npd
Is the designed occupancy
pi
Valve inlet pressure (Pa)
po
Valve outlet pressure (Pa)
Ps
Static pressure (Pa)
PT
Total pressure (Pa)
PV
Velocity pressure (Pa)
q1
Back shift operator
Q exld
External heat gains from solar radiation (W)
Q inld
Internal heat gains from occupants, and light (W)
Internal heat transfer rate of the jth zone (W)
Q_ inldj
Q max
Maximum heat transfer rate (W)
r
Re-circulation percentage
Rra
Return air ratio
T ain
Inlet air temperature (K)
T aout
Outlet air temperature (K)
T ca
Temperature of conditioned air (K)
T cout
Temperature of the outlet of the coil (K)
T ex
External temperature (K)
T mout
Temperature of the air exiting the mixing box (K)
Tr
Temperature of re-circulated (return) air (K)
T aex
Temperature of external air (K)
T amb
Ambient temperature (K)
T amix
Temperature of mixed air (K)
T insd
Temperature of internal supply duct (K)
T sa
Temperature of supply air (K)
T samax
Max possible supply air (K)
T spt
Set-point temperature (K)
T wdn
T win
T wln
T wout
Tz
T zj
UAahu
Temperature of nth window (K)

Inlet water temperature (K)
Temperature of nth wall (K)
Outlet water temperature (K)
Zone temperature (K)
Temperature of jth zone (K)
Overall transmittance area factor of the air handling
unit (W/K)
U exsd
Conduction heat transfer coefcient of external supply
duct (W/m2 K)
U insd
Conduction heat transfer coefcient of internal supply
duct (W/m2 K)
U sd
Conduction heat transfer coefcient of supply duct
(W/m2 K)
ut
Input signal
U wdn
Conduction heat transfer coefcient of nth window
(W/m2 K)
U wln
Conduction heat transfer coefcient of the nth wall
(W/m2 K)
U zj
The thermal transmittance of the exterior walls of jth
zone (W/m2 K)
V ahu
Volume of air handling unit (m3)
V on
Is the sum of space ventilation airows (m3/s)
vor
Specic volume of the refrigerant leaving the expansion valve (m3/kg)
V ot Y V st Required system outdoor airow (m3/s)
V st
Is the total supply airow (m3/s)
v_ sa
Volume ow rate of supply air (m3/s)
Vz
Zone volume (m3)
V zj
Volume of jth zone (m3)
v
Air velocity (m/s)
v
Is the average velocity of duct air in (m/s)
vin
Box inlet air velocity (m/s)
vout
Box outlet air velocity (m/s)
_
w
Water ow rate (m3/s)
W cout
Humidity ratio of air exiting the coil
W mout
Humidity ratio of air exiting the mixing box
W sa
Humidity ratio of supply air (WSA)
X V on =V st Is uncorrected outdoor airow fraction.
Y V ot =V st Is the corrected outdoor airow fraction
yt
Output signal
Z
Is the critical space ventilation fraction
Greek letters
,; ; ; Polynomials
a
Density of air (kg/m3)
r
Density of refrigerant (kg/m3)
w
Density of water (kg/m3)
Effectiveness of heat exchanger
Damper angular position (deg)
Time constant
P dp
Damper pressure change (Pa)
Abbreviations
a
C spt
compr
dp
elec
H spt
mech
Rspt
Air
Cooling set point
Compressor
Damper pressure
Electrical
Heating set point
Mechanical
Re-heating set point
r
rpm
st
Refrigerant
Revolutions per minute
Saturated
This study shall review literature on VAV systems modeling and

simulations, controls, airow properties, common faults' diagnosis
and control, energy consumption analysis, and current applications of VAV systems.
2. Modeling and simulation

VAV models can be broadly categorized as steady state models
or unsteady state models. Steady state (static) models are those in
which parameters are constant or do not vary with time while
unsteady state models are those in which parameters vary with
time. The unsteady state models are also known as dynamic or
transient models. These (transient models) are the most common
models used in VAV systems modeling.
The unsteady state (dynamic) models in VAV systems use three
approaches namely: physics based (white box/mathematical/forward) approach, data driven (black box/ inverse) approach, and
gray box (hybrid) approach [13,14].
Previously, cooling or heating loads were found using simple
hand-calculation methods. In mid 1960s, computer simulation
programs for HVAC systems were introduced. According to GATC
[15], the rst simulation algorithm imitated physical conditions of
the entire system by treating variable time as independent. Candy
[16], reports that mathematical models are the basis of most of the
earliest simulation methods. Even though mathematical models
received preference over empirical and the hybrid models since
they were easy to analyze, they suffered complexity problem
compared to the other two [14].
2.1. Model types
2.1.1. Mathematical models
The mathematical (white box) models can be distributed or
lumped parameters type. The lumped parameter method is
superior to the distributed parameter type because it is easier to
solve [14]. According to Orosa [17], universal laws, physical laws of
mass conservation, momentum conservation, and energy balance
form the foundation upon which process signals are constructed
using the mathematical methods. These model types are very
useful in understanding the behavior of indoor conditions since it
is able to describe the interactions between VAV system inputs and
outputs. The format of these kinds of models is usually of timedomain differential equations type but they can be easily transformed to frequency domain transfer functions [18] or time
domain state space representations [19,20]. Generally, the process
of modeling an entire VAV system is very complex. Therefore,
several sub-models are employed to describe the complete system
to avoid complexity [21].
The building model is the most complex part in the entire VAV
air conditioning system modeling process, since the components
to be modeled go beyond just parts of the building such as walls,
roof, oor, and windows; but puts into consideration internal
loads such as activity within the conditioned space, number of
occupants, and as well as modeling the heat gain from lighting
[14]. Due to the challenge of complexity associated with modeling
of an entire VAV air conditioning system, literature survey has
shown that studies that address the entire scope of building model
are rare if none. Just to mention but a few examples, in a study
sa
w
z
787
Supply air
Water
Zone
carried out by L [22] to determine thermal and humid characteristics of indoor air, he only considered heat and moisture
transmission across the walls, ceilings and roofs [14]. He ignored
heat and moisture transmission through ventilation, ltration via
windows and doors, and the internal load. In addition, the equation of moisture conservation he used assumed uniform temperature across all phases. He applied Fick's law and Darcy's law of
transmission of moisture to address the law of mass conservation.
For energy, the author used Fourier's law.
In another study, Ghiaus and Hazyuk [23] approximated space
load in transient simulation using a mathematical model. To
obtain the model parameters, they applied the theory of superposition for electrical circuits. The authors assumed a linear
building thermal model and used the lumped parameter method
to take care of the indoor air and thermal capacity of the wall. They
also assumed knowledge of series of disturbances which occur at
different times and occupational programs. This clearly shows that
the researchers made many assumptions in order to determine the
heat load which would result in large errors in the nal result.
Furthermore, the authors used a single-input single-output (SISO)
model that ignores moisture transmission. Moisture transmission
is an important factor in determining moisture comfort.
Concerning the air-handling unit (AHU) model, Wang et al.
[2427] used the law of conservation of energy and equation of
thermal balance on control volume to build models of heat
exchanger for air handling unit. The supply air and surface of heat
exchanger were also assumed to be at same temperature. They
went further to ignore the humidity of moist supply air since they
wanted to avoid consideration of the effect varying humidity on
thermal comfort.
Badescu and Sicre [28] developed a mathematical model to
compute the heating demand for a three-zone passive house (PH).
They used a 1D time-dependent conduction heat transfer equation
to analyze heat transfer through the high thermal inertia elements. They solved the heat transfer equation using a standard
Netlib solver (PDECHEEB). The authors used appropriate models
for conduction through low thermal inertia elements as well as a
simple approach for solar radiation transmission through windows. Finally, they considered how internal heat sources affected
the model. The above work was later improved by introducing two
new equations in the solar collector model. The two elements
added included useful heat ux and the mean temperature inside
collector area [29].
In another work by Badescu et al. [30], they developed a time
dependent model for a passive building in which they noted that
for proper evaluation of space cooling/heating demand we must
account for heat ux entering into the building by conduction,
passive solar heating as a result of light penetrating via the windows as well as internal heat sources and sinks. They also reported
that the thermal load changes continuously as a result of variations in both ambient temperature and solar irradiance.
In a later work by Badescu et al. [31], they developed a timedependent model to evaluate the thermal load of the AMVIC
building. They coupled the model with other the modules modeling the ventilation-heating system. The ensemble constituted
time-dependent Passive House Thermal Transients (PHTT) model.
During comparison of results, the authors found out that the PHTT
could estimate an annual relative total thermal load value which is
788
in reasonable agreement with the value predicted by the Passive

House Planning Package [31].
Mathematical (physical) based models can be used to model
several sub-systems in VAV air-conditioning systems. These are
discussed below.
2.1.1.1. The zone model. Addition and removal of heat into and from
the conditioned space maintains temperature by balancing various
kinds of heat gains and losses [32]. The addition and subtraction of
heat is done through the opening and closing of the damper [33].
Heat is transferred into the zone through the supply air, inltration, conduction through walls and windows, and the external and
internal gains caused by solar ux and zone occupants [34].
Modeling of heat transfer to the zone is normally through the use
of the equation of heat conduction model, heat balance method,
weighing factor and thermal-network method [14,35].
According to Tashtoush et al. [18], the zone model is a characteristic of three state variables. These variables include; the
temperature of the inner walls (Tinwl), the zone temperature (Tz)
and the zone humidity ratio (W). To simplify the zone model and
make it practical, the following assumptions are made during
modeling:
Uniform distribution of zone temperature due to proper mixing

of air. This enables the dynamics of the conditioned space to be

expressed as a lumped parameter model.
Effects of opposite walls on zone temperature are considered to
be the same while the space oor has no effect on zone
temperature.
Constant air density.
Negligible pressure losses across the zone and in the mixing
section.
Uncontrolled inputs comprise of occupants, lights, and extreme
weather conditions.
Under these assumptions, we can apply several energy governing equations to estimate the dynamics of air temperature of a
single zone as given in Eq. (2.1) below [36].

dT z
_ sa C pa T sa T z U wln Awln T wln T z
m
a V z C z
dt

2:1
U wdn Awdn T wdn T z Q inld Q exld
Eq. (2.1) above suggests that the change rate of zone energy is
equivalent to the deviation between the gained and lost energy
inside the zone through heat transfer processes [36].
For a multi-zone model, air temperature of each jth zone is
given by Eq. (2.2) below:

dT zj
_ aj C pa T sa T zj U zj Azj T ex T zj Q_ inldj
a V zj C pa
m
2:2
dt
Note: In Eq. (2.2) above, the inuence of wall capacitance is
neglected.
2.1.1.2. The heating and cooling coil model. In a VAV air conditioning system, the cooling or heating coil conditions the
supply air [13]. This facilitates maintenance of the conditioned
space within the desired temperature. In VAV systems, a reheat
coil that is usually located in inside the mixing box is used to
control temperature of supply air. The dynamic temperature
model of the heating/cooling coil can be derived from the
energy balance in the airwater heat exchanger [18,33,37,38].
During modeling, it is assumed that:
The mass ow rates of both water and air are constant.
The thermal resistance of the coil is negligible.

The thermodynamic properties of the uids are independent of
the temperature variations for the operating temperature range.
The coil mass temperature is taken to be constant within the

coil, and
Steady state heat transfer between the water and air.

The energy balance between the air and water is given by
Eq. (2.3) below;

dT cout
_ w C pw T win T wout UAahu T ex T cout
w
C ahu
dt

2:3
v_ sa a C pa T mout T cout
The mass balance is given by Eq. (2.4) below;

W cout
V ahu
v_ a W mout W cout
t
2:4
In a study by Tahmasebi et al. the authors modeled the heating

system by calculating the rate at which the radiators deliver heat
due to radiation and convection [39]. The heating or cooling coil
can also be assumed to have constant effectiveness [4042]. This
type of model is less complex and determines the outlet air and
outlet water temperature from the inlet air and inlet water temperatures as shown below;
T wout T win
T aout T ain
Q_ max
_ w C pw
m
Q_ max
_ a C pa
m
2:5
2:6
_ a C pa T ain T win when m

_ a C pa o m
_ w C pw ,
In which Q_ max m
_ w C pw T ain T win .
otherwise Q_ max m
_ w is mass ow rate of water (kg/s), is effectiveness of
Here; m
heat exchanger, Q max is the maximum heat transfer rate (W), T ain is
the inlet air temperature (K), and T aout is the outlet air
temperature (K).
2.1.1.3. Mixing air box model. As a system optimization strategy,
part of return air is mixed with fresh air from outside. It is
assumed that no frictional loses occur in the mixing box and the
system is under steady state and adiabatic conditions. Therefore,
the temperature of the mixed air is a linear combination of return
air temperature and fresh outside air temperature [18]. The
expressions that follow describe energy and mass balance;
_ amix m
_ ra m
_ aex ;
m
2:7
T mix rT r 1 rT ex ;
2:8
In which:
r
_ a2 T z2
_ ra
_ a1 T z1 m
m
m
;T
_ a2
_ amix r
_ a1 m
m
m
T amix
_ aex T aex m
_ sa m
_ aex T z
m
_ sa
m
2:9
2:10
The above equation can also be used to calculate CO2 concentration [43] and the humidity ratio [41] of mixed air by simply
replacing the temperature variables with their corresponding CO2
and humidity ratio variables respectively.
2.1.1.4. Duct and pipe model. Ducts aid in supplying of conditioned
air to the zone and extracting exhaust air from the zone [13]. There
is usually heat exchange between air passing through the ducts

and the ambient air. In this case, the rate of temperature change in
the supply duct model can be expressed as;
dT insd 4U sd T amb T insd
C sd Dsd a
dt
2:11
where T amb is the ambient temperature (K), T insd is the temperature of the internal supply duct (K), C sd is the thermal capacitance
of supply duct (J/K), Dsd is the diameter of supply duct (M), and
U sd is the conduction heat transfer coefcient of supply duct
(W/m2 K).
The model of supply air duct by Tashtoush et al. describes the
relationship between exit air temperature and the inlet air temperature [18]. This is shown below;
dT aout U insd U exsd ma C pa
T ain T aout
dt
U insd M sd C sd
2:12
The above models only considered cases when the supplying

duct is placed in the ambient air. In some cases supplying ducts
may be burried into the ground. For a detailed explanation of such
scenerios refer to works by [44,45]. In the work by [45], the author
notes that the quantity of interest for the designer is the outlet air
temperature of the underground pipe. This temperature is signicantly affected by different design parameters such as the
depth of the pipe from the ground surface, diamemter of the pipe
and pipe's material. The author modeled the thermal behavior of
the heat exchager by means of numerical transient bi-dimensional
approach. It was claimed that ground surface boundary condition
can be derived from an energy balance equation comprising of
convective energy exchange between the air and the soil, solar
radiation energy absorbed by the ground, the latent heat ux due
to evaporation at the surface of the ground as well as the longwave radiation.
2.1.2. Empirical models
Here models are constructed on the basis of experiments or
available data [22,46,47]. The entire data based approach consists
of eight different types of models [13]. Description of the
approaches is as follows:
2.1.2.1. Frequency domain model with dead time. Temperature and
humidity in VAV systems requires a substantial period of time to
undergo change. This is due to the heavy thermal inertia of a VAV
system. Therefore, dynamics of temperature and humidity in VAV
systems are modeled using rst and second order (over-damped)
models with dead time [22,4850] as shown below;
i) Second-order dead-time model
Gs
Y s
1
e Ls ;
U s as2 bs c
2:13
where
"
j L !#
e c
a c real
=2c ;
Gjc

b imag
e jc L
Gjc
1
c ;
K
K
y1
;
u1
2:14

;
2:15
2:16
2:17
where u1 is the amplitude of step change in process input
789
while y1 is the deviation between the original value and

the new steady-state value of the output process, K is the
static gain, is a constant of time and L is the dead time.
ii) Step-method
Gs
Y s
K
e Ls
U s s 1
2:18
2.1.2.2. Data mining algorithms (DMAs). DMAs are usually used to

model complex and non-linear system dynamics. Such algorithms
include Articial Neutral Network (ANN) and Support Vector
Machine (SVM). A study by Kalogirou [51] provided a comprehensive survey on the application of Articial Neutral Network
(ANN) in building energy systems. The authors reported that nonoptimum weights could lead to inaccurate results when a single
method (i.e. Articial Neutral Network (ANN) or multilayer perceptron (MLP)) is used [51]. Kusiak and Zhang [52] developed a
daily steam load model for HVAC system. They used ve data
mining algorithms to construct the model.
Tang [53] developed sensors for predicting energy consumption of AHU as well as IAQ using ANN models. Indoor air temperature, humidity and CO2 concentration were used to describe
indoor air quality. An improved model was achieved when using
clustering of the Articial Neutral Network (ANN) input variables
compared to a non-clustering model. Kusiak and Xu developed a
model based on dynamic ANN whose results showed signicant
improvement in system performance [54]. Morisot and Marchio
[55] modeled a cooling coil for application in VAV fault detection
using ANN. In their study, they were able to detect cooling coil
faults by nding the deviation between the outputs of the actual
system and the modeled system. They found out that in the
absence of any fault, there was a small margin difference while in
the presence of a fault the margin difference increased. Ding et al.
[56] used a single variable time series to estimate a cooling load.
This is an example of using SVM in load forecasting.
Some dominant advantages of data mining algorithms are; easy
implementation, can model noisy data with complex relationships
etc. These models suffer from the following disadvantages;
requires training and data testing to facilitate construction of
models, it is not possible to physically interpret the developed
models and it is difcult to implement the algorithms online [13].
2.1.2.3. Fuzzy-logic (FL) models. The development of fuzzy-logic
models involves the implementation of the if-then-else statements whose rules are expressed as a database or table [57]. These
kinds of models comprise of fuzzy adaptive network (FAN),
TakagiSugeno (TS), and adaptive network based fuzzy inference
system (ANFIS). Chen et al. [58] used FAN to model thermal
comfort. TS was used to model AHU [59] and thermal comfort
[32]. Soyguder and Alli [60] applied ANFIS to ambient and zone
temperature to achieve prediction of fan rotational speed and
degree of damper opening. ANN and Genetic Algorithm-based
ANFIS (GA-ANFIS) was used to project energy consumption in two
different buildings [55].
The main advantage of fuzzy-logic models is that they are
simple and easy to understand. However, these models also have
shortcomings. It is difcult to model due to lack of the huge
volume of data that needs to be known [13].
2.1.2.4. Statistical models. These are models built on the basis of
assumptions of how a sample of data can be generated from a
huge set by following a certain trend. Here we describe averages,
probability distributions, regressions and so on. Examples of these
kinds of models are; single and multivariate regression, output
error (OE), Box Jenkins (BJ), autoregressive integrated moving
average (ARIMA), autoregressive exogenous (ARX), autoregressive
790
moving average exogenous (ARMAX) and nite impulse response

(FIR) [13]. The simple input/output relationship of statistical data
was given by Yiu [62] as follows;

q1
q1
2:19
q 1 yt 1 ut 1 et
q
q
where , ; ; ; and are polynomials, q 1 is back shift operator,
yt , ut , and et are output, input and disturbance respectively.
The type of model developed is determined by numerator and
denominator polynomials of input u(t), output y(t) and disturbance e(t) that will be chosen.
Some examples of studies which have employed the use of
statistical model include; Dong et al. [63] used a second order
polynomial to model to study how the rotational speed of the
supply fan is related to the power consumption. They used the
method of curve tting on measured data to derive the coefcients
of the polynomial.
Yoshida and Kumar [64] studied the AHU of a VAV system using
ARX model and adaptive forgetting through multiple models
(AFMM) online based real time. Later on, Yoshida and Kumar [65]
used an ARX model based on off-line identication to develop a
model of VAV air handling unit.
Evident merits of statistical methods include; promotes steady
state modeling using linear and polynomial time series regression
(essential for quick calculations), and models can be developed
with limited data. Some setbacks include; the linear and polynomial time series models have low accuracy and cannot capture
the non-linear dynamics of VAV system.
2.1.2.5. Geometric models. These models use mathematical methods
to model real objects using computer graphics and computer aided
design (CAD). They use geometric principles to represent a system
through the use of curves, surfaces, and volumes. Geometric modeling mainly focuses on two-dimensional (2D) or three-dimensional
(3D) representation of shapes. The spline surfaces are very common
tools used in geometric modeling. Examples of spline surfaces are
Hermite-splines, Bezier-splines, B-splines and the thin plate spline
(TPS). The thin plate splines can effectively be used to interpolate
scattered data. The thin plate spline uses a special function to map
the input-output data [66]. TPS is not affected by initial conditions
this makes it superior to ANN. However, TPS is sensitive to data
noise. In addition, TPS becomes computationally expensive as the
size of data increases [66].
2.1.2.6. Case-Based Reasoning (CBR). These kinds of models make
use of related or similar past cases to build new models [67]. The
CBR modeling which only uses a part of the given data is referred
to as topological case-based modeling (TCBM) [13]. According to
Nishiguchi et al. [66], this model is based on the similarity principle i.e. it constructed around relating similar parameters. The
model has several advantages which include; suitable for nonlinear and ill-dened systems; minimizes measurement noise and
data [13]. Unseen cases pose a challenge to the models of this kind.
2.1.2.7. Stochastic models. These models are used to represent
processes with random variables in VAV AC systems such as fan
energy use. Zlatanovic et al. [68] used Gaussian probability density
functions (GPDF) to model the fan power consumption in a large
building. In their study, the authors assumed that variable speed
fans (VSFs) operated at the minimum power level during unoccupied hours while the energy used during the occupied hours
followed a normal distribution function.
The advantage of the stochastic approach is applicable to many
physical processes which can be approximated to standard normal
and uniform distributions.
The approach suffers from the fact that a large amount of data
is required in order to generate an accurate shape of the probability distribution function (PDF) of a random variable.
The most common VAV components which adapt empirical
modeling approach include dampers, expansion valves, fans, and
pumps. These can be modeled as shown below:(1) Damper model
The main function of a damper is to control airow rate in ducts
of a VAV AC system. It accomplishes control of ow rate by
varying resistance to air ow in the supply duct leading to a given
conditioned space. When the damper rotates in an air stream, it
regulates air passage to the space. According to Huang [69], the
signal ut 0; 1 controls damper opening, where 0 indicates
fully closed and 1 indicates fully opened. Thus, the signal
determines the rate of mass ow of air through the damper.
ma adp edp ut bdp
2:20
In which; adp , bdp , dp are damper coefcients and ut is the input

signal (position of the damper).
In some dampers, the rate of air mass ow through the damper
is dependent on the damper ow coefcient, pressure difference
and position is dependent on the damper ow cross-sectional
area [60]. An example of such model is as follows;
q
ma C dp a P dp Adp
2:21
In the above equation, the pressure drop (P dp ) is relatively
constant [70]. This is because static pressure in the loop is selfregulated due to fan characteristics that would maintain static
pressure at a specic value such that the fan power balances with
the rate of ow circulation in the loop [70]. In such a case the
_ a ) will be proportional to the ow area (Adp ).
mass ow rate (m
(2) Expansion valve model
Expansion valves control refrigerant and water ow rates in VAV
air conditioning systems. An expansion valve adjusts the
quantity of refrigerant at evaporator inlet thus controls the value
of superheat temperature of the refrigerant [71]. In addition,
expansion valves control the rate at which water ows into the
heating or cooling coiling of a VAV system. In an expansion
valve, the cooling load is assumed to be uncertain thus the most
effective variable for optimal operation of a valve is refrigerant
ow rate [72]. Since the expansion process is usually considered
to have a constant enthalpy, we can use the orice equation to
describe refrigerant ow through an expansion valve. The rate
of refrigerant mass ow is calculated as;
!
q

D2
mref C ref
r pi po
2:22
4
In which C ref , is a function of several parameters. These are,

valve geometry, refrigerant's inlet and outlet conditions of
temperature and pressure as well as its thermo-physical properties [73]. It can be described as follows;
p
C ref 0:02005 r 0:63vor
2:23
(3) Fan and pump model
Fans control air ow rate whereas pumps control water and
refrigerant ow rate through a VAV system [13]. The ow
rates, pressure difference between the inlet and outlet, and fan
efciency or pump efciency determine the power utilized by
the fan or pump [74,69,75,76].
If the fan motor is placed within the ow, there is heat
exchange between the fan motor and the air. According to

Tashtoush et al. [18], the air temperature may rise by a margin
of 12 C. Fan exit air temperature can be calculated using the
formula given by TRNSYS [40] as shown below;

W fan f
2:24
T aout T ain
ma C pa
where W fan is the fan power and f is the fraction of power
converted to thermal energy.
2.1.3. The hybrid models
The hybrid models are a combination of mathematical and
empirical models. They are aimed at improving the weaknesses of
individual approaches [14]. During hybrid modeling, physical or
thermodynamic principles are used to derive the structures of the
models while the catalog, commissioning, or operating data are
employed in the determination of the model parameters. This idea
has been supported by several works [77,78]. In their respective
studies, the authors developed effective models by considering the
specic heat of saturated air. The numbers of transfer units were
obtained from the catalog data at any operating condition.
Wang et al. [79] constructed a hybrid model for predicting
chiller performance under unchanging conditions. Their method
borrowed ideas from the approach previously used by Braun et al.
[77]. In another study, Ghiaus [80] used the hybrid model to
represent the AHU by making the following assumptions during
mathematical modeling; that difference in air temperature takes
place in the cooling coils while humidity ratio deviation is
restricted in the humidier. This implies that the cooling coil is
under dry operating conditions and that in the humidier air
temperature is constant. This was done to achieve separation of
elemental transfer functions so as to avoid coupling between
humidity and temperature. Experiments are used to determine
discrete parameters of these models. From the above, it is clear
that their assumptions are unachievable except in rare ideal
situations [14].
Jin et al. [81] developed a dynamic cooling coil unit (CCU)
model by extending the cooling coil unit engineering model and
combining the model with the mass energy balance equations.
They used commissioning information to estimate at most six
model parameters by a non-linear online identication method. In
their conclusion, they reported that the modeling method was
simple and exactly captured the non-linear characteristics over a
wide range of system operation. Vakiloroaya et al. [82] developed
a hybrid model of the cooling coil. Real-time experimental data
can be used to determine coil parameters.
The main advantages of the hybrid models are that they are
more accurate than black-box (physical) models. These models are
also easy to generalize.
The main shortcoming of the hybrid models is that they are the
most difcult to develop. Successful development of the hybrid
models requires both knowledge of underlying physical phenomenon and system input-output data. These may be difcult to
know and represent hence the difculty in building hybrid models.
2.2. Simulation
The term simulation is mainly used to refer to the process of
creating a computer model (an abstract representation) of existing
or proposed VAV systems in such a way that we get to know the
factors affecting the systems and the future behavior of the systems in the real environment. Simulation is an important powerful
tool that provides a means of evaluating VAV systems without
having to carry out experiments on the real system. Thus, it provides a convenient and cheap method of testing, commissioning
and evaluating performance of VAV systems [83,84]. In order to
791
carry out simulation, engineers employ the use of simulation tools

such as TRYNSYS, ESP-r, DOE-2, HVACSIM , Building Loads Analysis and System Thermodynamics (BLAST), Matlab/Simulink,
EnergyPlus, SIMBAD (SIMulator of Building And Devices) etc. Each
of these tools is used depending on its eld of application.
Most of the recent researches on simulation have focused on
dynamic modeling and simulation of VAV systems. There has been
a couple of simulation works carried out in the eld of VAV systems optimization and control. Such works include that by ZaheerUddin and Zhen [85] who developed transient model of a two
zone VAV air-conditioning system. Later on, during a study on
optimal control, House and Smith [86] simulated the building and
VAV air-conditioning system as a test facility.
Wang [83] developed dynamic models to carry out simulation
of various factors in VAV AC system under control of Energy
Management and Control Systems (EMCS). The author reported
that dynamic simulation is a convenient and suitable tool of
studying the dynamic, environmental and energy performances of
air conditioning systems.
Aynur et al. [87] carried out simulation of a VAV AC system in a
building. The mode of operation was of the system was cooling.
The authors used several parameters in the simulation package.
These included building design information, internal loads, and
manufacturer's data for VAV boxes and roof top unit (RTU). The
simulation results were validated with experimental data obtained
from the real system under similar conditions. The results revealed
that there was a variation of about 7 15% in more than 70% of all
simulated data on power consumption.
2.3. Summary
There are two main classications of models namely, mathematical models and empirical models. The third type of models
known as hybrid models is simply a combination of the two main
model classes. No single modeling approach can provide a perfect
system model due to the weaknesses associated with each individual modeling approach. Therefore, the hybrid models have
proved to be superior in design as they compensate weaknesses
for individual models. Most of the work that has been done in VAV
systems modeling has addressed sub-models since the model of
the entire VAV system is complex to analyze.
The survey of the literature has shown that most of the modeling and simulation research has focused under dynamic conditions. We also saw that simulation only gives an approximate
picture of what we expect but not exactly what we have in the
practical sense. This is because most of the simulation results
varied from the practical values.
Some of the challenges experienced with different modeling
methods include: at least each modeling approach has a major
shortcoming derived from assumptions, unmeasured disturbances, and uncertainties in some properties; there is no
complete system model with all the required parameters; and the
existing models for space zone do not address the effect of solar
rays incident on building walls, roof, and windows as well as lag
time cooling load. In addition, most of the AHU models analyzed
considered only the cooling coil and overlooked the precooling
coil while at the same time neglecting the effectiveness of the air
mixing chamber. For accurate modeling, we should try to minimize assumptions as much as possible.
3. Operations and control of VAV systems

The operation and control of VAV air-conditioning systems is
very critical in increasing the efciency of system energy usage.
Variables such as ambient temperatures, occupancy, and lighting
792
loads are dynamic throughout the system operation. Thus, deviations from the design set point result in drastic oscillations or
imbalances which cannot be ignored. Therefore, this gives us a
reason for implementing VAV system control which helps to eliminate zone overcool or overheat due to instability of the system.
Basically, a VAV control system has four basic elements namely;
sensor, controller, controlled device, and the source of energy. The
VAV system components which are in most cases under control
include, air handling unit (AHU), terminal unit, and Chiller plant
[88]. Basically, we have open loop (feed-forward) control as well as
closed loop (feedback) control. In feed-forward control, we cannot
monitor the accuracy of the control system whereas in feedback
control, we can compare measured parameters with set parameters and the corresponding signals are accordingly generated.
Thus, we are able to measure the effectiveness of the control
scheme. The feedback control scheme is the most widely used
control method in most VAV systems [89].
Fig. 3.1. Proportional control.
3.1. VAV controllers

In VAV air-conditioning systems, there are usually several
variables that can be controlled during the operation in order to
achieve the desired system conditions. These variables include; fan
speed, static pressure, zone air temperature, humidity, ventilation,
ow-rate of supply air, and ow-rate of chilled water entering the
cooling coil.
Generally controllers can be classied into two main categories
as;
i) Classical controllers and,
ii) Optimal, predictive, and adaptive controllers.
3.1.1. Classical controllers
Thermostats were initially used for feedback control of temperature [90]. But, the controllers were faced with a challenge of
frequent changes between two states of a thermostat. In order to
overcome this challenge, the introduction and use of thermostats
with dead zone was inevitable. The use of thermostats with dead
zone is known as bangbang control with dead zone. However,
that did not solve the problem of overshoots in the controlled
temperature as a result of which more energy was wasted. In
attempt to overcome the overshoot challenge, the use of
Proportional-Integrate-Derivative (PID) controllers was adopted
[90,91]. Therefore, in the following discussion we will look at the
development of different types of controllers.
3.1.1.1. Proportional controller. Proportional control (PC) regulates
the output capacity of an AHU system by closing or opening the
damper to satisfy the cooling/heating load of conditioned zone
[92]. Here, the displacement of the nal control variable is directly
proportional to the error between the value of controlled variable
and the set-point [92]. This phenomenon can be mathematically
expressed as:
u K p E u0
3:1
where u is the output signal, K p is the constant of proportionality

(gain), E is the error/deviation, and u0 is the output value in the
absence of an error (i.e. E 0). The reset of the generated control
signal is manually accomplished by adding/subtracting it from the
output value. PC maintains the temperature with variations above
or below the set-point. Figs. 3.1 and 3.2 below show proportional
logic diagram and a curve of proportional control used in cooling a
zone respectively.
Kulkarni and Hong [93] designed a proportional control (PC)
system for residential building. They achieved their objective by
setting up transient simulation for the building and control
Fig. 3.2. A curve of proportional control used in cooling a zone.
system. During their study, they modeled the building using statespace approach and implemented the simulation code using
MATLABTM. They concluded that proportional control is superior
to the two-position control for thermal comfort. But, the two
control schemes portrayed almost similar characteristics in terms
of energy consumption. The authors proposed that in proportional
control, and integral control element should be added in order to
do away with the steady-state error and improve the performance
of thermal comfort.
3.1.1.2. Proportional Integral (PI) Control. In PI controller, reset of
control mode is automatic [92]. In this system, deviation of the
controlled variable from the set point leads to an offset. This then
leads to a gradually and automatic shift of position of the proportional band thus restoring the variable to the set-point [92].
According to Haines [94] PI controllers are preferred in the
control of VAV systems compared to proportional controllers.
Therefore, PI controller adds a component proportional to the
control algorithm. Thus, can be mathematically expressed as:
Z
u K p E K i E:dT u0
3:2
where K i is the integral parameter, and dT is the time change.
According to Bai and Zhang [95], the transfer function of a PI
controller can be expressed as shown below:
Gc s
U s
K
Kp i
Es
s
3:3
Since the reset modes of PI controllers are automatic much of

the research in this eld has focused on design of efcient selftuning PI controllers [96]. Krakow and Lin [97] designed a
numerical model to determine the PI gains for digital controllers
which would maintain a constant discharge pressure at different
loads. They used an experimental system similar to the modeled
system to validate their results. Both systems gave close results
with some differences due to unaccounted inertial effects in the
numerical model.
Krakow [98] proposed the relationship between the sampling
and interval in digital PI control system performance by using
experimental and simulated response characteristics for a PI controlled mixing valve air heating system. The results showed that
long sampling intervals may yield satisfactory response
characteristics compared to short sampling intervals if the system

is well tuned.
Zhou and Liu [99] developed an online self-tuning algorithm for PI controllers used in heating or cooling coils in
buildings. They found out that the method could be easily
implemented in the VAV energy control systems when digital
lters are applied to do spectral analysis instead of Fourier
transformation.
Qu and Zaheer-Uddin [100] studied online adaptive method for
tuning parameters of proportional integrate (PI) controllers. They
investigated challenges related to online tuning of proportional
integrate controllers in HVAC systems. They used a rst-order plus
dead-time model to describe the discharge air temperature system. They estimated the model parameters online while maintaining control in the closed-loop. H 1 adaptive tuning rules were
applied and new PI parameters were estimated. Results found
after studying both simulation and experimental data showed that
their controller is able to track changes in set-point, cancel out
disturbances and is robust.
3.1.1.3. Proportional Integral Derivative (PID) controller. This controller adds a derivative function to PI controller. The aim of the
derivative is to oppose any change (works to bring the error to
zero) and is usually proportional to the rate of change [92]. The
corrective action initiated by the derivative is directly dependent
on the speed of change of the control point. Thus, PID control
improves the speed of response of a PI controller. The effect of the
deviation action is dependent on the rate time setting. The time
constants determine the PID setting status of the system under
control. The mathematical expression of the derivative part of PID
controller is given as [93]:
u K dT d
dE
dT
3:4
where
K d differential parameter
T d action time in relation to proportional action
K d T d rate gain value
dE
rate of change error signal
dT
Generally, components under control in HVAC systems follow a
PID-type control schematic that adjusts equipment settings based
on prevailing conditions. For instance in VAV systems control, the
damper opening is modulated based on the deviation between the
zone and set-point temperatures [101]. Therefore, the complete
PID controller can be mathematically expressed by Eq. (3.5) as
shown below;
Z
dE
u K p :E K i E:dT K d T d u0
3:5
dT
Fig. 3.3. Proportional-Integral-Derivative control curve.
793
Eq. (3.5) above can generate a curve similar the one shown in
Fig. 3.3 below.
There has been intensive research work done in the eld of
applications of Proportional Integrate Derivative (PID) controllers
in VAV systems. Here, we will briey review some techniques
proposed for tuning and designing PID based control structures as
well as methods for assessing their performance in VAV systems.
Despite the advancement in control methods, PID controller
design and ZieglerNichols (ZN) [102] techniques are still being
used in VAV air-conditioning systems industry. However, the
ZieglerNichols technique suffers from prolonged time of testing
and limited performance; thus, it is used as a rst cut for tuning
PID controllers [95]. Use of control methods such as PID controllers suffers from poor tuning of K p ; K i ; and K d parameters.
These methods are also unable to take into account the interactions between various loops [103,104].
Dexter [105] studied a four-level HVAC control scheme. The
author assigned the fourth order or supervisory level to focus on
maintaining the desired indoor temperature and relative humidity; the third order level was for temperature and relative
humidity control for supply air; the second order level maintained
the performance of plant actuators and local control loops; and,
nally, the rst order level was to maintain the desired controller
setting based on system model.
Later on, Geng and Geary [106] studied how the overshoot and
settling time are affected by the disturbances in PID controllers.
Their study revealed that the tuning rules based on the Ziegler
Nichols method can only be applied where we have small normalized delay time and that when this is coupled with normalized
gain, they can tune PID controllers.
Lin et al. [107] used the theory of adaptive interaction to
design a new adaptation algorithm for PID controllers. The
algorithm achieves auto-tuning objective by minimizing the error
function. The PID auto-tuning algorithm proposed in their study
has advantages such as simplicity and independent of system
model. Results from simulations showed that the algorithm can
perform well under linear and non-linear systems; with or
without noise; stable or unstable systems; and with or without
time delay.
Bi et al. [48] developed an advanced PID auto-tuner for both
single and multi-variable processes in HVAC systems. The autotuner incorporates recent PID design rules to achieve high control
performances with dynamic characteristics. Decoupling control is
used for multi-variable processes. Results from experiments
showed effective and superior performance of the proposed autotuner over the manually tuned PID controller and the standard
relay auto-tuner. Wang et al. [108] developed an advanced autotuner and its applications in HVAC systems. The auto-tuner rst
identies a second order plus dead-time model based on two
continuous relay feedback experiments. Then, the PID controller is
designed on the basis of gain and phase-margin specications.
Results from experiments showed that the studied auto-tuner was
effective and had superior performance compared with the standard relay auto-tuner.
Kasahara et al. [109] investigated stability limit analysis and
tuning of PID controllers in bi-linear systems with time-delay
feedback. The results showed that instabilities are not produced by
the non-linearity of VAV systems. However, PID parameters for
linear systems must be slightly modied by nding the gain
reduction factors to suit real VAV systems.
During their study on development of PID tuning algorithm
using optimization subject to applying constraints on control input
parameters for a single-zone cooling system, Ozawa et al. [110]
achieved an over damped response with zero overshoot, which
prevented long oscillations during the utilization of integral
squared time error criterion in the performance index.
794
Tashtoush et al. [18] described a procedure to derive a dynamic

model of an air-conditioning system. A PID controller was used to
maintain set-point values for indoor temperature and humidity. In
order to produce less oscillatory responses, the PID parameters
were carefully tuned using ZN method. Simulation results
showed that the system was capable of efciently controlling the
disturbance within a short time period with less error.
Lim et al. [111] presented recent advances in PID control design
and employed Hurwitz polynomials to improve performance and
robustness. The authors developed some simple yet intuitive
design tools and applied them to control supply air temperature of
a VAV system unit. They also used an electronic expansion valve to
control evaporator superheat temperature.
Yamazaki et al. [112] examined two PID controllers which use
xed and adjustable resets. Their control results for the controller
with adjustable reset were encouraging.
3.1.1.4. Advantages of PID controllers. PID controllers are still the
most widely used controllers in VAV systems across the world
[113,114]. This is because of several advantages which include:
i) As can be seen from the diagram above, PID controllers restrict
the control system to the set-point. This causes the narrowing
down of the range of operation a VAV system to a very small
fraction of a degree. Thus, it eliminates the large temperature
variations (oscillations) dominant in the proportional control
(PC) and maintains space temperature close to the set-point
[114].
ii) PID controllers are easy to understand and use as they have a
simple structure [115].
3.1.1.5. Disadvantages of PID controllers. Despite the advantages
outlined above, there are several disadvantages that emerge while
using PID controllers. These include:
i) PID controllers have rapid control but processes in VAV systems don't need rapid response.
ii) The operation of this control method is not very energy efcient since effective utilization of the method demands for a
substantial pressure drop under all operating conditions.
When applied to cooling or heating VAV systems, the required
valve pressure drop can largely increase the required
pump power.
iii) Inconvenient tuning parameters, faint anti-interference and
large overshoot [116].
iv) This method is largely dependent on model parameters and its
robustness is poor [117].
v) There is no assurance of satisfying all the loads at any given
time since PID loops operate in isolation to each other.
vi) Limited to single input single output systems (SISO).
Due to the above shortcomings, designers have recently
resorted to study optimal, predictive, or adaptive control
techniques.
3.1.2. Optimal, predictive, and adaptive controllers
Signicant studies in this eld were carried out during 1980s
and 1990s. The use of optimal control [118] or predictive control
[119] requires building models. Predictive control aims at establishing a model for future disturbances. The disturbances may be
from solar gains, presence of occupants etc., [119123]. Also, predictive control improves thermal comfort by reducing overheating
through night cooling [123]. However, during the application of
this method, mathematical analysis of the thermal characteristics
of a building results in non-linear models.
Adaptive controllers self-regulate and adapt to the climatic

conditions in various conditioned spaces (buildings).These controllers are believed to be the most promising building adaptive
control systems [125].
3.1.3. A combination of FLC, neural controller, adaptive and PID
controller
Due to the merits associated with Optimal, predictive, and
adaptive controllers, there has been increased research work done
to replace use of PID in buildings. However, there has been slow
response to replace PID controllers in industry because of a
number of reasons. The rst reason is that it can prove difcult to
guarantee robustness especially for non-linear methods that occur
in buildings [125]. In addition, some methods require complex
computer control algorithm for even simple low-cost hardware
used in buildings [50]. Finally, Industrialists fear adaptation of
something that would appear as empirical after development of
PID control [89].
In order to promote the acceptance of recent developments to
replace PID control, researchers have come up with hybrid solutions in which the PID element is retained [126129]. One of the
most widely adopted adaptive controller in building industry and
which falls under this hybrid classication is based on pattern
recognition [130]. Since this controller (pattern recognition adaptive controller, PRAC) is an add-on element on to a conventional
PID controller, it has been received well by Industrialists. In this
controller, the PID loop is maintained and pattern recognition
controller (PRAC) makes adjustments to the controller parameters
so as to maintain consistent control performance. This method is
based on detecting load and set-point changes and acquiring
features that characterize the response. The main limitation is that
PRAC cannot represent characteristics of the VAV AC system but
can only estimate VAV control process noise [113].
In a study by Federspiel [131] on ow control with electric
actuators, the author developed an adaptive controller that is
specically used for ow loops that contain constant speed
actuators. In these loops, the plant contains an integrator and the
static gain can vary considerably during normal operation. The
adaptive controller continually estimates the maximum static gain
and noise variance in order to determine the appropriate actuator
stroke times. They have the ability to self-regulate and adapt to the
climatic conditions in various applications [132].
3.2. Control strategies
The VAV air-conditioning system control strategies are the
approaches employed to achieve optimal system control. These
approaches are broadly divided into two classes namely; pressure
dependent and pressure independent control strategies [133].
Pressure dependent control uses the temperature sensor (thermostat) xed inside the conditioned zone to directly control
damper position. The damper inclination (position) determines
the actual air volume reaching the zone which is dependent on the
static pressure in the duct work at the terminal unit's inlet.
Therefore, the zone temperature sensor continually corrects the
damper position but its response can be slow and thus lead to
unacceptable variations of temperature within the zone. On the
other hand, in pressure independent control approach, the actual
volume of the primary air which ows into the conditioned zone is
under direct control. This is achieved through the employment of
two cascading control loops (i.e. the output of one loop establishes
the set-point for the other loop). The rst loop (space temperature
sensor) controls zone temperature. The output of from the rst
loop is fed to the second controller as a reset signal, setting the
airow required to cool/heat the space. The second controller
varies damper position in the VAV terminal unit to maintain the
795
airow at the set-point. Thus, the damper position is not under

direct control but it is a result of regulating airow through the
terminal unit. There are no studies which show clearly the merits
of one strategy over the other [133]. However, most operating
engineers prefer using pressure independent control (PIC) over
pressure dependent control (PDC) due to the fact that PIC gives
more stable VAV boxes operation controls [134].
The specic control strategies that are usually employed to
control VAV air-conditioning systems are discussed below.
3.2.1. Optimal start/stop control strategy
According to Murphy [135], optimal start makes use of building
automation system also known as BAS to determine the time
space needed to meet the set-point temperature of each zone.
Here, the system waits for the set point temperature to be realized
before starting. This ensures that zone temperature reaches
occupied set point in time [135]. This strategy minimizes system
operation time as well as energy saving by eliminating energy
usage when the building is unoccupied [136].
In optimal stop control strategy, the BAS determines the
shortest period within which each zone's heating and cooling
could be shut off [135]. This ensures a small drift of indoor temperature from occupied set point. In this case, only heating or
cooling are shut off while the supply fan continues to operate. At
the same time, the OA damper remains open to ventilate the
building [134]. This strategy also minimizes system operation time
and saves energy [136].
3.2.2. Duct static pressure reset control strategy
The set point value for duct static pressure can be reset by
varying system supply airow rate [137]. In VAV systems, the zone
load determines supply airow [138]. According to ASHRAE
(1999), modulation of airow is usually realized through the use of
variable frequency drive (VFD), inlet (entrance) guide vanes, an
eddy-current clutch, or outlet dampers [139]. The static pressure
sensor normally located about two thirds (2/3) downstream
within the main supply air duct controls the control devices [138].
Maintenance of a high level of constant static pressure set point
value as a control strategy is popular in use today. This ensures
sufcient airow under design conditions. The main shortcoming
of this strategy is high static pressure under part load conditions.
The fan curve in Fig. 3.4 below illustrates the operation of the
above control strategy:
The diagram above shows that damper modulation towards the
minimum position is directly proportional to decrease in cooling
load. This results into an increase in the static pressure from point
W to point X. The pressure increase is as a result of increased
system resistance. Then the controller reduces fan speed as can be
seen from n1 to n2. This results into maintenance of a xed static
pressure set point value (i.e. from point X to point Y) [138]. When
the fan attains new speed n2, the static pressure is kept at the
initial set point. However, at this instant in time, the position of
the damper in the terminal box is partially open signifying
increased system resistance. Thus, a high static pressure set-point
for this part loads condition.
To avoid changing airow rate under the existing condition, we
can reduce the speed of the fan to n3 while t the same time
allowing the damper to open up as shown by path YZ. This solves
the challenge of wasting power used by the fan that could be
caused by the excess pressure head (PYPZ) that has to be overcome by the fan. In order to attain this and minimize the power
consumed by the fan, the control strategy can be converted to one
in which we maintain the position of the damper in the terminal
box to be fully open by modulating speed of the fan.
Fig. 3.4. Variation of operating point in a VAV system [138].
We notice that in order to avoid extremely low fan speed, it is a

fundamental practice to set a minimum value for either fan speed
or static pressure.
Several works have been done on static pressure reset control
strategy. This is discussed in the sections that follow.
3.2.2.1. Resetting static pressure using terminal box feedback techniques. Hartman developed a concept known as Terminal Regulated Air Volume (TRAV). This concept uses real time terminal unit
airow demand to directly control supply fan speed instead of
satisfying duct static pressure set point value [140,141]. Therefore,
this concept relies on the difference between the terminal box set
point and measured airows. Later on, Hartman [142] developed a
fan speed control algorithm for his previous studies. Englander
and Norford [143] suggested two control strategies. They modulated fan speed by utilizing the error signal of primary airow
from one or more zones. They compute the error signal as the
maximum or average deviation of the set point from measured
airows. Rather than determining the difference between the set
point and the measured airows, Warren and Norford [144] proposed a control strategy in which terminal box controllers trigger
an alarm whenever the system cannot meet required airow. The
number of alarms determines set point value for adjusting static
pressure. When the ow sensors function properly, the system can
save up to 42% of fan power energy consumption.
Wei et al. designed improved reset algorithm for controlling air
volume in which the air volume control logic is connected parallel
to the reset schedule [138]. In this strategy, the position of the
damper and the static pressure of the highest zone form the basis
for comparing the output. The smaller output signal is then
selected and sent to the variable frequency drive (VFD) controller.
The shortcoming of this method is that it is not applicable in
buildings with pneumatic terminal unitcontrollers.
Nassif and Moujaes proposed split-signal damper control
strategy [145]. In this control strategy, the outdoor air is controlled
only by one damper while other dampers are kept fully opened.
They found that since the two modulating devices (dampers) are
usually fully opened under occupied conditions, the strategy can
facilitate realization of minimum static pressure drop in inlet air
dampers and this result in use of minimum energy in both supply
and return fans.
Haasl et al. [146] proposed the polling of terminal units at an
interval of ve minutes and that the adjustment of the set point
value of static pressure be carried out at a minimum and maximum value to ensure that at least one damper is always maintained at 95% open position. This type of static pressure reset
control was modied by Song et al. [147]. The authors carried out
the modication process by using VSD speed as the basis for
varying the minimum set point value. The speed range was from
796
Fig. 3.5. Resetting static pressure using terminal box feedback techniques.
30% to 70%. Pang et al. [148] polled each terminal box at intervals
of twenty minutes and used the maximum damper position using
a dead band from 85% to 95% as the basis for adjusting the set
point value.
Fig. 3.5 below depicts resetting static pressure using terminal
box feedback techniques.
3.2.2.2. Resetting static pressure without using terminal box feedback
techniques. Liu developed an airow control for VAV AC systems
[149]. The author created a fan airow station (FAS) whose role
was to calculate airow using measured fan speed and fan head.
This was done in order to avoid inaccurate airow measurements.
Liu and Liu [150] as well as Wu et al. [151] modied the previous
strategy developed by Liu [149] in their efforts to address system
stability. Their modications accounted for the system load distribution prole using a load factor that increases with zone
load ratio.
Liu et al. suggested that the set point of static pressure should
be linearly reset as a function of system airow [152]. They proposed the use of VFD speed in the absence of measured total airow. In their case study, they demonstrated that annual fan power
could be saved by between 68% and 75% through the application of
static pressure set-point reset based on VFD speed.
Zheng et al. [153] suggested that an effective control strategy
for envelop dominated buildings was linearly resetting the static
pressure set-point with reference to outdoor air temperature. In
such a strategy, energy savings are heavily dependent on the
minimum airow setting and the OA temperature range.
3.2.3. Air temperature reset control strategy
3.2.3.1. Supply air temperature reset control strategy (SAT). This
control strategy resets supply air temperature (SAT) in warm and
cool weather conditions. This is implemented to cut down the
energy associated with cooling systems in VAV systems conguration when zones may require reheating of central supply air for
maintenance of desired temperature [154]. Raising SAT during part
load conditions result into saving reheat energy in a VAV system
[135]. At any time when temperature of outdoor air is lower than
the set-point value of SAT, compressors are shut down and returnair (RA) and the outdoor air dampers adjust to provide the desired
SAT. Raising SAT whenever the supply airow is decreased to a
minimum setting of a VAV terminal unit can also decrease reheat
at zone level for zones with low cooling loads [135]. However,
since the supply air is at a slightly higher temperature, we need
more air to satisfy the cooling load and that will result into
increasing fan energy [11]. Therefore, SAT reset strategy should be
able to minimize overall VAV system energy use. To achieve
simultaneous minimization of compressor energy, reheat energy,
as well as fan energy several researchers [155157] have agreed on
general principles for balancing the competing issues as
outlined below:
Fig. 3.6. Supply air temperature reset control process.
When the outside air is warm keep the supply air cold so as to
benet from energy savings due to fan unloading. During mild
weather, raise the SAT set point in order to enhance the benets of
the economizer and reduce reheat energy [135]. For instance in
Fig. 3.6 below, the SAT set point is reset depending on the degree
by which outdoor dry-bulb temperature changes. Whenever the
dry-bulb temperature of outdoor air is warm or higher than 18 C
the set point value of SAT remains constant at 13 C (i.e. design
value). At this moment there is little or literary no cooling benet
from outdoor air, and no reheat is needed since the cooling load in
most zones is high enough. The absence of reheat at this point
eliminates chances of over-cooling. The air is kept cold due to
reduced airow emerging from the fan which is turned down at
this point. The turning down of the fan saves energy. The process
can be shown in Fig. 3.6.
When the outside air is cooler, the controls start to increase the
SAT set point value (i.e. upward reset). Under cooler conditions,
resetting enhances the benets of the economizer. Also, in case of
any zone level reheat it is minimized or in some cases eliminated.
In addition, there is a likelihood of signicant unloading of the
supply fan such that the extra energy used to deliver more air is
reduced. Finally, the maximum reset value is limited to 16 C. This
allows the systems to meet cooling load demands in interior zones
without the need to oversize VAV ductwork and terminals.
Literature survey on this control strategy shows that indeed the
method contributes to substantial energy saving. Norford et al.
studied SAT and outdoor airow control as well as its resulting
effects on energy consumption trend in a variable air volume
system [11]. They found out that varying the SAT led to a 10%
decrease in used energy during winter and between 11% and 21%
during summer.
Zaheer-Uddin and Zheng showed that there is an optimal SAT
in a climate. At this point, the supply air contains high relative
humidity and return air is used [158]. Zheng and Zaheer-Uddin
carried out a case study and reported 20% energy saving as a result
of increasing outdoor air under certain conditions in a system that
used return air [159].
Ke et al. [160] simulated eight control strategies for ventilation
in VAV systems. Among the eight strategies three included varying
supply air temperature. They concluded that supply air temperature as well as supply airow rate were the best parameters that
can be optimized in a VAV system. Ke and Mumma [161] presented
a method for determining the SAT set point in VAV systems. They
reported that the optimization concept simultaneously reduced
energy consumption and met ventilation requirements.
Engdahl and Johansson [6] investigated the inuence of optimal SAT on energy use in a VAV system. Their analyses showed
that optimal control of SAT is superior to xed supply air temperature as it results into signicantly lower VAV system energy
use. Engdahl and Svensson [162] presented a theory for optimal
SAT with regard to energy use and analyzed its ability to save
energy when optimal temperature is applied to a complete outdoor air VAV system in northern Europe climatic conditions. Liu
et al. [152] published a guideline for SAT reset schedule in a continuous Commissioning Guidebook and provided some case studies to support the need for SAT reset schedule.
Advantages of SAT reset control strategy include:(1) Decreasing compressor energy;
(2) more compressor or chiller shut off hours when economizer
provides all necessary cooling, and
(3) signicantly decreases reheat energy.
However, the SAT reset control strategy may increase fan
energy and cause increased levels of humidity in zones.
3.2.3.2. Discharge air temperature reset control strategy (DAT). In
this approach reset is realized through lowering or rising temperature at which air is discharged from the air handling unit. The raising
and lowering is carried out based on the corresponding fall or rise of
the temperature of return air. The objective of doing this is to ensure
that all conditioned space load demands are effectively met as well
as minimizing the total energy used in the entire process. Much of
the energy saving is realized when the actual load conditions are
less than design operating conditions. Such a scenario calls for
minimizing pumping, cooling and heating which are the major
points of energy consumption in the system. However, when damper opening degree of the VAV terminal unit increases in response to
load demand, energy consumed by the supply fan will increase. VAV
terminal unit damper position reset is discussed here.
Reset of discharge air temperature is carried out on the basis of
damper position in VAV terminal units. Modulating actuators give
the signal of the actual damper position whereas uctuating
actuators give the signal of an approximate damper position [163].
Whenever there is a decrease of heat load, air dampers start to
gradually close to reduce the supply air volume. This causes the
DAT to gradually increase until one or more dampers attain a
desired position. On the other hand, whenever there is an increase
of heat load, air dampers start to gradually open to increase the
supply air volume. This causes the DAT to gradually decrease until
one or more dampers attain a desired position [135]. Here, the set
point value of DAT is continually adjusted according to the prevailing load demands.
3.2.4. Demand controlled ventilation (DCV) control strategy
Demand controlled ventilation is dened as automatic ventilation that uses occupancy to make adjustments. This strategy
reduces outdoor air intake rates below design rates when the
actual occupancy of zones served by the system goes below the
designed value [164]. The approach comprises of hardware,
797
software, and strategy for control, and is an important part of

building's ventilation design [165]. The main shortcoming of DCV
is that its implementation may be too complicated and require
proper hardware and software to operate. Also, location and
number of CO2 sensors or real time data pose a challenge to the
efciency of the control algorithm if the system's outdoor air (OA)
intake is dynamic [166]. As part of their efforts to improve DCV, Yu
et al. designed new integrated demand controlled ventilation
(IDCV) for single duct VAV system with conference rooms [166].
This logic resets both the minimum and maximum airow rates of
the terminal boxes based on the occupancy. The authors used a
model for one ofce building to demonstrate energy savings and
show how the indoor air ventilation can be satised under different conditions. This methodology can ensure acceptable Indoor
Air Quality (IAQ) and energy savings with lower AO intake ratio.
3.2.4.1. Carbon dioxide based DCV. ANSI/ASHRAE [167] recommends that the acceptable ways to dene instantaneous occupancy are population counting, use of CO2 sensors, use of timers,
occupancy schedules and occupancy sensors. Among these
methods, CO2 based DCV is the most preferred design. After carrying out a review on several DVC sensors, Won and Yang
recommended CO2 sensors based on three criteria: performance,
cost and indoor air quality (IAQ) [168]. Scientically, people take in
oxygen and generate CO2 and odorous bio-efuents. CO2 concentration is considered to be the most dependable indicator of
odorous efuent concentration compared to other methods [167].
It should be well understood that CO2 based DCV is applicable in
buildings with unpredictable variations in occupancy, where
cooling and heating is required during most part of the year, and
pollutant emissions from non-occupants sources are negligible
[169].
3.2.4.1.1. CO2based DCV for singlezone VAV systems. In singlezone VAV systems, the rate of intake of outdoor air is controlled by
the CO2 concentration sensed at the zone level. Several research
works have been ongoing in this eld of study. Houghton [170]
proposed three control strategies as; CO2 set point control of
outdoor air (OA) damper to be at fully closed or fully opened
positions, proportional control (PC) where degree of opening of
the OA damper is proportional to the CO2 concentration, and PID
control which takes into account both change rate of CO2 concentration and concentration separately.
After the publication of Standard 62.1-2004 [171], Taylor [172]
illustrated how the control system designs should be modied in
compliance to this standard. The author established the relationship between CO2 concentration and the set-point of air ow rate
by assuming steady-state conditions. Therefore, Taylor concluded
that for single zone VAV systems, the ow rate of the required
minimum OA intake is related to the difference between the
indoor and outdoor CO2 concentrations [172].
In buildings with dynamic occupancy schedules which change
on hourly interval, it is difcult to attain steady-state conditions. In
acknowledgment to this fact, Lu et al. [174] devised a strategy to
cover such scenarios. The authors proposed a strategy to supply a
base ventilation rate during unoccupied periods and to supply a
calculated ventilation rate during occupied hours by solving a CO2
mass balance equation which maintains indoor CO2 concentration
close to the set-point.
3.2.4.1.2. CO2 based DCV for multizone VAV systems. Early studies in this eld were mostly done with simulations [174176].
Sorensen [176] modeled a two-zone ofce with one zone served
by a constant air volume (CAV) terminal box without DCV and the
other by a VAV system with CO2 based DCV. The author used both
temperature and CO2 concentration to control the supply fan
speed and OA damper position. Sorensen found energy savings of
798
up to 31% for the VAV system with CO2 based DCV compared to the
CAV system without DCV for cold ambient conditions.
A CO2 based DCV control strategy (S-CO2-DCV) to maintain a
low supply air CO2 concentration set-point enough to dilute CO2
produced in critical zone so as to meet occupant needs was proposed by Nassif et al. [177] and Nassif and Zaheer-Uddin [178]. In
this strategy, the set-point of supply air CO2 concentration is
reached at on the assumption that critical zone is fully occupied
[177,178]. This strategy has the shortcoming such that it can lead
to over-ventilation of zones, when it is not fully occupied. Overventilation will lead to low CO2 concentrations in both return and
supply air. In response to this, the system closes the air-handler OA
damper until supply air CO2 concentration is achieved thus
decreasing over-ventilation.
A strategy similar to the above was also reported by Warden
[179]. Nassif et al. [177] compared their strategy with four other
strategies. They concluded that the S-CO2-DCV strategy can dynamically reset the rate of outdoor air intake to be less than the peak
design ow-rate, yet the dynamic ow-rate of supply air is always
above that for the CO2 based DCV system. They reported that the
return air duct CO2 based DCV did not provide enough ventilation to
the critical zones [177]. Nassif [180] updated his S-CO2-DCV strategy
in response to the new requirements in Standard 62-1-2010. In his
study, he proposed a robust DCV strategy based on CO2 concentration. The proposed strategy offers great benets in terms of
better indoor air control and improved energy efciency.
Several possible approaches for dynamic ventilation reset with
or without DCV were summarized by Stanke [181]. In his study,
Stanke included several occupancy sensing methods other than
CO2 sensors. The advantages and disadvantages of each method of
dynamic ventilation reset were outlined but no single method
emerges as a clear best solution.
Xu et al. proposed a model based optimal ventilation control
method for multi-zone VAV systems [182]. The strategy uses
genetic algorithms to seek an optimized solution for temperature
set-points of critical zones, while considering the dynamic changes in occupancy.
Advantages of DCV mainly include energy cost savings in various types of buildings as was reported by Mansson [183] and Meir
[184]. However, DCV also suffers from disadvantages such as;
(1) Difculties in maintenance, frequent calibration, drift, temperature effects, and uncertain proper location of sensors.
(2) Sensor unreliability as was identied by [185188]. In the event
that the sensors are out of calibration such that measured CO2
concentration is higher than the actual value, the system will
allow inow of excess outdoor air into the air-handling system
to meet the inaccurately measured CO2 concentration in the
zones served by the air handler. Houghton [170] examined the
impact of free cooling by an economizer on DCV systems. He
found out that prolonged periods of economizing decrease the
potential energy savings from DCV.
3.3. Control algorithms for optimization tools in VAV AC systems
3.3.1. Genetic algorithms
Genetic algorithms (GAs) use Darwinian evolution theory to
realize the optimization objective [189]. According to Koza [189]
GAs use operations of evolution to produce new populations with
tness values of higher average. The higher the tness value an
individual possesses, the closer it is to the optimal result. These
algorithms were developed in early 1970s but their utilization in
HVAC systems is recent [190]. The phrase genetic algorithm was
rst brought into existence by Bargley [191]. GAs have so far
addressed issues to do with design of VAV systems, or control of
the systems to minimize cost, energy consumption and
maximizing comfort. The system performance is mainly evaluated

using algebraic equations. A typical genetic algorithm has the
following steps:
i) Generating a random initial population of potential solutions
which contains several individuals.
ii) Evaluation of the tness value of each individual with specied cost function.
iii) Check whether the population meets the prescribed criteria of
optimization; if not, apply genetic operations such as selection, crossover, and mutation to the population to create a new
generation of potential solution.
iv) Repeat step (ii) and (iii) until the optimization criterion is
satised.
3.3.1.1. Applications in VAV systems. Application of genetic algorithms to optimize HVAC systems has had its share in the research
eld. Lu et al. [192,193] employed a GA to minimize the total
power required for a HVAC system. In another study by Lu et al.
[194], they used a binary elitist GA to minimize power consumption of chillers, pumps and fans in VAV system. Nassif et al. [195]
reported simultaneous minimization of predicted percentage dissatisfaction (comfort zone) and energy use by a GA in the real-time
computation of the setting of HVAC system. The GA determined
optimal parameters such as set point values for duct static pressure, supply air temperature, temperature of water in the chiller,
the temperature set-points of all zones and the required reheat. Jin
et al. [196] reported minimization of used energy when a GA was
applied as an online control tool to control outdoor air for a VAV
air conditioning system.
Huang and Lam [197] optimized the proportional and integral
parameters of a PID controller for regulation of a HVAC system.
Their results showed that the genetic algorithm based optimization procedures yielded minimum overshoot and minimum settling time. In a study by Wang and Wang [198] the authors
designed a PID controller for a VAV air conditioning system. In
their design, the controller parameters were chosen by a GA to
minimize the deviation between the set-point and measured
temperature in a test room. The authors concluded that the
method of genetic algorithm is an effective way to optimize the
controller parameters of a VAV system.
Xue et al. [199] proposed a method that predicts and optimizes
the inlet ow conditions using various objective functions. They
reported that the program can accurately predict the inlet
boundaries conditions, with given space variables with reduced
CFD cases.
3.3.1.2. Advantages of GAs
(1) Easy to use and readily available [200].
(2) Have good robustness. This drawn from their probabilistic
nature which ensures that the initial guess has a low incidence
on the nal result of the optimization.
(3) Genetic Algorithms have features that allow the designer to
use a simplied model to perform a pre-optimization. This
aids to identify good designs, and then evaluate more precisely with a better model a certain number of the best
individuals produced by GAs [201].
(4) Genetic Algorithms are easily parallelized. The objective
function of several individuals of a population could be calculated simultaneous on different processors.
3.3.1.3. Disadvantages of GAs
(1) At times they can be slow to converge in case of complex
problems compared to other optimization methods that have
been developed specically for a problem. This problem can

be solved by coupling genetic algorithm with articial neutral
network [202].
(2) Difculty in adjustment of the algorithm. This due to the
reason that there are no rules for determining the number of
individuals in the population, the number of generation or
cross-over and mutation probability [203].
3.3.2. Fuzzy-logic control algorithms
The fuzzy logic controls (FLCs) have received increased attention in recent years due to their compatibility with multi-input
multi-output (MIMO) systems. The basic idea of this approach is to
incorporate the experience of human operators in the design of
controllers. Basically a fuzzy logic model entails mapping input
values to output values using simple IF - THEN statements for
instance; IF room temperature is high, THEN supply more cool air to
the room [204].
According to Ralston and Ward [205], a control algorithm can
be constructed from a set of linguistic rules describing the control
strategies of the operator. A fuzzy logic algorithm is basically
composed of three steps:
i) Fuzzication (input stage) here input values are obtained
and transferred to fuzzy type via transfer functions such as
Gaussian distribution curve, trapezoidal, triangular etc.,
ii) Fuzzy reasoning unit/inferencing/processing stage (linguistic
description) generates inferred results based on previously
fuzzied values and linguistic IF..THEN statements, and
iii) Defuzzication (Output stage) converts the fuzzy control
signals into the real control output.
These stages can be summarized in Fig. 3.7.
The fuzzy If-Then Rule:
Let, x and y, be two system inputs and z be the output. The
fuzzy if-then rule will be:
If
is
Ai
and
is
Bi ; then
is
Ci
where x, y and z are linguistic variables,Ai , Bi , and C i are fuzzy sets.

They are characterized by membership functions.
Another type of fuzzy if-then rule is TakagiSugeno (TS) type
which contains fuzzy set rules involved only on the antecedent
[206].
If
is
Ai
and
is
Bi;
then
z c0i c1i x c2i y
where, the consequent rule is a linear function of input variables.

When the order is zero, T-S fuzzy rule is reduced to,
If
is
Ai
and
is
Bi ; then
z zi
where zi is a crisply dened number.

3.3.2.1. Applications in VAV systems. FLC methods have been widely
applied in VAV AC systems over time. In a study by Kaymaz [207],
the FLC method employing two inputs was comparatively tested
with on/off control for a HVAC system in a hospital building. The
FLC method displayed decreased overshoot, oscillation, and power
consumption than the other three methods employed in the study.
In addition, the FL controllers displayed a better performance
when system parameters were changed. So et al. [208,209] applied
FLC to the control of an AHU plant in VAV systems. The authors
used triangular membership functions for error and error rate to
generate FLC outputs for the AHU plant air damper, fan, cooling
coil, humidier, and reheat coil. The performance was compared
with tuned and detuned PID control using computer simulations
and it was found that FLC compared well with tuned PID control
but was more robust and response time and offset of FLC was
superior to that of detuned PID control.
799
Karunakaran et al. [210] developed a simple VAV building

model and experimentally investigated the energy utilization of
the VAV air-conditioning system. Input data for fuzzy logic were
zone temperature and duct static pressure while the output was
supply fan speed. They claimed that the energy saving potential of
the VAV system was close to 30% when not operating at full load
conditions compared with a CAV system. They also reported
achievement of the required thermal comfort using FL controller.
In another study by Karunakaran et al. [211], they employed a
Fuzzy Logic Controller in variable air volume (VAV) AC system and
variable refrigerant volume (VRV) AC system to optimally control
temperature and air quality in which FLC generated compressor
speed, fan speed, and damper opening were the outputs. This
system provided improved air quality, thermal comfort, and
energy efciency, compared to the constant air volume air conditioning system.
3.3.2.2. Advantages of FLCs
(1) No mathematical model is needed as is the case of genetic
algorithms.
(2) Possible to use all available information about the process in
the design of the fuzzy approximator scheme.
(3) Can control non-linear processes.
(4) Capable of ne controlling and energy conservation.
3.3.2.3. Disadvantages of FLCs
(1) Consists of many local variables which many it difcult
to tune.
(2) Many unclear options as thousands of different fuzzy system
congurations may arise depending on conjunction, disjunction, implication and diffuzication choices.
3.3.3. Fuzzy-Genetic controllers
The properties of genetic algorithms make them suitable to use
in design and optimization of fuzzy systems. Genetic algorithms
are basically used to tune fuzzy logic controllers so as to enable the
implementation of complex control techniques [212]. FLCs are
applied in VAV systems to solve simple problems such as that of
thermal regulation [213]. However, in in order to minimize energy
consumption and maintain desired level of comfort, many variables have to be considered and this makes the strategy more
complex. Therefore, researchers have in the recent years resorted
to the combination of fuzzy logic and genetic algorithm controllers
to optimize VAV systems.
The applications of Fuzzy-Genetic controllers in VAV systems
have been widely studied by several scholars. Alcala et al. [212]
proposed the use of genetic algorithms to develop smartly tuned
fuzzy logic controllers to control energy performance and indoor
requirements of air-conditioning systems. The proposed technique
yielded much better results than OnOff controller showing the
good behavior that the FLCs can achieve.
Parameshwaran et al. [213] developed a fuzzy based genetic
algorithm program to optimize the energy saving potential for
VAV air-conditioning systems. The authors concluded that VAV GA
optimization system with FLC is superior to CAV system in terms of
energy saving. In addition, the optimal static pressure obtained
through this approach leads to improved air distribution. Finally,
the approach improved the merits of uniform air distribution,
thermal comfort and improved energy savings potential [213].
In another study, Parameshwaran et al. [215] combined the
fuzzy and genetic algorithms to experimentally investigate energy
conservation in VRF system and VAV system. The method gave
good results for thermal comfort and IAQ requirements with
substantial energy saving. Homod et al. [216] proposed an online
800
Fig. 3.7. Block fuzzy logic controller.
Fig. 3.8. Dual maximum logic operation modes.
control strategy of HVAC system whose main objective is to

effectively deal with characteristics which are coupled and tackle
nonlinearities. The results showed superior performance of the
studied system compared to normal fuzzy control schemes.
3.3.4. Direct Digital Control (DDC)
A Direct Digital Control (DDC) uses a single computer or
interconnected computers to control the environmental conditions of a building. The DDC system consists of a microprocessor
based closed control loop. The microprocessor performs the control function by directly transmitting a control signal (can be
electronic or pneumatic) to the damper [217]. The main function of
VAV systems control algorithms is to decide the control inputs in
order to promote maintenance of good thermal comfort and IAQ in
a conditioned space. The control inputs are usually supply air
_ sa ) [218].
temperature (T sa ) and mass ow rate of supplied air (m
This section will review four VAV systems DDC control algorithms
namely as discussed below;
3.3.4.1. Baseline (BL) control. Generally, there are two categories of
baseline control known as single maximum and dual maximum BL
control logics [219]. Here, the dual maximum BL control algorithm
will be discussed. Dual maximum baseline controller uses the zone
temperature measurements obtain temperature and ow rate of
supply air. The temperature of conditioned air as well as the return
air ratio is held constant. This control algorithm is specically used
to control single zones in places where VAV air conditioning systems are employed.
Under this control logic, we can have four operation modes

[220]. These are, Re-heating, Heating, Dead-Band and cooling as
shown in Fig. 3.8 that follows.
The discussion of control logic for dual maximum can be
summarized into the following operation as given in Ref. [221];
If the temperature of the zone is below the re-heating set-point
(Rspt ) longer than a specic set time, the reheating mode is automatically turned on. If the temperature of the zone is held above
the cooling set-point (C spt ) longer than a specic set time, cooling
takes place. If the zone temperature remains between re-heat set
point (Rspt ) and heating set-point (H spt ) longer than a specic set
time, heating is activated. If the zone temperature remains
between the H spt and C spt longer than a specic set time, the deadband mode is turned on. During re-heat, temperature of supply air
is given a maximum possible value (T samax ) and the rate of ow of
supply air is changed using PID controller in order to maintain
actual space temperature as close as possible to the set-point
(T spt ). During heating, the rate of ow of supply air is given a
minimum permissible value while the temperature of supply air is
gradually adjusted by a PID controller in order to maintain actual
space temperature as close as possible to the set-point (T spt ). The
minimum permissible value for the rate of airow is chosen
according to ASHRAE ventilation standard 62.1. 2010 [164] as follows;
_ sa=p N pd m
_ samin
Minimum Allowed Flow RateK m
3:6
_ aout=p =1 Rra ; m
_ samin m
_ aout=z Az =1 Rra ;
_ sa=p m
m
3:7
_ aout=p is volume of OA needed per occupant,

where m
is the
_ aout=z is volume of
designed occupancy (number of occupants), m
N pd
_ samin is the minimum volume

outside air needed per given area, m
of supply air when the zone is not occupied, Az is the area of the
oor, and Rra is the return air ratio.
But since BL method doesn't use the occupancy data, we use
designed occupancy (N pd ) to compute the minimum permissible
ow rate. In the Dead-band mode, no re-heating is done i.e. value
of supply air temperature (T sa ) is equivalent to the temperature of
conditioned air (T ca ), and supply air ow rate is continually
adjusted to a minimum allowable value as given in the
equation above.
During the cooling mode, the value of supply air temperature
(T sa ) is equivalent to the value of conditioned air temperature (T ca )
but, the supply air ow rate is changed in order to match the setpoint T spt in the zone.
The main advantages of dual maximum baseline (BL) controller
are as follows:
(1) Maintains code ventilation requirements at low airow rates
in the dead-band and the rst stage of heating. This results
into reduced reheat and fan energy;
(2) The reduction of the minimum airow rate in the dead-band
eliminates overcooling of spaces, and
(3) since the reheat valve can be controlled to maintain discharge
temperature of supply air rather than zone temperature, the
temperature of supply air can be limited in order to eliminate
stratication and short-circuiting of supply to return [220].
However, dual maximum baseline (BL) controller suffers from
the following disadvantages:
(1) There are high chances of dumping and poor air distribution
in case of improper diffusers which might lead to greater
airow turndown.
(2) Despite meeting ventilation code requirements, the reduction
in airow rates results into increased indoor contaminants.
3.3.4.2. Zone-Level Feedback Control (Z-FC). This control algorithm
requires measurements of occupancy and zone temperature. The
main differences between Z-FC and BL control logics are; rst, the
actual occupancy is used in place of the design occupancy to
compute the minimum permissible ow as shown below:
Minimum Allowed Flow RateK msa=p Np k msamin
3:8
where Np k is the occupancy measured at time index k.

The second difference is that the temperature set-points are
determined by the zone occupancy status (i.e. occupied or unoccupied). The choice of design temperature variables involves a
tradeoff between energy saving and thermal comfort. The range
between the maximum and minimum allowable zone temperature set points should be well chosen to ensure that temperature
of the zone is within the right range in order to guarantee comfort
to the occupants. A wide range means decreased energy usage due
to reduced re-heating under small heat load and decreased airow
when the heat load is large. However, too wide range comes with
increased occupant discomfort. Therefore, the parameters for
unoccupied period should be selected to unsure that the range
between minimum and maximum zone temperature during
unoccupied period should be greater than the range between
minimum and maximum zone temperature during occupied
period.
A number of researchers have studied this eld and given
different ndings. Yu et al. [222] designed a smart logic for
conference room VAV terminal unit of single duct system. Evaluation of their control algorithm was via simulation. They used
the simulated results of thermal performance and energy consumption to compare conventional and improved control logic
sequences. Results showed that the proposed strategy can
improve both IAQ and energy saving.
801
Taylor [223] during his study on increasing efciency with VAV

air conditioning system static pressure set point reset claimed that
static pressure set point reset logic requires knowledge of required
static pressure in each zone since the critical zone varies
depending on thermal loads the zones and their location in the
distribution system. The author further reported that the highly
interactive relationship that exists between duct static pressure
and damper position can pose a challenge in realizing the
control logic.
Cho [224] developed an algorithm for terminal unit control
with variable minimum rate of airow and used it in conventional single duct VAV terminal box control sequences. Validation of simulation results was done through evaluation of the
actual building for indoor air quality (IAQ), comfort, and energy
usage. The energy consumption and thermal performance of
terminal units operating under two control algorithms were
compared. The results showed that the ratio of constant minimum rate of airow causes signicant concurrent heating and
cooling cycles; the terminal box can maintain room thermal
comfort conditions to meet the various load changes in addition
to reducing fan power and saving reheat energy; and the energy
usage of the variable minimum rate of airow ratio is smaller
compared to that of the conventional constant minimum rate of
airow ratio.
3.3.4.3. AHU-Level Feedback Control (A-FC). This feedback strategy
computes four inputs namely; the supply air temperature (T sa ),
_ sa ), conditioned air temperature (T ca ), as
supply air ow rate (m
well as return air ratio (Rra ). The algorithm comprises of four steps
as follows; at every time index k,
(1) Obtain measurements.
(2) Then determine the return air (Rra ) by carrying out exhaustive
search.
(3) Determine the conditioned air temperature (T ca ) based on the
enthalpy of mixed air (MA), and
(4) Finally, re-calculate return air (Rra ) to ensure satisfaction of
zone humidity constraints.
Two special cases can be considered while evaluating the performance of A-FC algorithm. The rst case is that conditioned air
temperature (T ca ) remains unchanged at the minimum value,
T camin . This means that, T camax T camin . The second case is that the
return air ratio (Rra ) remains unchanged implying, Rramax Rramin .
Several research works has been accomplished in this eld of
study. Rajkumar et al. [225] carried out a study on optimization of
AHU control strategy. They designed a strategy for controlling
chilled water valve. They compared this strategy with the existing
control strategy to realize the desired results. Their results showed
a signicant decrease in power consumption of the AHU.
Yu et al. [226,227] proposed an integrated AHU (IAHU) operation for ofce buildings where there exists a continuous air path.
Its operation is different from those of previous AHUs where the
control of units is independent without interactions; IAHU brings
interdependence among the AHUs based on the varying conditions
of outdoor air and modes of system operation in order to realize an
energy performance that is synergized and maintain acceptable
indoor air quality. The strategy for synergization allows intake of
the outdoor air and airows be properly relocated among the air
handling units (AHUs) during appropriate conditions. To implement the strategy, the authors adopted a zoned-AHU system layout with interior and exterior regions virtually separated in order
reduce thermal compromising due to difference in cooling and
heating modes across the spaces. The authors concluded that their
strategy achieve high energy efciency when implemented in
buildings with open spaces.
802
Fig. 3.9. Schematic diagram of a model-based predictive controller [216].
3.3.4.4. AHU-Level Model Predictive Control (A-MPC). This algorithm

also determines all the four inputs just as in A-FC. However, its
strategy is based on model predictive control that is able to obtain
control inputs by solving an optimization problem with constraints. This requires extra information such as a model of thermal dynamics, hygro-dynamics as well as predictions of the
environmental weather conditions [229]. Occupancy prediction
helps to determine acceptable zone temperature range when the
zone is occupied or not. The process of model predictive control
[216] is as shown in Fig. 3.9.
We can analyze the control logic under two scenarios namely
occupied and unoccupied modes. The occupied mode is invoked if
the occupancy is measured at the initiation of the process when
the time interval is at least 1 [218]. The unoccupied mode is
invoked if the occupancy measured at the initiation of the process
is measured to be zero.
Literature survey shows that Model Predictive Controllers
(MPC) have been have widely developed for energy saving and
building comfort [230]. It controls the Predictive Mean Vote (PMV)
[231], a criteria for ensuring thermal comfort in buildings. Chen
[232] carried studies on use of MPC to control a system for heating
building oor with a large thermal lag. Several researchers [233
236] have used the method to improve performance of VAV systems. MPC combined with adaptive neuro-fuzzy inference systems
[237] has been applied in the eld of building energy
management.
3.4. Summary
Control of VAV systems helps to increase the working and
energy efciencies of the systems. The VAV system components
which are usually under control include, VAV air handling unit
(AHU), VAV terminal box, Boiler plant, and Chiller plant [88]. In
VAV systems, variables that are usually controlled include; fan
speed, static pressure, zone air temperature, humidity, ventilation,
ow-rate of air entering zones, and ow-rate of chilled water
entering the cooling coil.
PID controllers are simple to understand and use as their
structure is less complex compared to optimal, predictive and
adaptive controllers. The control process in PID controllers is rapid
and this makes them unfavorable to use in VAV AC systems which
don't need rapid response. In addition, these controllers (PID) are
limited to SISO systems. Optimal, predictive adaptive controllers
are superior in control compared to PID controllers but they
require building models to carry out control. Recent control systems are hybrid types which combine PID elements with adaptive
and predictive elements. An example of a hybrid controller is the
pattern recognition adaptive controller (PRAC).
Duct static pressure set point can be reset by varying the supply
airow rate in the system. According to ASHRAE [141], airow is
usually modulated by means of a variable frequency drive (VFD),
an eddy-current clutch, inlet guide vanes, outlet dampers, or a
vane axial fan with adjustable pitch blades. The following points
are worthy noting here, that;
The improved air volume control logic reset algorithm elim-

inates the possibility of high speed operation of the fan due to

faulty conditions. However, it is not suitable in buildings with
pneumatic terminal box controllers.
The split-signal damper control strategy can lead to minimum
static pressure drop in economizer dampers which results in
use of minimum energy in both return and supply fans.
The set point of static pressure should be linearly reset as a
function of system airow.
An effective control strategy for envelop dominated buildings
was linearly resetting the static pressure set-point on the basis
of outside air temperature.
Many scholars have agreed those supply air temperature and
supply airow rates are the best parameters that can be optimized in a VAV system [6,160,161] as they greatly minimize
energy consumption.
The genetic algorithms have good robustness drawn from their

probabilistic nature, and can be easily parallelized. However they
suffer from shortcomings such as slow convergence rates under
some conditions, difculty in adjustment of algorithms since there
are no rules for determining the number of individuals in populations. The fuzzy-logic algorithms entail mapping input values on
to output values using the IF..THEN statements. They are used in
VAV systems to solve difculties in thermal regulations and
maintenance of temperature set-point. These algorithms have
several advantages which include less overshoot, oscillation and
power consumption compared to PID controllers, can be used in
MIMO systems and do not require models as they can control nonlinear processes.
The direct digital control algorithms reviewed in this study are
namely, Baseline (BL) control, Zone-Level Feedback Control (Z-FC),
AHU-Level Feedback Control (A-FC), and AHU-Level Model Predictive Control (A-MPC). Among these, A-MPC has recorded more
interest in recent research works as it proves superiority to
the rest.
4. Air properties
The air properties comprise of velocity/airow rate, duct static
pressure, supply air temperature, and air humidity in the
conditioned space. This section basically looks at two main areas

namely; airow rate measurement and indoor air quality (IAQ).
4.1. Airow rate measurement
Measurement of airow rate in buildings' air conditioning
systems is an important exercise that should not be overlooked.
This is because different airow rates greatly affect other important VAV system parameters such as static pressure and conditioned room temperature. Also, airow measurement can signicantly improve the performance of a building served by a VAV
AC system since airow measurement provides stable compartmentalized net pressure control [238]. In addition, air ow measurement can reduce VAV system energy consumption by
increasing system efciency and simultaneously minimize energy
wasted on conditioning the excess outside and supply air.
Airow rate measurement in VAV systems can be done in two
ways. The rst approach is to calculate total volume ow rate
through a duct by using the relationship below;
Q vA
4:1
where v is the average air velocity in m/s and A is the crosssectional area in contact with the air in the duct in m2. The average
velocity (v) can be determined by dividing the duct opening into a
number of equal areas and taking the velocity reading of each area
then calculating the average.
The second approach of determining airow rate is by using
pressure drop method. The basic principle in this approach is that
any obstruction in the air path will cause a pressure drop in the
moving air due to frictional resistance and turbulence. By characterizing the obstruction in the air duct and calibrating the
pressure drop against known ow rates, the value of pressure
difference may be used to determine volume airow rate.
According to Bernoulli's principle, the square of air velocity is
directly proportional to the pressure difference between tubes
when measurements are done using a standard pitot-tube as
shown in the equation below;
PT Ps
v2
PV
2
4:2
where P T total pressure (Pa), P s static pressure (Pa), air

density (kg/m3), v air velocity (m/s), and P V velocity pressure.
The above equation can be applied to measure pressure drops
in VAV terminal boxes as shown below [239];
P T P s P v
4:3
vin 2 vout 2
P T P s

4005
4005
4:4
"
P T P s
4Q
4005D2
2
4Q
4005WH
2 #
4:5
where vin and vout are the box inlet and outlet velocities, Q is the
volume airow rate, D is the inlet diameter of the box, W is the
inside width and H inside height of the box outlet.
4.1.1. Airow rate measuring devices
Proper selections of airow rate measuring devices are a very
important aspect for efcient operation of VAV AC systems [238].
The performances of these devices differ even under similar
operating conditions. VAV systems airow rate measuring devices
include but not limited to.
4.1.1.1. Pitot-static tubes. These are mostly used to measure airow
in terminal boxes. The Pitot tube is placed into the air stream when
measuring the velocity of an air in ducts, so that the sensing tip
803
points directly into the moving air stream [240]. Then the Pitot
tube along with a differential pressure measuring device will
generate a signal which represents the difference between the
total pressure and the static pressure [241]. The resultant pressure
magnitude represents the true pressure caused by air movement.
Pitot-static tubes have a disadvantage in that their accuracy
usually falls off at low end of the range.
4.1.1.2. Hot-wire anemometer. Hot-wire (thermal) anemometer can
accurately measure airow in buildings when adequate number of
sensors are used across a duct section [242]. The main advantage
of hot-wire anemometers is that they can provide analog outputs
that are proportional to ow and doesn't need calculation of
square root to measure airow. However, the application of this
device is unpopular due to the following disadvantages. Hot-wire
anemometers can only measure one point in the duct crosssection and may also require periodic calibration [243].
The set-up for the airow measurement is as shown in Fig. 4.1.
4.1.2. Airow control
Airow control in VAV systems is a very important aspect. The
main objective of airow control is to ensure each zone or space
receives sufcient airow, proper control of outside air intake and
maintenance of positive building pressure [149]. We can use
damper position control as well as static pressure control to
ensure delivery of sufcient air to each conditioned space [149].
Intake of outdoor air often utilizes the direct method [245], the
xed damper position, the plenum pressure control, and the CO2
demand control.
In many applications, control of outdoor air is achieved by
setting both the outside air damper and return air damper at xed
positions. The positioning of dampers is done during the process
of system balancing and testing. The prevailing conditions during
testing dictate intake of outdoor air. This can be higher or lower
than the set-point value. The outside air damper is usually
modulated by a controller in order to maintain the needed outdoor
airow. Accurate measurement of airow requires provision of a
minimum outdoor air fan or duct [246]. The direct method can
provide efcient outside air control as long as measurement of
airow is accurate. However, cases of air leakage in most buildings
can lead to reduced accuracy of measurement of outside air. This
may occur when the maximum outdoor air damper by-passes the
ow station [149].
The plenum-pressure control method is able to maintain the
desired level of pressure inside the air mixing box by regulating
the speed of return air fan or the position of return air damper
[247]. In this control method, the ow meter is the air damper.
4.1.3. Improvement of airow control
There have been several studies done to improve airow control strategies. Cheong [248] studied airow measurements balancing of air distribution system. The author studied the applicability and precision of tracer-gas method. Results showed that
tracer-gas method can be a good substitute of pitot-static transverse method in measuring ow during balancing.
Tan and Dexter [249] investigated the design of simple airow
actuators or control signals. The authors carried out simulation of
the air-circuits of a VAV AC system and experimentally validated
their ndings. They used simulation to investigate the relationship
between supply airow, inlet airow, extract airow. They further
used simulation to draw relationship between fan control signals
and mixing air-box dampers in the AHU [249]. Linear estimators
were proposed for determining airow rates on the basis of results
from simulation. Chow and Liu [250] developed and implemented
practical terminal box control sequences for resetting minimum
airow in a single duct terminal boxes in VAV systems. A
804
Fig. 4.1. Schematic diagram of airow measurement and control set-up [243].
measured data was used to carry out a comparison of the thermal

comfort and energy usage between conventional and improved
control sequences. Results showed that the improved sequences
for terminal box control can stabilize temperature of air in the
room and lead to substantial energy saving compared to the
conventional sequences.
Recently, Kang et al. [251] proposed a control algorithm for
minimum airow rate in a VAV system terminal unit for an ofce
building. They analyzed the temperature of vertical air in order to
determine the value of supply air temperature that does not lead
to stratication. After carrying out comparisons of the proposed
control algorithms and the existing control algorithms, they found
out that the proposed method was superior to the existing
methods. Liu et al. [252] investigated the solution of using a VAV
ow conditioner to control the upstream velocity prole in a VAV
system airow sensor and increasing the measurement accuracy
of the VAV airow. They used computational uid dynamics (CFD)
modeling to evaluate several ow conditioners. For all the inlet
conditions that were tested, the prototype reduced the VAV airow reading error when installed immediately before the VAV box
inlet, irrespective of upstream duct conditions [252].
4.2. Indoor climate requirements
The main goal of air-conditioning systems is provision of good
indoor climate for building occupants with reference to indoor air
quality (IAQ), thermal comfort and the acoustic environment
[253]. The indoor environment is affected by several factors such
as movement of air, temperature, rate of air exchange, humidity,
ventilation, gaseous, particulate and biological pollutants [254].
Indoor Air Quality (IAQ) can be dened as the chemical, biological
and physical characteristics of air inside a conditioned space.
Before 1970, IAQ was not a matter of concern to HVAC designers
but after the energy crisis of 1973, attention was given to indoor
air quality in buildings. Good indoor air quality can be realized in
VAV system by provision of a sufcient volume of fresh air to the
occupied zone while avoiding drafts large air temperature gradients, and large radiant asymmetry [255]. Poor indoor air quality
can lead to diseases such as sick building syndrome (SBS)
[256,257].
4.2.1. Indoor climatic problems related to VAV systems
Common indoor climatic problems associated with VAV air
supply include:
i) Insufcient supply of outdoor air

This problem is prevalent in places where circulated air is used
in exhaust air in an outside air economizer cycle for cooling/
heat recovery system. Several studies have been carried out in
attempt to determine the best ratio of outside air to supply air
in order to realize both proper indoor air quality and realization of energy savings [11,159,258260].
ii) Stagnant air
This problem usually occurs in events where the minimum
airow rates from the supply air device do not guarantee
sufcient air movement. This condition is dominant during
heating whenever the temperature of supply air is equal to or
raised above room temperature level. So far there are no clear
values given for minimum air motion needed for comfort.
However, it is known that increasing the speed of air motion
may increase comfort if the air temperature is too high [262].
iii) Draught
Draught has been reported to be a common problem associated
with VAV systems air distribution [2]. Draught in simple terms is
the unwanted cooling of human body caused by movement of air
in the room. The thermal balance of human body is affected by
draught in the sense that convective heat loss of the body is
increased as well as the evaporation of the skin moisture.
The draught problem in rooms is caused by the use of improper
diffusers or low supply air temperatures.
iv) System noise
Varying airow rates result into varying static pressures in
ducts. VAV system characteristics changes as the airow rates
at the terminal units/diffusers decrease. Consequently, an
increase in the total pressure of the fan occurs when fan speed
is held constant. In addition, as rate of airow is reduced, the
pressure drop in the duct system is reduced. The combined
effect of the above factors will lead to higher static pressure at
supply air terminal units as airow is decreased. The excess
pressure must be throttled off somewhere within the system.
As a consequence, the noise levels of the system devices will
increase [262].
4.2.2. Indoor air quality (IAQ) parameters
According to Novoselac and Srebric [255], the level of concentration of contaminants in an occupied space indicates IAQ. The
air quality for a given location in a room can be determined by the
degree of contamination [263,264]. This degree of contaminant
concentration is also known as dimensionless concentration [265]

or contaminant removal efciency [266] calculated as shown in
Eq. (4.6) below;
Ci Cs
;
Ce Cs
4:6
where C i is pollution level in a given location i, C s pollution level in

the supply air and, C e is pollution level in the exhaust air.
The degree of contaminant concentration is dependent on
spatial distribution of the contaminant sources as well as airow
pattern within a space. A typical value for degree of contamination
for perfect mixing ventilation is 1.0. In such a case, the exhaust
contaminant tends to be equal to the local room concentration.
4.2.3. Indoor air quality (IAQ) control
Control of level of concentration of indoor air pollutants and
improvement of IAQ is very critical in the provision of a comfortable and healthy indoor air environment. The recent approaches
used to control IAQ consist of control of pollution sources, ventilation and indoor air purication.
4.2.3.1. Control of pollution sources. According to Guo et al. [267],
control of pollution sources is the most economical and effective
approach in improving IAQ. The best way of controlling indoor air
pollution is to use lters during the design of a VAV air conditioning system. These help in preventing external pollutants
from entering the conditioned space. In addition, since dust and
liquid drops are breeding medium for bacteria, it is necessary to
carry out cleaning on components that are easily infected (e.g.
lter, heat exchanger, and mufer) in VAV AC systems [253].
4.2.3.2. Dilution of indoor air contaminants through ventilation.
Another control strategy is dilution of indoor air contaminants by
ventilation. Ventilation is dened by ASHRAE as the process of
supplying or removing air by natural or mechanical means to or
from any space. VAV air-conditioning systems supply air at a
constant temperature and ensure thermal comfort by varying
airow to occupied spaces. Most early VAV systems did not allow
control of outdoor air quantity, such that a decreasing amount of
outdoor air was provided as supply airow was reduced. Some
more recent designs ensure a minimum supply of outdoor air
stream.
ASHRAE Standard 62-2001 suggests Eqs. (4.7) and (4.8) below
for calculating the minimum outdoor airow required at the system level.
X
1 X Z
4:7
V ot Y V st
4:8
805
4.3. Summary
The most important properties of air in VAV air-conditioning
systems are airow rate, duct static pressure, humidity, outdoor
and supply air temperature, space temperature and return air
temperature of the conditioned space. Airow rate measurement
can be done using hot-wire anemometer, pitot-static, ow stations, trace-gas technique etc.
Airow control has the main objective of ensuring sufcient
airow capacity to each conditioned zoned, proper control of
outside air intake, and maintaining positive building pressure.
Airow control in VAV systems can be achieved through controlling static pressure and damper position. Accurate measurement
of airow requires a minimum outside air duct or fan.
Literature survey shows that balancing and distribution of airow in VAV air-conditioning systems is one of the main challenging areas of research concerning VAV system control. So far, there
is no standard method given for balancing and distribution of
airow through zones in VAV air-conditioning systems.
5. VAV system faults detection and diagnosis

VAV air conditioning systems' faults can occur at different
system levels [269]. Potential fault areas may include subsystems,
components, sensors and controllers. Absence of VAV system
faults can lead to energy saving of up to 30% [270]. Research in this
eld in HVAC systems began more than 30 years ago [268,271,272]
and have since then become a popular eld of study.
According to Pimenta and Mendoza [273], faults can be dened
as an unfortunate/unexpected condition that leads to malfunctioning of a system. Faults found in VAV systems are mostly
due to improper system design, application, operation or maintenance of the VAV systems [274,275]. Haves [276] noted that
these faults can be categorized as abrupt (e.g. a sudden failure of
sensor) or they can develop over a period of time. Those faults
occurring over a period of time are also known as degradation
faults. Wang et al. [277] suggested that we could also classify faults
as soft and hard faults depending on whether their occurrence is
abrupt or gradual.
5.1. Some common VAV system faults
Several researchers have agreed on some very common faults
in VAV systems. These scholars have documented a collection of
reports on common faults in HVAC system during their studies on
fault detection and diagnosis at different times and regions across
the world [278282]. These faults are as shown in table 5.1 that
follows.
5.2. Methods of fault detection and diagnosis
where
X V on =V st is uncorrected outdoor airow fraction.
Y V ot =V st is the corrected outdoor airow fraction
Z is the critical space ventilation fraction
V on is the sum of space ventilation airows.
V ot Y V st required system outdoor airow
V st is the total supply airow.
The challenge of VAV design is to properly ventilate all spaces
at all load conditions. Supply air in VAV systems is a mixture of
ventilation air and re-circulated air that is carried to the space by a
single duct. This therefore means that the proportion of air in this
supply air mixture must change as the supply airow changes to
maintain proper ventilation levels in conditioned spaces [268].
FDD systems carry out timely detection of faults and diagnose

their possible causes, enabling correction before additional
damage occurs to the system [283]. Usually faults are detected by
comparing set-point values and measured values of variables.
When an abnormal deviation is present, we conclude detection of
a possible fault. Thereafter, a detailed description of the type of the
fault, its magnitude (i.e. energy penalty, discomfort etc.), location
and cause is given. This stage is called fault diagnosis [273,287].
According to Isermann [284] there are two basic approaches
used to detect faults. These are state variable estimation and
parameter estimation. The state variable estimation approach
assumes process parameters are known then it monitors the signals. On the other hand, parameter estimation method aims at
directly monitoring the process, detecting any variation in
806
Table 5.1
A list of some common VAV system faults [288].
Faults
Symptoms
Clogging of pipes in the plant loop
1.
2.
1.
1.
Fouling of water heating/cooling coils

Stuck/leaking outside air economizer dampers
2.
Zone temperature sensor offset
1.
Fan/pump stuck at full/intermediate speed and fails to

respond to control signal
1.
2.
Increased energy consumption due to increased pressure drops in the plant loop.
Occupant discomfort due to insufcient water ow rate in the plant loop.
Reduction in coil capacity due to reduced overall conventional heat transfer coefcient (UA).
In case of stuck OA damper, no air modulation. This results into energy penalty when OA conditions allow
free cooling or minimum OA is demanded for mechanical cooling/heating mode.
In case of leaking OA damper, an energy penalty when rate of leakage is higher than the needed OA
ow rate.
Energy penalty and thermal discomfort due to improper control of outside air damper, return air damper,
and cooling/heating valve.
Higher energy consumption by the fan/pump in case it is stuck at higher speed than that required.
Reduced indoor air quality when the fan is operating at low speed than that required.
Fig. 5.1. (a) Flow diagram of state variable estimation approach, and (b) parameter estimation approach [284].
parameters with regard to the theoretical ones. The two processes

can be depicted as shown in Fig. 5.1 below;
Generally, the classication of FDD methods can be summarized as depicted by the diagram in Fig. 5.2 below.
5.3. Studies on fault detection and diagnosis (FDD) in VAV AC
systems
Literature survey suggests that early research on FDD in VAV
systems focused on main VAV system components namely, AHUs,
water pumps, and fans. During that era, several FDD methods were
studied, developed and applied to various VAV systems.
Dexter and Benouarets [285] investigated the application of
fuzzy models to detect and diagnose faults on AHUs and terminal
units. Faults were detected in the cooling coil. In another study,
Ngo and Dexter [286] developed a robust model-based strategy for
simultaneously detecting faults and avoiding false alarms which
can arise due to sensor bias in AHUs. No false alarms were
reported in practice. Later, Dexter et al. [287,288] focused on the
process of coil heat exchange in AHUs. In their studies, they made
analyses of ve modes of faulting namely; fouled coil, valve stuck
open, valve stuck midway, valve stuck closed and valve leak. A
robust fuzzy model for AHU fault diagnosis which considered the
temperature sensor error was designed.
Yoshida et al. [289291] investigated VAV damper faults and

tested their occurrence as both abrupt and successive faults. They
showed the robustness of RARX method. Lee et al. [292] described
more than 10 system faults and the utilization of a two-stage ANN
to diagnose faults in a simulated air handling unit (AHU). The
authors found out that the estimated temperature of supply air
could be implemented in a feed-back control loop to restore the
supply air temperature (SAT) value to the set-point value.
Katipamula et al. [293,294] designed an OA/economizer diagnostician to monitor AHU performance and automatically detect
problems with operation of economizer or automatically detect
ventilation problems for systems without economizers using a
decision tree method. The authors noted that it is difcult to
detect economizer failure and in many cases faults may be completely unnoticed. Dodier et al. [295] investigated both damper
failure and power failure in fan powered mixing box containing
two electric reheat stages. They adopted probabilistic inference
methods in real-time diagnostic system (RTDS). After presenting
laboratory data for application of real-time diagnostic system to
heating, ventilation and air-condition (HVAC), they reported that
the system yielded positive results but there were also some
shortcomings of misdiagnoses of states of failure.
Seem et al. [296] studied online control of terminal unit of
variable air volume (VAV) system. The authors described control
807
Fig. 5.2. A summary of classications of FDD methods [283].
loop performance assessment indices and showed how the indices

work during fault detection in both VAV terminal units and AHUs
[296]. The performance indices are integrated into VAV terminal
unit controllers and they are able to detect terminal boxes with
faults through comparison among a number of terminal units.
House et al. [297] examined a number of AHU faults. They used
ve classiers to detect and diagnose seven faults of a simulated
air handling unit (AHU). The authors went further to propose an
expert rule set with 28 simple rules to use to detect faults in AHUs
[298].
McGhee et al. [299] made a summary and classication of
typical faults in valves and actuators. They used experiments and
simulation tests to validate an ANN method applied to valve fault
diagnosis process. Han et al. [300] proposed an overall modelbased FDD system to address problems from diversied levels.
Their FDD method assumed that all system sensors were reliable
and fault free. However, sensor faults which could not be detected
by a simple pre-test would make the FDD rules malfunctional.
Wang and Chen [301] investigated faults related to airow
sensors in sensor-based demand control ventilation systems. They
reported that it was possible to apply the fault-tolerant control
using neutral networks for outdoor ventilation airow rate in
buildings.
Norford et al. [302,303] during their investigation of both
abrupt and gradual faults on several sections of three air handling
units, they proposed two FDD methods. These methods are physical and hybrid (gray) box methods. Nearly all faults in the two
matched AHUs were detected by using both methods. Field trials
of expert set rules were able to identify two occurrences of faults
in the air mixing box damper. They noted that there was limited
effort given to the development of diagnostic capabilities of VAV
boxes when compared to the effort given to AHUs and other types
of HVAC components [304].
Later on, Schein and House [305] proposed a diagnostic method
used to assess the performance of VAV terminal units. After their
study, the authors reported that their strategy gave encouraging
results as far as fault detection was concerned [305].
Qin and Wang [269] investigated faults in VAV terminal boxes.
They proposed a strategy that can use hybrid method to
automatically detect and diagnose faults in VAV air conditioning

systems. They reported that the method developed can effectively
detect faults in VAV systems. They concluded that, Principal
Component Analysis (PCA) based approach using system models at
different levels is a good strategy for for detecting and diagnosing
faults in ow sensor at the VAV terminal unit. Yi and Chen [306]
proposed a PCA method to detect sensor and identify sensor fault
in VAV systems. The PCA method partitions the measurement
vector space into principal component subspace (PCS) and residual
rubspace (RS). In case of occurrence of sensor fault, the sample
vector projection on RS will have a remarkable increase and consequently the value of square prediction error (SPE) will increase
and beyond the condence limit a fault will be detected. Their
results showed that PCA method can effectively detect and identify sensor faults, and then reconstruct sensor fault data by PCA
model of VAV system. Du et al. [307] designed two complementary
methods for single sensor fault detection and diagnosis in VAV
systems. The PCA-FDA method presented by the authors has
merits such as; easy of setting up in VAV systems, FDA does not
need complicated knowledge to commission, PCA has strong
capacity of detecting faults, and FDA is good at isolation of faults.
Xu et al. [308] developed an enhanced sensor AFDD strategy for
centrifugal chillers using PCA method and wavelet analysis
method. Their results showed that the proposed strategy performed well when compared to other related technologies.
Brambley et al. [309] studied self-correcting controls for VAV
system faults on lters, fans, coils, and terminal boxes. They
developed a set of algorithms to aid automatic correction of faults
in VAV systems. The algorithms were able to effectively detect,
diagnose and correct faults in most of the tested cases.
Recently, Zhu et al. [310] designed a method of diagnosing
faults in sensors located in AHU. The authors applied wavelet
analysis at 3-levels to decompose data from measurements, and
then extracting fractal measurements of each frequency band and
uses these to depict the characteristics of sensors with failure.
Then extract the signal into an eigenvector comprising of a number of fractal dimensions. Finally, is the introduction of the
eigenvector into a neutral network and then tuned to diagnose the
808
faults in sensors. The results suggest that the method can efciently diagnose various types of faults.
Wang et al. [311] proposed an online model-based FDD strategy
to diagnose sudden faults of VAV air-handling units (AHU). The
proposed FDD strategy is a hybrid method. Faults in AHU are
detected using a self-tuning model. The adjustment of model
parameters is accomplished by an optimization strategy based on
a genetic algorithm. If the fault detection threshold is less than the
corresponding residual, this suggests presence of faults in air
handling unit system. The results showed that the strategy can
effectively detect and diagnose faults of AHUs.
Yang et al. [312] designed a hybrid model-based strategy for
detecting faults occurring in sensors located in air handling units.
The authors combined the statistical residual-based method and
the fractal correlation dimension (FCD) algorithm. They claimed
that, considering the strengths and weaknesses associated with
each individual approaches, the hybrid fault detection technique
can give mutual strengths and complement each other.
Sterling et al. [313] studied a model-based detection and
diagnostic solution underpinned by modelica models that uses a
qualitative approach to diagnose that was applied to diagnosis of
AHU. The authors pointed out that modelica-based approach can
adapt to different plants and to changes in the same plant. Xiao
et al. [314] proposed a diagnostic Bayesian network (DBN) for
detecting and diagnosing VAV system terminal faults. Two rules
were adopted to isolate the fault on the basis of the fault probabilities to make the method more robust. When compared with
previous rule-based methods, the proposed method can work well
with uncertain and incomplete information since the faults are
with probabilities rather than in the Boolean format. Results
revealed that ten typical VAV terminal faults could be effectively
diagnosed.
mainly focused on the air handling unit as one of the main components in VAV systems. Recently, attention has been given to the
study of faults in VAV sub-systems like heating/cooling coils,
dampers etc. A few points to note here are that;
It is difcult to detect economizer failure and in many cases

faults may be completely unnoticed.
RARX as FDD method is robust and can even detect various

faults including stuck dampers [289291].
Principal Component Analysis (PCA) based approach using

system models at different levels is a good strategy for VAV
terminal ow sensor FDD [269].
It is evident from literature available on recent studies in this
eld that hybrid methods have taken the lead in FDD as they prove
to be the most efcient. Finally, VAV system faults contribute to a
substantial amount of energy wastage as has been reported by a
number of authors [283,315318].
6. VAV systems energy consumption analysis

The analysis of energy consumption of any HVAC system is
essential for design engineers to make sound choice decisions on
which kind of air conditioning systems they ought to recommend
for installation in buildings. In this section, we shall review
available literature on energy saving potential of VAV air conditioning systems compared to other existing systems, optimization of energy efciency of VAV air conditioning systems, and the
tools used for energy simulation in this eld of study.
6.1. Energy saving potential (efciency) of VAV AC systems compared
to other systems
5.4. Energy penalty due to VAV systems' faults

Madjidi [315] carried out a research on the dynamics in energy
use and comfort, as result of presence of faults in VAV systems. It
was reported that 20% or 30% of energy consumption can be saved
in small VAV systems of commercial buildings after debugging and
correction of faults [316]. Roth et al. [317] reported preliminary
results during their study to quantify the national energy impacts
of buildings faults in the U.S. Their results showed that key faults
can be estimated to cause annual energy wastage of between 4%
and 18% of the total energy used in commercial buildings. Katipamula and Brambley [283] reported that a reasonable amount of
energy in buildings can be lost due to poor ineffective control, bad
maintenance, and degradation of VAV system components. Lee
and Yik [318] presented the results of the energy penalty related to
various air-side VAV system faults. They found out that extra
cooling energy may be required for a single VAV terminal unit
damper that is stuck open on every oor.
5.5. Summary
The main function of a fault detection and diagnosis system is
timely detection of faults as well as diagnosis of their possible
causes, enabling correction before additional damage occurs to the
system [283]. VAV air conditioning systems' faults can occur at
four levels. Absence of VAV system faults can lead to energy saving
of up to 30% [270]. VAV faults can either be abrupt (e.g. a sudden
failure of a temperature sensor) or they can develop over a period
of time (e.g. a temperature or humidity sensor that drifts over time
or fouling of heat exchange devices).
The two basic approaches used to detect faults in VAV systems
are state variable estimation and parameter estimation. Most of
the studies done on FDD in VAV systems in earlier years were
There has been continued research works geared towards the

evaluation of energy efciency of VAV AC systems in relation to
other HVAC systems. Sekhar [319] made a critical comparison of
performance of a VAV system with an equivalent CAV system in
ve different conditioned spaces in hot and humid environments.
The VAV system was found to have better performance than the
CAV in terms of energy savings.
Wang et al. [311] proposed an online model-based FDD strategy
to diagnose sudden faults of VAV air-handling units (AHU). The
author reported that a minimum of 11.5% energy savings is
expected with a VAV system. This energy saving is closely associated with building envelope variations for identical oor plan
and similar internal loads. A year later, Sekhar and Yat [320]
evaluated the thermal comfort performance and energy consumption of ve different types of AC systems using an energy
simulation program. The systems evaluated were VAV system, CAV
system, Two Pipe Fan Coil System (TPFCS), Two Pipe Induction
Unit System (TPIUS), and Packaged Variable Air Volume System
(PVAVS). Results from simulation suggested that the VAV AC system has one of the lowest energy consumption characteristics,
second only to TPIUS.
Yang and Ting [321] developed an innovative simple energy
calculation methodology for estimating energy savings. The
authors compared VAV and CAV systems. The results of a full scale
experiment procedure performed to validate this calculation
methodology implied that the energy saving effect could be as
high as 49.7%. In addition, the year round experiments demonstrated a 3050% energy savings when VAV systems are employed
compared to CAV systems.
Pan et al. [322] investigated two high rise ofce buildings in
installed with VAV systems in Shanghai, China. They selected four
separate spaces on a standard ofce oor in each building facing
different directions (i.e. north, south, east and west) in order to

carry out on-site measurements of outdoor airow rates and IAQ
on an hourly basis during the four seasons. They recorded the
monthly electrical energy consumption of the two buildings. The
authors then carried out computer simulations of building electrical energy consumption and cost using two different outdoor air
control strategies of a typical VAV system as well as a fan coil unit
(FCU) system. A comparison of on-site recorded data and simulation results revealed that the VAV system consumed 5.4% less
energy compared to the FCU system.
Yao et al. [12] developed energy simulation models for comparing three kinds of HVAC systems namely, VAV, CAV, and FCU in
small ofce buildings for six different cities in China. They
obtained the energy model parameters either from the eld testing data or the performance data from the manufacturer. Their
results revealed that the VAV system has the potential to save
between 17% and 37% of the energy when compared to the CAV
system, between 4.6% and 10.2% when compared to the fan coil
system. These energy savings are dependent on the prevailing
climatic conditions.
6.2. Improving the energy performance of VAV systems
Zheng and Zaheer-Uddin [159] developed steady-state models
of HVAC system components in their effort to study optimization
of thermal processes in VAV systems. Their results showed that
optimization of economizers resulted into signicant energy savings of up to 20% without compromising IAQ.
Westphalen [323] developed an improved energy efcient roof
top VAV AC system. The system used blower modulation rather
than cooling to control SAT. This technique can be called reverse
VAV. In this approach, the adjustment of compressor speed was
executed on the basis of space temperature and the speed of
blower was adjusted to maintain desirable SAT. The author made a
comparison of the developed prototype and the conventional VAV
system in a eld performance test. Experimental results showed
that the proposed VAV system possessed better performance while
using 25% less energy compared to the conventional roof top VAV
system.
Engdahl and Johansson [6] investigated how optimal supply air
temperature (SAT) affects energy consumption of a VAV system.
They examined the energy saving potential of a controlled SAT of a
VAV system as compared to a constant SAT. Their results showed
that varying SAT had 827% energy saving potential compared to
the constant SAT scenario.
Gao et al. [324] used building experimental data to compare
how electric and hydronic reheat modes perform in VAV systems.
Their results showed energy saving of about 24% when hydronic
reheat is applied in AHU as compared to using electric reheat. The
claimed that the energy saving is caused by demand for low
minimum airow across the reheat coil.
Ke et al [325] developed an innovative system to improve the
shortcomings of energy saving and indoor comfort found in VAV
systems. Experimental results of the innovative system prototype
showed that, compared to the traditional unit, we can prevent the
overdraft through the adjustment of by-pass damper which leads
to a decrease of the supply or return air differential.
Yao and Wang [326] studied the inuence of different economizers on energy use of a VAV system in six ofce buildings
scattered in different climate zones in China. The two types of
economizers used were based on temperature and enthalpy cycles.
Comparison of energy simulation results showed that the economizers will lead to better energy savings ration in hot humid climates as compared to when they are applied to dry cold climatic
conditions.
809
In another study to enhance energy conservation in space air

conditioning, Parameshwaran et al. [327] developed a concept
involving the combination of VAV-based chilled water AC system
with thermal energy storage [TES] system. They observed that the
total energy used for cooling and ventilation was conserved due to
the good charging and discharging characteristics that were
exhibited by the phase change material (PCM). Results obtained
from experimental investigation of the system during summer and
winter climatic conditions under DCV and DCV integrated with the
economizer cycle were employed to validate the energy saving
capability of the proposed system. Based on the results, the system
is capable of achieving 28% energy saving each day under DCV and
47% energy saving each day under combined DCV-ECV mode,
when compared to the chilled water AC system.
In a recent research, Kusiak et al. [328] proposed a method of
minimizing energy consumption of AHU with computational
intelligence approach. The authors combined data-mining with
optimization. The multi-layer perceptron ensemble algorithm was
applied to develop predicative models of the energy consumption
of pump, chiller and the supply fans as well as return fans. Results
showed that chiller energy could be signicantly reduced by
slightly increasing the energy consumed by fans and pumps. The
total energy consumption of AHU was reduced by almost 23%.
6.3. Summary
VAV AC systems are one of the most energy efcient systems in
use today as has also been reported by several researchers
[12,319322]. Despite their current strengths, VAV systems energy
saving potential can still be improved. This can be done through
utilization of economizers resulting into signicant energy savings
without compromising IAQ [159], use of blower modulation rather
than cooling to control SAT [323], varying SAT [6], using hydronic
reheat [324], reducing supply/return air differential via adjustment of bypass damper to prevent overdraft [325], use of air-side
economizers in hot humid climates, increasing the temperature
set point of indoor air [326] etc.
7. Current applications of VAV AC systems

7.1. Working principle
The main aim of using VAV air-conditioning systems is provision of thermal comfort. This objective can be achieved by heating,
ventilation, and cooling processes. VAV AC systems are of different
kinds. They can be broadly divided into central VAV systems and
distributed VAV systems. These systems can vary the amounts of
outside air supplied by the fan unit and also vary volume of
delivered air through a given supply air diffuser. As such, it is able
to save energy and maintain comfortable environment in multiple
individual zones or areas. Each end of a supply duct branch contains variable air volume dampers which control the volume of air
delivered to a zone. Temperature sensors located in the zone
control VAV dampers. When the temperatures in the zones are
high, temperature sensors call for cooling and the dampers
respond by opening to allow the required amount of cool air to
ow into the zones to bring the temperatures to the set-point.
When heating is needed (i.e. temperature in the zones is low), the
temperature sensors call for increase of temperature and the
dampers respond by closing to reduce the amount of cool air that
ows into zones to bring the temperatures to the set-point.
In order to provide favorable working conditions to VAV
terminal units, a constant pressure is maintained at the outlet of
supply fan. This pressure must be sufcient to compensate for all
810
Fig. 7.1. Schematic diagram of a multi-zone VAV system.
pressure loses occurring in the air duct network during very

severe conditions (maximum airow rate).
Fig. 7.1 shows schematic diagram of a multi-zone VAV system
showing airow sequence and control parameters.
VAV systems have found application in various elds of life.
They are most commonly used in modern high rise large commercial ofce buildings. These systems have also been useful in
areas such as modern residential apartments, hospitals, factories,
auditoriums, arenas, hotels, libraries, shopping malls, educational
facilities (schools and universities), marine vessels (ships, submarines), trains, aerospace industry (planes) etc.
The application and scope of use of these systems is dependent
on number of zones served, and zone load requirements [329].
Central VAV air-conditioning systems are mainly used in most
commercial and industrial buildings. These come in different types
such as dual duct, all air systems, and air-water systems. The all-air
systems are the most commonly used central VAV systems
because of their simplicity and effective control. However, these

systems suffer from the fact that air is not an efcient medium of
heat transfer and as such, all-air VAV systems may require
extensive building volume for distribution of ductwork. In circumstances where it is difcult to accommodate ductwork in
building design, air-water systems are preferred.
Some VAV systems are used for cooling only. Such systems
work well in areas with uctuating cooling loads for instance in
conference rooms where the loads uctuate due to occupancy, or
exterior zones of a building where loads uctuate due to solar
orientation [330].
VAV induction systems have widely found use in hospitals
[329]. According to Trane [331] increased focus to reduce energy
usage has promoted the use of single-zone VAV systems in smaller
zones. As the popularity of application of VAV single zone systems
in smaller zones increase, they are also being employed in smaller
air conditioning facilities. According to Chen et al. [332] direct
expansion (DX) VAV systems are favorable for applications in light

commercial and low rise residential buildings. Multi-zone VAV
systems which have a single air handler with a hot and cold deck
which has a series of automatic hot and cold deck dampers can be
used to condition multi-zone buildings. In these systems different
zone requirements are met by mixing hot and cold air through the
zone dampers at the central air handler. System powered VAV airconditioning systems which use supply duct air to power the
controls on the terminal unit instead of pneumatically or electrically powered control systems are popular in smaller ofce
buildings which range from 10 to 50 tons.
7.2. Examples of VAV systems in practice
VAV systems have been widely employed in practice. Pan et al.
[322] reported the installation of VAV systems in two high-rise
ofce buildings in the Lujiazui Finance & Trade Zone in the Pudong
New Area in Shanghai, China. They claimed that in VAV systems,
the outdoor airow rate distributed to each zone varies greatly,
especially during part-load hours, making it difcult to ensure that
the zones meet ventilation requirements every hour. Nassif et al.
[333] evaluated a simplied process for nding set point values in
a variable air volume (VAV) system installed in an existing system.
The results showed that the VAV system developed lead to signicant energy savings.
Yee et al. [334] carried out a study in a commercial ofce
building at TaiKoo Place, Quarry Bay, in Hong Kong. Rohdin and
Moshfegh [335] studied energy use and thermal climate for
existing VAV systems in a large Swedish light alloy foundry. This
study showed that a VAV system is an interesting HVAC control
technique for the foundry industry. In this case the technical
potential for reducing energy use in terms of both heat and electricity was predicted to be about 30% (heat 30.3% and electricity
28.9%). It also revealed that neither the thermal comfort nor the
ventilation efciency is negatively affected. In addition, the VAV
system reduces power usage during the cold season, when the
demand for district heating and electricity is highest. Han et al.
[336] reported the existence of a VAV air-conditioning system in
an ofce building called Beijing Fortunate building in China during
their study to improve control performance of the system in use.
Aynur et al. [87] reported existence of a VAV system in the Chesapeake building located on the campus of University of Maryland.
Murphy [337] studied dehumidication performance of a VAV
system in a 30-person classroom in Jacksonville, Fla. The author
reported that VAV systems dehumidify well over a wide range of
indoor loads. Yao and Wang [326] reported the installation of VAV
systems with and without air-side economizers in Shenyang,
Beijing, Xian, Chengdu, Shanghai, and Guangzhou cities in China.
The results of their study showed good performance of VAV systems installed compared to other air-conditioning systems.
Alvarez et al. [338] studied an existing VAV system operating
under tropical conditions in Colombia. The authors modeled and
simulated the VAV system in order to quantify its energy consumption. In-Hout [339] gives a report of a recently completed
study on single-duct VAV terminal unit with hot water reheat and
DDC systems installed in different locations in Northern California.
The settings of VAV terminals units were modied in response to
observed system behavior.
Different applications have different challenges. In applications
such as auditoriums, problems with latent heat loads are likely to
arise. In equipment located just above the ceiling, control of sound
pollution can be a challenge. Lastly, when installed in hospitals
and nursing facilities it requires an alternative means of maintaining pressurization.
811
8. Conclusion
In this review we have discussed VAV systems modeling and
simulations, control strategies and optimization tools, the airow
characteristics of VAV systems, some common VAV systems' faults,
detection and diagnosis, energy consumption and analysis, and the
current applications of VAV air-conditioning systems.
There are two main classications of models namely, mathematical models and empirical models. The third type of models
known as hybrid models is simply a combination of the two main
model classes. Some of the challenges experienced with different
modeling methods include: at least each modeling approach has a
major shortcoming derived from assumptions, unmeasured disturbances, and uncertainties in thermal properties of elements
that make up the structure; there is no model that combines both
the AHU and building with all the required parameters; and the
existing models for space zone do not address the lag time cooling
load and the effect of solar rays incident on building walls, roof,
and windows. For accurate modeling, we should try to minimize
assumptions as much as possible.
Control of VAV systems helps to increase the working and
energy efciencies of the systems. The most common controllers
used in VAV systems are the PID controllers. The most common VAV
control strategies are the static pressure reset control strategy and
supply air temperature control strategy. Duct static pressure set
point can be reset by varying the supply airow rate in the system.
Modulation of airow is usually realized means of a variable frequency drive (VFD), inlet guide vanes, an eddy-current clutch, or
outlet dampers. The following were noted on VAV control, that;
The improved air volume control logic reset algorithm elim-

inates the possibility of high speed operation of the fan due to

faulty conditions. However, its shortcoming is that it cannot be
used in buildings with pneumatic terminal box controllers.
The split-signal damper control strategy can provide minimum
static pressure drop in economizer dampers which results in
use of minimum energy in both return and supply fans.
The set point of static pressure should be linearly reset as a
function of system airow.
An effective control strategy for envelop dominated buildings
was linearly resetting the static pressure set-point on the basis
of outside air temperature.
There is an optimal supply air temperature in a climate where
the relative humidity is high and return air is used.
Many scholars have agreed that, supply air temperature and
supply air ow rate are the best parameters that can be optimized in a VAV system as they greatly minimize energy
consumption.
The genetic algorithms have good robustness drawn from their

probabilistic nature, and can be easily parallelized. However they
suffer from shortcomings such as slow convergence rates under
some conditions, difculty in adjustment of algorithms since there
are no rules for determining the number of individuals in populations. The FLCs are used in VAV systems to solve difculties in
thermal regulations and maintenance of temperature set-point.
Their advantages include less overshoot, oscillation and power
consumption compared to PID controllers, can be used in MIMO
systems and do not require models as they can control non-linear
processes. AHU-Level Model Predictive Controllers (A-MPCs) are
the most common direct digital control algorithms in use today.
Airow control has the main objective of ensuring sufcient
airow capacity to each conditioned zoned, proper control of
outside air intake, and maintaining positive building pressure.
Airow control in VAV systems can be achieved through controlling static pressure and damper position. Literature survey shows
812
that balancing and distribution of airow in VAV air-conditioning

systems is one of the main challenging areas of research concerning VAV system control. So far, there is no standard method
given for balancing and distribution of airow through zones in
VAV air-conditioning systems.
The main function of fault detection and diagnosis system is
timely detection of faults as well as diagnosis of their possible causes,
enabling correction before additional damage occurs to the system.
Common faults in VAV systems include clogging of pipes in the
plant loop, fouling of water heating/cooling coils, stuck/leaking
outside air economizer dampers, zone temperature sensor offset,
fan/pump stuck at full/intermediate speed and fails to respond to
control signal etc. A few points to note here are that;
It is difcult to detect economizer failure and in many cases

faults may be completely unnoticed.
RARX as FDD method is robust and can even detect various

faults including stuck dampers.
Principal Component Analysis (PCA) based approach using

system models at different levels is a good strategy for VAV

terminal ow sensor FDD.
The PCA method can effectively detect and identify sensor
faults, and then reconstruct sensor fault data by PCA model of
VAV system while the FDA technique can reduce the dimensionality of data by maximizing the scatter between classes
while minimizing the scatter within classes. Therefore, a hybrid
of the two methods (i.e. PCA-FDA) method has merits such as;
ease of setting up in VAV systems, FDA does not need complicated knowledge to commission, PCA has strong capacity of
detecting faults, and FDA is good at isolation of faults.
The use of wavelet neutral network as FDD method increases
the efciency of detecting and diagnosing faults since when the
two methods are combined they compensate for each other's
weaknesses.
A fault diagnosis method for sensors in AHU based on neutral
network pre-processed by wavelet and fractal (NNPWF) can
efciently diagnose different kinds of faults.
A hybrid approach integrating model-based FDD method and
rule-based FDD method can provide an effective tool for
detecting and diagnosing the faults in AHUs.
Finally, we conclude that VAV AC systems are one of the most

energy efcient systems in use today as has also been reported by
several researchers. Despite their current strengths, VAV systems
energy saving potential can still be improved. This can be done
through optimization of outdoor air economy cycle results into
signicant energy savings without compromising IAQ, use of
blower modulation rather than cooling to control SAT, varying SAT,
using hydronic reheat, reducing supply/return air differential via
adjustment of bypass damper to prevent overdraft, use of air-side
economizers in hot humid climates, increasing the temperature
set point of indoor air, etc.
Acknowledgment
This work was supported by the National Natural Science
Foundation of China under Contract no. 51076100.
Appendix
Denition of terms used
Control point The actual condition (temperature, humidity,
pressure etc.) that is being sensed.
Error/offset This is the amount of deviation from the setpoint. It is the difference between the set-point and the
control point.
Flow-rate of air entering the zone This is the rate at which
cooled or heated (conditioned) air enters the zone in a given
time period.
Set point - This is the desired condition (which must be maintained) of a variable under control such as pressure, temperature, among others.
Static pressure This is the pressure of air that is not in motion
within the duct. It can also be understood as the amount of
resistance to air ow within the duct work. A high static pressure or resistance means more energy is needed to move air
within the duct work.
Throttling range (TR) This is the change in measured variable
(pressure, temperature, humidity etc.) that causes the controller
to travel from one end of its stroke to the other. An example of
this can be the amount of temperature range that causes an
actuator to travel from a completely open position to a completely closed position.
References
[1] Urban RA. Design considerations and operating characteristics of variable
volume systems. ASHRAE J 1969;12(2):7784.
[2] Shepherd K. VAV air conditioning systems. United Kingdom: Blackwell Science Ltd; 1999.
[3] ASHRAE Handbook. HVAC systems and equipment [Chapter 4]. In: Air
handling and distribution. Atlanta: American Society of Heating, Refrigerating and Air conditioning Engineers, Inc; 2008, 4.11, http://app.knovel.com/
web/toc.v/cid:kpASHRAE07/viewerType:toc/root_slug:ashrae-handbookheating/url_slug:ashrae-handbook-heating?b-cat-name=Mechanics%20%
26%20Mechanical%20Engineering&b-cat-slug=mechanics-mechanicalengineering&b-cat-id=219&b-order-by=name&b-sort-by=ascending&b-offset=0&b-lter-by=all-content.
[4] Dodd M. Comparing energy savings of different VAV systems. EnergySoft
LLC; 2012. p. 1 9.
[5] Bearg DW. Indoor air quality and HVAC systems. Boca Raton, FL: Lewis
Publishers; 1993.
[6] Engdahl F, Johansson D. Optimal supply air temperature with respect to
energy use in a variable air volume system. J Energy Build 2004;36:20518.
[7] Inoue U, Matsumoto T. A study on energy savings with variable air volume
systems by simulation and eld measurement. J Energy Build 1979;2:2736.
[8] Kloostra L. VAV systems save 38% of energy use. Heat Pip Air Cond 1979;51
(12):613.
[9] Mull TE. Energy conservation measures for air distribution and HVAC systems. J Plant Eng 2004;58(10):604.
[10] Mysen M, Rydock IP, Tjelaat PO. Demand controlled ventilation for ofce
cubicles-can it be protable? J Energy Build 2003;35:65762.
[11] Norford LK, Rabl A, Socolow RH. Control of supply air temperature and
outdoor airow and its effect on energy use in a variable air volume system.
ASHRAE Trans 1986;92:3045.
[12] Yao Y, Lian Z, Liu W, Hou Z, Wu M. Evaluation program for the energy-saving
of variable-air-volume systems. J Energy Build 2007;39(5):55868.
[13] Afram A, Janabi-Shari F. Review of modeling methods for HVAC systems.
J Appl Therm Eng 2014;67:50719.
[14] Homod RZ. Review on the HVAC system modeling types and the shortcomings of their application. J Energy 2013;2013:110. http://dx.doi.org/
10.1155/2013/768632 [Article ID 768632].
[15] GATC. Computer program for analysis of energy utilization in postal facilities,
vol. 1. Niles, USA: General American Transportation Corporation; 1967.
[16] Candy JV, Model-Based. Signal processing. New York, NY, USA: John Wiley &
Sons; 2006.
[17] Orosa JA. A new modeling methodology to control HVAC systems. J Expert
Syst Appl 2011;38(4):450513.
[18] Tashtoush B, Molhim M, Al-Rousan M. Dynamic model of an HVAC system
for control analysis. J Energy 2005;30:172945.
[19] Herzog S, Atabay D, Jungwirth J, Mikulovic V. Self-adapting building models for
model predictive control. In: Proceedings of the 13th conference of the international building performance simulation association; 2013. p. 248993.
[20] Thosar A, Patra A, Bhattacharyya S. Feedback linearization based control of a
variable air volume air conditioning system for cooling applications. ISA
Trans 2008;47:33949.
[21] Agachi PS, Nagy ZK, Cristea MV, Lucaci AI. Model based control. New York,
NY, USA: John Wiley & Sons; 2006.
[22] L X. Modelling of heat and moisture transfer in buildings: I. Model program.
J Energy Build 2002;34(10):103343.
[23] Ghiaus C, Hazyuk I. Calculation of optimal thermal load of intermittently

heated buildings. J Energy Build 2010;42(8):124858.
[24] Wang J, Zhang C, Jing Y. Hybrid CMAC-PID controller in heating ventilating
and air- conditioning system. In: Proceedings of IEEE international conference on mechatronics and automation (ICMA'07); August 2007. p. 3706
11.
[25] Wang J, Zhang C, Jing Y. Research of cascade control with an application to
central air-conditioning system. In: Proceedings of the IEEE conference on
digital object identier; 2007. p. 498503.
[26] Wang J, Zhang C, Jing Y. Analytical design of decoupling control for variableair-volume air-conditioning system. In: Proceedings of the IEEE conference
on digital object identier; 2008. p. 630635.
[27] Wang J, Jing Y, Zhang C. Robust cascade control system design for central airconditioning system. In: Proceedings of the world congress on intelligent
control and automation (WCICA 08); June 2008. p. 150611.
[28] Badescu V, Sicre B. Renewable energy for passive house heating II model.
Energy Build 2003;35:108596.
[29] Badescu V, Staicovic MD. Renewable energy for passive house heating Model
of the active solar heating system. EnergyBuild 2006;38:12941.
[30] Badescu V, Laaser N, Crutescu R. Warm season cooling requirements for passive
buildings in Southeastern Europe (Romania). Energy 2010;35:3284300.
[31] Badescu V, Laaser N, Crutescu R, Crutescu M, Dobrovicescu A, Tsatsaronis G.
Modeling, validation and time-dependent simulation of the rst large passive building in Romania. Renew Energy 2011;36:14257.
[32] Homod RZ, Sahari KSM, Almurib HAF, Nagi FH. RLF and TS fuzzy model
identication of indoor thermal comfort based on PMV/PPD. J Build Environ
2012;49:14153.
[33] Beghi A, Cecchinato L, Paggiaro F, Rampazzo M. VAVAC systems modeling
and simulation for FDD applications. In: Proceedings of the 9th IEEE international conference on control and automation ((ICCA)ICCA Santiago, Chile;
2011. p. 80005.
[34] Wu S, Sun JQ. A physics-based linear parametric model of room temperature
in ofce buildings. J Build Environ 2012;50:19.
[35] Energy estimating and modeling methods. In: ASHRAE handbook fundamentals, American society of heating, refrigerating and air-conditioning
engineers; 2009.
[36] Wang S, Xu X, Huang G. Robust MPC for temperature control of air conditioning systems concerning on constraints and multi-type uncertainties.
J Build Serv Eng Res Technol 2010;31:3955.
[37] Zajic I, Larkowski T, Sumislawska M, Burnham KJ, Hill D. Modelling of an air
handling unit: a Hammerstein-bilinear model identication approach. In:
Proceedings of the 21st international conference system engineering
(ICSEng), IEEE, Las Vegas, Nevada, United States; 2011. p. 5963.
[38] van Schijndel AWM, Schellen HL, Timmermans WJ. Simulation of the climate
system performance of a museum in case of failure events. J Energy Build
2010;42:17906.
[39] Tahmasebi F, Zach R, Schub M, Mahdavi A. Simulation model calibration: an
optimization-based approach. In: Proceedings of the Fourth GermanAustrian IBPSA Conference. Berlin, Germany: IBPSA; 2012. p. 38691.
[40] TRNSYS 16 A Transient Energy Simulation Program, Mathematical Reference,
Solar Energy Lab, University of Visconsin-Madison; 2005.
[41] Pascual JAF, Garay RS, Torrens JI, Keane MM. Heat ventilation and air conditioning modeling for model based fault detection and diagnosis. In: Proceedings of the 13th international conference building performance simulation association, BS2013, IBPSA, Chambery, France; 2013. p. 351320.
[42] Matsushita N, Yoshida H, Suzuki H. Application of the air-conditioning system energy simulation for commissioning (ACSES/CX) tool to HVAC system
commissioning Part 2: application to the substation of a heat source system
with bleed-in control. In: Proceedings of the 13th international conference
building performance simulation association, IBPSA, Chambery, France;
2013. p. 177885.
[43] Mossolly M, Ghali K, Ghaddar N. Optimal control strategy for a multi-zone
air conditioning system using a genetic algorithm. J Energy 2009;34:5866.
[44] Badescu V. Simple and accurate model for the ground heat exchanger of a
passive house. Renew Energy 2007;32:84555.
[45] Badescu V, Isvoranu D. Pneumatic and thermal design procedure and analysis of earth-to-air heat exchangers of registry type. Appl Energy
2011;88:126680.
[46] L X, Viljanen M. Controlling building indoor temperature and reducing
heating cost through night heating electric stove. J Energy Build 2011;33
(8):86573.
[47] Ogonowski S. Modeling of the heating system in small building for control.
J Energy Build 2011;42(9):15106.
[48] Bi Q, Cai WJ, Wang QG, Hang CC, Lee EL, Sun Y, Liu KD, Zhang Y, Zou B.
Advanced controller auto-tuning and its application in HVAC systems. Control Eng Pract 2000;8:63344.
[49] Rehrl J, Horn M, Temperature control for HVAC systems based on exact
linearization and model predictive control. In: Proceedings of the international conference on control application (CCA), IEEE, Denver, Colorado,
United States; 2011; p. 111924.
[50] Huang G, Wang S, Xu X. Robust model predictive control of VAV air-handling
units concerning uncertainties and constraints. HVAC&R Res 2010;16:1533.
[51] Kalogirou SA. Articial neural networks and genetic algorithms in energy
applications in buildings. J Adv Build Energy Res 2009;3:83120.
[52] Kusiak A, Li M, Zhang Z. A data-driven approach for steam load prediction in
buildings. J Appl Energy 2010;87:92533.
813
[53] Tang F. HVAC system modeling and optimization: a data-mining approach.

In: Department of industrial engineering. Iowa City, Iowa, United States: The
University of Iowa; 2010. p. 92.
[54] Kusiak A, Xu G. Modeling and optimization of HVAC systems using a
dynamic neural network. J Energy 2012;42:24150.
[55] Morisot O, Marchio D, Fault detection and diagnosis on HVAC variable air
volume system using articial neural network. In: Proceedings of the
building performance simulation, IBPSA. Kyoto, Japan, 1999. p. 102734.
[56] Ding L, Lv J, Li X, Li L, Support vector regression and ant-colony optimization
for HVAC cooling load prediction. In: Proceedings of the international symposium on computer, communication, control and automation (3CA). Tainan,
Taiwan: IEEE; 2010. vol. 1. p. 53741.
[57] Pal A, Mudi R. Self-tuning fuzzy PI controller and its applications to HVAC
systems. Int J Comput Cognit 2008;6:2530.
[58] Chen K, Jiao Y, Lee ES. Fuzzy adaptive networks in thermal comfort. J Appl
Math Lett 2006;19:4206.
[59] Lu H, Jia L, Kong S, Zhang Z. Predictive functional control based on fuzzy TS
model for HVAC systems temperature control. J Control Theory Appl
2007;5:948.
[60] Soyguder S, Alli H. Predicting of fan speed for energy saving in HVAC system
based on adaptive network based fuzzy inference system. J Expert Syst Appl
2009;36:86318.
[62] Yiu CMJ. Statistical modelling and forecasting schemes for air-conditioning
system. In: Department of Building Services Engineering. Hung Hom, Kowloon, Hong Kong: The Hong Kong Polytechnic University; 2008. p. 181.
[63] Dong B, ONeill Z, Luo D, Trevor B, Development and calibration of a reducedorder energy performance model for a mixed-use building. In: Proceedings
of the 13th conference on international building performance simulation
association BS2013, IBPSA. Chambery, France; 2013. p. 63643.
[64] Yoshida H, Kumar S. ARX and AFMM model-based on-line real-time data
base diagnosis of sudden fault in AHU of VAV system. Energy Convers Manag
1999;40:1191206.
[65] Yoshida H, Kumar S. Development of ARX model based off-line FDD technique for energy efcient buildings. J Renew Energy 2001;22:539.
[66] Nishiguchi J, Konda T, Dazai R, Data-driven optimal control for building
energy conservation. In: Proceedings of the SICE annual conference, IEEE,
Taipei, Taiwan; 2010. p. 11620.
[67] Watson I. A case study of maintenance of a commercially elded case-based
reasoning system. J Comput Intell 2001;17:38798.
[68] Zlatanovic I, Gligorevic K, Ivanovic S, Rudonja N. Energy-saving estimation
model for hypermarket HVAC systems applications. J Energy Build
2011;43:33539.
[69] Huang G. Model predictive control of VAV zone thermal systems concerning
bi-linearity and gain nonlinearity. J Control Eng Pract 2011;19:70010.
[70] McQuiston F, Parker J. Heating, ventilating, and air conditioning analysis and
design. New York: Wiley; 1997.
[71] Vakiloroaya V, Ha QP, Samali B. Energy-efcient HVAC systems: simulation
empirical modeling and gradient optimization. J Autom Construct
2013;31:17685.
[72] Vakiloroaya V, Zhu JG, Ha QP. Modeling and optimization of direct expansion
air conditioning system for commercial building energy saving. In: Proceedings of the 28th international symposium on automation and robotics in
construction (ISARC 2011), Seoul, Korea; 2011. p. 198202.
[73] Zhifang X, Lin S, Hongfei O. Refrigeration ow characteristics of electronic
expansion valve based on thermodynamic analysis and experiment. J Appl
Therm Eng 2008;28(23):23843.
[74] Wemhoff AP, Frank MV. Predictions of energy savings in HVAC systems by
lumped models. J Energy Build 2010;42:180714.
[75] Cui P, Yang H, Spitler JD, Fang Z. Simulation of hybrid ground-coupled heat
pump with domestic hot water heating systems using HVACSIM. J Energy
Build 2008;40:17316.
[76] Yan X, Ren Q, Meng Q. Global optimization of VAV air conditioning system.
In: Proceedings of the 8th world congress on intelligent control and automation (WCICA), IEEE. Jinan, China; 2010. p. 507781.
[77] Braun JE, Klein SA, Mitchell JW. Effectiveness models for cooling towers and
cooling coils. ASHRAE Trans 1989;95(2):16474.
[78] Wang X, Xia J, Zhang X, Shiochi S, Peng C, Jiang Y. Modelling and experimental analysis of variable refrigerant ow air-conditioning systems. In:
Proceedings of the IBPSA conference on building simulation; 2009. p. 3618.
[79] Wang YW, Cai WJ, Soh YC, Li SJ, Lu L, Xie L. A simplied modeling of cooling
coils for control and optimization of HVAC systems. J Energy Convers Manag
2004;45(1819):291530.
[80] Ghiaus C, Chicinas A, Inard C. Grey-box identication of air-handling unit
elements. J Control Eng Pract 2007;15(4):42133.
[81] Jin GY, Cai WJ, Wang YW, Yao Y. A simple dynamic model of cooling coil unit.
J Energy Convers Manag 2006;47:265972.
[82] Vakiloroaya V, Samali B, Madadnia J, Ha QP, Component-wise optimization
for a commercial central cooling plant. In: Proceedings of the annual conference IEEE industrial electronics society (IECON), IEEE. Melbourne, Victoria,
Australia; 2011. p. 2769774.
[83] Wang S. Dynamic simulation of building VAV air-conditioning system and
evaluation of EMCS on-line control strategies. J Build Environ 1999;34:681
705.
[84] Wang SW. Dynamic simulation of a building central chilling system and
evaluation of EMCS on-line control strategies. J Build Environ 1997;32(2).
814
[85] Zaheer-Uddin M, Zheng GR. A dynamic model of multi-zone VAV system for
control analysis. ASHRAE Trans 1994;100(l).
[86] House JM. A system approach to optimal control for HVAC and building
systems. ASHRAE Trans 1995;l0l(2).
[87] Aynur TN, Hwang Y, Radermacher R. Simulation of a VAV air conditioning
system in an existing building for the cooling mode. J Energy Build
2009;41:9229.
[88] Goldschmidt I. Temperature controls design. ASHRAE J 2006:48.
[89] Salsbury TI. A survey of control technologies in the building automation
industries. J international federation of automatic control; 2005.
[90] Levermore GJ. Building energy management systems: an application to
heating and control. London: E & FN SPON; 1992.
[91] Dounis AI, Bruant M, Santamouris MJ, Guarrancino G, Michel P. Comparison
of conventional and fuzzy control of indoor air quality in buildings. J Intell
Fuzzy Syst 1996;4(2):13140.
[92] Honeywell Inc., Engineering manual of automatic control for commercial
buildings; 1997.
[93] Kulkarni MR, Hong F. Energy optimal control of a residential space conditioning system based on sensible heat transfer modeling. J Build Environ
2004;39:318.
[94] Haines RW. HVAC system design handbook, Blue Ridge Summit, PA, Tab
Professional and Reference Books; 1988.
[95] Bai J, Zhang X. A new adaptive PI controller and its application in HVAC
systems. J Energy Convers Manag 2007;48:104354.
[96] Naidu DS, Rieger CG. Advanced control strategies for heating, ventilation, airconditioning, and refrigeration systems an overview: Part I: Hard control.
HVACR Res 2011;17(1):221.
[97] Krakow KI, Lin S. PI control of fan speed to maintain constant discharge
pressure. ASHRAE Trans Res 1995;101(Part 2):398407.
[98] Krakow KI. Sampling interval magnitude relation to digital PI control system
performance. ASHRAE Trans 1998;104(2):24656.
[99] Zhou Q, Liu M. An on-line self-tuning algorithm of pi controller for the
heating and cooling coil in buildings. In: Proceedings of the eleventh symposium on improving building systems in hot and humid climates. Fort
Worth, TX; June 12, 1998.
[100] Qu G, Zaheer-uddin M. Online H1 adaptive tuning of PI controllers for
discharge air temperature system. J Energy Convers Manag 2010;51:1179
95.
[101] Kreider JF, Curtiss PS, Rabl A. Heating and cooling of buildings; design for
efciency. 2nd ed.. New York, NY: McGraw-Hill; 2002.
[102] Ziegler J, Nichols N. Optimum settings for automatic controllers. Trans ASME
1942;65:43344.
[103] Nesler C, Stoecker W. Selecting the proportional and integral constants in the
direct digital control of discharge air temperature. ASHRAE Trans
1984;90:83445.
[104] Nesler C. Adaptive control of thermal processes in buildings. IEEE Control
Syst Mag 1986;6(4):913.
[105] Dexter A. Control system simulation - computer control. J Energy Build
1988;10:20311.
[106] Geng G, Geary G. On performance and tuning of PID controllers in HVAC
systems. In: Proceedings of the second IEEE conference on control applications. Vancouver, BC; 1993; 2. p. 81924.
[107] Lin F, Brandt RD, Saikalis G. Self-tuning of PID controllers by adaptive interaction. In: Proceedings of the American control conference. Chicago, Illinois;
June 2000.
[108] Wang YG, Shi ZG, Cai WJ. PID auto-tuner and its applications in HVAC systems. In: Proceedings of the American control conference. Arlington, VA;
June 2527, 2001.
[109] Kasahara M, Yamazaki T, Kazuu Y, Hashimoto Y, Kamimura K, Matsuba T,
Kurosu S. Stability analysis and tuning of PID controller in VAV systems.
ASHRAE Trans 2001;107(1):28596.
[110] Ozawa K, Noda Y, Yamazaki T, Kamimura K, Kurosu S. A tuning method for
PID controller using optimization subject to constraints on control input.
ASHRAE Trans Res 2003;109:7988.
[111] Lim D, Rasmussen B, Swaroop D. Selecting PID control gains for nonlinear
HVAC&R systems. HVACR Res 2009;15(6):9911019.
[112] Yamazaki T, Yamakawa Y, Kamimura K, Kurosu S. Air-conditioning control
system using PID control with adjustable manual reset, SICE Annual Conference 2010. p. 9738.
[113] Visioli A. Research trends for PID controllers. Acta Polytech 2012;52(5):144
50.
[114] strm K, Hgglund T. PID controllers: theory, design and tuning. Instrum
Soc Am 1995;164166:1968.
[115] Norman SN. Control system engineering. 5th ed.. Pomona: California State
Polytechnic University John Wiley & Sons Interprise; 2007. p. 3181.
[116] Xu Z, Liang Y, Han N. Development of a fuzzy PID controller of airconditioning unit. In: Proceedings of the 3rd international conference on
advanced computer theory and engineering (ICACTE); 2010. p. V6-559V6562.
[117] Wenju Y. A self-tuning fuzzy PI temperature controller. In: Proceedings of the
international conference on new trends in information and service science,
NISS '09; 2009. p. 33638.
[118] Wang S, Jin X. Model-based optimal control of VAV air-conditioning system
using genetic algorithms. J Build Environ 2000;35:47187.
[119] Curtis PS, Shavit G, Kreider K. Neural networks applied to buildings - a

tutorial and case studies in prediction and adaptive control. ASHRAE Trans
1996;102(1).
[120] Chen T. Real-time predictive supervisory operation of building thermal systems with thermal mass. J Energy Build 2001;33(2):14150.
[121] Morel N, Bauer M, El-Khoury M, Krauss J. Neurobat A predictive and adaptive
heating control system using articial neural networks. Int J Sol Energy
2001;21:161201.
[122] Nygard A. Predictive thermal control of building systems [Ph.D thesis]. Lausanne, Switzerland: Swiss Federal Institute of Technology; 1990.
[123] Lam HN. Stochastic modeling and genetic algorithm based optimal control of
air conditioning systems. J Build Simul 1993:43541.
[125] Dexter AL, Geng G, Haves P. Application of self-tuning PID control to HVAC
systems. IEE Colloquium (Digest) 1990;93:3.
[126] Hepworth SJ, Dexter AL. Adaptive neural control with stable learning. J Math
Comput Simul 1996;41(12):39.
[127] Rahmati A, Rashidi F, Rashidi M. A hybrid fuzzy logic and PID controller for
control of nonlinear HVAC systems. In: Proceedings of the IEEE international
conference on systems, man and cybernetics; 2003. 3. p. 224954.
[128] Zaheer-Uddin M, Tudoroiu N. Neuro-PID tracking control of a discharge air
temperature system. J Energy Convers Manag 2004;45(1516):240515.
[129] Salsbury TI. Temperature controller for VAV air-handling units based on
simplied physical models. HVACR Res 1998;4(3):265.
[130] Seem JE. New pattern recognition adaptive controller with application to
HVAC systems. Automatica 1998;34(8):969.
[131] Federspiel CC. Flow control with electric actuators. HVACR Res 1997;3
(3):265.
[132] Dounis AI, Caraiscos C. Advanced control systems engineering for energy and
comfort management in a building environment a review. Renew Sustain
Energy Rev 2009;13:124661.
[133] McDowall R. Fundamentals of HVAC control systems. ASHRAE eLearning,
Elsevier; 2009. p. 2379.
[134] Fan Z. Dynamic performance of control loops and their interactions in VAV
AC system [Masters' thesis]. Canada: Concordia University Montreal, Quebec;
2002.
[135] Murphy J. High performance VAV systems. ASHRAE J, October 2011;1829.
[136] Wilkinson P, Olmsted C. Modeling HVAC optimization control strategies for
high performance buildings. In: Proceedings of the fth national conference
of IBPSA-USA Madison. Wisconsin; August 13, 2012.
[137] Shim G, Song L, Wang G. Comparison of different fan control strategies on
variable air volume systems through simulations and experiments. Build
Environ 2014;72:21222.
[138] Wei GH, Liu MS, Claridge DE, Sakurai Y. Improved air volume control logic for
VAV systems. In: Proceedings of the twelfth symposium on improving
building systems in hot and humid climates. San Antonio, TX; May 1517,
2000.
[139] ASHRAE handbook HVAC Applications. American Society of heating,
refrigerating and air-conditioning engineers, Inc. 1791 Tullie Circle, N.E.,
Atlanta, GA; 1999.
[140] Hartman T. TRAV - a new HVAC concept. Heat/Pip/Air Cond Eng HPAC
1989;61(7):6973.
[141] Hartman T. Terminal regulated air volume (TRAV) systems. ASHRAE Trans
1993;99(1):791800.
[142] Hartman T. Global optimization strategies for high-performance controls.
ASHRAE Trans 1995;101(2):67987.
[143] Englander SL, Norford LK. Saving fan energy in VAV systems Part 2: Supply
fan control for static pressure minimization using DDC zone feedback.
ASHRAE Trans 1992;98(1):1932.
[144] Warren M, Norford LK. Integrating VAV zone requirements with supply fan
operation. ASHRAE J 1993;35(4):436.
[145] Nassif N, Moujaes S. A new operating strategy for economizer dampers of
VAV system. J Energy Build 2008;40:28999.
[146] Haasl T, Potter A, Irvine L, Luskay L. Retro-commissioning's greatest hits. In:
Proceedings of the 1st international conference for enhanced building
operations. Austin, TX; 2001. http://esl.eslwin.tamu.edu/digital-library.
html.
[147] Song LI, Joo D, Dong M, Liu J, Wang K, Hansen LQ, A. Swiatek. optimizing
HVAC control to improve building comfort and energy performance. Proceedings of the 3rd international conference for enhanced building operations. Berkeley, CA. 2003. http://esl.eslwin.tamu.edu/digital-library.html.
[148] Pang X, Zheng B, Liu M. Case study of continuous commissioning in an ofce
building. In: Proceedings of the 6th international conference for enhanced
building operations. Shenzhen, China; 2006. http://esl.eslwin.tamu.edu/
digital-library.html.
[149] Liu M. Variable speed drive volumetric tracking (VSDVT) for airow control
in variable air volume (VAV) systems. J Sol Energy Eng 2003;125:31823.
[150] Liu G, Liu M. Supply fan control methods for VAV systems using a fan airow
station. ASHRAE Trans 2008;114(2):4517.
[151] Wu L, M. Liu, Wang G, Pang X. Integrated static pressure reset with fan airow station in dual-duct VAV system control. In: Proceedings of the energy
sustainability conference, Long Beach, CA; 2007. p. 44149.
[152] Liu M, Claridge DE, Turner WD. Continuous Commissioning guidebook for
federal energy managers. Federal Energy Management Program, U.S.
Department of Energy, Washington, DC; 2002. www1.eere.energy.gov/femp/
operations_maintenance/om_ccguide.html.
[153] Zheng K, Li H, Yang H. Application of wireless sensor network (WSN) technologies in optimal static pressure reset in variable air volume (VAV) system.
In: Proceedings of the 7th international conference for enhanced building
operations. San Francisco, CA; 2007. http://esl.eslwin.tamu.edu/digitallibrary.html.
[154] Kao JY. ASHRAE Trans 1985;91(2B):810.
[155] Murphy J, Bakkum B. Chilled- Water VAV Systems. La Crosse, Wis.: Trane;
2009.
[156] Energy design resources. Advanced variable air volume system design guide.
Sonoma, California; 2009.
[157] Wei G, Liu M, Claridge D. Optimizing the supply air temperature reset
schedule for a single-duct VAV system. In: Proceedings of the twelfth symposium on improving building systems in hot and humid climates; 2000.
[158] Zaheer-Uddin M, Zheng GR. A VAV system model for simulation of energy
management control functions: of normal operation and duty cycling. Energy
Convers Manag J 1994;35(11):91731.
[159] Zheng GR, Zaheer-Uddin M. Optimization of thermal process in a variable air
volume HVAC system. Energy 1996;21(5):40720.
[160] Ke YP, Mumma SA, Stanke D. Simulation results and analysis of eight ventilation control strategies in VAV systems. ASHRAE Trans 1997;103(Part
2):38192.
[161] Ke YP, Mumma SA. Optimized supply air temperature (SAT) in variable-airvolume (VAV) systems. J Energy 1997;22(6):60114.
[162] Engdahl F, Svensson A. Pressure controlled variable air volume system theory. J Energy Build 2003;35(11):116172.
[163] Johnson Controls Inc., VAV terminal control applications. technical bulletin;
April 2003.
[164] ANSI/ASHRAE/IESNA. Standard 90.12010 Energy standard for buildings
except lowrise residential buildings. international code Council, Inc.,
Washington, D.C. and American Society of heating, refrigerating and air
conditioning engineers, Inc., Atlanta, GA; 2010.
[165] Stipe M. Demandcontrolled ventilation: a design guide. Salem, OR: Oregon
Ofce of Energy; 2003.
[166] Yu Y, Liu M, Cho Y, Xu K. Integrated demand controlled ventilation for single
duct VAV system with conference rooms. In: Proceedings of ICEBO; 2007.
California.
[167] ANSI/ASHRAE. Standard 62.12010 ventilation for acceptable indoor air
quality. American society of heating, refrigerating and airconditioning
engineers, Atlanta, GA; 2010.
[168] Won D, Yang W. The state oftheart in sensor technology for demand
controlled ventilation. IRCRR243. Ottawa, Canada: National Research
Council Canada; 2005.
[169] Emmerich SJ, Persily AK. Stateoftheart review of CO2 demand controlled
ventilation technology and application NISTIR 6729. Gaithersburg, MD:
National Institute of Standards and Technology; 2001.
[170] Houghton D. Demandcontrolled ventilation. Teaching buildings to Breathe.
Tech Update, TU9010, ESource, Boulder, CO; 1995.
[171] ANSI/ASHRAE. Standard 62.12004 ventilation for acceptable indoor air
quality. American Society of heating, refrigerating and airconditioning
engineers, Atlanta, GA; 2004.
[172] Taylor S. CO2based DCV using 62.12004. ASHRAE J 2006;48(5):6775.
[174] Knoespel PD, Mitchell JW, Beckman WA. Macroscopic model of indoor air
quality and automaticcontrol of ventilation airow. ASHRAE Trans 1991;97
(Pt 2):102030.
[175] Emmerich SJ, Mitchell JW, Beckman WA. Demandcontrolled ventilation in a
multizone ofce building. J Indoor Environ 1994;3(6):33140.
[176] Sorensen BR. Simulation of a small VAV plant. In: Proceedings of 7th international conference on indoor air quality and climate, J Indoor Air; 1996. 2. p.
199204.
[177] Nassif N, Kajl S, Sabourin R. Ventilation control strategy using the supply CO2
concentration set-point. HVACR Res 2005;11(2):23962.
[178] Nassif N, Zaheer-Uddin M. Simulated performance analysis of a multizone
VAV system under different ventilation control strategies. ASHRAE Trans
2007;113(1):61729.
[179] Warden D. Supply air CO2 control of minimum outdoor air for multiple space
systems. ASHRAE J 2004;46(10):2635.
[180] Nassif N. A robust CO2based demandcontrolled ventilation control strategy
for multizone HVAC systems. J Energy Build 2012;45:7281.
[181] Stanke D. Dynamic reset for multiplezone systems. (Standard 62.12007).
ASHRAE J 2010;52(3):2235.
[182] Xu X, Wang S, Sun Z, Xiao F. A modelbased optimal ventilation control
strategy of multizone VAV airconditioning systems. J Appl Therm Eng
2009;29(1):91104.
[183] Mansson L. Demand Controlled ventilation systems in nonindustrial buildings. In: Proceedings of the European conference on energy performance and
indoor climate in buildings. Lyon, France; 1994. Vol. 3. p. 895900.
[184] Meier S. Demandcontrolled ventilation requirements and control strategies. In: Mundt E, Malmstrom TG, editors. Proceedings of room vent 98, 6th
international conference on air distribution in rooms. Stockholm, Sweden;
June 1417, 1998. Vol. 1. p. 20512.
[185] Jones J, Meyers D, Singh H, Rojeski P. Performance analysis for commercially
available CO2 sensors. J Archit Eng 1997;3(1):2531.
[186] Fisk WJ, De Almeida AT. Sensorbased demandcontrolled ventilation: a
review. J Energy Build 1998;29(1):3545.
[187] Fisk WJ, Faulkner D, Sullivan DP. A pilot study of the accuracy of CO2 sensors
in commercial buildings. In IAQ 2007: healthy and sustainable buildings,
[188]
[189]
[190]
[191]
[192]
[193]
[194]
[195]
[196]
[197]
[198]
[199]
[200]
[201]
[202]
[203]
[204]
[205]
[206]
[207]
[208]
[209]
[210]
[211]
[212]
[213]
[215]
[216]
[217]
[218]
[219]
[220]
[221]
815
Baltimore, MD, Oct.1417. Atlanta, GA: American society of heating, refrigerating and airconditioning engineers; 2007.
Frank D. Roombased occupancy sensors M&V grouse mountain lodge
whitesh, MT. Portland, OR: Bonneville Power Administration; 2011.
Koza JR. Genetic programming: on the programming of computers by means
of natural selection. Cambridge: Massachusetts Institute of Technology Press;
1992.
Gosselin L, Tye-Gingras M, Mathieu-Potvin F. Review of utilization of genetic
algorithms in heat transfer problems. Int J Heat Mass Transf 2009;52:2169
88.
Bagley JD. The behavior of adaptive systems which employ genetic and correlation algorithms [Ph.D. thesis]. Ann Arbor: University of Michigan; 1967.
Lu L, Cai WJ, Chai YS, Xie L. Global optimization for overall HVAC systems
Part I problem formulation and analysis. J Energy Convers Manag 2005;46(7
8):9991014.
Lu L, Cai WJ, Soh YC, Xie L. Global optimization for overall HVAC systemsPart
II problem solution and simulations. Energy Convers Manag 2005;46(7
8):101528.
Lu L, Cai W, Xie L, Li S, Soh YC. HVAC system optimization-in building section.
J Energy Build 2005;37(1):1122.
Nassif N, Kajl S, Sabourin R. Optimization of HVAC, control system strategy
using two-objective genetic algorithm. HVACR 2005;11(3):45986.
Jin X, Ren H, Xiao X. Prediction-based online optimal control of outdoor air of
multi-zone VAV air conditioning systems. Energy Build 2005;37(9):93944.
Huang W, Lam HN. Using genetic algorithms to optimize controller parameters for HVAC systems. J Energy Build 1997;26(3):27782.
Wang J, Wang Y. Performance improvement of VAV air conditioning system
through feed forward compensation decoupling and genetic algorithm.
J Appl Therm Eng 2008;28(56):56674.
Xue Y, Zhai ZQ, Chen QY. Inverse prediction and optimization of ow control
conditions for conned spaces using a CFD-based genetic algorithm. J Build
Environ 2013;64:7784.
Chippereld A, Fleming P, Pohlheim H, Fonseca C. Genetic algorithm toolbox
for use with Matlab. IEE Colloq Appl Control Tech Matlab 1995;14:10-14.
Kim T, Song D, Kato S, Murakami S. Two-step optimal design method using
genetic algorithms and CFD-coupled simulation for indoor thermal environments. J Appl Therm Eng 2007;27(1):311.
Magnier L, Haghighat F. Multi-objective optimization of building design
using TRNSYS simulations, genetic algorithm, and articial neural network.
J Build Environ 2010;45:73946.
Foucquier A, Robert S, Suard F, Stephan L, Jay A. State of the art in building
modeling and energy performances prediction: a review. J Renew Sustain
Energy Rev 2013;23:27288.
Zheng XQ. Self tuning fuzzy controller for air-conditioning systems [Masters'
thesis]. National University of Singapore; 2002.
Ralston PA, Ward TL. Fuzzy control of industrial process. In: applications of
fuzzy set methodologies in industrial engineering. North-Holland Amsterdam: Elsevier Science Publishers; 1985. p. 2945.
Sugeno M. Industrial applications of fuzzy controller. Amsterdam, Netherlands: Elsevier Science; 1985 . p. 2319.
Kaymaz E. Adaptive environmental control for optimal energy consumption
in hospitals, computer based medical systems. In: Proceedings of the eighth
IEEE symposium; 1995. p. 165 72.
So ATP, Chow TT, Chan WL, Tse WL. Fuzzy air handling system controller.
J Build Serv Eng Res Technol 1994;15(2):95105.
So ATP, Chow TT, Chan WL, Tse WL. Self-learning fuzzy air handling system
controller. J Build Servs Eng Res Technol 1997;18(2):99108.
Karunakaran R, Parameshwaran R, Senthilkumar A, Iniyan S. Efcient variable
air volume air conditioning system based on fuzzy logic control for buildings.
Thammasat Int J Sci Techol 2009;14(1).
Karunakaran R, Iniyan S, Goic R. Energy efcient fuzzy based combined
variable refrigerant volume and variable air volume air conditioning system
for buildings. J Appl Energy 2010;87:115875.
Alcala R, Casillas J, Castro JL, Gonzalez A, Herrera F. A multi-criteria genetic
tuning for fuzzy logic controllers. Mathw Soft Comput 2001;8(2):179201.
Huang S, Nelson RM. Rule development and adjustment strategies of a fuzzy
logic controller for an HVAC systemParts I and II (analysis and experiment).
ASHRAE Trans 1994;100(1):8516.
Parameshwaran R, Karunakaran R, Vinu Raja Kumar C, Iniyan C. Energy
conservative building air conditioning system controlled and optimized
using fuzzy-genetic algorithm. J Energy Build 2010;42:74562.
Homod RZ, Sahari KSM, Almurib HAF, Nagi FH. Gradient auto-tuned Takagi
Sugeno fuzzy forward control of a HVAC system using predicted mean vote
index. J Energy Build 2012;49:25467.
Panke RA. Energy management systems and direct digital control. Lilburn,
GA: The Fairmont Press, Inc.; 2002.
Goyal S, Ingley H, Barooah P. Occupancy-based zone climate control for
energy efcient buildings: complexity vs. performance. J Appl Energy
2012;106:20921.
The ASHRAE handbook - HVAC applications (SI Edition); 2011.
Taylor ST, Stein J, Paliaga G, Cheng HK. Dual maximum VAV box control logic.
ASHRAE J 2012;54(12):1624.
Goyal S, Barooah P. Energy-efcient control of an air handling unit for a
single-zone VAV system. In: Proceedings of the 52nd IEEE conference on
decision and control; 2013.
816
[222] Yu Y, Xu K, Cho YH, Liu M. A Smart Logic for Conference Room Terminal Box
of Single Duct VAV System. In: Proceedings of the 7th international conference for enhanced building operations. San Francisco, CA; 2007.
[223] Taylor ST. Increasing efciency with VAV system static pressure set-point
reset. ASHRAE 2007:2432.
[224] Cho YH. Development of a terminal control system with variable minimum
airow rate. Energies 2012;5:464364.
[225] Rajkumar CV, Srihari DP. Optimization of AHU control strategy. Int J Innov
Technol Res 2013;1(2):1249.
[226] Yu Y, Liu M, Li H, Yu D, Loftness V. Synergization of air handling units for high
energy efciency in ofce buildings: the theory and analysis. Energy Build
2012;45:26473.
[227] Yu Y, Liu M, Li H, Yu D, Loftness V. Synergization of air handling units for high
energy efciency in ofce buildings: implement methodology and performance evaluation. Energy Build 2012;54:42635.
[229] American Society of Heating, Refrigerating and air conditioning engineers,
The ASHRAE handbook fundamentals (SI Edition); 2005.
[230] Eynard J, Bourdais R, Gueguen H, Dumur D. Tractable predictable control
strategies for heating systems in buildings. In: Proceedings of 13th conference of international building performance simulation association,
BS2013. Chambry, France; August 2628, 2013.
[231] Freire RZ, Oliveira GH, Mendes N. Predictive controllers for thermal comfort
optimization and energy savings. J Energy Build 2008;40(7):135365.
[232] Chen TY. Application of adaptive predictive control to a oor heating system
with a large thermal lag. J Energy Build 2002;34(1):4551.
[233] Yuan S, Perez R. Multiple-zone ventilation and temperature control of a
single-duct VAV system using model predictive strategy. J Energy Build
2006;38(10):124861.
[234] Huang G, Wang S, Xu X. A robust model predictive control strategy for
improving the control performance of air-conditioning systems. J Energy
Convers Manag 2010;50(10):26508.
[235] Hadjiski M, Deliiski N, Boshnakov K. Thermal comfort-based MPC of air
handling unit. In: Proceedings of the IFAC workshop on energy saving control
in plants and buildings, international federation of automatic control. 1;
2006. p. 8186.
[236] Paris B, Eynard J, Grieu S, Talbert T, Polit M. Heating control schemes for
energy management in buildings. J Energy Build 2010;42(10):190817.
[237] Shaikh PH, Nor NBM, Nallagownden P, Elamvazuthi I, Ibrahim T. A review on
optimized control systems for building energy and comfort management of
smart sustainable buildings. J Renew Sustain Energy Rev 2014;34:40929.
[238] McWilliams J. Review of airow measurement techniques. LBNL-49747.
California: Lawrence Berkeley National Laboratory; 2002.
[239] Taylor ST, Stein J. Sizing VAV boxes. ASHRAE J 2004.
[240] Klopfenstein R. Air velocity and ow measurement using a pitot tube. ISA
Trans 1998;37(4).
[241] Miller R. Flow measurement engineering handbook. New York: McGraw Hill;
1996.
[242] Mueller H, Vogel Pbibinf UK. Air Inltration and Ventilation Centre. The Role
of ventilation. In: Proceedings of 15th AIVC conference. Buxton, UK, 1; 2730
September 1994. p. 1018.
[243] Schwenk D. Air ow measurement accuracy, US Army corps of engineers,
construction engineering research laboratories. Conference Proceedings.
Washington D.C: HQUSACE; June 1998.
[245] Atkinson GV. VAV system volume control using electronic strategies. ASHRAE
Trans 1986;Part 2B:4657.
[246] Janu GJ, Wenger JD, Nesler CG. Outside air ow control for VAV systems.
ASHRAE J 1995;37(4):628.
[247] Cohen T. Providing constant ventilation in variable air volume systems.
ASHRAE J 1994;34(7):3850.
[248] Cheong KW. Airow measurements for balancing of air distribution system
tracer-gas technique as an alternative? J Build Environ 2001;36:95564.
[249] Tan H, Dexter A. Estimating airow rates in air-handling units from actuator
control signals. J Build Environ 2006;41:12918.
[250] Cho YH, Liu MS. Minimum airow reset of single duct VAV terminal boxes.
J Build Environ 2009;44:187685.
[251] Kang SH, Kim HJ, Cho YH. A study on the control method of single duct VAV
terminal unit through the determination of proper minimum air ow.
J Energy Build 2014;69:46472.
[252] Liu R, Wen J, Waring MS. Improving airow measurement accuracy in VAV
terminal units using ow conditioners. J Build Environ 2014;71:8194.
[253] Yu BF, Hu ZB, Liu M, Yang HL, Kong XQ, Liu YH. Review of research on airconditioning systems and indoor air quality control for human health. Int J
Refrig 2009;32:320.
[254] Graudenz GS, Oliveira CG, Tribess A, Mendes C, Latorre MRDO, Kalil J.
Association of air-conditioning with respiratory symptoms in ofce workers
in tropical climate. Indoor Air 2005;15:626.
[255] Novoselac A, Srebric J. A critical review on the performance and design of
combined cooled ceiling and displacement ventilation systems. J Energy
Build 2002;34:497509.
[256] Seppanen O, Fisk WJ. Association of ventilation system type with SBS
symptoms in ofce workers. Indoor Air 2002;12:98112.
[257] Niu JL. Some signicant environmental issues in high-rise residential building design in urban areas. J Energy Build 2004;36:125963.
[258] Ardehali MM, Smith TF. Evaluation of variable volume and temperature
HVAC system for commercial and residential buildings. J Energy Convers
Manag 1996;37(9):146979.
[259] Filardo MJ. Outdoor air: how much is enough? ASHRAE J 1993;35(1):368.
[260] Maki KS, Li Z, Chamberlin GA, Christianson LL. VAV systems performanceeld characterization of ow, system diagnosis tools, and operational design
implications. ASHRAE Trans 1997;103(2):83042.
[262] Daryanani S, McKay W, Shataloff N, Straub H. VAV Variable air volume air
conditioning. Air Cond Heat And Vent 1966;63:5678.
[263] Fitzner K. Displacement ventilation and cooled ceilings, results of laboratory
tests and practical installations. Proc Indoor Air 1996;1:4150.
[264] Kruhne H. Effect of cooled ceilings in rooms with displacement ventilation
on the air quality. Proc Indoor Air 1993;5:395400.
[265] Kulpmann RW. Thermal comfort and air quality in rooms with cooled
ceilings-results of scientic investigations. ASHRAE Trans 1993;99(2):488
502.
[266] Behne M. Indoor air quality in rooms with cooled ceilings. Mixing ventilation
or rather displacement ventilation? J Energy Build 1999;30:15566.
[267] Guo H, Murray F, Lee SC. The development of low volatile organic compound
emission house - a case study. J Build Environ 2003;38:141322.
[268] Stanke DA. Ventilation where it's needed. ASHRAE J 1998:3947.
[269] Qin J, Wang S. A fault detection and diagnosis strategy of VAV airconditioning systems for improved energy and control performances.
J Energy Build 2005;37:103548.
[270] Hyvarinen J, Karki S. Building optimization and fault diagnosis source book,
IEA Annex; 1996. p. 25.
[271] Usoro PB, Schick IC, Negahdaripour S. An Innovation-based methodology for
HVAC system fault detection. J Dyn Syst Meas Control 1985;107:2845.
[272] Anderson D, Graves L, Reinert W, Kreider JF, Dow J, Wubbena H. A quasi-realtime expert system for commercial building HVAC diagnostics. ASHRAE
Trans 1989;95(2):95460.
[273] J.M.D. Pimenta O.S.H. Mendoza Overview of fault detection and diagnosis
methods applied to HVAC systems. In: Proceedings of 15th Brazilian congress
of mechanical engineering. Aguas de Lindola, Sao Paulo; November 2223,
1999.
[274] Linder R, Dorgan CB. VAV systems work despite some design and application
problems. ASHRAE Trans 1997;103(2):80713.
[275] Yoshida H. VAV air handling unit, IEA-Annex 25 source book. Espoo, Finland:
Technical Research Centre of Finland; 1996.
[276] Haves P. Fault modeling in component-based HVAC simulation. In: Proceedings of Building Simulation; 1997.
[277] Wang S, Zhou Q, Xiao F. A system-level fault detection and diagnosis strategy
for HVAC systems involving sensor faults. J Energy Build 2010;42(4):47790.
[278] IEA. Building optimization and fault diagnosis source book, Annex 25.
International energy agency energy conservation in buildings and communities programme; 1996.
[279] Shun L. A model-based fault detection and diagnostic methodology for secondary HVAC systems [Doctoral thesis]. Drexel University; 2009.
[280] Siegel JA. Particulate fouling of HVAC heat exchangers [Doctoral thesis].
Berkeley: University of California; 2002.
[281] Braun JE, Comstock MC. Development of analysis tools for the evaluation of
fault detection and diagnostics for chillers. Deliverable for ASHRAE research
project 1043-RP; 1999.
[282] Basarkar M, Pang X, Wang L, Haves P, Hong T. Modeling and simulation of
HVAC results in energyplus. LBNL-5564; November, 2011.
[283] Katipamula S, Brambley MR. Methods for fault detection, diagnostics and
prognostics for building systems- a review, Part I. IJHVACR Res 2005;11(1):3
25.
[284] Isermann R. Process fault detection based on modeling and estimation
methodsa survey. Automatica 1984;20(4):387404.
[285] Dexter AL, Benouatets M. A generic approach to identifying faults in HVAC
plants. ASHRAE Trans 1996;102(1):5506.
[286] Ngo D, Dexter AL. A robust model-based approach to diagnosing faults in airhandling units. ASHRAE Trans 1999;105(1):107886.
[287] Dexter AL, Ngo D. Fault diagnosis in air-conditioning systems: a multi-step
fuzzy model-based approach. HVACR Res 2001;7(1):83102.
[288] Liu XF, Dexter A. Fault-tolerant supervisory control of VAV air conditioning
systems. J Energy Build 2001;33:37989.
[289] Yoshida H, Iwami T, Yuzawa H, Suzuki M. Typical faults of air conditioning
systems and fault detection by ARX model and extended Kalman Filter.
ASHRAE Trans 1996;102(1):55764.
[290] Yoshida H, ARX Kumar S. and AFMM model-based on-line real time data base
diagnosis of sudden fault in AHU of VAV system. Energy Convers Manag
1999;40:1191206.
[291] Yoshida H, Kumar S, Morita Y. Online fault detection and diagnosis in VAV air
handling unit by RARX modeling. J Energy Build 2001;33(4):391401.
[292] Lee WY, House JM, Shin DR. Fault diagnosis and temperature sensor recovery
for an air handling unit. ASHRAE Trans 1997;103(1):62133.
[293] Brambley MR, Pratt RG, Chassin DP, Katipamula S, Hatley D. Diagnosis for
outdoor air ventilation and economizers. ASHRAE J 1998:4955.
[294] Katipamula S, Pratt RG, Chassin DP, Taylor ZT, Gowri K, Brambley MR.
Automated fault detection and diagnosis for outdoor-air ventilation system
and economizers: methodology and results from eld testing. ASHRAE Trans
1999;105(1):55567.
[295] Dodier RH, Curtiss PS, Kreider JF. Small-scale on-line diagnosis for an HVAC
system. ASHRAE Trans 1998;104(1):5309.
[296] Seem JE, House JM, Monroe RH. On-line monitoring and fault detection.
ASHRAE J 1999:216.
[297] House JM, Lee WY, Shin DR. Classication techniques for fault detection and
diagnosis of an air-handling unit. ASHRAE Trans 1999;105(1):1087100.
[298] House JM, Vaezi-Nejad H, Whitcomb JM. An expert rule set for fault detection in air-handling units. ASHRAE Trans 2001;107(1):85871.
[299] McGhee J, Henderson IA, Baird A. Neural networks applied for the identication and fault diagnosis of process valves and actuators. Measurement
1997;20(4):26775.
[300] Han CY, Xiao YF, Ruther CJ. Fault detection and diagnosis of HVAC systems.
ASHRAE Trans 1999;105(1):56878.
[301] Wang SW, Chen YM. Fault-tolerant control for outdoor ventilation air ow
rate in buildings based on neural network. J Build Environ 2001;37:691704.
[302] Shaw SR, Norford LK, Luo D, Leeb SB. Detection and diagnosis of HVAC faults
via electrical load monitoring. HVACR Res 2002;8(1):1340.
[303] Norford LK, Wright JA, Buswell RA, Luo D, Klaassen CJ, Suby A. Demonstration
of fault detection and diagnosis methods for air handling units (ASHRAE
1020-RP). HVACR Res 2002;8(1):4171.
[304] House JM, Lee KD, Norford LK. Controls and diagnostics for air distribution
systems. J Sol Energy Eng Trans ASME 2003;125(3):3107.
[305] Schein J, House JM. Application of control chart for detecting faults in
variable-air-volume box. ASHRAE Trans 2003;109(2):67182.
[306] Yi X, Chen Y, Wu L. Sensor fault detection and diagnosis for VAV systems
based on principal component analysis. Proceedings: building simulation;
2007. p. 131318.
[307] Du Z, Jin X. PCA-FDA-based fault diagnosis in VAV systems. HVAC Res
2007;13(2):34967.
[308] Xu X, Fu X, Wang S. Enhanced chiller sensor fault detection, diagnosis and
estimation using wavelet analysis and principal component analysis methods. J Appl Therm Eng 2008;28(23):22637.
[309] Brambley MR, Fernandez N, Wang W, Cho H, Ngo H, Goddard J. Self-correcting controls for VAV system faults lter/fan/coil and VAV box sections.
Richland, WA: Pacic Northwest National Laboratory; 2011.
[310] Zhu Y, Jin X, Du Z. Fault diagnosis for sensors in air handling unit based on
neural network pre-processed by wavelet and fractal. J Energy Build
2012;44:716.
[311] Wang H, Chena Y, Chan CWH, Qin J, Wang J. Online model-based fault
detection and diagnosis strategy for VAV air handling units. J Energy Build
2012;55:25263.
[312] Yang X, Jin X, Du Z, Zhu Y, Guo Y. A hybrid model-based fault detection
strategy for air handling unit sensors. J Energy Build 2013;57:13243.
[313] Sterling R, Struss P, Febres J, Sabir U, Keane MM. From modelica models to
fault diagnosis in air handling units. In: Proceedings of the 10th international
modelica conference. Lund, Sweden; March 1012, 2014. p. 447 54.
[314] Xiao F, Zhao Y, Wen J, Wang S. Bayesian network based FDD strategy for
variable air volume terminals. Autom Constr 2014;41:10618.
[315] Madjidi M. Impact of operational faults in a VAV system on energy consumption, thermal comfort and monitored process variables. Pre-prints of
the proceedings of the 4th international conference - system simulation in
buildings. Lige Belgium: University of Lige; 1994.
[316] Annex 34. Computer-aided evaluation of HVAC system performance, nal
report; 2001.
[317] Roth KW, Westphalen D, Feng MY, Llana P, Quartararo L. Quartararo energy
impact of commercial building controls and performance diagnostics: market
characterization, energy impact of building faults and energy savings
potential. Reference no. 02140-2390, TIAX, Lexington, Massachusetts. Prepared for the U.S. Department of Energy Washington, D.C; 2005.
817
[318] Lee SH, Yik FWH. A study on the energy penalty of various air-side system
faults in buildings. J Energy Build 2010;42(1):210.
[319] Sekhar SC. A critical evaluation of variable air volume system in hot and
humid climates. J Energy Build 1997;26(2):22332.
[320] Sekhar SC, Yat CJ. Energy simulation approach to air-conditioning system
evaluation. J Build Environ 1998;33(6):397408.
[321] Yang K, Ting M. An innovative analysis and experimental investigation on
energy savings of a VAV system in hot and humid climates. J Build Environ
2000;35(1):2731.
[322] Pan Y, Zhou H, Huang Z, Zeng Y, Long W. Measurement and simulation of
indoor air quality and energy consumption in two Shanghai ofce buildings
with variable air volume systems. J Energy Build 2003;35(9):87791.
[323] Westphalen D. New approach to energy savings for rooftop AC. ASHRAE J
2004;46(3):3846.
[324] Gao J, Sun Y, Wen J, Smith TF. An experimental study of energy consumption
and thermal comfort for electric and hydronic reheats. J Energy Build
2005;37:20314.
[325] Ke MT, Weng KL, Chiang CM. Performance evaluation of an innovative fancoil unit: low-temperature differential variable air volume FCU. J Energy
Build 2007;39(6):7028.
[326] Yao Y, Wang L. Energy analysis on VAV system with different air-side economizers in China. J Energy Build 2010;42:122030.
[327] Parameshwaran R, Harikrishnan S, Kalaiselvam S. Energy efcient PCM-based
variable air volume air conditioning system for modern buildings. J Energy
Build 2010;42(8):135360.
[328] Kusiak A, Zeng Y, Xu G. Minimizing energy consumption of an air handling
unit with a computational intelligence approach. J Energy Build
2013;60:35563.
[329] Wendes H. Variable air volume manual. Lilburn, GA: The Fairmont Press, Inc.,
Jan 1, 1994.
[330] Bhatia A. HVAC design aspects: choosing a right system; 2012.
[331] Trane, Understanding single-zone VAV systems, providing insight for today's
HVAC systems design. Engineers Newsletter, 42(2); 2013.
[332] Chen W, Li Z, Deng S. Capacity control of a DX VAV system and its modeling.
In: Proceedings of the international refrigeration and air conditioning conference; 2004. p. 648 http://docs.lib.purdue.edu/iracc/648.
[333] Nassif N, Kajl S, Sabourin R. Simplied model-based optimal control of VAV
air-conditioning system. In: Proceedings of the ninth international IBPSA
conference, Montral, Canada; August 1518, 2005.
[334] Yee KF, Chan CWH, Yik FWH, Chan HYT. A study on the energy saving of
converting a VAV system from inlet guide vane to variable fan speed control.
Hong Kong Inst Eng HKIE Trans 2002;9(2):1824.
[335] Rohdin P, Moshfegh B. A study of variable air volume (VAV) systems in
foundries. In: Proceedings of the Clima 2007 WellBeing Indoors; 2007.
[336] Han Z, Zhu YX, Xia CH, Xu GH. Application of simulation to solve the hunting
problem in VAV system. In: Proceedings of building simulation; 2007. p.
130712.
[337] Murphy J. Using to V.A.V.limit humidity at part load. ASHRAE J 2010.
[338] Alvarez O, Sanjuan M, Amaya F, Bula A. Bula VAV system operating in an
educational building under tropical conditions modeling and simulation.
In: Proceedings of the ASME 2013 international mechanical engineering
congress and exposition IMECE2013, San Diego, California, USA; November
1521 2013.
[339] In-Hout D. Research validates VAV, HVAC applications. ASHRAE J 2013:96.

Renewable and Sustainable Energy Reviews Volume 59 Issue 2016 (Doi 10.1016/j.rser.2015.12.328) Okochi, Godwine Swere Yao, Ye - A Review of Recent Developments and Technological Advancements of Var

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Renewable and Sustainable Energy Reviews Volume 59 Issue 2016 (Doi 10.1016/j.rser.2015.12.328) Okochi, Godwine Swere Yao, Ye - A Review of Recent Developments and Technological Advancements of Var

Încărcat de

Drepturi de autor:

Formate disponibile

Renewable and Sustainable Energy Reviews 59 (2016) 784817

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

A review of recent developments and technological advancements

Corresponding author. Tel.: 86 21 13641943577; fax: 86 21 34206814.

Control strategies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 794

VAV systems work on the principle of opening or closing

Temperature of nth window (K)

Effectiveness of heat exchanger

Damper angular position (deg)

This study shall review literature on VAV systems modeling and

2. Modeling and simulation

in reasonable agreement with the value predicted by the Passive

Uniform distribution of zone temperature due to proper mixing

of air. This enables the dynamics of the conditioned space to be

The mass ow rates of both water and air are constant.

The thermal resistance of the coil is negligible.

The coil mass temperature is taken to be constant within the

Steady state heat transfer between the water and air.

In a study by Tahmasebi et al. the authors modeled the heating

_ a C pa T ain T win when m

is usually heat exchange between air passing through the ducts

The above models only considered cases when the supplying

where u1 is the amplitude of step change in process input

while y1 is the deviation between the original value and

2.1.2.2. Data mining algorithms (DMAs). DMAs are usually used to

moving average exogenous (ARMAX) and nite impulse response

In which; adp , bdp , dp are damper coefcients and ut is the input

In which C ref , is a function of several parameters. These are,

exchange between the fan motor and the air. According to

carry out simulation, engineers employ the use of simulation tools

3. Operations and control of VAV systems

Fig. 3.1. Proportional control.

3.1. VAV controllers

where u is the output signal, K p is the constant of proportionality

Fig. 3.2. A curve of proportional control used in cooling a zone.

Since the reset modes of PI controllers are automatic much of

characteristics compared to short sampling intervals if the system

Fig. 3.3. Proportional-Integral-Derivative control curve.

Tashtoush et al. [18] described a procedure to derive a dynamic

Adaptive controllers self-regulate and adapt to the climatic

airow at the set-point. Thus, the damper position is not under

Fig. 3.4. Variation of operating point in a VAV system [138].

We notice that in order to avoid extremely low fan speed, it is a

Fig. 3.6. Supply air temperature reset control process.

software, and strategy for control, and is an important part of

maximizing comfort. The system performance is mainly evaluated

been developed specically for a problem. This problem can

where x, y and z are linguistic variables,Ai , Bi , and C i are fuzzy sets.

z c0i c1i x c2i y

where, the consequent rule is a linear function of input variables.

where zi is a crisply dened number.

Karunakaran et al. [210] developed a simple VAV building

Fig. 3.7. Block fuzzy logic controller.

Fig. 3.8. Dual maximum logic operation modes.

control strategy of HVAC system whose main objective is to

Under this control logic, we can have four operation modes

_ aout=p is volume of OA needed per occupant,

_ samin is the minimum volume

where Np k is the occupancy measured at time index k.

Taylor [223] during his study on increasing efciency with VAV

Fig. 3.9. Schematic diagram of a model-based predictive controller [216].