ArticlePDF Available

Advanced control strategies for heating, ventilation, air-conditioning, and refrigeration systems—An overview: Part I: Hard control

Authors:

Abstract and Figures

A chronological overview of the advanced control strategies for heating, ventilation, air-conditioning, and refrigeration (HVAC&R) is presented in this article. The overview focuses on hard-computing or control techniques, such as proportional-integral-derivative, optimal, nonlinear, adaptive, and robust; soft-computing or control techniques, such as neural networks, fuzzy logic, genetic algorithms; and on the fusion or hybrid of hard- and soft-control techniques. Thus, it is to be noted that the terminology “hard” and “soft” computing/control has nothing to do with the “hardware” and “software” that is being generally used. Part I of a two-part series focuses on hard-control strategies, and Part II focuses on soft- and fusion-control in addition to some future directions in HVAC&R research. This overview is not intended to be an exhaustive survey on this topic, and any omission of other works is purely unintentional.
Content may be subject to copyright.
PLEASE SCROLL DOWN FOR ARTICLE
This article was downloaded by:
[Naidu, D Subbaram]
On:
21 February 2011
Access details:
Access Details: [subscription number 933676802]
Publisher
Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK
HVAC&R Research
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t929587710
Advanced control strategies for heating, ventilation, air-conditioning, and
refrigeration systems—An overview: Part I: Hard control
D. Subbaram Naidu
a
; Craig G. Rieger
b
a
Department or Electrical Engineering and Computer Science, Idaho State University, Pocatello, ID,
USA
b
Idaho National Laboratory, Idaho Falls, ID, USA
Online publication date: 18 February 2011
To cite this Article Naidu, D. Subbaram and Rieger, Craig G.(2011) 'Advanced control strategies for heating, ventilation,
air-conditioning, and refrigeration systems—An overview: Part I: Hard control', HVAC&R Research, 17: 1, 2 — 21
To link to this Article: DOI: 10.1080/10789669.2011.540942
URL: http://dx.doi.org/10.1080/10789669.2011.540942
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Advanced control strategies for heating, ventilation,
air-conditioning, and refrigeration systems—An overview:
Part I: Hard control
D. Subbaram Naidu
1,
and Craig G. Rieger
2
1
Department or Electrical Engineering and Computer Science, Idaho State University, Pocatello, ID, USA
2
Idaho National Laboratory, Idaho Falls, ID, USA
Corresponding author e-mail: naiduds@isu.edu
A chronological overview of the advanced control strategies for heating, ventilation, air-conditioning,
and refrigeration (HVAC&R) is presented in this article. The overview focuses on hard-computing or
control techniques, s uch as proportional-integral-derivative, optimal, nonlinear, adaptive, and robust;
soft-computing or control techniques, such as neural networks, fuzzy logic, genetic algorithms; and on
the fusion or hybrid of hard- and soft-control techniques. Thus, it is to be noted that the terminology
“hard” and “soft” computing/control has nothing to do with the “hardware” and “software” that is being
generally used. Part I of a two-part series focuses on hard-control strategies, and Part II focuses on soft-
and fusion-control in addition to some future directions in HVAC&R research. This overview is not intended
to be an exhaustive survey on this topic, and any omission of other works is purely unintentional.
Introduction
In large commercial and residential buildings,
energy management control systems (EMCS) play a
major role to maintain good control of temperature,
human comfort, and overall operational and energy
efficiency. In a typical situation, heating, ventilation,
air-conditioning, and refrigeration (HVAC&R) sys-
tems provide a central air supply at a controlled
temperature and a flow rate for heating or cooling
a particular unit or zone or entire building com-
plex. The two main requirements of any HVAC&R
system is first to provide satisfactory indoor com-
fort (temperature and relative humidity) conditions
to the building housing both humans and equip-
ment and, at the same, time minimize the overall en-
ergy consumption. Another important requirement
Received April 5, 2010; accepted November 1, 2010
D. Subbaram Naidu, PhD, is Director, School of Engineering. Craig G. Rieger, PhD, is ICIS Distinctive Signature Lead.
is to prevent the spread of any chemical or biolog-
ical species from any point where these species are
released to the rest of the building. The primary
professional organization responsible for all activ-
ities of HVAC&R systems is the American Soci-
ety of Heating, Refrigerating and Air-Conditioning
Engineers (ASHRAE), which is located in Atlanta,
Georgia, USA, and has been engaged in publish-
ing and updating industry-wide standard handbooks
such as ASHRAE (2005, 2006, 2007, 2008). It has
been reported that the energy consumption due to
HVAC&R systems in commercial and industrial
buildings constitutes nearly half of the total world
energy consumption according to Payne (1984) and
Imbabi (1990b).
The minimization of energy consumption and
maximization of indoor user comfort can be
2
HVAC&R Research, 17(1):2–21, 2011. Copyright
C
2011 American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.
ISSN: 1078-9669 print / 1938-5587 online
DOI: 10.1080/10789669.2011.540942
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 3
combined into a single minimization problem by
using the predicted mean value (PMV) and the
predicted percentage dissatisfied (PPD) value and
defining discomfort instead of comfort value as the
square of the difference between set-point temper-
ature and internal room temperature, normalized
with respect to set-point temperature. Minimizing
the discomfort value gives the optimal (maximum)
comfort value given by Fanger (1972, 1984), ISO-
1984 (1984), ISO-1995 (1995), ISO-2005 (2005),
and Pargfrieder and Jorgl (2002). One of the earli-
est applications of automatic control is thermostatic
control of a building, which is basically an on/off
control. Some of the books and handbooks in the
area of HVAC&R are Haines (1971, 1987), Hart-
man (1993), Rogers and Douglas (1993), Leven-
hagen and Spethmann (1993), Newman and Mor-
ris (1994), Levenhagen (1999), Underwood (1999),
Hordeski (2001), Haines and Hittle (2003), Brum-
baugh (2004a, 2004b, 2004c, 2005), Honeywell
(1997), and McDowall (2009). A previous survey by
Harrold and Lush (1988) examined HVAC&R con-
trols as a part of larger building services and covered
a variety of topics, including those relating to auto-
matic controls. A recent review article by Wang and
Ma (2008) published in HVAC&R Research took
a different approach, focusing exclusively on su-
pervisory and optimal control techniques arising in
HVAC&R in building automation systems (BAS)
and building energy management systems (BEMS).
In particular, the review addressed various control
methods, such as local or decentralized control (DC)
with sequencing control, such as ON/OFF con-
trol; process control, such as proportional-integral-
derivative (PID) control; and supervisory or cen-
tralized control (CC) with a model-based method
(physical model, gray-box model, black-box model,
hybrid model, etc.).
Further, various optimization techniques were
discussed in this review by Wang and Ma (2008).
From the perspective of the topics on energy,
comfort, and control, a review was conducted in
Dounis and Caraiscos (2009) of the work ini-
tially on conventional (optimal, predictive, and
adaptive) control schemes and then the state-
of-the art intelligent (neural, fuzzy, neuro-fuzzy,
proportional-integral (PI)-fuzzy, adaptive fuzzy
proportional-derivative (PD) and PID) control sys-
tems for improving the efficiency and indoor en-
vironment in buildings, with a particular empha-
sis on multi-agent control systems (MACS) with
simulations using TRNSYS/MATLAB software.
Also, see the previous literature review works
by Dexter (1988), Kelly (1988), and Sane et al.
(2006).
Overview and terminology: hard control
(HC) and soft control (SC)
There are various ways of conducting an
overview, such as a chronological—or topical—
overview. The main purpose of this overview is
to provide the reader with a summary of the re-
cent results on the topic of control techniques for
HVAC&R systems so that this forms a staging point
for further research in the field. Thus, a topical
survey based on various topics or control method-
ologies and within each topic is presented, and a
chronological order of the published results follows.
However, note that there are a number of situations
involving multiple topics; the focus is on
1. HC, such as basic controls involving PID con-
trol, optimal control (Anderson and Moore 1990;
Naidu 2003; Lewis et al. 2008), nonlinear con-
trol (Kristic et al. 1995), robust or Hcontrol
(Zhou and Doyle 1998), and adaptive control
(Tao 2003);
2. SC, involving neural networks (NNs), fuzzy
logic (FL), genetic algorithms (GAs), and
other evolutionary methods (Jang et al. 1997;
Tsoukalas and Uhrig 1997; Nguyen et al. 2003;
Karray and De Silva 2004; Konar 2005; Kasabov
2007; Sumathi et al. 2008); and
3. hybrid control resulting from the fusion of SC
and HC to achieve a better performance (Ovaska
et al. 2002; Tettamanzi and Tomassini 2001;
Konar 2005; Kasabov 2007; Sumathi et al. 2008).
It is to be noted that the new terminology, the
“hard” in HC and “soft” in SC, has been used re-
cently in the control systems community (Ovaska et
al. 2002; Karray and De Silva 2004) and has noth-
ing to do with the “hardware” and “software” that
is generally used.
Modeling, testing, and validation
Modeling
A generic piping/process and instrumentation di-
agram (P&ID) of a typical HVAC&R system avail-
able in Department of Energy (DOE) facilities is
shown in Figure 1 (courtesy of Idaho National Lab).
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
4 VOLUME 17, NUMBER 1, FEBRUARY 2011
Figure 1. HVAC generic P&ID diagram.
It shows three zones, two access areas, and the nec-
essary instrumentation and control architecture.
In a typical HVAC&R system for large build-
ings, there are three subsystems: boiler and chiller,
constituting a primary subsystem; heat pumps and
airflow ductwork, called the distribution subsystem;
and the subsystem consisting of the environmen-
tal zones; the whole configuration is also called a
multi-zone space heating (MZSH) system (Zaheer-
Uddin et al. 1993; Saboksayr et al. 1995). Here,
using the principles of energy conservation and bal-
ance, a seventh-order, bilinear, state-space model
for a two-zone space heating system was developed
with
r
seven state variables: boiler temperature, temper-
ature of the evaporator for heat pump 1, tem-
perature of the evaporator for heat pump 2, tem-
perature of the condenser coil f or heat pump 1,
temperature of the condenser coil for heat pump
2, zone 1 temperature, and zone 2 temperature;
r
three output variables: boiler temperature, zone 1
temperature, and zone 2 temperature,
r
five control (input) variables: air flow rate for zone
1 controller 1, air flow rate for zone 2 controller 2,
boiler fuel firing rate controller 3, input energy for
heat pump 1 via controller 4, and input energy for
heat pump 2 via controller 5, essentially grouped
into three controllers.
In another detailed study by Zaheer-Uddin and
Zheng (1994), as many as 328 nonlinear, time-
varying equations were developed for the dynamic
models for a two-zone variable airflow volume
(VAV) system in terms of subsystem models for en-
vironmental zones, cooling and dehumidifier coil,
variable air flow rates in the duct, fan motor, and
chiller and storage tank, described by nine control
input variables—six dampers (for zone 1, zone 2,
fan, outdoor air, exhaust air, and recirculating air),
fan and chiller energy inputs, and mass flow rate of
chilled water to control the temperatures and humid-
ity ratios in the two zones, discharge air conditions,
chilled water temperature, outdoor and supply air-
flow rates, static pressure in the duct system, and
fan speed. The transient analysis of the open-loop
system performed on the linearized system showed
the dynamics of the overall VAV system being com-
posed a slow phenomenon due to the chiller-coil-
zone thermal subsystem and a fast phenomenon
due to the fan-airflow subsystem. This interaction of
slow and fast phenomena gives rise to an interesting
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 5
avenue for analysis and design of the system using
the singular perturbation and time scales (SPaTS)
methodology; surprisingly, no significant work has
been done (except in Dexter 1988; Zaheer-Uddin
and Patel 1995; Zaheer-Uddin and Zheng 2000) in
this direction to apply the SPaTS methodology to
HVAC&R systems, in spite of the attractive features
of SPaTS in terms of order reduction and decoupling
of slow and fast dynamics, with the rich literature
in this field (see Naidu and Rao 1985; Kokotovi
´
c
et al. 1986; Naidu 1988, 2002).
Distributed-parameter model
In modeling most HVAC&R systems, it was as-
sumed that the system variables, such as temper-
ature and air flow velocity, are not dependent on
spatial coordinates within the thermal zone, leading
to the lumped-parameter models. However, in real-
ity, these variables, and hence the thermal comfort
level, are not homogeneous within the zone/room. A
more realistic model with spatial dependency within
the room leading to the distributed-parameter model
was developed in Liu and He (1994) based on non-
isothermal airflow pattern showing downward de-
flection of a cool jet due to the balance between
buoyancy and gravity forces. However, the optimal
control settings for the system were obtained for
steady-state operations only with the objective of
maximizing the comfort level, not only at one point
but at several working points within the room while
minimizing the total power consumption, resulting
in a more accurate prediction of thermal comfort
level of a room.
Testing and validation
Issues relating to the optimal location of hard-
ware components influencing the hydronic net-
work design and thermal control task were stud-
ied by Franco et al. (2005) using a mathematical
model for the thermal-hydraulic network consist-
ing of the hardware elements of a control valve and
two booster pumps, leading to the results that the
network and the control system characteristics are
strongly influenced by the location of the hardware
components. The testing of various monitoring and
control strategies on the existing buildings seems
time consuming and expensive. Alternatively, stud-
ies by Imbabi (1990a, 1990b) using integral, reset,
compensated, and optimized controllers concluded
that the results of small-scale experimental mod-
els could be extended to full-scale real HVAC&R
systems.
A simulator within MATLAB
©
SIMULINK
©
environment, including a building zone and
HVAC&R plant emulating the environment of the
real controller, was developed in Lahrech et al.
(2002), linking the real and simulated conditions via
mainly sensor side and actuator side interfaces. The
simulated environment contained a building block,
an HVAC&R system block, a “bench in” block, and
a “bench out” block; the testing method validation
was performed using two types of tests: open-loop
tests (without a real controller) and closed-loop
tests, showing good agreement between the mea-
sured and emulated testing methods for heating, fan
coil, and chilled ceiling applications.
Dedicated software for HVAC&R systems
Several dedicated software packages have been
developed over the years for simulating the build-
ings and energy usage; see, for example, BLAST
(building loads analysis and system thermodynam-
ics; MATLAB and SIMULINK are registered trade-
marks of The Mathworks, Inc., Natick, MA, USA),
originally developed by University of Illinois at Ur-
bana Champaign (UIUC), Urbana, IL (Blast-UIUC
1983). BLAST features are now being incorporated
into the DOE-2 of EnergyPlus. DOE-2 is a com-
puter program that predicts energy usage and costs
of a building given the description of the building
and its HVAC&R equipment, besides other details.
The DOE-2 software was developed by James J.
Hirsch & Associates (JJH) in collaboration with
Lawrence Berkeley National Laboratory (LBNL),
with LBNL DOE-2 work performed mostly un-
der funding from the United States DOE (USDOE;
http://www.doe2.com). Also, see other programs
such as HVACSIM+ (HVAC Simulation; HVAC-
SIM 1986) and TRNSYS (Transient Energy System
Simulation Tool; TRNSYS 1996). It is to be noted
that BLAST and DOE-2 have more or less merged
to form EnergyPlus (E+), combining the best fea-
tures and capabilities of BLAST and DOE-2, and
it is now more widely used among researchers and
practitioners in the HVAC&R field, although DOE-2
is still available (Crawley et al. 2000).
CC and DC
There are two basic types of controls—local con-
trol for a single unit and super visory control for
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
6 VOLUME 17, NUMBER 1, FEBRUARY 2011
managing several units or zones in a large b uild-
ing complex (Payne 1984; Zaheer-Uddin and Zheng
2000; Wang and Ma 2008). The function of the lo-
cal controller is to ensure stability first and then
to ensure good set-point tracking, where the super-
visory control is i n charge of coordinating various
local controllers for the various subsystems and, at
the same time, maintain the overall operation of the
entire HVAC&R system.
The HVAC&R systems possess special char-
acteristics of distribution networks with more or
less identical basic modular structures for different
zones or cells in the building, satisfy the local com-
fort requirements in each zone or cell, and make an
idea candidate for implementation of DCs (Zaheer-
Uddin 1992). However, an HVAC&R system has a
CC scheme where a single zone temperature is con-
trolled by three controllers: (i) damper controller
C1 to regulate the rate of airflow to the zone with
temperature Tz, (ii) valve controller C2 regulating
the mass flow rate of chilled water (with temperature
Ts) flowing in the coil, or (iii) a controller to regulate
the input energy to the chiller (with temperature Tc)
with three controllers each receiving feedback sig-
nals from all three temperature sensors. On the other
hand, with three separate controllers each receiving
a feedback signal from its related (or nearest) ther-
mal sensor or control point, the HVAC&R system
has a DC system.
With the objective of investigating the effect of
replacing the rule-based control algorithm with one
based on a combination of a b uilding/system model
with the optimal supervisory control algorithm by
Zhang and Hanby (2008), a prototype building was
considered, consisting of a ventilated photovoltaic
array, solar air and water heating, a biomass-fired
boiler, and a stratified thermal store with the objec-
tive of minimizing the net energy consumption by
the b uilding that resulted in the substantial improve-
ment of overall system operation and performance.
The performance of a central variable water volume
(VWV) chiller system in HVAC was analyzed by
Jin et al. (2007) with three optimal control strate-
gies with respect to the supervisory control for the
optimal resetting of the supply head of a secondary
pump (called Delt-P), a supply chilled water tem-
perature (SWT), and a combination of the previous
two strategies.
PID control, gain scheduling, and
state feedback
PID control
The PID controller is the most widely used con-
troller in the industry. Basically, the PID actions
relate to present (proportional), past (integral), and
future (derivative). Although there are many im-
proved methods of PI/PID controller design, the tra-
ditional Ziegler–Nichols (Z-N) techniques (Ziegler
and Nichols 1942) are still being used by many
HVAC&R control engineers. However, the Z-N
method suffers from a long testing time and lim-
ited performance; hence, it is best used as a first
cut for tuning PID controllers. There are a num-
ber of excellent books on this subject (see
˚
Astr
¨
om
and H
¨
agglund 1995, 2006; O’Dwyer 2003; Visioli
2006).
Early investigations using basic single-
input–single-output (SISO) control methods, such
as PID controllers, faced the problem of tuning
the KP, KI, and KD parameters in addition to
the inability to take into account the interactions
between the various loops (Nesler and Stoecker
1984; Nesler 1986). A four-level HVAC&R
control scheme was investigated in Dexter (1988),
with the highest (fourth), or supervisory, level
focusing on maintaining the desired levels of the
internal temperature and relative humidity of the
overall building; the next (third) level taking care
of temperature and relative humidity of the air
supplied by the HVAC&R plant; the second level
assuming the responsibility of maintaining the
desired performance of the plant actuators and
local control l oops; and, finally, the lowest (first) or
local level taking care of maintaining the desired
controller setting based on the system model.
Problems associated with discrete-time simulation
of an HVAC&R system with the four-level structure
were investigated by Dexter (1988) with special
reference to the selection of parameters and
self-tuning of PID controllers using parameter
estimator and control algorithm. The effects of
disturbances on the overshoot and settling time
using PID controllers were studied in Geng and
Geary (1993), and it revealed that tuning rules
based on the Z-N method (Ziegler and Nichols
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 7
1942) are valid for small normalized time delay
and that this, combined with the normalized gain,
can be used for further tuning PID controllers.
Deriving a dynamic model of an HVAC system
consisting of a zone, heating coil, cooling and de-
humidifying coil, humidifier, ductwork, fan, mix-
ing box, and PID controller with the Z-N r ule was
obtained in Tashtoush et al. (2005) to account for
disturbances and to reduce energy consumption and
improve the quality of the indoor environment. Us-
ing recursive least squares (RLS) with exponen-
tial forgetting and model matching for a first-order,
dead-time model, an adaptive PI controller was de-
signed by Bai and Zhang (2007) with an auto-
matic adjustment of controller parameters for an
HVAC system, with simulations showing superior
performance compared to the H adaptive PI con-
troller. Using a recursive least square (RLS) algo-
rithm along with the z-domain fitting method by Bai
et al. (2008) for estimating the parameters (such as
time delay) of an air-conditioning (A/C) process, a
Smith predictor was adopted to reduce the effects of
the time delay, leading to a new self-tuning PI con-
troller using the integral of time and absolute error
(ITAE) with an experimental validation showing a
better system performance compared to an adaptive
PI controller.
Recent works by Bi et al. (2000) and Visioli
(2006) show advanced controller auto-tuning strate-
gies with successful application to both SISO and
multi-input–multi-output (MIMO) systems, where
decoupling control is used for MIMO systems.
In particular, Bi et al. (2000) developed an auto-
tuner with special features consisting of differ-
ent tuning tests—relay plus step test and step
test, different first- and second-order plant mod-
els with dead-time, and implementation with a
commercial, internet-enabled distributed computer-
controlled system featuring graphical user interface
(GUI). Three control schemes—hot-gas by-pass
scheme, cylinder unloading scheme, and suction-
gas throttling scheme—Yaqub and Zubair (2001)
were investigated for a hydro fluoro carbon (HFC)-
134 refrigeration system, and it was found that
the cylinder-unloading scheme is the most suitable
because of its higher coefficient of performance
(COP). Several tuning methods, in general, for PID
controllers (Kamimura et al. 2002) and, in particu-
lar, a tuning method (Ozawa et al. 2003) with con-
straints on control input for a single-zone environ-
mental space cooling system, showed a particularly
over damped response with no overshoot, thus pre-
venting the long oscillations when using integral
squared time error criterion in the performance in-
dex.
Using some of the recent results in PID con-
trollers involving Hurwitz polynomials to improve
the performance and robustness, some simple and
intuitive design tools are developed and applied to
two examples on the temperature control of a VAV
unit and evaporator super-heat control using an elec-
tronic expansion valve (Lim et al. 2009). A hybrid
of mechanical and electronic controls for evaporator
super-heat control was proposed by Elliott and Wal-
ton (2009) in terms of inner control f or regulating
the evaporator pressure and outer control to regulate
evaporator super-heat, resulting i n improved tran-
sient performance compared to mechanical valves,
more robust performance to large changes in oper-
ating conditions and motor failure, and less actuator
effort compared to electronic valves.
Considering the dynamics near an operating
point, a simple linear model was obtained by Flesch
and Normey-Rico (2010) for control of the output
temperature of a calorimeter for evaluating the per-
formance of refrigerant compressors, leading to the
design of a dead-beat compensator (DBC).
Gain scheduling
Recently in Rasmussen and Alleyne (2010), va-
por compression cycle of an A/C system was consid-
ered with an alternative local model network gain-
scheduling strategy using Youla parameterization
to improve stability and linear matrix inequalities
(LMIs) for guaranteeing global asymptotic stability,
resulting in a nonlinear controller that can “effec-
tively regulate evaporator super-heat while meeting
changing demands for cooling capacity with guar-
anteed closed-loop stability (p. 1224).
State feedback control
This section includes the topics of controllabil-
ity and observers as well as state feedback. Using
a third-order, bilinear, single-zone HVAC&R model
(with three states [temperature of supply air, temper-
ature of thermal space, and humidity ratio of ther-
mal space], two outputs [thermal space temperature
and humidity ratio], and two controls [volumetric
flow rate of air and flow rate of chilled water]),
linearized around an operating point, an observer-
based, disturbance rejection, state feedback, non-
linear controller for this bilinear model (Mohler
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
8 VOLUME 17, NUMBER 1, FEBRUARY 2011
1991a, 1991b) was designed based on the Lyapunov
approach involving the algebraic Riccati equation
while estimating online the thermal loads acting as
disturbances. It was found that the controller mini-
mizes the effect of large thermal loads maintaining
the comfort level in the thermal space. A multi-
variable controller was designed by Zaheer-Uddin
(1993) to control boiler temperature and hot water
temperature for a heating system for hot water and
space heating using the pole placement technique
(Patel and Munro 1982).
For a single-zone VAV HVAC&R nonlinear
model described by three state variables (temper-
ature of thermal space, humidity ratio of thermal
space, and temperature of supply air), two control
variables (volumetric flow rate of air and flow rate
of chilled water), and two outputs (temperature and
humidity ratio of thermal space), a state feedback
controller was developed in Arg
¨
uello-Serrano and
V
´
elez-Reyes (1999) to maintain thermal comfort in
the thermal space in the presence of time-varying
thermal loads (disturbances) by first linearizing the
model around an operating point and designing a
state and thermal load observer, and then design-
ing a disturbance-rejection state feedback controller
based on Lyapunov stability and the algebraic Ric-
cati equation. The methodology resulted in reducing
the effect of thermal loads by extracting more heat
from the thermal space, thereby reducing the s upply
air t emperature.
Two control strategies of condenser super-heat
regulation (CSR) and evaporator super-heat regu-
lation (ESR) were examined in Yeh et al. (2009)
for a cascade structure consisting of slow and fast
dynamics due to vapor compression cycle and in-
door dynamics. The two strategies were combined
with an experimental setup t o achieve both transient
performance and steady-state power savings.
Distributed parameter control
Using the distributed parameter model in Liu and
He (1994) for a room rather than a lumped parame-
ter model to take care of the thermal comfort level
at different locations within the room by describ-
ing the spatial distribution of air temperature and
velocity at steady-state conditions, a recursive opti-
mization algorithm was developed for a set of com-
fort indices at different locations within the room at
steady-state conditions rather than using a dynamic
model involving time and spatial coordinates.
Optimal control
First of all, in a recent editorial of HVAC&R
Research publication by Radermacher and Abde-
laziz (2008), it was clearly articulated that “opti-
mization, a process or methodology, is to “provide
a significant reduction in energy consumption and
material utilization. Thus, it appears that out of all
control techniques, optimal control took the lion’s
share of application to the HVAC&R field, which
is not surprising due to the nature of optimization
in terms of energy savings and human comfort.
Keeping in view the chronological nature of the
overview, what follows is divided into three subsec-
tions: early developments focusing on modeling and
optimization (up to 1989), continuing developments
(1990–1999), and recent developments focusing on
experimentation and applications to specific real-
world situations (2000–2010).
Early developments: up to 1989
One of the earliest applications of dynamic and
static optimization (Kaya 1978; Hartman 1980;
Kaya et al. 1982) was to find optimal control poli-
cies to minimize the overall energy expenditure and
simultaneously control room temperature, room hu-
midity, and outside wind velocity to keep the room
at a comfort zone, as recommended by ASHRAE. A
comparative study was made with conventional con-
trol methods consisting of a heating/cooling ther-
mostat and humidistat, claiming energy savings of
38.5% with an optimal control strategy. A similar
treatment was given by Nizet et al. (1984) by using
the airflow rate as a control variable for a simplified
model to minimize the total energy cost and thermal
comfort penalty using discretization of the original
continuous optimal control problem, and by solving
the resulting problem using a conjugate gradient
method (Fletcher 1980) to realize energy savings of
12% to 30% compared to the case without using the
conjugate method.
A combination of PID, feed-forward, and optimal
control f or regulating the temperature within a ther-
mal space was developed in Cherchas et al. (1985)
to maintain a set-point value with a linear cost func-
tion. In particular, a mathematical model for a single
environmental space was developed in terms of both
exact and simplified equations, using the principles
of conservation of mass, humidity, and energy (Bor-
resen 1981). The model had as the two state variables
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 9
a dry bulb temperature (T(t)) and a moisture con-
tent or humidity ratio (w(t)), and as the three control
variables, it had an air inlet volumetric flow rate
(f (t)) and a space-sensible load (q(t)) to control the
temperature and a supply air humidity ratio (wi(t))
to control the moisture content. With a simple cost
function in terms of f (t) and q(t), a feed-forward
and feedback control algorithm was developed and
implemented in discrete form. In a follow-up work
by Townsend et al. (1986), for a single zone de-
scribed by zone temperature and moisture content
in the zone as the two state (output) variables, inlet
supply air volumetric flow rate and heat input rate as
the two control variables, and the objective function
composed of energy cost, comfort, and start-up, an
optimal bang-bang switching control strategy was
obtained using the Pontryagin maximum principle
(Kirk 1970; Naidu 2003), yielding lower operating
costs compared to the previous work in Cherchas
et al. (1985). An optimal control strategy using the
Pontr yagin maximum principle (Pontryagin et al.
1962) was applied to a heat pump system, when
the storage capability was available and time-of-day
energy incentives were offered by electrical utility
company, using a single-zone model and a simple
space load and cost function to be minimized as
the cost of purchasing electrical energy (Rink et al.
1988), yielding extremal trajectories with one bang-
bang interval and one singular interval. Also see the
related works by Le et al. (1987) and Zaheer-Uddin
(1989) for a more accurate single-zone model and
Zaheer-Uddin (1991) for digital control. With a per-
formance criterion to maximize human comfort and
to minimize the energy and operating costs, a fuzzy
optimal controller was developed by Shoureshi and
Rahmani (1989).
Continuing developments: 1990–1999
A group of researchers (House et al. 1991) devel-
oped an optimal control methodology for a repre-
sentative HVAC&R system. The procedure involved
discretizing the continuous optimal control problem
and was called the discrete time method (DTM). In
particular, the temperatures within heat exchanger
and thermal space were the two state variables, and
heat input to heat exchanger and volumetric air-
flow rate were the two control variables. Assum-
ing the system is initially at the steady-state condi-
tion, the performance criterion to be minimized was
composed of five terms: disturbance rejection as a
penalty for the system not at steady state, thermal
comfort penalty, fuel usage, penalty for the fan as-
sociated with airflow rate, and the power required
to operate the fan. For solving the optimal con-
trol problem, a sequential quadratic programming
(SQP) was implemented with the DTM. It was found
that with the bang-bang (like in a simple room ther-
mostat), fuel is 49% higher than that with the opti-
mal control. Further, using tge performance index
in terms of a comfort penalty for deviations of the
zone air temperature from the set-point value and
the cost for energy consumption, an optimal control
strategy is compared with a conventional control
strategy for a two-zone building and HVAC&R sys-
tem with two zone envelope temperatures as state
variables and energy usage for cooling and heating
coils. It was found by House and Smith (1995b)
that there is a cost savings of 11% with the optimal
strategy. Here, the optimal control was obtained by
discretizing the continuous optimal control prob-
lem (Tseng and Arora 1989). Also see House and
Smith (1995a) for the same problem treated using a
systems approach, accommodating a variety of sys-
tem conditions, constraints, and different set-points
for different zones, resulting in energy savings of
24%.
The application of modern techniques, such as
optimal and adaptive control, to HVAC&R systems
was explored in Zaheer-Uddin (1993), which also
included a very good literature review. Next, us-
ing the linearized model, which consisted of zone
air temperature and zone air humidity ratio as the
two state variables and mass flow rate of supply
air as the control variable, the linear quadratic reg-
ulator (LQR) theory (Anderson and Moore 1971)
was applied to regulate the temperature and hu-
midity, at the same time rejecting any disturbances.
For an HVAC&R system consisting of two chillers
(one for direct chilling and the other ice-storage
charging), a cooling tower, an air handler, a con-
denser and chilled water pumps, and an ice-storage
tank, a comprehensive analysis of the optimal con-
trol was presented by Kintner-Meyer and Emery
(1995) for minimizing the operating cost over a
24-hour period. The cost function consisted of the
costs for electricity consumption and the demand
charge, and the main result showed significant re-
duction in operating costs (compared to a conven-
tional control scheme) due to “free cooling” through
early morning ventilation and by shifting cool-
ing loads from peak to off-peak hours. In another
work by Zaheer-Uddin and Patel (1995), a nonlinear
multi-zone environmental system was modeled as a
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
10 VOLUME 17, NUMBER 1, FEBRUARY 2011
linear reduced-order (second-order) model based on
the fast (second-order) and slow (fifth-order) sub-
systems using SPaTS analysis (Naidu 1988; Koko-
tovi
´
c et al. 1986; Naidu 2002). A reduced-order fast
model was retained by neglecting the slow subsys-
tem, although in the normal SPaTS literature, the
fast subsystem is neglected while retaining the slow
subsystem and an optimal tracking control was de-
signed for set-point changes. Two techniques of dif-
ferential dynamic programming (DDP) and nonlin-
ear programming (NLP) were compared as applied
to three cases of optimal control of an HVAC&R
system, described by temperature of air exiting the
heat exchanger and temperature of thermal space
being controlled as the two state variables, heat in-
put to the system and air flow rate from outside as
the two control variables, and the cost f unction. It
was found that DDP was more efficient than NLP,
although NLP is more robust (Kota et al. 1996).
Using an autoregressive (AR) model with three in-
put (manipulated) variables (opening of cooling wa-
ter valve, heater current, and humidifier current)
and two outputs (indoor temperature and humid-
ity) for an HVAC&R system model, a new opti-
mal preview control using linear quadratic Gaus-
sian (LQG) optimal control with a feed-forward
compensation (Stengel 1986) was implemented in
Kasahara et al. (1998) to improve tracking and to
suppress the interaction between different loops.
An optimal control for HVAC&R and related sys-
tems include Zheng and Zaheer-Uddin’s (1999) dis-
charge air system (DAS), consisting of a cooling
and de-humidifying coil, a chilled storage tank, and
an electric coil for re-heating. Simulation results
were compared with experimental data using opti-
mal control strategies to step changes in set-points
for the two cases of heating with temperature control
and cooling with temperature and humidity control.
The optimal cost function was formulated in terms
of the discharge air temperature and its set-point as
the state variable; and the mass flow rate of chilled
water as the control variable for the heating case;
discharge air temperature, discharge air humidity
ratio, and their set-points as the state variables;
and input energy to chiller, the mass flow rate of
chilled water, and electric heater energy as the three
control variables. Simulations using the numerical
search (gradient) method (Mufti 1970) for opti-
mal control showed smoother and rapid responses
in discharge air temperature and air humidity ra-
tio, while tracking performance showed significant
improvement.
Recent developments: 2000–2010
For the time-scheduled operation of an HVAC&R
system, optimal control strategies were developed in
Zaheer-Uddin and Zheng (2000), using a two-zone
VAV heating (VAVH) model consisting of a heat
pump, a s torage tank, water and airflow networks,
and two environmental zones.
Here, the time schedule of the b uilding opera-
tion was divided into three stages/modes: the set-
back mode between 5 PM and 7 AM of the next day,
the start-up mode between 7 AM and 8 AM, and
the working normal mode between 8 AM and 5 PM
of the day, thus leading to the three-stage optimiza-
tion control. Recognizing that the airflow subsys-
tem is faster than the environmental zones (slow)
subsystem, the optimal supervisory control prob-
lem was cast in the singularly perturbed structure
and solved using the SPaTS methodology (Koko-
tovi
´
c and Sannuti 1968; Naidu 1988). A similar ap-
proach by Zaheer-Uddin and Zheng (2001), involv-
ing multi-stage optimization and SPaTS methodol-
ogy, was developed for a single-zone space heating
(SZSH) system with different operating strategies
of constant volume (CV; where zone temperature
alone is modulated), VAV (where airflow rate is
modulated) and a more general VAV called gen-
eral variable-air-volume (VAVN) (air-supply tem-
perature and flow rate are continuously modulated),
and with three-mode building time-scheduled oper-
ation. The results showed that the VAVN strategy
offered a operating costs savings of 25% compared
to the CV strategy.
A simple second-order model was used for the
HVAC&R system in T
˘
ıgrek et al. (2002) with two
state variables—temperature immediately following
the heat exchanger and temperature of the ther-
mal zone, two control variables—the heat input
to the heat exchanger and the volumetric airflow
rate, and a nonquadratic (cubic) term in the per-
formance criteria to apply the optimal control the-
ory to obtain state and co-state equations and lin-
earize them to remove non-causality. Further, an
adaptive controller was designed using RLS to take
care of changes in external temperature and ther-
mal load variance over time and the difficulty of
accurately measuring these variables. A combined
utilization of active ( ice storage) and passive (pre-
cooling) inventory for the reduction of electrical
utility costs using common time-of-use rate differ-
entials led to the design of an optimal controller,
resulting in utility cost savings and substantial
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 11
on-peak electrical demand reductions (Henze et al.
2004).
An existing HVAC&R system installed at the
Montreal campus of the Ecole de Techologie
Sup
´
erieure (ETS) has 70 interior zones and has
more than 60 set-point variables to be optimized
with the multi-objective criteria of minimizing the
energy consumption (resulting from fan energy use
and chiller energy use) and maximizing the b uild-
ing thermal comfort (represented by PPD in terms
of PMV) by choosing the optil set-points for the su-
pervisor control strategy. Two different evolutionary
algorithms were given by Deb (2001): the nondom-
inated sorting GA (NSGA) and the elitist NSGA
(NSGAII), which were evaluated and compared
with Pareto-optimal solutions (Engwerda 2005). It
was concluded that NSGAII performs better with
energy demand lessened by 18.8% for July 29 and
by 19.5% for July 25–31.
In addition to using a direct digital control (DDC)
system and HVAC set-points as control variables, a
method was developed in Xu et al. (2005) combin-
ing Lagrangian relaxation (LR), NNs, stochastic dy-
namic programming (SDP), and heuristics with nu-
merical testing and prototype implementation. Here
LR, a decomposition and coordination approach, is
used to transform the original problem into a sep-
arable structure by introducing two new variables;
NNs are used to predict system dynamics; uncon-
trollable loads and SDP are used to solve the HVAC
unit subproblems with a thermal load described by
a single-state Markov chain; and heuristics are used
to obtain feasible solutions. A mathematical basis
for a complete simulation-based SQP (CSB-SQP)
methodology was developed and applied to a sim-
ple model arising in determining the optimal con-
trol for the operation of HVAC&R building systems
(Sun and Reddy 2005).
An experimental study was carried out on a com-
mercial building’s passive and active thermal stor-
age inventory using a hybrid control consisting of
model-based optimal control and model-free rein-
forced learning control; the proposed method, com-
pared with model-based predictive optimal control,
was implemented on a full-scale laboratory facility
(Liu and Henze 2006a, 2006b). A simulated rein-
forcement learning consists of two phases: a simu-
lated learning phase (where a learning controller is
trained by a simulator without using the actual re-
sponse) and an implemented learning phase (where
the learning controller is expected learn and improve
while in direct contact with the environment). The
hybrid approach was validated by an experimental
study, and it was found that the hybrid approach
achieved cost savings of 8.3% compared to the case
using measured data, while the quality of the simu-
lator remained a key disadvantage. A simple near-
optimal method was developed by Braun (2007) for
controlling the charging and discharging thermal
storage systems having real-time pricing (RTP) by
determining the effective on-peak and off-peak pe-
riods. The performance was measured relative to a
benchmark optimization problem, resulting in an-
nual costs of about 2% of the costs with optimal
control with the additional features of low hard-
ware costs and a simplified controller architecture.
Another survey by Sane et al. (2006) of literature fo-
cused on HVAC&R controls and optimization pro-
vided an over view on building dynamics (air-side
dynamics, chilled water dynamics, loads and dis-
turbances, energy costs) and control problems in
chilled water dynamics (control related to flow con-
trol, supply temperature, resource allocation with
chilled sequence) with a presentation of an example
on dither-based optimal control. For an HVAC&R
system consisting of an air-to-air heat exchanger
and a water-to-air heat exchanger (Komareji et al.
2007), an objective function to be minimized was
formed in terms of powers due to primary and ter-
tiary pumps, fan, and wheel rotation and thermal
power. Optimal set-points were computed for two
cases of unequal water flow of the tertiary circuit
and the supply (primary/secondary) circuit. See the
related work by Komareji et al. (2009) that used a
simplified optimal control structure using the lin-
earized model from the primary water flow to the
inlet air temperature.
Realizing that the modern buildings are complex,
highly uncertain nonlinear, and multi-dimensional
dynamic systems with wide varieties of distur-
bances, the estimation and control problem for a dis-
tributed parameter model of a multi-room building
was addressed by Borggaard et al. (2009), using the
distributed parameter LQR theory combined with fi-
nite elements to compute both feed-back functional
gains and observer functional gains for thermal con-
trol of a 3D room in a typical HVAC&R system. For
simultaneously controlling the indoor air tempera-
ture and humidity for thermal comfort and indoor
air quality (IAQ) by varying the speeds of both the
compressor and supply fan in an experimental direct
expansion (DX) A/C system, a MIMO control strat-
egy involving the LQG technique was developed
by Qi and Deng (2009) to take care of disturbance
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
12 VOLUME 17, NUMBER 1, FEBRUARY 2011
rejection and command tracking and improve super-
heat control, as reported in Qi et al. (2010). A re-
ceding horizon optimal control (RHOC) at a super-
visory level was applied in Lukasse et al. (2009) to
control the climate of storage facilities for potatoes
and onions with the results from both simulation
and full-scale experimentation. An online optimal
control strategy consisting of a model-based perfor-
mance predictor to get simplified air-handling unit
(AHU) models, cost function, optimization tech-
nique, and supervisory and local control schemes
for various units of chillers, variable-speed pumps,
and heat exchangers was presented in Ma and Wang
(2009) to realize energy efficiency, robustness, and
tracking with a simulated environment. An opti-
mal control strategy was evaluated by Yu and Chan
(2010) for controlling the cooling towers and con-
denser water pumps in a water-cooled chiller sys-
tem, claiming that the load-based speed control to
the cooling tower fans resulted in 8.6% annual en-
ergy and 9.9% operating cost savings compared to
the case without using the proposed methodology.
Model predictive control (MPC)
A novel supervisory controller was successfully
executed in Henze et al. (2005) using a three-step
procedure for a model-based predictive optimal con-
trol for building a thermal storage inventory in a test
facility in real time using time-of-use differentiated
electricity prices without demand charges based on
their earlier works (Henze et al. 1997). A MPC strat-
egy was used in Yuan and Perez (2006) to maintain
ventilation air requirements and temperature of mul-
tiple zones for VAV systems to achieve acceptable
IAQ. The results were demonstrated by experimen-
tation with four different typical weather conditions.
A hierarchical structure based on minimizing the
generalized predictive control (GPC) criterion to
tune conventional PID controller parameters was
proposed in Xu et al. (2006), which was applica-
ble to a wide range of operating conditions for a
cooling coil unit in an HVAC system. In Xu and Li
(2007), using sequential step changes, the original
coupled HVAC&R plant was decoupled into four
subsystems, and a practical GPC technique along
with a novel parameter identification was presented,
thereby providing less computation compared to the
original coupled system, and was demonstrated by
simulations of an HVAC&R plant. A nonlinear MPC
technique along with an optimization algorithm was
applied in Xi et al. (2007) to control the tempera-
ture and humidity of an HVAC system, which was a
two-by-two nonlinear dynamic model using support
vector regression (SVR) and constraints on control
variables to generate real-time control signals. It
showed good performance in tracking a reference
trajectory and steady-state errors and superior per-
formance compared to the neuro-fuzzy controller.
Using the PMV and a psychrometric chart to charac-
terize occupants’ thermal comfort, different strate-
gies based on MPC were proposed in Freire et al.
(2008) for optimization of thermal comfort and min-
imization of energy consumption. The results were
demonstrated by simulations for two case studies re-
lating to controller performance analysis and vary-
ing metabolic rate (ASHRAE 2005) and clothing
index (Parsons 1988). A robust MPC method was
presented by Huang and Wang (2008) for control-
ling the temperature of AHUs using two loops—one
outer loop using Lyapunov analysis and another in-
ner loop with an integral controller—to improve
system performance in the presence of model un-
certainties and constraints. The novelty of this ap-
proach is the adaptation of overlapping modes to
make sure that the controller does not switch of-
ten among the operating modes. The application of
this methodology to a typical AHU and a compari-
son of the results with an anti-windup PI controller
shows higher robustness and overall performance
improvement without the need for online tuning.
A related work on the application of an LMI-
based robust MPC strategy was reported in Huang
and Wang (2009) for a discrete-time constraint pro-
cess, with time delay treated as an uncertainty for
a typical AHU. Another investigation from Huang
et al. (2009) presented a robust MPC strategy for
the temperature control VAV A/C (VAVAC ) s y s t e m
to take care of uncertainties and nonlinearities us-
ing the Takagi–Sugeno (T-S) fuzzy model with a
local controller consisting of two loops (an inner-
loop integral controller and an outer-loop min-max
predictive controller) and a global controller based
on the parallel distributed compensation technique
with experimental justification for acceptable con-
trol performance without any on-site tuning. See a
related work by Huang et al. (2009) for a first-order
plus time-delay model for the AHU with uncertain
time-delay and system gain, using an offline LMI-
based robust MPC algorithm and giving a com-
parison with a traditional PI control technique. A
novel technique for temperature control of tankless
water heaters (TWHs) was developed by Henze et
al. (2009) based on MPC to minimize the outlet
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 13
temperature error with an experimental demonstra-
tion of a physical prototype tankless heater system
(THS) with an integral performance criterion.
Describing the dynamics of VAV AHUs as a first-
order, a time-delay model with uncertainties, a con-
trol input rate-limit, and saturation constraints, a
new robust temperature control strategy was pre-
sented in Huang et al. (2010), using an offline ro-
bust MPC. Simulations of the VAV AHU showed
enhancement of robustness with less operator inter-
vention.
Robust or H control
The robust control methodology takes care of
model uncertainty and the nonlinearities associated
with the system (see Green and Limebeer 1995;
Zhou and Doyle 1998; Chen 2000; Sinha 2007).
For a two-zone space heating system (Zaheer-
Uddin et al. 1993), an analytical bilinear model
was developed with seven state variables, three
output variables, and five control variables; it was
linearized about an operating point to get a linear
state space model. A decentralized rob ust controller
was designed to asymptotically regulate the two
zone temperatures to the desired set-points in the
presence of unknown disturbances in the outdoor
temperatures and any nondestabilizing perturba-
tions in the parameters of the system. In particular,
two different controller designs were obtained, con-
sidering the system as a linear, constrained, robust
servomechanism problem (LCRSP) (Davison and
Ferguson 1981) and as a nonlinear, constrained,
servomechanism problem (NCSP) (Davison 1976).
A comparative study of simulations showed that the
decentralized controllers are good for multi-zone
space-heating systems.
A simple model for hot water heating of
fresh air to a controlled downstream duct tem-
perature and a benchmark controller were devel-
oped in Underwood (2000a) using MATLAB
©
and
SIMULINK
©
. Next, after providing some basic
concepts of a robust control theory for a fifth-order
state space model of an air heating plant with three
outputs and one input, a fixed-parameter robust con-
troller based on H design with structured uncer-
tainty was developed by Underwood (2000b), and
the simulation results showed a good comparison
with a locally optimized PID controller. A robust
control strategy combining freezing, gain schedul-
ing, I-term reset, and feedback transition control
was developed in Wang and Xu (2002, 2004) to
overcome the instability during transition processes
between different control modes while combining
the demand-controlled ventilation (DCV) and econ-
omizer control, providing savings in energy require-
ment.
Although a comparison of PID and H con-
trollers for the temperature control of a single-zone
room showed better performance with large pertur-
bations, the PID controllers will continue to play
a leading role for control of HVAC systems (Noda
et al. 2003).
It was shown by both theoretical and experimen-
tal results in Anderson et al. (2007, 2008) that the ap-
plication of the MIMO robust control methodology
vastly (as much as 300%) improves performance
and constraints, as was demonstrated by building
an experimental setup to test a variety of HVAC&R
controllers, including the well-known SISO PI con-
trollers. The experimental system, in particular, con-
sisted of five subsystem models—blower, mixing
box, boiler, water flow control valve, and heating
coil—and the control software used included MAT-
LAB
©
with tool boxes Simulink, Real-Time Work-
shop and Windows Target. The four key control
variables in a DAS are the input air and water tem-
peratures and the corresponding flow rates supplied
to the heating coil. The Hrobust controller was
designed based on the structured singular value (µ)
and implemented with the experimental system.
Nonlinear and adaptive control
Nonlinear control
For the single-zone VAV HVAC&R system by
Semsar-Kazerooni et al. (2008), a bilinear model
was considered for describing both temperature and
humidity dynamics; a back-stepping controller was
designed for the feedback linearized model, consid-
ering heat and moisture loads as measurable dis-
turbances; and a stable observer was designed for
nonmeasurable disturbances backed by simulation
results for optimal energy consumption. Also see
related results by He and Asada (2003). A feedback
linearization technique available for the design of
nonlinear control systems was applied by Thosar et
al. (2008) to design a controller for a VAVAC m odel.
It was simulated with a laboratory-scale plant
and showed superior performance in keeping com-
fort and optimal energy compared to the conven-
tional PI controller. In a typical room temperature
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
14 VOLUME 17, NUMBER 1, FEBRUARY 2011
control system using a standard controller, such as a
PI controller, it was found that temperature periodi-
cally oscillated around the constant reference value
with an unacceptablly high amplitude. The combi-
nation of well-known techniques (linearization, in-
ward approach, and describing function) proved to
be a valuable tool to find controllers with superior
performance compared to the PI-controller (Rehrl
et al. 2009).
Adaptive control
One of the earliest works by Farris and McDon-
ald (1980) to apply adaptive control for HVAC&R
systems focused on DDC for solar-heated b uild-
ings, with a single-zone air space and room air
temperature as the output of the system. In par-
ticular, an adaptive optimal control (AOC) strategy
was designed using a linearized model of the orig-
inal nonlinear HVAC&R system and closed-loop
optimal obtained via the matrix Riccati equation
in Zelikin (2000). With zone temperature and hot
water temperature as the two state variables, and
heat pump input as the control variable, an adap-
tive control strategy by
˚
Astr
¨
om and Wittenmark
(1989) was applied to a discharge air temperature
model (Zaheer-Uddin 1993) for the discharge air
temperature to track the optimal reference temper-
ature in the presence of disturbances. Another class
of adaptive systems, known as model-following or
model-reference adaptive control (MRAC), was ap-
plied to a VAV system with zone, coil, and water
temperatures as the three state variables; mass flow
rate of supply air, mass flow rate of chilled water,
and input energy to the chiller as the three control
variables; and a second-order model as the refer-
ence model for the VAV system. The simulations
showed good adaptability of the actual zone temper-
ature with its reference value. For a fan-coil heating
(FCH) with two thermal zones described by three
outputs (two zone temperatures and boiler temper-
ature), three inputs (two mass flow rate controllers,
and one flow rate of natural gas to the burner), and
a fifth-order dynamic disturbance, a robust adaptive
controller was designed in Singh et al. (2000) using
the continuous-time least squares (CTLS) algorithm
(
˚
Astrom and Wittenmark 1989) to estimate a linear
model of the original nonlinear FCH system. Then,
this estimated linear model was used to design a
linear feedback controller based on the LQR tech-
nique (Naidu 2003). The results showed that the ro-
bust adaptive controller was able to reject rapidly the
effect of both static and dynamic disturbances and
take care of unmodeled dynamics in the actuators.
Various control strategies relating to air-bypass, re-
set, setback, improved start–stop times, economizer,
and CO2 were investigated by Mathews et al. (2001)
using QuickControl
©
software developed by Mo-
tion Control, resulting in improving the comfort
and saving energy (60%) for a particular HVAC
system.
The single-zone VAV HVAC&R nonlinear
model in Arg
¨
uello-Serrano and V
´
elez-Reyes
(1999), consisting of three state variables (temper-
ature of thermal space, humidity ratio of thermal
space, and temperature of supply air), two control
variables (volumetric flow rate of air and flow rate
of chilled water), and two outputs (temperature and
humidity ratio of thermal space), was generalized
as a class of an interconnected MIMO system
consisting of several dynamical subsystems. A de-
centralized nonlinear adaptive controller (DNAC)
was developed by Huaguang and Cai ( 2002) in
terms of a state feedback FL controller (SFLC) for
the inner loop and a frequency-domain adaptive
compensator (FDAC) f or the outer loop with the
global objective of minimizing the error between
the actual output and desired trajectory in the
presence of disturbances, approximated as a Fourier
integration function. The resulting DNAC provided
higher precision, smaller overshoot, and shorter
settling time, with better overall performance than
the SFLC and other controllers.
An energy management control (EMC) strategy
was developed by Huang et al. (2006) for a single-
zone VAV HVAC&R system composed of a zone
model, cooling coil model, heat pump and stor-
age model, heat pump COP model, and fan-motor
model. The EMC consisted of five functions: out-
door air economizer cycle, programmed start/stop
lead time, load reset, and occupied time adaptive
control strategy. The objective function to be opti-
mized was the system power composed of heat pump
input power, fan power, and pump power, providing
the optimal set-points that are used as tracking sig-
nals for the adaptive controllers. This control strat-
egy resulted in 17% energy savings compared with
the no-EMC strategy.
An adaptive controller was designed in Al-
bieri et al. (2009) for a single scroll compres-
sor, using packaged air-cooled water chillers with
MATLAB
©
/SIMULINK
©
simulations and a state-
of-the-art experimental facility to achieve excellent
regulation performance.
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 15
Concluding remarks on Part I:
HC techniques
The main HC techniques discussed above are
1. PID control, gain scheduling, and state feedback;
2. optimal control;
3. model predictive control;
4. robust or H control; and
5. nonlinear and adaptive control.
The contributions of the HC techniques to the
HVAC&R field are summarized below:
1. The traditional control t echniques, such as the
PID control, have been a dominant area of re-
search and applications to the field of HVAC&R.
2. The area of optimal control dominated the re-
search efforts in HVAC&R, mainly due to the
attractive feature of energy savings.
3. On the other hand, the MPC has the advantage
of arriving at a control strategy when the com-
plete knowledge of model is not readily avail-
able; there are some notable contributions in this
HVAC&R field.
4. Robust control provides attractive features for
the situation model parameter uncertainty and
external disturbances and needs further investi-
gation for the HVAC&R field.
5. Finally, the area of nonlinear and adaptive con-
trol gives an entirely different approach to deal
with the nonlinear model and the fact that the pa-
rameters of the system are slowly time-varying or
uncertain. Surprisingly, there are not many works
dealing with nonlinear control approaches to the
HVAC&R field, although adaptive control has a
good record of applications to the field.
6. It appears that most control techniques discussed
here did not take into account the various con-
straints on states and controls to reflect the realist
situations.
Part II of the overview will address SC tech-
niques and the fusion of HC and SC techniques
as applicable to the HVAC&R field. Some of the
future directions of research to achieve better mod-
eling, analysis, design (control), and security will
be discussed for overall performance enhancement
of HVAC&R systems.
Nomenclature
A/C = air-conditioning
AHU = air-handling unit
ANF = adaptive neuro-fuzzy
ANFIS = adaptive neuro-fuzzy inference sys-
tem
ANN = artificial neural networks
AOC = adaptive optimal control
AR = autoregressive
ASHRAE = American Society of Heating, Re-
frigerating and Air-Conditioning En-
gineers
BACS = building automation and control sys-
tems
BAS = building automation s ystems
BEMS = building energy management system
BLAST = building loads analysis and system
thermodynamics
CC = centralized control
CGI = common gateway interface
CI = computational intelligence
CIBSE = Chartered Institution of Building
Services Engineers
COP = coefficient of performance
CRIM = commercial refrigerator/incubator
module
CSB-SQP = complete s imulation-based sequen-
tial quadratic programming
CSR = condenser super-heat regulation
CTLS = continuous time least squares
CV = constant volume
DAS = discharge air system
DBC = dead-beat compensator
DC = decentralized control
DCV = demand-controlled ventilation
DDC = direct digital control
DDP = differential dynamic programming
DNAC = decentralized nonlinear adaptive
controller
DOE = Department of Energy
DTM = discrete time method
DX = direct expansion
EMC = energy management control
EMCS = energy management control systems
ESR = evaporator super-heat regulation
ETS = Ecole de Techologie Sup
´
erieure
FCH = fan-coil heating
FDAC = frequency-domain adaptive compen-
sator
FIS = fuzzy inference system
FL = fuzzy logic
FLC = fuzzy logic control or controller
FPID = fuzzy proportional, integral, deriva-
tive
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
16 VOLUME 17, NUMBER 1, FEBRUARY 2011
GA = genetic algorithm
GENESYS = generic embedded system
GFC = Gupta fuzzy controller
GPC = generalized predictive control
GUI = graphical user interface
HC = hard computing or control
HFC = hydro fluoro carbon
HVAC = heating, ventilation, and air-
conditioning
HVAC&R = heating, ventilation, air-
conditioning, and refrigeration
IAQ = indoor air quality
ICIS = instrumentation, control, and intelli-
gent systems
ICS = industrial control s ystems
INL = Idaho National Laboratory
ITAE = integral of time and absolute error
ISU = Idaho State University
JJH = James J. Hirsch & Associates
LBNL = Lawrence Berkeley National Labo-
ratory
LCRSP = linear, constrained, robust ser-
vomechanism problem
LDRD = laboratory-directed research and de-
velopment
LMIs = linear matrix inequalities
LQG = linear quadratic Gaussian
LQR = linear quadratic regulator
LQT = linear quadratic tracker
LR = Lagrangian relaxation
MACS = multi-agent control systems
MFC = Mamdani fuzzy controller
MICNO = mixed-integer, constraint, nonlinear
optimization
MIMO = multi-input, multi-output
MOGA = multi-objective genetic algorithm
MPC = model predictive control
MRAC = model-reference adaptive control
MZSH
= multi-zone space heating
NCSP = nonlinear, constrained, servomecha-
nism problem
NLP = nonlinear programming
NN = neural networks
NNDC = neural network based decentralized
controller
NSGA = non-dominated sorting genetic algo-
rithm
NSGAII = elitist non-dominated sorting genetic
algorithm
P&ID = piping/process and instrumentation
diagram
PD = proportional-derivative
PI = proportional-integral
PID = proportional-integral-derivative
PMES = performance map and exhaustive
search
PMV = predicted mean value
PPD = predicted percentage dissatisfied
RBF = radial basis function
REA = robust evolutionary algorithm
RHOC = receding horizon optimal control
RLS = recursive least squares
RTP = real-time pricing
SC = soft computing or control
SDP = stochastic dynamic programming
SFLC = state feedback fuzzy logic controller
SISO = single-input, single-output
SPaTS = singular perturbation and time scales
SQP = sequential quadratic programming
SVR = support vector regression
SWT = supply chilled water temperature
SZSH = single-zone space heating
TCP/IP = transmission control proto-
col/internet protocol
TES = thermal enclosure system
THS: = tankless heater system
TWH =
tankless water heater
UIUC = University of Illinois at Urbana
Champaign
USDOE = United States D epartment of Energy
VAV = variable airflow volume
VAVAC = variable airflow volume air-
conditioning
VAV H = variable air volume heating
VAV N = general variable-air-volume
VC = visual comfort
VWV = variable water volume
Z-N = Ziegler-Nichols
Acknowledgment
The funding provided for this research ac-
tivity, performed under subcontract support of
a laboratory-directed research and development
(LDRD) project focusing on areas of both energy
science and national security at the Idaho National
Laboratory (INL), Idaho Falls, is gratefully ac-
knowledged.
References
Albieri, M., A. Beghi, C. Bodo, and L. Cecchinato. 2009. Ad-
vanced control systems for single compressor chiller units.
International Journal of Refrigeration 32:1068–76.
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 17
Anderson, B., and J. Moore. 1971. Linear Optimal Control.En-
glewood Cliffs, NJ: Prentice-Hall.
Anderson, B., and J. Moore. 1990. Optimal Control: Linear
Quadratic Methods. Englewood Cliffs, NJ: Prentice Hall.
Anderson, M., M. Buehner, P. Young, D. Hittle, C. Anderson, J.
Tu, and D. Hodgson. 2008. MIMO robust control for HVAC
systems. IEEE Transactions on Control Systems Technology
16(3):475–83.
Anderson, M., M. Buehner, P. Young, D. Hittle, C. Anderson,
J. Tu, T. Jilin, and D. Hodgson. 2007. An experimental sys-
tem for advanced heating, ventilating and air conditioning
(HVAC) control. Energy and Buildings 39(2):136–47.
Arg
¨
uello-Serrano, B., and M. V
´
elez-Reyes. 1999. Nonlinear con-
trol of a heating, ventilating, and air conditioning system with
thermal load estimation. IEEE Transactions on Control Sys-
tems Technology 7:56–63.
ASHRAE. 2005. 2005 ASHRAE Handbook—Fundamentals (SI).
Atlanta, GA: American Society of Heating, Refrigerating and
Air-Conditioning Engineers.
ASHRAE. 2006. 2006 ASHRAE Handbook—Refrigeration (SI).
Atlanta, GA: American Society of Heating, Refrigerating and
Air-Conditioning Engineers.
ASHRAE. 2007. 2007 ASHRAE Handbook—HVAC Applications
(SI). Atlanta, GA: American Society of Heating, Refrigerat-
ing and Air-Conditioning Engineers.
ASHRAE. 2008. 2008 ASHRAE Handbook—HVAC Systems and
Equipment (SI). Atlanta, GA: American Society of Heating,
Refrigerating and Air-Conditioning Engineers.
˚
Astr
¨
om, K., and T. H
¨
agglund. 1995. PID Controllers: Theory,
Design, and Tuning, 2nd ed. Research Triangle Park, NC:
Instrument Society of America.
˚
Astr
¨
om, K., and T. H
¨
agglund. 2006. Advanced PID Control.
Research Triangle Park, NC: Instrument Society of America.
˚
Astr
¨
om, K., and B. Wittenmark. 1989. Adaptive Con-
trol. Reading, MA: Addison-Wesley Publishing Company,
Inc.
Bai, J., S. Wang, and X. Zhang. 2008. Development of an adap-
tive Smith predictor-based self-tuning PI controller for an
HVAC system in a test room. Energy and Buildings 40:2244–
52.
Bai, J., and X. Zhang. 2007. A new adaptive PI controller and
its application in HVAC systems. Energy Conversion and
Management 48:1043–54.
Bi, Q., W.-J. Cai, Q.-G. Wang, C.-C. Hangc, E.-L. Lee, Y. Sun,
K.-D. Liu, Y. Zhang, and B. Zou. 2000. Advanced controller
auto-tuning and its application in HVAC systems. Control
Engineering Practice 8(6):633–44.
Blast-UIUC. 1983. BLAST 3.0 Building Loads Analysis and
Thermodynamics Program User Manual.IL:Dept.of
Mechanical Engineering, University of Illinois, Urbana-
Champaign, IL.
Borggaard, J., J. Burns, A. Surana, and L. Zietsman. 2009.
Control, estimation and optimization of energy efficient
buildings. American Control Conference 2009 (ACC ‘09),
St. Louis, MO, pp. 837–41.
Borresen, B. 1981. Thermal room models for control analysis.
Transactions of the ASHRAE Annual Meeting Cincinnati,
OH, pp. 251–61.
Braun, J. 2007. A near-optimal control strategy for cool storage
systems with dynamic electric rates (RP-1252). HVAC&R
Research 13(4):557–80.
Brumbaugh, J. 2004a. Audel HVAC Fundamentals: Volume 1:
Heating Systems, Furnaces and Boliers, all new 4th ed. In-
dianapolis, IN: Wiley Publishing Inc..
Brumbaugh, J. 2004b. Audel HVAC Fundamentals: Volume 2:
Heating System Components, Gas and Oil Burners, and Au-
tomatic Controls, all new 4th ed. Indianapolis, IN: Wiley
Publishing Inc..
Brumbaugh, J. 2004c. Audel HVAC Fundamentals: Volume 3:
Air Conditioning, Heat Pumps, and Distribution Systems, all
new 4th ed. Indianapolis, IN: Wiley Publishing Inc..
Brumbaugh, J. 2005. Audel HVAC Pocket Reference. Indianapo-
lis, IN: Wiley Publishing Inc..
Chen, B. 2000. Robust and H Control. London, UK: Springer-
Verlag.
Cherchas, D., A. Abdelmessih, and M. Townsend. 1985. A direct
digital control algorithm for control of a single environmen-
tal space. Transactions of the ASME: Journal of Dynamic
Systems, Measurement, and Control 107(4):324–31.
Crawley, D., L. Lawrie, C. Pedersen, and F. Winkelmann. 2000.
EnergyPlus: Energy simulation program. ASHRAE Journal
42(4):49–56.
Davison, E. 1976. The robust decentralized control of a gen-
eral servomechanism problem. IEEE Transactions Automatic
Control AC-21:14–24.
Davison, E., and I. Ferguson. 1981. The design of controllers for
the multivariable robust servomechanism problem using pa-
rameter optimization methods. IEEE Transactions Automatic
Control AC-26:93–110.
Deb, K. 2001. Vol. 5 of Prentice Hall Series on Environmen-
tal and Intelligent Manufacturing Systems, Multi-objective
optimization using evolutionary algorithms. Chichester, UK:
John Wiley & Sons.
Dexter, A. 1988. Control system simulation—computer control.
Energy and Buildings 10:203–11.
Dounis, A., and C. Caraiscos. 2009. Advanced control systems
engineering for energy and comfort management in a build-
ing environment: A review. Renewable and Sustainable En-
ergy Reviews 13(6–7):1246–61.
Elliott, M., and Z. Walton. 2009. Superheat control: A hybrid
approach. HVAC&R Research 15(6):1021–43.
Engwerda, J., 2005. LQ Dynamic Optimization and Differential
Games. John Wiley & Sons, New York, NY.
Fanger, P. 1972. Thermal Comfort: Analysis and Applications in
Environmental Engineering. New York: McGraw-Hill.
Fanger, P., 1984. Moderate thermal environments— Determina-
tion of the PMV and PPD indices and specification of the
conditions for thermal comfort. ISO-7730, International Or-
ganization for Standardization, Geneva, Switzerland.
Farris, D., and T. McDonald. 1980. Adaptive optimal control—an
algorithm for direct digital control. Transactions of the
ASHRAE 86:880–93.
Flesch, R.C., and J.E. Normey-Rico. 2010. Modelling, identi-
fication and control of a calorimeter used for performance
evaluation of refrigerant compressors. Control Engineering
Practice 18(3):254–61.
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
18 VOLUME 17, NUMBER 1, FEBRUARY 2011
Fletcher, R. 1980. Practical Methods of Optimization.NewYork:
John Wiley & Sons.
Franco, W., M. Sen, and K. Yang. 2005. Optimization of con-
trol hardware placement in a thermal-hydraulic network.
HVAC&R Research 14(1):73–84.
Freire, R., G. Oliveira, and N. Mendes. 2008. Predictive con-
trollers for thermal comfort optimization and energy savings.
Energy and Buildings 40(7):1353–65.
Geng, G., and G. Geary. 1993. On performance and tuning of
PID controllers in HVAC systems. Second IEEE Conference
on Control Applications. Vol. 2. Vancouver, BC, pp. 819–24.
Green, M., and D. Limebeer. 1995. Linear Robust Control.En-
glewood Cliffs, NJ: Prentice-Hall.
Haines, R. 1971. Control Systems for Heating, Ventilating and Air
Conditioning. New York: Van Nostrand Reinhold Company.
Haines, R. 1987. Control Systems for Heating, Ventilating and
Air Conditioning, 4th ed. New York: Van Nostrand Reinhold
Company.
Haines, R., and D. Hittle. 2003. Control Systems for Heating, Ven-
tilating and Air Conditioning, 6th ed. Boston, MA: Kluwer
Academic Publishers (2nd) printing, 2006).
Harrold, M., and D. Lush. 1988. Automatic controls in building
services. IEE Proceedings B: Electric Power Applications
135(3):105–33.
Hartman, T. 1980. Dynamic control of heat called key to saving
energy. Air-Conditioning, Heating and Refrigeration News
151(1):14–6.
Hartman, T. 1993. Direct Digital Controls for HVAC Systems.
New York: McGraw-Hill, Inc..
He, X.-D., and H. Asada. 2003. A new feedback lineariza-
tion approach to advanced control of multi-unit HVAC sys-
tems. Proceedings of the 2003 American Control Conference
3:2311–6.
Henze, G., A. Coward, and D. Yuill. 2009. Development of
a model predictive controller for tankless water heaters.
HVAC&R Research 15(1):3–33.
Henze, G., R. Dodier, and M. Krarti. 1997. Development of
a predictive optimal controller for thermal energy storage
systems. HVAC&R Research 3(3):233–64.
Henze, G., C. Felsmannb, and G. Knabeb. 2004. Evaluation of op-
timal control for active and passive building thermal storage.
International Journal of Thermal Sciences 43(2):173–83.
Henze, G., D. Kalz, S. Liu, and C. Felsmann. 2005. Experimental
analysis of model-based predictive optimal control for active
and passive building thermal storage inventory. HVAC&R
Research 11(2):189–213.
Honeywell. 1997. Engineering Manual of Automatic Control for
Commercial Buildings, 6th ed. Morristown, NJ: Honeywell.
Hordeski, M. 2001. HVAC Control in the New Millennium. Lil-
burn, GA: The Fairmont Press, Inc..
House, J., and T. Smith. 1995a. A system approach to optimal
control for HVAC and building systems. ASHRAE Transac-
tions 101:647–60.
House, J., and T. Smith. 1995b. Optimal control of building and
HVAC systems. Proceedings of the American Control Con-
ference. Vol. 6 Seattle, WA , pp. 4326–30.
House, J., T. Smith, and J. Arora. 1991. Optimal control of a
thermal system. ASHRAE Transactions 97:991–1001.
Huaguang, Z., and L. Cai. 2002. Decentralized nonlinear adaptive
control of an HVAC system. IEEE Transactions on Systems,
Man, and Cybernetics, Part C: Applications and Reviews
32:493–98.
Huang, G., A. Dexter, and S. Wang. 2009. A robust model predic-
tive control strategy for improving the control performance
of air-conditioning systems. Energy Conversion and Man-
agement 50(10):2650–8.
Huang, G., and S. Wang. 2008. Two-loop robust model predic-
tive control for the temperature control of air-handling units.
HVAC&R Research 14(4):565–80.
Huang, G., and S. Wang. 2009. Use of uncertainty polytope to
describe constraint processes with uncertain time-delay for
robust model predictive control applications. ISA Transac-
tions 48(4):503–11.
Huang, G., S. Wang, and X. Xu. 2010. Robust model predictive
control of VAV air-handling units concerning uncertainties
and constraints. HVAC&R Research 16(1):15–33.
Huang, W., M. Zaheeruddin, and S. Cho. 2006. Dynamic sim-
ulation of energy management control functions for HVAC
systems in buildings. Energy Conversion and Management
47(7–8):926–43.
HVACSIM. 1986. Building systems and equipment simulation
program reference manual. Technical Report, National Tech-
nical Information Service (NTIS), U.S. Department of Com-
merce, Washington, DC.
Imbabi, M., 1990a. A general procedure for the small-scale mod-
eling of buildings. International Journal of Energy Research
14(3):311–21.
Imbabi, M. 1990b. Computer validation of scale model tests for
building energy simulation. International Journal of Energy
Research 14(7):727–36.
ISO-1984. 1984. Moderate thermal environments— Determina-
tion of the PMV and PPD indices and specification of the
conditions for thermal comfort. Geneva, Switzerland: Inter-
national Organization for Standardization.
ISO-1995. 1995. Moderate thermal environments— Determina-
tion of the PMV and PPD indices and specification of the
conditions for thermal comfort. Geneva, Switzerland: Inter-
national Organization for Standardization.
ISO-2005. 2005. Moderate thermal environments— determina-
tion of the PMV and PPD indices and specification of the
conditions for thermal comfort. Geneva, Switzerland: Inter-
national Organization for Standardization.
Jang, J.-S., C.-T. Sun, and E. Mizutani. 1997. Neuro-Fuzzy and
Soft Computing: A Computational Approach to Learning and
Machine Intelligence. Upper Saddle River, NJ: Prentice Hall
PTR.
Jin, X., Z. Du, and X. Xiao. 2007. Energy evaluation of optimal
control strategies for central VWV chiller systems. Applied
Thermal Engineering 27(5–6):934–41.
Kamimura, K., Y. Hashimoto, T. Yamazaki, Y. Noda, and S.
Kurosu. 2002. A comparison of controller tuning methods
from a design viewpoint of the potential for energy savings.
ASHRAE Transactions: Research 108:155–65.
Karray, F., and C . De Silva. 2004. Soft Computing and Intelligent
Systems Design: Theory, Tools and Applications. Harlow,
England, UK: Pearson Educational Limited.
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 19
Kasabov, N. 2007. Evolving Connectionist Systems: The Knowl-
edge Engineering Approach, 2nd ed. London, UK: Springer-
Verlag.
Kasahara, M., T. Matsuba, Y. Hashimoto, I. Murasawa, A. Kim-
bara, K. Kaminura, and S. Kurosu. 1998. Optimal preview
control for HVAC system. Proceedings of the ASHRAE Win-
ter Meeting, St. Louis, MO, pp. 502–13.
Kaya, A. 1978. Modeling of an environmental space for optimal
control of energy use. Proceedings of the 7th IFAC Triennial
World Congress, Helsinki, Finland, pp. 327–34.
Kaya, A., C. Chen, S. Raina, and S. Alexander. 1982. Optimum
control policies to minimize energy use in HVAC systems.
ASHRAE Transactions 88(2):235–48.
Kelly, G. 1988. Control system simulation in North America.
Energy and Buildings 10:193–202.
Kintner-Meyer, M., and A. Emery. 1995. Optimal control of
an HVAC system using cold storage and building thermal
capacitance. Energy and Buildings 23:1931.
Kirk, D. 1970. Optimal Control Theory. Englewood Cliffs, NJ:
Prentice Hall.
Kokotovi
´
c, P., H. Khalil, and J. O’Reilly. 1986. Singular Pertur-
bation Methods in Control: Analysis and Design. London,
UK: Academic Press (republished as Volume 25 of Classics
in Applied Mathematics by the Society of Industrial and Ap-
plied Mathematics (SIAM), Philadelphia, PA, in 1999).
Kokotovi
´
c, P., and P. Sannuti. 1968. Singular perturbation method
for reducing model order in optimal control design. IEEE
Transactions on Automatic Control AC-13:377–84.
Komareji, M., J. Stoustrup, H. Rasmussen, N. Bidstrup, P. Svend-
sen, and F. Nielsen. 2007. Optimal set-point synthesis in
HVAC systems. Proceedings of the 2007 American Control
Conference, New York, NY, pp. 5076–81.
Komareji, M., J. Stoustrup, H. Rasmussen, N. Bidstrup, P. Svend-
sen, and F. Nielsen. 2009. Simplified optimal control in
HVAC systems. 2009 IEEE Control Applications (CCA) In-
telligent Control (ISIC), St. Petersburg, Russia, pp. 1033–8.
Konar, A. 2005. Computational Intelligence: Principles, Tech-
niques and Applications. Berlin, Germany: Springer-Verlag.
Kota, N., J. House, J. Arora, and T. Smith. 1996. Optimal control
of HVAC systems using DDP and NLP techniques. Optimal
Control Applications & Methods 17(1):7178.
Kristic, M., I. Kanellakopoulos, and P. Kokotovi
´
c. 1995. Nonlin-
ear and Adaptive Control Design. New York: John Wiley &
Sons.
Lahrech, R., P. Gruber, P. Riederer, P. Tessier, and J. Visier. 2002.
Development of a testing method for control HVAC systems
by emulation. Energy and Buildings 34:906–16.
Le, H., M. Zaheer-Uddin, V. Gourishankar, and R. Rink.
1987. Near-optimal control of a bilinear, solar-assisted heat
pump system. ASME Journal of Solar Energy Engineering
109(4):259–66.
Levenhagen, J. (Ed.): 1999. HVAC Control System Design Dia-
grams. New York: McGraw-Hill.
Levenhagen, J., and D. Spethmann. (Eds.). 1993. HVAC Controls
and Systems. New York: McGraw-Hill.
Lewis, F., L. Xie, and K. Popa. 2008. Optimal and Robust Esti-
mation: With an Introduction to Stochastic Control Theory,
2nd ed. Boca Raton, FL: Taylor & Francis Group.
Lim, D., B. Rasmussen, and D. Swaroop. 2009. Selecting PID
control gains for nonlinear HVAC&R systems. HVAC&R Re-
search 15(6):991–1019.
Liu, S., and X. He. 1994. A distributed-parameter-model ap-
proach to optimal comfort control in air conditioning sys-
tems. Proceedings of the American Control Conference, Bal-
timore, MD, pp. 3454–8.
Liu, S., and G. Henze. 2006a. Experimental analysis of simu-
lated reinforcement learning control for active and passive
building thermal storage inventory. Part 1 theoretical foun-
dation and Part 2: Results and analysis. Energy and Buildings
38(2):142–61.
Liu, S., and G. Henze. 2006b. Experimental analysis of simulated
reinforcement learning control for active and passive building
thermal storage inventory Part 2: Results and analysis. Energy
and Buildings 38(2):148–61.
Lukasse, L., J. van Maldegem, E. Dierkes, A.-J. Van Der Voort,
J. de Kramer-Cuppen, and G. Van Der Kolk. 2009. Opti-
mal control of indoor climate in agricultural storage facili-
ties for potatoes and onions. Control Engineering Practice
17(9):1044–52.
Ma, Z., and S. Wang. 2009. An optimal control strategy for
complex building central chilled water systems for prac-
tical and real-time applications. Building and Environment
44:1188–98.
Mathews, E.H., C.P. Botha, D.C. Arndt, and A. Malan. 2001.
HVAC control strategies to enhance comfort and minimize
energy usage. Energy and Buildings 33(8):853–63.
McDowall, R. 2009. Fundamentals of HVAC Control Systems.
New York: Elsevier.
Mohler, R. 1991a. Nonlinear Systems, Volume I, Dynamics and
Control. Englewood Cliffs, NJ: Prentice Hall.
Mohler, R. 1991b. Nonlinear Systems, Volume II, Applications
to Bilinear Control. Englewood Cliffs, NJ: Prentice Hall.
Mufti, I. 1970. Computational Methods in Optimal Control Prob-
lems. New York: Springer-Verlag.
Naidu, D. 1988. Vol. 34 of IEE Control Engineering Series, Singu-
lar per turbation methodology in control systems. Stevenage
Herts, UK: Peter Peregrinus Limited.
Naidu, D. 2002. Singular perturbations and time scales in con-
trol theory and applications: Overview. Dynamics of Con-
tinuous, Discrete and Impulsive Systems (DCDIS) Journal
9(2):233–278 (invited survey paper with 467 references in
a special issue on singularly perturbed dynamic systems in
control technology, edited by Z. Gajic).
Naidu, D. 2003. Optimal Control Systems. Boca Raton, FL: CRC
Press.
Naidu, D., and A. Rao. 1985. Vol. 1154 of Lecture Notes in
Mathematics, Singular perturbation analysis of discrete con-
trol systems. New York: Springer-Verlag.
Nesler, C. 1986. Adaptive control of thermal processes in build-
ings. IEEE Control Systems Magazine 6(4):9–13.
Nesler, C., and W. Stoecker. 1984. Selecting the proportional and
integral constants in the direct digital control o f discharge air
temperature. ASHRAE Transactions 90:834–45.
Newman, H., and M. Morris. 1994. Direct Digital Control of
Building Systems: Theory and Practice. New York: Wiley-
Interscience.
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
20 VOLUME 17, NUMBER 1, FEBRUARY 2011
Nguyen, H., N. Prasad, C. Walker, and E. Walker. 2003. AFirst
Course in Fuzzy and Neural Control. Boca Raton, FL: Chap-
man & Hall/CRC.
Nizet, J., J. Lecomte, and F. Litt. 1984. Optimal control ap-
plied to air conditioning in buildings. ASHRAE Transactions
90:587–600.
Noda, Y., T. Yamazaki, T. Matsuba, K. Kamimura, and S. Kurosu.
2003. Comparison in control performance between PID and
H controllers for HVAC control. ASHRAE Transactions
109:3–11.
O’Dwyer, A. 2003. Handbook of PI and PID Controller Tuning
Rules. Singapore: World Scientific, Singapore.
Ovaska, S., H. VanLandingham, and A. Kamiya. 2002. Fusion
of soft computing and hard computing in industrial applica-
tions: An overview. IEEE Transactions on Systems, Man, and
Cybeernetics—Part C: Applications and Reviews 32(2):72–9.
Ozawa, K., Y. Noda, T. Yamazaki, K. Kamimura, and S. Kurosu.
2003. A tuning method for PID controller using optimization
subject to constraints on control input. ASHRAE Transac-
tions: Research 109:79–88.
Pargfrieder, J., and H. Jorgl. 2002. An integrated control system
for optimizing the energy consumption and user comfort in
buildings. Proceedings of the 2002 IEEE International Sym-
posium on Computer Aided Control System Design, Glasgow,
Scotland, pp. 127–32.
Parsons, K. 1988. Protective clothing: Heat exchange and physi-
ological objectives. Ergonomics 31(7):991–1007.
Patel, R., and N. Munro. 1982. Multivariable System Theory and
Design. Oxford, UK: Pergamon Press.
Payne, F. 1984. Energy Management and Control Systems Hand-
book. Lilburn, GA: Fairmont Press.
Pontryagin, L., V. Boltyanskii, R. Gamkrelidze, and E.
Mishchenko. 1962. The Mathematical Theory of Optimal
Processes. New York: Wiley-Interscience (translated from
Russian).
Qi, Q., and S. Deng. 2009. Multivariable control of indoor
air temperature and humidity in a direct expansion (DX)
air conditioning (A/C) system. Building and Environment
44(8):1659–67.
Qi, Q., S. Deng, X. Xu, and M. Chan. 2010. Improving degree
of superheat control in a direct expansion (DX) air condi-
tioning (A/C) system. International Journal of Refrigeration
33(1):125–34.
Radermacher, R., and O. Abdelaziz. 2008. Optimization and
HVAC&R: Editorial. HVAC&R Research 14(6):817–8.
Rasmussen, B., and A. Alleyne. 2010. Gain scheduled control of
an air conditioning system using the youla parameterization.
IEEE Transactions on Control Systems Technology 1:1216–
1225.
Rehrl, J., M. Horn, and M. Reichhartinger. 2009. Elimination
of limit cycles in HVAC systems using the describing func-
tion method.Proceedings of the 48th IEEE Conference on
Decision and Control 2009 (held jointly with the 2009 28th
Chinese Control Conference (CDC/CCC 2009), Shanghai,
China, pp. 133–9.
Rink, R., V. Gourishankar, and M. Zaheer-Uddin. 1988. Optimal
control of heat pump/heat storage system with time-of-day
energy price incentives. Journal of Optimization Theory and
Applications 58(1):93–108.
Rogers, W., and C. Douglas. 1993. Control Systems for Heat-
ing, Ventilation, and Air Conditioning, 5th ed. New York:
Chapman & Hall.
Saboksayr, S., R. Patel, and M. Zaheer-Uddin. 1995. Energy-
efficient operation of HVAC systems using neural network-
based decentralized controllers. Proceedings of the American
Control Conference, Seattle, WA, pp. 4321–5.
Sane, H., C. Haugstetter, and S. Bortoff. 2006. Building HVAC
control systems—role of controls and optimization. Proceed-
ings of the 2006 American Control Conference, Minneapolis,
MN
, pp. 1121–6.
Semsar-Kazerooni, E., M. Yazdanpanah, and C. Lucas. 2008.
Nonlinear control and disturbance decoupling of HVAC sys-
tems using feedback linearization and backstepping with load
estimation. IEEE Transactions on Control Systems Technol-
ogy 16(5):918–29.
Shoureshi, R., and K. Rahmani. 1989. Intelligent control of build-
ing systems. Proceedings of the ASME Winter Meetingm San
Francisco, CA, pp. 7–15.
Singh, G., M. Zaheer-Uddin, and R. Patel. 2000. Adaptive control
of multivariable thermal processes in HVAC systems. Energy
Conversion & Management 41:1671–85.
Sinha, A. 2007. Linear Systems: Optimal and Robust Control.
Boca Raton, FL: Taylor & Francis Group.
Stengel, R. 1986. Stochastic Optimal Control: Theory and Ap-
plication. New York: Wiley-Interscience.
Sumathi, S., T. Hamsapriya, and P. Surekha. 2008. Evolutionary
Intelligence: An Introduction to Theory and Applications with
MATLAB
©
. Berlin, Germany: Springer-Verlag.
Sun, J., and A. Reddy. 2005. Optimal control of building
HVAC&R systems using complete simulation-based sequen-
tial quadratic programming (CSB-SQP). Building and Envi-
ronment 40(5):657–69.
Tao, G. (Ed.): 2003. Adaptive Control Design and Analysis.New
York: Wiley-Interscience.
Tashtoush, B., M. Molhim, and M. Al-Rousan. 2005. Dynamic
model of an HVAC system for control analysis. Energy
30(10):1729–45.
Tettamanzi, A., and M. Tomassini. 2001. Soft Computing: In-
tegrating Evolutionary, Neural, and Fuzzy Systems. Berlin,
Germany: Springer-Verlag.
Thosar, A., A. Patra, and S. Bhattacharyya. 2008. Feedback lin-
earization based control of a variable air volume air con-
ditioning system for cooling applications. ISA Transactions
47(3):339–49.
T
˘
ıgrek, T., S. Dasgupta, and T. Smith. 2002. Nonlinear op-
timal control of HVAC systems. Proceedings of The
15th IFAC Triennial World Congress, Barcelona, Spain,
pp. 1–6.
Townsend, M., D. Cherchas, and A. Abdelmessih. 1986. Optimal
control of a general environmental space. Transactions of
the ASME: Journal of Dynamic Systems, Measurement and
Control 108:330–9.
TRNSYS. 1996. A transient system simulation program—
volume 1 reference manual. Technical Report, Solar Energy
Laboratory, University of Wisconsin, Madison, WI.
Tseng, C., and J. Arora. 1989. Optimum design of systems for
dynamics and controls using sequential quadratic program-
ming. AIAA Journal 27(12):1793–800.
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
HVAC&R RESEARCH 21
Tsoukalas, L., and R. Uhrig. 1997. Fuzzy and Neural Approaches
in Engineering. New York: John Wiley & Sons, Inc..
Underwood, C. 1999. HVAC Control Systems: Modelling, Anal-
ysis and Design. London, UK: E & FN SPON.
Underwood, C. 2000a. Robust control of HVAC plant I: Mod-
elling. Proceedings of CIBSE A: Building Services Engineer-
ing Research and Technology 21:53–61.
Underwood, C., 2000b. Robust control of HVAC plant II: Con-
troller design. Proceedings of CIBSE A: Building Services
Engineering Research and Technology 21:63–71.
Visioli, A. 2006. Practical PID Control. Advances in Industrial
Control. New York: Springer.
Wang, S., and Z. Ma. 2008. Supervisory and optimal control
of building HVAC systems: A review. HVAC&R Research
14(1):3–32.
Wang, S., and X. Xu. 2002. A robust control strategy for com-
bining DCV control with economizer control. Energy Con-
version and Management 43(18):2569–88.
Wang, S., and X. Xu. 2004. Optimal and robust control of outdoor
ventilation airflow rate for improving energy efficiency and
IAQ. Building and Environment 39(7):763–73.
Xi, X.-C., A.-N. Poo, and S.-K. Chou. 2007. Support vector
regression model predictive control on a HVAC plant. Control
Engineering Practice 15(8):897–908.
Xu, J., P. Luh, W. Blankson, R. Jerdonek, and K. Khalil. 2005. An
optimization-based approach for facility energy management
with uncertainties. HVAC&R Research 11(2):215–37.
Xu, M., and S. Li. 2007. Practical generalized predictive control
with decentralized identification approach to HVAC systems.
Energy Conversion and Management 48(1):292–9.
Xu, M., S. Li, W. Jian Cai, and L. Lu. 2006. Effects of a GPC-PID
control strategy with hierarchical structure for a cooling coil
unit. Energy Conversion and Management 47(1):132–45.
Yaqub, M., and S. M. Zubair. 2001. Capacity control for refrig-
eration and airconditioning systems: A comparative study.
International Journal of Energy Research 123:92–9.
Yeh, T.-J., Y.-H. Chen, and J.-L. Lin. 2009. Control of air-
conditioning systems in heating mode to enhance transient
performance and steady-state efficiency. Energy and Build-
ings 41(12):1391–400.
Yu, F., and K. Chan. 2010. Economic benefits of optimal control
for water-cooled chiller systems serving hotels in a subtrop-
ical climate. Energy and Buildings 42(2):203–9.
Yuan, S., and R. Perez. 2006. Multiple-zone ventilation and tem-
perature control of a single-duct VAV system using model
predictive strategy. Energy and Buildings 38(10):1248–
61.
Zaheer-Uddin, M. 1989. Sub-optimal controller for a space heat-
ing system. ASHRAE Transactions: Research 95: 152–159.
Zaheer-Uddin, M. 1991. Digital control of a heating system for
hot water and space heating. Energy 16(10):1247–57.
Zaheer-Uddin, M. 1992. Decentralized control systems for
HVAC. ASHRAE Transactions: Research 98:114–26.
Zaheer-Uddin, M. 1993. Optimal, sub-optimal and adaptive con-
trol methods for the design of temperature controllers for in-
telligent building. Building and Environment 28(3):311–22.
Zaheer-Uddin, M., and R. Patel. 1995. Optimal tracking control
of multi-zone indoor environmental spaces. Transactions of
the ASME: Journal of Dynamic Systems, Measurement, and
Control 117(3):292–303.
Zaheer-Uddin, M., R. Patel, and A. Al-Assadi. 1993. Design of
decentralized robust controllers for multizone spaceheating
systems. IEEE Transactions on Control Systems Technology
1(4):246–61.
Zaheer-Uddin, M., and G. Zheng. 1994. A dynamic model of
multi-zone VAV system for control analysis. ASHRAE Trans-
actions: Research 100:219–29.
Zaheer-Uddin, M., and G. Zheng. 2000. Optimal control of time-
scheduled heating, ventilating and air conditioning processes
in buildings. Energy Conversion & Management 41:49–60.
Zaheer-Uddin, M., and G. Zheng. 2001. Multistage optimal op-
erating strategies for HVAC systems. ASHRAE Transactions:
Research 107:346–52.
Zelikin, M. 2000. Control Theory and Optimization I: Homo-
geneous Spaces and the Riccati Equation in the Calculus of
Variations. Berlin, Germany: Springer-Verlag.
Zhang, Y., and V. Hanby. 2008. Model-based control of re-
newable energy systems in buildings. HVAC&R Research
12(3a):739–60.
Zheng, G., and M. Zaheer-Uddin. 1999. Discharge air system:
modelling and optimal control. International Journal of En-
ergy Research 23(8):727–38.
Zhou, K., and J. Doyle. 1998. Essentials of Robust Control. Upper
Saddle River, NJ: Prentice Hall.
Ziegler, J., and N. Nichols. 1942. Optimum settings for automatic
controllers. Transactions of the ASME 65:433–44.
Downloaded By: [Naidu, D Subbaram] At: 16:16 21 February 2011
... Many works have been focused on continuously controlling individual components of an HVAC system. The classic method is the proportional-integralderivative (PID) control that minimizes the error signal through feedback loop using PID controllers [8]- [10]. Nevertheless, they fail to minimize the total energy consumption of the entire HVAC system in a coordinated manner. ...
... Incorporating (6c) into (6b), the definition of WB becomes a constraint of second-stage problem, which is given by the left-hand-side (LHS) of (10). ...
Preprint
Full-text available
The heating, ventilation and air condition (HVAC) system consumes the most energy in commercial buildings, consisting over 60% of total energy usage in the U.S. Flexible HVAC system setpoint scheduling could potentially save building energy costs. This paper first studies deterministic optimization, robust optimization, and stochastic optimization to minimize the daily operation cost with constraints of indoor air temperature comfort and mechanic operating requirement. Considering the uncertainties from ambient temperature, a Wasserstein metric-based distributionally robust optimization (DRO) method is proposed to enhance the robustness of the optimal schedule against the uncertainty of probabilistic prediction errors. The schedule is optimized under the worst-case distribution within an ambiguity set defined by the Wasserstein metric. The proposed DRO method is initially formulated as a two-stage problem and then reformulated into a tractable mixed-integer linear programming (MILP) form. The paper evaluates the feasibility and optimality of the optimized schedules for a real commercial building. The numerical results indicate that the costs of the proposed DRO method are up to 6.6% lower compared with conventional techniques of optimization under uncertainties. They also provide granular risk-benefit options for decision making in demand response programs.
... MPC -Hard controllers use optimum control, robust control, nonlinear control Model Predictive Control (MPC), and adaptive control to control systems [14], [13]. Hard controllers are typically very simple to interpret. ...
Book
Full-text available
The ever-increasing population and advancing municipal business demands for new building construction are often regarded as the most significant contributors to energy management. For many practical applications, machine learning is a promising method. In this perspective, we demonstrate machine learning development and application. One best way of decreasing energy usage in new buildings is to look at energy efficiency at an early stage in the design. Efficient power management and intelligent renovation can improve existing stock's energy performance. For optimal decision-making, all of these systems include an exact energy forecast. In recent years, machine learning (ML) and artificial intelligence (AI) technology have been introduced to predict building energy consumption and performance in specific terms. This article addresses hard, soft, and hybrid control systems, as well as machine learning techniques including artificial neural networks, clustering, and support vector machines which are frequently utilized in building energy performance forecasts and improvement. The hybrid approach suggestion significantly expands machine learning applications.
... MPC -Hard controllers use optimum control, robust control, nonlinear control Model Predictive Control (MPC), and adaptive control to control systems [14], [13]. Hard controllers are typically very simple to interpret. ...
Conference Paper
Full-text available
The ever-increasing population and advancing municipal business demands for new building construction are often regarded as the most significant contributors to energy management. For many practical applications, machine learning is a promising method. In this perspective, we demonstrate machine learning development and application. One best way of decreasing energy usage in new buildings is to look at energy efficiency at an early stage in the design. Efficient power management and intelligent renovation can improve existing stock's energy performance. For optimal decision-making, all of these systems include an exact energy forecast. In recent years, machine learning (ML) and artificial intelligence (AI) technology have been introduced to predict building energy consumption and performance in specific terms. This article addresses hard, soft, and hybrid control systems, as well as machine learning techniques including artificial neural networks, clustering, and support vector machines which are frequently utilized in building energy performance forecasts and improvement. The hybrid approach suggestion significantly expands machine learning applications.
... Whether it is soft controls or hard controls, as defined by Naidu and Rieger [7,8], or a combination of the two, the optimal solution must be adapted for the industrial system of the study. The exact type of control to implement should be selected after a thorough analysis of the system. ...
Article
Full-text available
In a context of energy abundance for industrial applications, industrial systems are exploited with minimal attention to their actual energy consumption requirements to meet the loads imposed on them. As a result, most of them are used at maximal capacity, regardless of the varying operational conditions. First, the paper studies pneumatic conveying systems and thoroughly reviews previously published work. Then, we overview simulations and operating data of the experimental parameters and their effects on the flow characteristics and transport efficiency. Finally, we summarize with a conclusion and some suggestions for further work. The primary goal of this study is to identify the parameters that influence the energy consumption of industrial dust collector systems. It is differentiated from previously published overviews by being concentrated on wood particles collection systems. The results will permit a better selection of an appropriate methodology or solution for reducing an industrial system’s power requirements and energy consumption through more precise control. The anticipated benefits are not only on power requirement and energy consumption but also in reducing greenhouse gas emissions. This aspect shows more impacts in regions that rely on electricity supplied by thermal power stations, especially those that use petrol or coal.
... Besides, in the practical engineering application, the timeperiodical flow rate oscillation is commonly used in an evaporator to assess the time-varying thermal load in an energy efficient and cost effective manner for variable frequency refrigeration and air conditioning systems [18][19][20] . These systems with changing refrigerant flow are exposed to varying thermal loads. ...
Article
In this study, the time-periodical flow rate oscillation for R-410A in horizontal narrow annuli is experimentally investigated in details for its heat transfer feature. The influence of average level of refrigerant mass flux G¯, saturation temperature Tsat, applied heat flux q, and amplitude ΔG/G¯ and period tp of oscillating flow rate on the evaporative heat transfer of R-410A in a horizontal narrow annulus is presented. For the experimentation, the gap of the employed annular pipe is 2 mm, whereas the ranges of G¯, Tsat, and q are between G¯ = 300 to 400 kg/(m²s), Tsat = 5°C, 10°C and 15°C, and q = 5 to 15 kW/m², respectively. Particularly, the focus is on the mean level of flow rate oscillations G¯, and the impact of ΔG/G¯ and tp on time periodic evaporation. The experimental results disclose that the time-averaged hr has minor reliance on the ΔG/G¯ and tp, but the instantaneous Tw and hr periodically oscillate with time in the oscillation frequency as the oscillating of the mass flux since the bubble formation and growth are highly dependent on the mass flow rate. Different from the single-phase heat transfer, the Tw decreases with the reducing refrigerant flow rate, indicating the better the evaporative heat transfer on the heating surface in the periodic evaporation. This trend is reversed for the single-phase convection. These unusual variations in the Tw and hr with the oscillating mass flux are ascribed to the change of the steam quality and liquid film thickness. Besides, the amplitude of the Tw oscillation changes non-monotonously in association with the vapor quality for a particular G¯, Tsat, q, and ΔG/G¯ and tp. Finally, the correlations are provided to explore the dependency of hr on the oscillating mass flux and vapor quality of refrigerant R-410A under the mass flux oscillation.
... Hybrid control methods are developed by integrating soft and hard control methods. Adaptive-fuzzy control and fuzzy-PID control are examples of hybrid control techniques [92,93]. In this essence, hard control methods are counted as model-based control techniques, and soft and hybrid control strategies are defined as learning-based techniques. ...
Thesis
Full-text available
Cyber-physical systems (CPSs) are composed of heterogeneous, and networked hardware and software components tightly integrated with physical elements [72]. Large-scale CPSs are composed of complex components, subject to uncertainties [89], as though their design and development is a challenging task. Achieving reliability and real-time adaptation to changing environments are some of the challenges involved in large-scale CPSs development [51]. Addressing these challenges requires deep insights into control theory and machine learning. This research presents a learning-based control approach for CPSs management, considering their requirements, specifications, and constraints. Model-based control approaches, such as model predictive control (MPC), are proven to be efficient in the management of CPSs [26]. MPC is a control technique that uses a prediction model to estimate future dynamics of the system and generate an optimal control sequence over a prediction horizon. The main benefit of MPC in CPSs management comes from its ability to take the predictions of system’s environmental conditions and disturbances into account [26]. In this dissertation, centralized and distributed MPC strategies are designed for the management of CPSs. They are implemented for the thermal management of a CPS case study, smart building. The control goals are optimizing system efficiency (lower thermal power consumption in the building), and improving users’ convenience (maintaining desired indoor thermal conditions in the building). Model-based control strategies are advantageous in the management of CPSs due to their ability to provide system robustness and stability. The performance of a model-based controller strongly depends on the accuracy of the model as a representation of the system dynamics [26]. Accurate modeling of large-scale CPSs is difficult (due to the existence of unmodeled dynamics and uncertainties in the modeling process); therefore, modelbased control approach is not practical for these systems [6]. By incorporating machine learning with model-based control strategies, we can address CPS modeling challenges while preserving the advantages of model-based control methods. In this dissertation, a learning-based modeling strategy incorporated with a model-based control approach is proposed to manage energy usage and maintain thermal, visual, and olfactory performance in buildings. Neural networks (NNs) are used to learn the building’s performance criteria, occupant-related parameters, environmental conditions, and operation costs. Control inputs are generated through the model-based predictive controller and based on the learned parameters, to achieve the desired performance. In contrast to the existing building control systems presented in the literature, the proposed management system integrates current and future information of occupants (convenience, comfort, activities), building energy trends, and environment conditions (environmental temperature, humidity, and light) into the control design. This data is synthesized and evaluated in each instance of decision-making process for managing building subsystems. Thus, the controller can learn complex dynamics and adapt to the changing environment, to achieve optimal performance while satisfying problem constraints. Furthermore, while many prior studies in the filed are focused on optimizing a single aspect of buildings (such as thermal management), and little attention is given to the simultaneous management of all building objectives, our proposed management system is developed considering all buildings’ physical models, environmental conditions, comfort specifications, and occupants’ preferences, and can be applied to various building management applications. The proposed control strategy is implemented to manage indoor conditions and energy consumption in a building, simulated in EnergyPlus software. In addition, for comparison purposes, we designed and simulated a baseline controller for the building under the same conditions.
... In general, some residential electrical appliances, such as detectors, should always be activated, and some, such as televisions, are uncontrollable because turning them on is entirely unpredictable. Furthermore, the different automation levels can be considered manual, semi-automated, and automated DR in this sector [179]. The use of generation units in the home sector has also been growing. ...
Article
Full-text available
The need to increase network efficiency, enhance reliability, and reduce environmental effects, as well as advances in communication infrastructures, have led to demand response (DR) becoming an essential part of smart grid operation. DR can provide power system operators with a range of flexible resources through different schemes. From the operational decision-making viewpoint, in practice, each scheme can affect the system performance differently. Therefore, categorizing different DR schemes based on their potential impacts on the power grid, operational targets, and economic incentives can embed a pragmatic and practical perspective into the selection approach. In order to provide such insights, this paper presents an extensive review of DR programs. A goal-oriented classification based on the type of market, reliability, power flexibility and the participants’ economic motivation is proposed for DR programs. The benefits and barriers based on new classes are presented. Every involved party, including the power system operator and participants, can utilize the proposed classification to select a proper plan in the DR-related ancillary service ecosystem. The various enabling technologies and practical strategies for the application of DR schemes in various sectors are reviewed. Following this, changes in the procedure of DR schemes in the smart community concept are studied. Finally, the direction of future research and development in DR is discussed and analyzed.
... Obviously, the control methods are effective approaches with minimal additional cost. With the decreased costs of Data Collecting System (DCS), Cloud Service (CS), and Remote Control (RC), etc. in recent years, the design and implementation of more complex control techniques have become feasible [4][5][6]. ...
Article
Full-text available
To enhance the energy performance of a central air-conditioning system, an effective control method for the chilled water system is always essential. However, it is a real challenge to distribute exact cooling energy to multiple terminal units in different floors via a complex chilled water network. To mitigate hydraulic imbalance in a complex chilled water system, many throttle valves and variable-speed pumps are installed, which are usually regulated by PID-based controllers. Due to the severe hydraulic coupling among the valves and pumps, the hydraulic oscillation phenomena often occur while using those feedback-based controllers. Based on a data-calibrated water distribution model which can accurately predict the hydraulic behaviors of a chilled water system, a new Model Predictive Control (MPC) method is proposed in this study. The proposed method is validated by a real-life chilled water system in a 22-floor hotel. By the proposed method, the valves and pumps can be regulated safely without any hydraulic oscillations. Simultaneously, the hydraulic imbalance among different floors is also eliminated, which can save 23.3% electricity consumption of the pumps.
... In this context, smart buildings (SB) were introduced to provide higher flexibility and sustainability, by managing and controlling the energy generation/consumption/storage in the building (European Commission, 2019b). It has been reported that the final energy consumption can be cut down significantly in an SB by using building automation technologies (Dar, Sartori, Georges, & Novakovic, 2014;Subbaram Naidu & Rieger, 2011a). Thus, the utilization of automated technologies is mandatory in the development of the SB. ...
Article
Full-text available
High share of energy consumption in buildings and subsequent increase in greenhouse gas emissions along with stricter legislations have motivated researchers to look for sustainable solutions in order to reduce energy consumption by using alternative renewable energy resources and improving the efficiency in this sector. Today, the smart building and socially resilient city concepts have been introduced where building automation technologies are implemented to manage and control the energy generation/consumption/storage. Building automation and control systems can be roughly classified into traditional and advanced control strategies. Traditional strategies are not a viable choice for more sophisticated features required in smart buildings. The main focus of this paper is to review advanced control strategies and their impact on buildings and technical systems with respect to energy/cost saving. These strategies should be predictive/responsive/adaptive against weather, user, grid and thermal mass. In this context, special attention is paid to model predictive control and adaptive control strategies. Although model predictive control is the most common type used in buildings, it is not well suited for systems consisting of uncertainties and unpredictable data. Thus, adaptive predictive control strategies are being developed to address these shortcomings. Despite great progress in this field, the quantified results of these strategies reported in literature showed a high level of inconsistency. This is due to the application of different control modes, various boundary conditions, hypotheses, fields of application, and type of energy consumption in different studies. Thus, this review assesses the implementations and configurations of advanced control solutions and highlights research gaps in this field that need further investigations.