Hindawi Publishing Corporation

Journal of Sensors
Volume 2015, Article ID 954159, 15 pages
http://dx.doi.org/10.1155/2015/954159

Research Article
Design and Implementation of Energy
Efficiency in HVAC Systems Based on Robust PID Control for
Industrial Applications

Muharrem Imal
Mechanical Engineering Department, Kahramanmaras Sutcu Imam University, 46100 Kahramanmaras, Turkey

Correspondence should be addressed to Muharrem Imal; [email protected]

Received 11 October 2014; Revised 16 March 2015; Accepted 23 March 2015

Academic Editor: Qing-An Zeng

Copyright © 2015 Muharrem Imal. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Energy efficiency in heating, ventilating, and air-conditioning (HVAC) systems is a primary concern in process projects, since
the energy consumption has the highest percentage in HVAC for all processes. Without sacrifice of thermal comfort, to reset
the suitable operating parameters, such as the humidity and air temperature, would have energy saving with immediate effect.
In this paper, the simulation-optimization approach described the effective energy efficiency for HVAC systems which are used in
industrial process. Due to the complex relationship of the HVAC system parameters, it is necessary to suggest optimum settings
for different operations in response to the dynamic cooling loads and changing weather conditions during a year. Proportional-
integral-derivative (PID) programming was developed which can effectively handle the discrete, nonlinear and highly constrained
optimization problems. Energy efficiency process has been made by controlling of alternative current (AC) drivers for ventilation
and exhaust fans, according to supplied air flow capacity and differential air pressure between supplied and exhaust air. Supervisory
controller software was developed by using programmable controllers and human machine interface (HMI) units. The new designed
HVAC control system would have a saving potential of about 40% as compared to the existing operational settings, without any
extra cost.

1. Introduction studies have been conducted on the quantification and com-

parison of control strategies in process industry for industrial
Several studies have been carried out in the past few years for plants.
minimizing the energy consumption related with the control The HVAC systems are composed of a large number of
of HVAC systems. HVAC units used in industrial processes subsystems; each of them may exhibit nonlinear characteris-
have high energy consumptions. It has been estimated that tics. The parameters of the systems change with weather, load,
40% of total energy is consumed in HVAC systems for fabric and process occupancy. In many situations, the exact model
plants. So, the energy efficiency is the most important role of the system cannot be obtained, but the approximate model
played in HVAC systems. Therefore, a well-designed energy can be derived. These complexities can be eliminated by
control system improves the energy efficiency [1–6]. using a well-designed control technique, operating strategies,
In recent years, many researchers have studied dynamic and optimum conditions of different HVAC equipment and
models of HVAC systems and their components using the- subsystems.
oretical or experimental approaches for buildings on the Modern control methods such as proportional-integral-
ground. However, ground buildings and semiopen built space derivative (PID) controllers are widely used in the process
have become major spaces of use for metropolitan residents industry; their effectiveness is often limited due to poor tun-
[7, 8]. While many control strategies have been suggested in ing; on the other hand they can be easily implemented due to
previous reports [9], to the best of our knowledge only a few low cost and being reliable in harsh field conditions. The PID
2 Journal of Sensors

control techniques can be used on model-free and model-

based control systems [5–8].
It is very useful to consider the simulation-optimization
approach in response to the problem of effective energy effi-
ciency. In the context of the plant services and HVAC systems,
a variety of transfer functions have been developed according
to the problems, like all year around energy consumption,
lifecycle cost, thermal comfort, plant scheduling, design
parameters, routing, and distribution. In addition, there are
increasing applications of the plant simulation models for
decision making purpose, and a multicriterion optimization Figure 1: General block diagram of the controlled HVAC system for
plants.
approach would be adopted for a number of changeable
system capacities instead of a single solution which would be
provided for decision making.
This paper describes a novel simulation-optimization humidification, temperature, humidity, air flow, and differ-
technique that was applied to devise a reset scheme of air flow ential air pressure sensors and controllers. Figure 1 shows
for air temperatures and differential air pressure for supply the general schematic diagram of the single-zone HVAC
and exhaust air of the HVAC control system in fabric man- controlled system for fabric plants. The ideal temperature and
ufacturing plants in industry. The designed control system humidity levels are generally 25∘ C and 70%, respectively, in
has been used in a fabric group which is one of the biggest fabric air condition systems.
fabric companies in Turkey. The whole system was firstly con- Cooling process is performed by utilizing the cold water
structed by LUWA Corporation. The mechanical system was used for humidification by mixing fresh air supplied from
not changed, but the complete control system was removed outdoor damper in the fabric air condition systems. Heating
and the new designed control system was placed instead of processes have been made by circulation of indoor air that
old conventional control system. After the designed control generated heat from machines, lights, and people, using
system was operated, the energy consumption decreased bypass dampers. According to desired temperature value, the
approximately 40% in the four unit plants. The energy effi- temperature controller directs the damper actuators which
are BELIMO trademarked proportional servomotors having
ciency has been carried out by the controlling of ventilation
the 40 Nm torque, 2 to 10 VDC control signal, and feedback
and exhaust driver inverters depending on differential air
signals. If the controlled zone temperature is to be raised, the
pressure between supply and exhaust air and air flow mea-
controller commands to the bypass dampers in the direction
surement for air changes per second. The controller system
of opening position, and conversely the fresh and exhaust
consists of the four programmable controller modules, four dampers in direction of closing position. The rate of opening
human machine interface (HMI) units, and computer sys- and closing values has to be the same values to equilibrate air
tems. Every HVAC system has been controlled depending on pressure between indoor and outdoor.
four parameters: humidifier pump controller for humidifi-
Humidity controlling has been performed by using
cation, fresh and bypass damper controllers for temperature
humidifier pumps and its driver inverters having the power of
control, exhaust fan controllers for stability of differential air
22 kW, spray water flow rate of 78 m3 /h, and nozzle pressure
pressure, and ventilation fan controllers for air flow capacity.
of 2.1 bar. The task of the humidifier pumps is to provide
These parameters were controlled by four loop PID con- necessary humidity by vaporizing water supplied from water
trollers implemented in programmable controllers. Supervi- pool. The vaporized water has been mixed by the supplied air
sory control software was developed to analyze and to control from ventilating fans and then it has been carried to indoor
the whole system from a HMI interface. In addition, energy space for humidification.
and mass balance equations are applied to derive a dynamic
Air flow controller unit controls the desired volume flow
model of single-zone HVAC system and a simulation example rate of supplied air which is determined from the air changes
has been proposed to investigate different PID control algo- per second for the need of controlled plant. The air flow
rithms on the derived mathematical model of the controlled capacity changes depending on the plant’s size, the number of
zone. The results from the proposed simulation-optimization machines, and the type of the plant such as fabric or yarn. Air
technique were used to evaluate the existing operational set- flow controller has been done by the ventilation fans with the
tings and suggest the possible saving potential from the newly power of 18.5 kW and the capability of 103.000 m3 /h air flow
optimized information. capacity. The desired air flow capacity is determined by the
expert operators.
2. Material and Method The air pressure difference between indoor and outdoor
has to be equilibrated in the product type HVAC systems. The
2.1. System Structure. The system has such main components differential pressure controller compensates the air pressure
as an air conditioned room of fabric plant, the fresh, bypass, based upon the differential pressure transmitter using the
and exhaust air dampers, damper actuators, ventilation and 37 kW exhaust fan drivers. Humidifier pumps, nozzles, fresh
exhaust air fans, humidifiers pumps, driver inverters for air dampers, exhaust air dampers, ventilating, and exhaust
Journal of Sensors 3

Figure 2: General view of humidification and air supplier units of controlled system.

The old conventional and the new designed controller

system have been shown in Figure 4 for a single zone of
HVAC system.
Four independent controllers were designed for each
HVAC unit as shown in Figure 5. These controllers were con-
nected by the Ethernet communication units. The whole sys-
tem has been controlled by the developed supervisory control
software from a single control point as shown in Figure 6.
Each unit has an operator panel and the set points can be
entered from this panel or HMI unit. The developed super-
visory control software is able to perform online data visual-
ization, data storage, data trends, and animation graphs and
Figure 3: The machine over and the plant air flow channels of the full control of complete system.
controlled zone.

2.2. Dynamic Control Model of HVAC System. Since the

HVAC zones have the structure of complex thermal system
fans of controlled system have been shown in Figure 2. The structure and the actual air-conditioning system is more
supplied air has been directed to different air channels as variable, it is very difficult to determine for its exact model,
shown in Figure 3. These channels are the machine over and but the approximate overall dynamic model of a single-zone
the plant channels, respectively. These channels are controlled HVAC system consisting of all its component models can be
by the different controller units as shown in Figure 1. derived. The whole plant model could reflect the HVAC sys-
The formerly designed controller by LUWA incorpora- tem with complete dynamic operation of different equipment
tion had only temperature and humidity controller. Ventila- and subsystems for the cooling demand throughout the yearly
tion and exhaust fans had operated at fixed rotation speed. operating hours. In the hourly control and operation of differ-
Energy consumed by the ventilation and exhaust fans corre- ent equipment and subsystems, there were four sets of input
sponds to 40% of total energy spent in a plant. To decrease data files: total and space loads, weather conditions, water
energy consumption, the driver inverters have been used for temperature, and that specifically for the component model
speed control of ventilation and exhaust fans drivers accord- of water-cooled air conditioning systems. The total cooling
ing to desired air flow capacity and differential air pressure load was used to determine the water temperature and water
in the designed control system. pump pressure in operation. In addition, enthalpy control
4 Journal of Sensors

(a) (b)

Figure 4: (a) The old conventional control system for single zone, (b) the new designed control system for single zone.

HMI

Switch

Ethernet (TCP/IP)

OP panel Plant 1 Plant 2 Plant 3 Plant 4

PLC
Humidifier pump driver

Humidifier pump driver

Humidifier pump driver

Humidifier pump driver

Air flow anemometer
Diff. pressure trans.

Air flow anemometer

Air flow anemometer

Air flow anemometer

Exhaust fan drivers

Exhaust fan drivers

Diff. pressure trans.

Damper actuators

Exhaust fan drivers

Diff. pressure trans.

Exhaust fan drivers

Diff. pressure trans.

Temperature sensor
Humidity sensor

Damper actuators
Temperature sensor
Humidity sensor

Damper actuators
Temperature sensor
Humidity sensor

Damper actuators
Temperature sensor
Humidity sensor
Ventilating fan drivers

Ventilating fan drivers

Ventilating fan drivers

Ventilating fan drivers

Figure 5: The structure of the designed control system.

was utilized to decide the implementation of free cooling 2.3. Zone Model. According to thermodynamic energy bal-
mode, which would call for the free cooling fan in operation, ance principles, a simple energy equation for a single-zone
with the same speed of the other operating fans. This dynamic HVAC system can be written in (1)
control and operation algorithm was essential to provide the
necessary system response to different loading and climatic
conditions, although this would increase the degree of the 𝑄input + 𝑄generation = 𝑄out + 𝑄accumulation + 𝑄consumption , (1)
nonlinear, constrained, and discrete nature of the system
performance, in particular to the energy efficiency of both
the water side and air side systems. The HVAC optimization where 𝑄input is heat from supply air input, 𝑄generation is gen-
problem hence requires an effective method that could handle erated heat from people, woven machines, light, and so forth,
the nonlinear, discrete, and highly constrained characteris- 𝑄out is heat from exhaust air, 𝑄accumulation is indoor space heat,
tics. and 𝑄consumption is dissipated heat from indoor to outdoor.
Journal of Sensors 5

where ℎ is convection heat transfer coefficient (W/m2 ⋅∘ C), 𝑘 is

conduction heat transfer coefficient (W/m⋅∘ C), 𝜀 is emissivity
constant, and 𝜎 is Boltzmann constant (W/m2 ⋅∘ C). The heat
dissipates into the zone mostly through the conduction heat,
so that the convection and radiation heats can be ignored. The
temperature zone equation can be written as

𝑑𝑇𝑖 V𝑧 𝑐pa ACH

V𝑧 𝑐pa 𝜌𝑎 = (𝑇0 − 𝑇𝑖 )
𝑑𝑡 3600 (6)
+ 𝑓sa 𝑐pa 𝜌𝑎 (𝑇sa − 𝑇𝑖 ) + 𝑄room .

Temperature controlling has been performed by the three

damper controller units in the fabric HVAC systems which
are fresh damper 𝐷1, exhaust damper 𝐷2, and return air
damper 𝐷3. Therefore, supply air volume flow rate is com-
posed of mixing the returned air from inner (exhaust air)
zone and outdoor air. This mixing equation can be written
from mass balance equation

Figure 6: The developed program interface for control and analysis

from one HMI unit. 𝑚𝑟 𝑐pa 𝑇𝑖 + 𝑚𝑜 𝑐pa 𝑇𝑜 = 𝑚𝑠 𝑐pa 𝑇sa , (7)

The heat equation is derived from the first law of thermo- where 𝑚𝑟 , 𝑚𝑜 , and 𝑚𝑠 are mass flow rates of the return,
dynamics and conservation of energy. Any change in energy outdoor, and supplied airs, respectively. This condition has to
Δ𝑄 is proportional to the change in temperature Δ𝑇. That is, be fulfilled by the controlling of these three damper positions.
If the position of 𝐷1 and 𝐷2 dampers is opened at the rate
Δ𝑄 = 𝑓𝑎 𝑐𝑝 𝜌𝑎 Δ𝑇, (2)
of 40%, the position of 𝐷3 damper must be opened at the
3 ∘
where 𝑓𝑎 (m /s) is volume flow rate of air, 𝑐𝑝 (kJ/kg⋅ C) is the rate of 60%. That is, 𝐷1 = 𝐷2 = (100 − 𝐷3). Humidification
is a mass transfer process of water vapor to atmospheric air,
constant pressure specific heat capacity of air, and 𝜌𝑎 (kg/m3 )
which results in an increase of water vapor in the mixture.
is the mass density of air. From the thermodynamic heat
Humidifier model is the same as temperature model and can
equations, the fresh and exhaust heat equations can be written
be derived from mass and energy balance principles as given
as
in [4]
𝑄input = 𝑓sa 𝑐pa 𝜌𝑎 𝑇sa ,
(3)
𝑄out = 𝑓sa 𝑐pa 𝜌𝑎 𝑇𝑖 , 𝑑𝑇ℎ
𝐶ℎ = 𝑓sa 𝑐pa (𝑇sa − 𝑇ℎ ) + 𝛼ℎ (𝑇𝑜 − 𝑇ℎ ) , (8)
3
where 𝑓sa (m /s) is volume flow rate of supplied air: 𝑑𝑡

𝑄generation = 𝑄room ,
where 𝐶ℎ is overall thermal capacitance of the humidifier, 𝑇ℎ
𝑑𝑇 is supply air temperature (in humidifier), and 𝛼ℎ is overall
𝑄accumulation = 𝜌𝑎 𝑐pa V𝑧 𝑖 , (4) transmittance area factor of the humidifier (kJ/s⋅∘ C). Simi-
𝑑𝑡
larly, the rate change of moisture content in the zone is equal
V𝑧 𝑐pa ACH to the difference between the vapor added to and removed
𝑄consumption = (𝑇𝑖 − 𝑇0 ) ,
3600 from the zone. Therefore, humidity zone equation can be
where 𝑄room (W) is heat gains from machines, light, people, written in
and so forth and V𝑧 (m3 ) is volume of the zone. ACH is the air
changes per hour. There are three modes of consumption heat 𝑑𝑊ℎ ℎ (𝑡)
from indoor to outdoor: conduction, convection, and radia- V𝑧ℎ = 𝑓sa (𝑊sa − 𝑊ℎ ) + , (9)
𝑑𝑡 𝜌𝑎
tion. The convection heat equation is defined as follows:
𝑄convection = ℎΔ𝑇,
where V𝑧ℎ is the volume of humidifier, 𝑊ℎ is supply air humid-
𝑄conduction = 𝑘Δ𝑇, (5) ity ratio (in humidifier) in kg/kg (dry air), 𝑊sa is humidity
ratio of supply air (to the humidifier) in kg/kg (dry air), and
𝑄radiation = 𝜀𝜎 (Δ𝑇)4 , ℎ(𝑡) is the rate of moisture air produced in air humidifier.
6 Journal of Sensors

In order to find the transfer function of the controlled Integral terms are
zone, the temperature zone (6)-(7) should be rearranged
according the parameters of 𝑇𝑖 and 𝑇𝑜 : 𝑘𝑐 𝑡
𝑈𝐼 (𝑡) = ∫ (𝑦 (𝜏) − 𝑦 (𝜏)) 𝑑𝑡. (17)
V𝑧 𝑐pa ACH 𝜏𝐼 𝑜 𝑠
𝑑𝑇
V𝑧 𝑐pa 𝜌𝑎 𝑖 + ( + 𝑓sa 𝑐pa 𝜌𝑎 ) 𝑇𝑖
𝑑𝑡 3600 Derivative terms are
(10)
V𝑧 𝑐pa ACH
= 𝑇0 + 𝑓sa 𝑐pa 𝜌𝑎 𝑇sa + 𝑄room , 𝑑 (𝑦𝑠 (𝑡) − 𝑦 (𝑡) )
3600 𝑈𝐷 (𝑡) = 𝑘𝑐 𝜏𝑑 , (18)
𝑑𝑡
𝑑𝑇𝑖 V𝑧 𝑐pa ACH
V𝑧 𝑐pa 𝜌𝑎 +( + 𝑓sa 𝑐pa 𝜌𝑎 ) 𝑇𝑖 where 𝑦𝑠 (𝑡) is the set point of the process output, 𝑦(𝑡) is the
𝑑𝑡 3600
process output, and 𝑈𝑃 (𝑡), 𝑈𝐼 (𝑡), and 𝑈𝐷(𝑡) are the controller
V𝑧 𝑐pa ACH 𝑚𝑟 𝑇𝑖 + 𝑚𝑜 𝑇0 (11) outputs of the proportional, integral, and derivative parts,
= 𝑇0 + 𝑓sa 𝑐pa 𝜌𝑎 ( ) respectively. The constants 𝑘𝑐 , 𝜏𝐼 , and 𝜏𝑑 in (18) are the
3600 𝑚𝑠
proportional gain, integral time, and derivative time, respec-
+ 𝑄room . tively. The total output of the PID controller is a summation
of the three individual outputs in (16)–(18); that is,
In order to simplify (11), 𝜏 and 𝑘 can be defined as follows:
𝑈 (𝑡) = 𝑈𝑝 (𝑡) + 𝑈𝐼 (𝑡) + 𝑈𝐷 (𝑡)
3600V𝑧 𝜌𝑎 𝑚𝑠
𝜏= ,
𝑚𝑠 (V𝑧 ACH + 3600𝑓sa 𝜌𝑎 ) − 3600𝑓sa 𝜌𝑎 𝑚𝑟 = 𝑘𝑐 (𝑦𝑠 (𝑡) − 𝑦 (𝑡))
(12)
𝑚𝑠 V𝑧 ACH + 3600𝑓sa 𝜌𝑎 𝑚𝑜 𝑘𝑐 𝑡
𝑘= . + ∫ (𝑦 (𝜏) − 𝑦 (𝜏)) 𝑑𝑡
(19)
𝑚𝑠 (V𝑧 ACH + 3600𝑓sa 𝜌𝑎 ) − 3600𝑓sa 𝜌𝑎 𝑚𝑟 𝜏𝐼 𝑜 𝑠
Finally, temperature zone (6) has been simplified as 𝑑 (𝑦𝑠 (𝑡) − 𝑦 (𝑡))
+ 𝑘𝑐 𝜏𝑑 .
𝑑𝑇𝑖 𝑑𝑡
𝜏 + 𝑇𝑖 = 𝑘𝑇𝑜 + 𝑄room . (13)
𝑑𝑡 The three constants are the proportional gain, integral time,
𝑄room is heat gains from machines, lights, people, and so forth and derivative time, respectively.
and is a constant. The Laplace transform of (13) is expressed 𝑘𝑐 , 𝜏𝐼 , and 𝜏𝑑 are usually called the parameters of the PID
as controller and should be determined under the dynamic con-
dition of the process. So, the parameters for the improvement
𝜏𝑠𝑇𝑖 (𝑠) − 𝑇𝑖 (0) + 𝑇𝑖 (𝑠) = 𝑘𝑇𝑜 (𝑠) . (14) of the control performance have to be tuned.
Because of the distance between supply air and the measure-
ment point of temperature, there is pure delay time (𝐿) in 2.5. Parameter Tuning. Up to now, numerous tuning meth-
the process. The plant heat transfer function is expressed as ods have been developed, including the Ziegler-Nichols (Z-
follows: N) method, internal model control (IMC), and the integral
of the time-weighted absolute value of the error tuning rule
𝑇𝑖 (𝑠) 𝑘𝑒−𝐿𝑠 for a first-order plus time delay model (ITAE). In this paper,
𝐺𝑝 (𝑠) = = . (15)
𝑇𝑜 (𝑠) 𝜏𝑠 + 1 all three methods were implemented to tune the controller in
the energy efficient HVAC control system. If the model of the
The system is considered as a first-order plus dead time system can be obtained, it is easy to design the PID controller
(FOPDT) structure as shown in (15), where 𝑘 (∘ C⋅s/kg) is the for it.
gain coefficient, 𝜏 is the time constant, and 𝐿 is the time delay The gain coefficients 𝑘𝑝 and 𝑇𝑖 can be achieved model
of the controlled HVAC system. based algorithms such as Ziegler-Nichols (Z-N) or variant
of Z-N tuning rules (Chien-Hrones-Reswick (CHR), Cohen-
2.4. Proportional-Integral-Derivative (PID) Control. The pro- Coon), Wang-Juang-Chan, Zhuang-Atherton optimum PID
portional-integral-derivative (PID) method is used to control controller design, and internal model control (IMC) [7–9, 16–
the air flow and differential pressure by manipulating the 19].
fan speed of the ventilation system. PID controllers are The response of an FOPDT model is shown in Figure 7.
widely employed in feedback control loops in most industrial A large number of systems can be approximately modeled by
process due to their simple structure, robustness, easy imple- (15). If the step response of the system can be experimentally
mentation, and good performance [10–15]. As indicated by measured, the parameters of the 𝑘, 𝐿, 𝑇, and a can be
the name, PID controllers consist of a proportional part, an determined as shown in Figure 7(a). In the same way, if
integral part, and a derivative part as follows [16]. the frequency response of the system can be experimentally
Proportional terms are observed, the crossover frequency 𝜔𝑐 and critical gain 𝑘𝑐 can
be obtained from the Nyquist plot as shown in Figure 7(b),
𝑈𝑝 (𝑡) = 𝑘𝑐 (𝑦𝑠 (𝑡) − 𝑦 (𝑡)) . (16) where 𝑎 = 𝐾𝐿/𝑇 and 𝑇𝑐 = 2𝜋/𝜔𝑐 .
Journal of Sensors 7

y(t) Imaginary

k
1
kc

Real

L T t

(a) (b)

Figure 7: (a) The time response of FOPDT system, (b) the Nyquist plot of FOPDT system.

Table 1: Z-N tuning formulas.

Step response Frequency response

Controller type
𝑘𝑝 𝑇𝑖 𝑇𝑑 𝑘𝑝 𝑇𝑖 𝑇𝑑
P 1/𝑎 0.5𝑘𝑐
PI 0.9/𝑎 𝐿/0.3 0.45𝑘𝑐 0.83𝑇𝑐
PID 1.2/𝑎 2𝐿 𝐿/2 0.6𝑘𝑐 0.5𝑇𝑐 0.125𝑇𝑐

Table 2: CHR method tuning formulas.

Set point regulation Disturbance rejection

Controller type
𝑘𝑝 𝑇𝑖 𝑇𝑑 𝑘𝑝 𝑇𝑖 𝑇𝑑
P 0.3/𝑎 0.3/𝑎
PI 0.35/𝑎 1.2𝑇 0.6/𝑎 4𝐿
PID 0.6/𝑎 𝑇 0.5𝐿 0.95/𝑎 2.4𝐿 0.42𝐿

The critical gain 𝑘𝑐 at the crossover frequency of 𝜔𝑐 is the Table 3: Cohen-Coon method tuning formulas.
first intersection of a Nyquist plot with the negative part of
the real axis as shown in Figure 7(b): Set point regulation
Cont. type
𝑘𝑝 𝑇𝑖 𝑇𝑑
𝑘 (cos 𝜔𝑐 𝐿 𝑖 − 𝜔𝑐 𝑇 sin 𝜔𝑐𝐿 ) 1 1 0.35𝜏
=− . (20) P (1 + )
1 + 𝜔𝑐 2 𝑇2 𝑘𝑐 𝑎 1−𝜏
0.9 0.92𝜏 3.3 − 3𝜏
The Z-N tuning method is a very useful empirical tuning PI (1 + ) 𝐿
𝑎 1−𝜏 1 + 1.2𝜏
formula. The PID controller parameters in Z-N tuning algo-
rithms can be found from Table 1. 1.35 0.18𝜏 2.5 − 2𝜏 0.37 − 0.37𝜏
PID (1 + ) 𝐿 𝐿
The Chien-Hrones-Reswick (CHR) method emphasizes 𝑎 1−𝜏 1 − 0.39𝜏 1 − 0.81𝜏
the set point regulation or disturbance rejection. The CHR
PID controller tuning formulas are summarized in Table 2
with the condition of no overshoot. criterion. If the 𝑘, 𝐿, and 𝑇 parameters of the plant model are
Another Z-N type tuning algorithm is the Cohen-Coon known, the controller parameters are given by
tuning formula. In this algorithm time constant 𝜏 is calculated (0.7303 + (0.5307𝑇/𝐿)) (𝑇 + 0.5𝐿)
as 𝜏 = 𝐿/(𝐿 + 𝑇) and PID parameters are tuned from Table 3. 𝑘𝑝 = ,
𝑘 (𝐿 + 𝑇)
Wang-Juang-Chan tuning method is based on an opti-
mum ITAE (the integral of time multiplied by absolute error) 𝑇𝑖 = 𝑇 + 0.5𝐿,
8 Journal of Sensors

Table 4: Zhuang-Atherton method optimum tuning formulas.

𝐿/𝑇 0.1–1 1.1–2

Criterion ISE ISTE IST2 E ISE ISTE IST2 E
𝑎1 1.048 1.042 0.968 1.154 1.142 1.061
𝑏1 −0.897 −0.897 −0.904 −0.567 −0.579 −0.583
𝑎2 1.195 0.987 0.977 1.047 0.919 0.892
𝑏2 −0.368 −0.238 −0.253 −0.220 −0.172 −0.165
𝑎3 0.489 0.385 0.316 0.490 0.384 0.315
𝑏3 0.888 0.906 0.892 0.708 0.839 0.832

0.5𝐿𝑇 3. Simulation Method

𝑇𝑑 = .
𝑇 + 0.5𝐿 (21)
The objectives of this study are to assess the energy efficient
control of HVAC system and to determine an optimal control
Zhuang and Atherton proposed the optimum PID controller strategy and its tuning parameters in order to maintain an
design for various criteria. This method uses the integrated efficient temperature, humidity, and pressure condition in
square error (ISE) function an industrial fabric space. The first step, the temperature,
∞
humidity, and pressure control system is implemented using
2
𝐽𝑛 (𝜃) = ∫ [𝑡𝑛 𝑒 (𝜃, 𝑡)] 𝑑𝑡, (22) Matlab/Simulink. In the second step, four temperature con-
0 trol structures of the system are designed Z-N based tuning
where 𝑒(𝜃, 𝑡) ≜ 𝑟(𝑡) − 𝑦(𝑡, 𝜃) is the error signal of the closed algorithm model. The Z-N based tuning algorithm model
loop system and 𝜃 ≜ [𝑘𝑝 , 𝑇𝑖 , 𝑇𝑑 ]𝑇 is a vector containing establishes the kinetics of a process using a mathematical
the controller parameters. For the optimum PID controller expression. One such expression is a transfer function that is
parameters can be represented as the ratio of the Laplace transform of the process output and
process input.
𝑎1 𝐿 𝑏1 Totally four of the plants having the area of 22.000 m3
𝑘𝑝 = ( ) ,
𝑘 𝑇 have been controlled by using PI controller. The parameters
𝑇 of the biggest zone are as follows.
𝑇𝑖 = , (23) The volume of the zone is V𝑧 = 60 ∗ 20.5 ∗ 4.5 = 5535 m3 ,
𝑎2 + 𝑏2 (𝐿/𝑇)
the volume flow rate of supplied air is 𝑓sa = 18.05 m3 /s, the
𝐿 𝑏3 air density is 𝜌𝑎 = 1.25 kg/m3 , the air specific heat is 𝑐pa =
𝑇𝑑 = 𝑎3 𝑇 ( ) , 1.00 kJ/kg⋅∘ C, air change per hour is ACH = 9, mass flow rates
𝑇
of the return air is 𝑚𝑟 = 8.66 kg/s, mass flow rates of the out-
where (𝑎, 𝑏) pairs are found from Table 4 according to the
door air is 𝑚𝑜 = 5.78 kg/s, and mass flow rates of the supplied
integral squared error (ISE) criterion, integral squared time-
air is 𝑚𝑠 = 14.44 kg/s. Substituting these parameters to the
weighted error (ISTE) criterion, and the integral squared
calculation of 𝜏 and 𝑘 constants, the transfer function of the
time-squared weighted error (IST2 E) criterion.
plant is determined as follows:
The internal model control (IMC) algorithm is based on
the fact that an accurate model of the process can lead to the 1𝑒−20𝑠
design of a robust controller both in terms of stability and 𝐺𝑝 (𝑠) = . (25)
performance. 300𝑠 + 1
The first step in using the IMC (lambda) tuning corre- PI controller has been applied to the system having the feed
lations is to compute the closed loop time constant 𝜏𝑐 . The forward transfer function of (25) as shown in Figure 8:
closed loop time constant 𝜏𝑐 is the single adjustable IMC
tuning parameter that allows the desired levels of perfor- 1 1𝑒−20𝑠
mance and robust stability to be specified. IMC tuning PID 𝐺 (𝑠) = 𝐺𝑝 (𝑠) ⋅ 𝐺𝑐 (𝑠) = 𝐾𝑝 (1 + )⋅ . (26)
𝑇𝑖 𝑠 300𝑠 + 1
parameters are chosen as the following equations:
𝐿 + 2𝑇 Figure 9 shows the simulation of four different PID tuning
𝑘𝑝 = , methods for the unit step responses according to PID param-
2𝑘 (𝜏𝑐 + 𝐿)
eters given in Table 5.
𝐿 Z-N and Cohen-Coon method has fast response time,
𝑇𝑖 = 𝑇 + , (24)
but approximately 20% overshoot has occurred as shown in
2
Figure 9. Wang-Juang-Chan and IMC tuning method have
𝑇𝐿 not seen overshoot, but rise time is more slowly comparing
𝑇𝑑 = ,
2𝑇 + 𝐿 to other methods. It is well known that, the rise time depends
where 𝑇𝑐 is a design parameter that affects the tradeoff on the proportional constant but overshoot depends on the
between performance and robustness. integral time constant in PID controller.
Journal of Sensors 9

Table 5: Different types of PI controller parameters. Step response

1.4
Tuning methods 𝑘𝑝 𝑇𝑖 1.2
1

Amplitude
Z-N 13.5 66.6 0.8
Cohen-Coon 14.3 57.9 0.6
0.4
Wang-Juang-Chan 8.4 310 0.2
IMC 7.5 310 0
0 50 100 150 200 250 300 350
Time (s)

Ziegler-Nichols Wang-Juang-Chan
R(s) E(s) U(s) Y(s) Cohen-Coon
+ kp ∑ Gp (s) IMC
−
Figure 9: Comparison of four PID controller methods.

1
Ti s Gc (s)
31∘ C in 12 minutes at the heating position. At same condition,
Figure 8: PI controller diagram for controlled parameters. the temperature decreased from 31∘ C to 28∘ C in only 5
minutes at the cooling position. The heating process was
performed by return air circulation, so the heating process
happened more slowly. The cooling process was performed
3.1. Experimental Results of the Real System Controller. by the fresh air. Therefore, the system response depends on
Modern process control problems in the process industries the outdoor temperature as shown in Figure 12.
are dominated by nonlinear, time-varying behavior, distur- If the outdoor temperature is low, then the cooling
bances, and uncertainties. However, more than 90% of plants process takes place as quickly; on the other side, if the outdoor
are controlled by the well-established PID controllers in temperature is high, then the cooling process is more slowly.
industrial automation and process until today [8, 9]. The If the set point of the temperature is determined as 32∘ C while
control block diagram of the single-zone HVAC system is the outdoor temperature is 34∘ C, then the cooling processes
shown in Figure 8. Four types of PI controllers were designed take place very slowly and heating processes are more rapid as
to control temperature, humidity, air flow, and differential air shown in Figure 12.
pressure. Response of the control system depends on the out-
The cooling process utilizes the temperature of sprayed
door temperature and humidity. If the outdoor temperature is
water by mixing the supplied air from outdoor. Humidity
high, decreasing the temperature of the zone will be very slow
controller has been affected from the outdoor humidity ratio.
and limited. If the outdoor temperature is low, decreasing the
But indeed humidification performance depends generally
temperature is very quick. Humidity controller is altered by
on the capacity of pumps, the structure of control systems,
temperature controller. If the system operates by the returned
and units. Technical specification of the humidifier pumps
air, the increase in humidity becomes quickly. If the system
in designed system is: the flow rate is 78 m3 /h, pump head
runs at the fresh air position the humidity increment becomes
slowly. rate is 3 bar, spray water flow rate is 78 m3 /h, nozzle pressure
Totally 22 PI controllers were developed in a single is 2.1 bar, maximum rotating speed is 3000 rpm, and nozzle
programmable controller. These controller parameters were diameter of each unit is 6 mm.
adjusted by using Z-N tuning method. A software program Two experiments were done to test the performance of the
was developed for quickly tuning the PID parameters of the humidity controller. The first experiment has been performed
controllers. The Z-N based tuning algorithm of the developed at the condition of 36% outdoor humidity ratios. The desired
software is shown in Figure 10. The developed software is able set point humidity has been obtained in 15 minutes. The
to communicate with the programmable controller, selectable humidity of the zone rose from 62% to 70% in this time
of control type (auto/manual control), and able to store interval. Decreasing the humidity again from 70% to 62% was
control output data into database system and perform online carried out in 10 minutes as shown in Figure 13. The second
analysis graphically as shown in Figure 11. experiment has been done at the 28% outdoor humidity ratio.
The complete HVAC system can be controlled by a HMI The desired set point value has been reached in 18 minutes
unit. An operator can enter the temperature, humidity, air and the old humidity ratio has been reached in 8 minutes.
flow, and differential air pressure set points values from the Eventually, the humidity ratio is affected by outdoor humidity
software. After the set points are determined by the operators, ratio as shown in Figure 13.
the controllers hold the system parameters at the desired set Differential pressure controller has been carried out by
point values. The following experiments have been carried setting the speed of 37 kW exhaust fans according to mea-
out to observe the performance of the controlled system. sured value from differential air pressure transmitter. The
As stated early, the responses of the controller depend on differential pressure transmitter measures the difference
the outdoor climatic conditions. While the outdoor tempera- between indoor and outdoor air pressure. Figure 14 shows the
ture was 21∘ C, the system temperature increased from 28∘ C to air pressure deviation measured from differential air pressure
10 Journal of Sensors

(i) Set proportional band (PB) = 100%

(ii) Set integral effect = 0
(iii) Set differential effect = 0
(iv) Adjust the set point to a small value
(v) Wait, until the processes value is equal to set point value
(vi) Adjust the set point to desired value

Decrease the Increase the

proportional proportional
band (PB) Observe the process response
band (PB)

Tc
Process variable

Process variable

Process variable
t t t

(i) Determine kc and Tc

(ii) Calculate PID parameters from Table 1

Figure 10: Z-N tuning based PID controller algorithm.

3.2. Comparison of Real and Simulation Systems. The unit

step graphical responses of the derived approximate transfer
function of the system were shown in Figure 9 for temper-
ature controller. The designed real system temperature con-
troller responses were shown in Figure 12. The desired level of
the unit exchange of temperature has been obtained approx-
imately within 180 seconds depending on the outdoor tem-
perature by the designed controller. Model-based controller
reached the same unit temperature level within 150–200 sec-
onds depending on the PID parameters as shown in Figure 9.
So, the results of the derived model and the real application
Figure 11: The developed Z-N tuning based PID tuning program. system are approximately corresponding.

4. Energy Saved in the Systems

transmitter during nine hours. The difference of the air pres-
sure results from the event of clogging of exhaust filter and In the old conventional control system, exhaust and ventilat-
the difference air flow capacity between supply air from ven- ing fans had been driven at constant speed. The differential
tilation fans and exhaust air from exhaust fans. If the exhaust pressure sensors, anemometers, and driver inverters for
filter is clogged, the differential pressure transmitter warns exhaust and ventilating fans have been added to the new
operators and then the operator solves the filter problem and designed control system. In this manner, the exhaust and
the filter clogging process does not affect differential pressure ventilating fans’ speed have been controlled depending on
controller for a long time. That is, the differential pressure the requirements of the plant conditions. The expert operator
controller generally operates according to difference between only enters the desired parameters by the operator panel on
air capacity of supply air and exhaust air. the supervisory control system. Before the design of control
Air flow capacity controller operates depending on the system, the overall energy consumption was measured by
requirement of air changes of plant per second (m3 /s). The using energy analyzer to determine the rate of energy effi-
air supplies using ventilation fans by adjusting its rotational ciency of the designed four HVAC systems through Tables 6
speeds. Air flow capacity controller has the fast responses as to 9 which shows the amount of energy efficiency data for
shown in Figure 15. The desired maximum air capacity has each controlled plant. Table 6 shows the energy consump-
been reached during only 10 seconds. tions for plant 1, before and after the controller design. The
Journal of Sensors 11

Temperature
Temperature control Temperature control
Temperature

31 37
30 35

(∘ C)
(∘ C)

29 33
28
27 31
0 10 20 30 40 50 60 70 80 90 100 110 120 130 0 10 20 30 40 50 60 70 80 90 100 110
(min) (min)
(a) (b)
∘
Figure 12: The temperature controller response of the outdoor condition at 21 C (a) and the temperature controller response of the outdoor
condition at 34∘ C (b).

Humidity control Humidity control

72 72
Humidity

Humidity
68 68
(%)

(%)
64 64
60 60
5 15 25 35 45 55 65 75 85 95 5 15 25 35 45 55 65 75 85 95 105
(min) (min)
(a) (b)

Figure 13: The humidity controller response of the outdoor condition at 36% humidity (a) and the humidity controller response of the outdoor
condition at 28% humidity (b).

Table 6: Energy saving in plant 1.

Power consumption in old conventional system Power consumption after the designed control system
Date Hour kW Date Hour kW
14.11.2013 10:30:00 108.21 05.01.2014 19:00:00 59.08
14.11.2013 11:00:00 109.53 05.01.2014 19:15:00 67.05
14.11.2013 11:30:00 108.87 05.01.2014 19:30:00 65.72
14.11.2013 12:00:00 109.2 05.01.2014 19:45:00 65.06
14.11.2013 12:30:00 108.87 05.01.2014 20:00:00 60.74
14.11.2013 13:00:00 108.21 05.01.2014 20:15:00 65.72
14.11.2013 13:30:00 108.87 05.01.2014 20:30:00 65.72
14.11.2013 14:00:00 108.54 05.01.2014 20:45:00 65.39
14.11.2013 14:30:00 108.87 05.01.2014 21:00:00 65.72
14.11.2013 15:00:00 108.54 05.01.2014 21:15:00 65.06
14.11.2013 15:30:00 108.54 05.01.2014 21:30:00 67.05
14.11.2013 16:00:00 109.2 05.01.2014 21:45:00 66.05
14.11.2013 16:30:00 108.87 05.01.2014 22:00:00 63.06
14.11.2013 17:00:00 109.53 05.01.2014 22:15:00 65.72
14.11.2013 17:30:00 108.87 05.01.2014 22:30:00 65.39
14.11.2013 18:00:00 109.53 05.01.2014 22:45:00 66.38
14.11.2013 18:30:00 109.2 05.01.2014 23:00:00 67.05
14.11.2013 19:00:00 105.55 05.01.2014 23:15:00 66.38
14.11.2013 19:30:00 105.88 05.01.2014 23:30:00 65.39
14.11.2013 20:00:00 106.88 05.01.2014 23:45:00 66.05
14.11.2013 20:30:00 109.2 05.01.2014 00:00:00 66.05
14.11.2013 21:00:00 108.87 05.01.2014 00:15:00 65.39
14.11.2013 21:30:00 108.87 05.01.2014 00:30:00 65.72
Average 108.55 Average 65.26
Average power consumption: old = 108.55, new = 65.26, and profit = 39.88%.
12 Journal of Sensors

Table 7: Energy saving in plant 2.

Power consumption in old conventional system Power consumption after the designed control system
Date Hour kW Date Hour kW
13.07.2013 10:00:00 107.7 06.01.2014 10:30:00 79.99
13.07.2013 11:00:00 106.7 06.01.2014 10:45:00 80.32
13.07.2013 12:00:00 104.2 06.01.2014 11:00:00 79.66
13.07.2013 13:00:00 107.5 06.01.2014 11:15:00 82.65
13.07.2013 14:00:00 107.4 06.01.2014 11:30:00 76.67
13.07.2013 15:00:00 107,7 06.01.2014 11:45:00 75.35
13.07.2013 16:00:00 108.3 06.01.2014 12:00:00 75.01
13.07.2013 17:00:00 108 06.01.2014 12:15:00 71.36
13.07.2013 18:00:00 108.4 06.01.2014 12:30:00 72.03
13.07.2013 19:00:00 108.2 06.01.2014 12:45:00 71.69
13.07.2013 20:00:00 107.9 06.01.2014 13:00:00 71.69
13.07.2013 21:00:00 107.9 06.01.2014 13:15:00 72.03
13.07.2013 22:00:00 108.4 06.01.2014 13:30:00 74.02
13.07.2013 23:00:00 107.8 06.01.2014 13:45:00 74.68
13.07.2013 00:00:00 107.9 06.01.2014 14:00:00 74.68
13.07.2013 01:00:00 107.2 06.01.2014 14:15:00 75.68
13.07.2013 02:00:00 106.7 06.01.2014 14:30:00 75.01
13.07.2013 03:00:00 106.3 06.01.2014 14:45:00 76.34
13.07.2013 04:00:00 106.5 06.01.2014 15:00:00 74.68
13.07.2013 05:00:00 106.8 06.01.2014 15:15:00 75.35
13.07.2013 06:00:00 106.8 06.01.2014 15:30:00 75.35
13.07.2013 07:00:00 106.2 06.01.2014 15:45:00 75.01
13.07.2013 08:00:00 106.2 06.01.2014 16:00:00 75.01
Average 107.25 Average 75.4
Average power consumption: old = 107.25, new = 75.40, and profit = 29.69%.

15
Differential pressure of 32 kW has been obtained for plant 2, 41 kW in plant 3,
Pressure

10
(mbar)

5
and 24 kW in plant 4, respectively. The money spent for plant
0 1, plant 2, and plant 3 was approximately 39450 USD and
−5
−10 8331 USD for plant 4. The total energy efficiency has been
20 60 100 140 180 220 260 300 340 380 420 460 500
measured as 140 kW. The total cost of 47781 USD has been
(min)
spent of the whole system.
Figure 14: The differential air pressure of supply and exhaust air in
the period of 9 hours.
5. Conclusion
Air flow
20
This paper describes a novel simulation-optimization tech-
Air flow
(m3 /s)

15 nique that was applied to devise a reset scheme of air flow for
10
5 air temperatures and differential air pressure for supply and
0
10 40 70 100 130 160 190 220 250 280 310 340 370 400 exhaust air of the HVAC control system in fabric manufactur-
(s) ing plants in industry. Due to the complex relationship of the
HVAC system parameters, it is necessary to suggest optimum
Figure 15: The air flow capacity controller response. settings for different operations in response to the dynamic
cooling loads and changing weather conditions during a year.
Proportional-integral-derivative (PID) programming was
energy consumption in old conventional control system for developed which can effectively handle the discrete, nonlin-
plant 1 was measured as the average of 108 kW. For the energy ear, and highly constrained optimization problems. To solve
consumption in the new designed system, this value was this problem, a HVAC control system has been successfully
decreased to average of 65 kW. The energy efficiency has been designed and the dynamic model of the controlled zone has
obtained as 43 kW per hour for plant 1. The energy efficiency been derived accurately for an integrated big plant in Turkey.
Journal of Sensors 13

Table 8: Energy saving in plant 3.

Power consumption in old conventional system Power consumption after the designed control system
Date Hour kW Date Hour kW
16.11.2013 10:30:00 85.64 06.01.2014 18:00:00 43.81
16.11.2013 11:00:00 85.3 06.01.2014 18:30:00 43.81
16.11.2013 11:30:00 85.64 06.01.2014 19:00:00 43.48
16.11.2013 12:00:00 84.97 06.01.2014 19:30:00 42.82
16.11.2013 12:30:00 84.97 06.01.2014 20:00:00 43.15
16.11.2013 13:00:00 84.97 06.01.2014 20:30:00 48.13
16.11.2013 13:30:00 84.97 06.01.2014 21:00:00 43.48
16.11.2013 14:00:00 83.64 06.01.2014 21:30:00 43.81
16.11.2013 14:30:00 84.31 06.01.2014 22:00:00 40.83
16.11.2013 15:00:00 83.98 06.01.2014 22:30:00 42.15
16.11.2013 15:30:00 83.64 06.01.2014 23:00:00 49.12
16.11.2013 16:00:00 83.64 06.01.2014 23:30:00 43.81
16.11.2013 16:30:00 83.64 06.01.2014 00:00:00 43.15
16.11.2013 17:00:00 83.64 06.01.2014 00:30:00 44.81
16.11.2013 17:30:00 83.64 06.01.2014 01:00:00 39.5
16.11.2013 18:00:00 83.64 06.01.2014 01:30:00 43.81
16.11.2013 18:30:00 83.31 06.01.2014 02:00:00 39.5
16.11.2013 19:00:00 83.98 06.01.2014 02:30:00 44.15
16.11.2013 19:30:00 83.64 06.01.2014 03:00:00 36.51
16.11.2013 20:00:00 83.64 06.01.2014 03:30:00 38.83
16.11.2013 20:30:00 83.98 06.01.2014 04:00:00 41.16
16.11.2013 21:00:00 84.31 06.01.2014 04:30:00 42.15
16.11.2009 21:30:00 83.98 06.01.2014 05:00:00 38.83
Average 84.22 Average 42.64
Average power consumption: old = 84.22, new = 42.64, and profit = 49.37%.

The simulation and the real controller responses have been Nomenclature
observed, respectively. When the results have been compared,
either of the system responses has been obtained similarly. 𝑓𝑎 : Volume flow rate of air (m3 /s)
Supervisory control software was developed to control overall 𝑐𝑝 : Specific heat capacity of air (kJ/kg⋅∘ C)
system using a host computer and programmable controllers. 𝜌𝑎 : Mass density of air (kg/m3 )
All of the control units could communicate with the host V𝑧 : Volume of the zone (m3 )
computer by the MODBUS TCP/IP Ethernet protocol. Thus, ACH: Air changes per hour
the system parameters could be observed and controlled on ℎ: Convection heat transfer coefficient,
the single central point by the host computer and an operator. (W/m2 ⋅∘ C)
The approximately 40% energy efficiency was gained by ℎ𝑤 : Rate of moisture air produced in air
the designed control system. This energy efficiency amount humidifier
was clearly verified by measuring of energy consumption of 𝑘ℎ : Conduction heat transfer coefficient
each plant as shown in Table 6 through Table 9. Finally the (W/m⋅∘ C)
following beneficial results have been obtained: 𝑘: Gain coefficient (∘ C⋅s/kg)
𝜀: Emissivity constant
(i) total money spent to control system = 47781 $; 𝜎: Boltzmann constant (W/m2 ⋅∘ C)
𝑚: Mass flow rates
(ii) total energy efficiency per hour = 140 kW;
𝐶ℎ : Overall thermal capacitance of the
(iii) 1 kW electricity cost = 0.087 $/kWh; humidifier
𝑇ℎ : Supply air temperature (∘ C)
(iv) energy efficiency per month = 140 kWh ∗ 24 hours ∗ 𝛼ℎ : Overall transmittance area factor of the
30 days ∗ 0.087 $/kWh = 8770 $; humidifier (kJ/s⋅∘ C)
(v) depreciation time of the system = 47781/8770 = 5.44 V𝑧ℎ : Volume of humidifier
months. 𝑤ℎ : Supply air humidity ratio (kg/kg)
14 Journal of Sensors

Table 9: Energy saving in plant 4.

Power consumption in old conventional system Power consumption after the designed control system
Date Hour kW Date Hour kW
17.11.2013 10:30:00 46.8 07.01.2014 11:00:00 22.24
17.11.2013 11:00:00 46.8 07.01.2014 11:15:00 22.24
17.11.2013 11:30:00 46.47 07.01.2014 11:30:00 22.24
17.11.2013 12:00:00 46.47 07.01.2014 11:45:00 22.24
17.11.2013 12:30:00 46.47 07.01.2014 12:00:00 20.25
17.11.2013 13:00:00 47.8 07.01.2014 12:15:00 19.92
17.11.2013 13:30:00 44.81 07.01.2014 12:30:00 19.92
17.11.2013 14:00:00 46.47 07.01.2014 12:45:00 22.24
17.11.2013 14:30:00 46.47 07.01.2014 13:00:00 22.24
17.11.2013 15:00:00 45.81 07.01.2014 13:15:00 22.24
17.11.2013 15:30:00 45.81 07.01.2014 13:30:00 22.24
17.11.2013 16:00:00 46.47 07.01.2014 13:45:00 22.24
17.11.2013 16:30:00 46.14 07.01.2014 14:00:00 22.24
17.11.2013 17:00:00 45.81 07.01.2014 14:15:00 22.24
17.11.2013 17:30:00 43.81 07.01.2014 14:30:00 23.9
17.11.2013 18:00:00 45.81 07.01.2014 14:45:00 22.24
17.11.2013 18:30:00 45.14 07.01.2014 15:00:00 22.24
17.11.2013 19:00:00 45.14 07.01.2014 15:15:00 22.24
17.11.2013 19:30:00 45.14 07.01.2014 15:30:00 22.24
17.11.2013 20:00:00 45.47 07.01.2014 15:45:00 19.92
17.11.2013 20:30:00 45.47 07.01.2014 16:00:00 20.25
17.11.2013 21:00:00 43.48 07.01.2014 16:15:00 21.78
17.11.2013 21:30:00 45.14 07.01.2014 16:30:00 21.78
Average 45.79 Average 21.8
Average power consumption: old = 45.79, new = 21.80, and profit = 52.39%.

𝑤sa : Humidity ratio of supply air (kg/kg) [5] J. Teeter and M.-Y. Chow, “Application of functional link neural
𝜏: Time constant network to hvac thermal dynamic system identification,” IEEE
𝐿: Time delay. Transactions on Industrial Electronics, vol. 45, no. 1, pp. 170–176,
1998.
[6] J. Wang, C. Zhang, and Y. Jing, “Fuzzy immune self-tuning pid
Conflict of Interests control of HVAC system,” in Proceedings of the IEEE Interna-
The author declares that there is no conflict of interests tional Conference on Mechatronics and Automation (ICMA ’08),
pp. 678–683, Takamatsu, Japan, August 2008.
regarding the publication of this paper.
[7] K. F. Fong, V. I. Hanby, and T. T. Chow, “HVAC system opti-
mization for energy management by evolutionary program-
References ming,” Energy and Buildings, vol. 38, no. 3, pp. 220–231, 2006.
[8] A. P. Wemhoff, “Application of optimization techniques on
[1] G. P. Henze, D. E. Kalz, S. Liu, and C. Felsmann, “Experimental
lumped HVAC models for energy conservation,” Energy and
analysis of model-based predictive optimal control for active
Buildings, vol. 42, no. 12, pp. 2445–2451, 2010.
and passive building thermal storage inventory,” HVAC&R
Research, vol. 11, no. 2, pp. 189–213, 2005. [9] S. Lee, M. J. Kim, S. H. Pyo, J. T. Kim, and C. K. Yoo, “Evaluation
[2] W. Z. Huang, M. Zaheeruddin, and S. H. Cho, “Dynamic simu- of an optimal ventilation IAQ control strategy using control
lation of energy management control functions for HVAC sys- performance assessment and energy demand,” Energy and
tems in buildings,” Energy Conversion and Management, vol. 47, Buildings, 2014.
no. 7-8, pp. 926–943, 2006. [10] R. C. Panda, “Synthesis of PID controller for unstable and inte-
[3] S. Soyguder and H. Alli, “An expert system for the humidity grating processes,” Chemical Engineering Science, vol. 64, no. 12,
and temperature control in HVAC systems using ANFIS and pp. 2807–2816, 2009.
optimization with Fuzzy Modeling Approach,” Energy and [11] S. Skogestad, “Simple analytic rules for model reduction and
Buildings, vol. 41, no. 8, pp. 814–822, 2009. PID controller tuning,” Journal of Process Control, vol. 13, no.
[4] B. Tashtoush, M. Molhim, and M. Al-Rousan, “Dynamic model 4, pp. 291–309, 2003.
of an HVAC system for control analysis,” Energy, vol. 30, no. 10, [12] K. J. Åström and T. Hägglund, “The future of PID control,”
pp. 1729–1745, 2005. Control Engineering Practice, vol. 9, no. 11, pp. 1163–1175, 2001.
Journal of Sensors 15

[13] K. J. Åström and T. Hägglund, “Revisiting the Ziegler-Nichols

step response method for PID control,” Journal of Process
Control, vol. 14, no. 6, pp. 635–650, 2004.
[14] W. K. Ho, O. P. Gan, E. B. Tay, and E. L. Ang, “Performance and
gain and phase margins of well-known PID tuning formulas,”
IEEE Transactions on Control Systems Technology, vol. 4, no. 4,
pp. 473–477, 1996.
[15] S. W. Sung, J. Lee, and I. B. Lee, Process Identification and PID
Control, John Wiley & Sons, New York, NY, USA, 2009.
[16] L. Hongli, D. Peiyong, and J. Lei, “A novel fuzzy controller
design based-on PID Gains for HVAC systems,” in Proceedings
of the 7th World Congress on Intelligent Control and Automation
(WCICA ’08), pp. 736–739, IEEE, Chongqing, China, June 2008.
[17] K. F. Fong, V. I. Hanby, and T. T. Chow, “System optimization
for HVAC energy management using the robust evolutionary
algorithm,” Applied Thermal Engineering, vol. 29, no. 11-12, pp.
2327–2334, 2009.
[18] D. Xue, Y. Q. Chen, and D. P. Atherton, Linear Feedback Control,
SIAM’s Advances in Design and Control, SIAM, Philadelphia,
Pa, USA, 2007.
[19] N. J. Killingsworth and M. Krstic, “PID tuning using extremum
seeking: online, model-free performance optimization,” IEEE
Control Systems Magazine, vol. 26, no. 1, pp. 70–79, 2006.
 International Journal of

Rotating
Machinery

International Journal of
The Scientific
Engineering Distributed
Journal of
Journal of

Hindawi Publishing Corporation

World Journal
Hindawi Publishing Corporation Hindawi Publishing Corporation
Sensors
Hindawi Publishing Corporation
Sensor Networks
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of

Control Science
and Engineering

Advances in
Civil Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Submit your manuscripts at

http://www.hindawi.com

Journal of
Journal of Electrical and Computer
Robotics
Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

VLSI Design
Advances in
OptoElectronics
International Journal of

International Journal of
Modelling &
Simulation
Aerospace
Hindawi Publishing Corporation Volume 2014
Navigation and
Observation
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
in Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com
http://www.hindawi.com Volume 2014

International Journal of
International Journal of Antennas and Active and Passive Advances in
Chemical Engineering Propagation Electronic Components Shock and Vibration Acoustics and Vibration
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014