Applied Thermal Engineering 113 (2017) 1386–1395

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Dynamic target high pressure control of a VRF system for heating energy
savings
Geun Young Yun a,⇑, Je Hyeon Lee b, Inhan Kim c
a
Department of Architectural Engineering, Kyung Hee University, 1732, Deogyeong-daero, Giheung-gu, Yongin si, Gyeonggi-do 17104, South Korea
b
Department of Digital Appliances R&D Team, Samsung Electronics, 129, Samsung-ro, Yeongtong-gu, Suwon-si, Gyeonggi-do 443-742, South Korea
c
Department of Architecture, Kyung Hee University, 1732, Deogyeong-daero, Giheung-gu, Yongin si, Gyeonggi-do 17104, South Korea

h i g h l i g h t s

We developed the dynamic target high pressure control of a VRF system.

We created the VRF control model using the EnergyPlus runtime language.

Multicalorimeter experimental results indicate that the energy efficiency improved by 21%.

EnergyPlus simulations demonstrate that the annual heating energy consumption was lowered by 22%.

a r t i c l e i n f o a b s t r a c t

Article history: Variable refrigerant flow (VRF) systems are widely used because of their ability to provide individualized
Received 20 July 2016 comfort control with energy-saving potential. This study develops load responsive high pressure control
Revised 18 November 2016 of a VRF system with the aim of reducing the heating energy consumption of a VRF system under part
Accepted 20 November 2016
load conditions. The developed control consists of two parts: one part that determines the level of heating
Available online 22 November 2016
load, and the other that assigns a target high pressure based on the level of heating load. In this way, the
compressor speed can be accurately matched to heating load, which improves the energy performance of
Keywords:
the VRF system. A series of multicalorimeter experiments revealed that the heating capacity of the VRF
Building energy
Heat pump
system varied by 45% by modulating the target high pressure and that its efficiency was enhanced by 21%
Energy efficiency by changing the high pressure from 30 kgf/cm2 to 25 kgf/cm2. To evaluate the annual heating energy per-
Simulation formance of the VRF system with the developed control, a custom computer code was developed to
EnergyPlus implement the developed control using a programming language called EnergyPlus Runtime Language.
Variable refrigerant flow system Simulation outcomes showed that the annual heating energy consumption of a medium-size office build-
ing was reduced by 22% when the developed control was applied.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction active methods to reduce energy consumption for space condition-

ing are of critical importance.
Buildings are one of the biggest consumers of energy world- Electricity is one of the fastest growing delivered energy
wide; they are responsible for 32% of the world total energy con- sources. In the US, electricity consumption in buildings has
sumption [1]. A large portion of electricity consumed in buildings increased 28% since 1990 because of increased use of space heating
is used for air conditioning and heating [2]. Globally, space and air conditioning [5]. World electricity consumption is pro-
conditioning to meet thermal comfort requirements accounted jected to rise by 2.2% per year from 2010 to 2040, with this
for 34–40% of the final energy consumption in both residential increase driven mainly by Asia with an average growth of 3.1%
and commercial buildings in 2010 [3]. In the European Union, per year [6]. One reason for this rapid growth is the increased
space conditioning was the largest energy end-use in the building installations of heat pump units in Asia [7]. In particular, the mar-
sector, accounting for 69% of residential energy consumption and ket of the variable refrigerant flow (VRF) system increases at an
45% of commercial energy use in 2010 [4]. Therefore, passive and annual rate of 11% in Asia [8]. Growing reliance on heat pumps
for space heating could increase the risk of winter blackouts, which
have devastating consequences for individuals and society. Thus,
⇑ Corresponding author.
E-mail address: [email protected] (G.Y. Yun).

http://dx.doi.org/10.1016/j.applthermaleng.2016.11.148
1359-4311/Ó 2016 Elsevier Ltd. All rights reserved.
 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395 1387

the improvement of their energy efficiency is critical to decrease 2. Thermal load based control of high pressure in a VRF system
electricity demands in winter.
One feature of the VRF system is that it enables concurrent The developed control in this study consists of two parts: one
heating and cooling modes of different indoor units through mode part that determines the level of heating load, and the other that
change units or heat recovery units [9,10]. A key technology is to assigns a target high pressure based on the level of heating load.
vary refrigerant flow rates between outdoor units and multiple A unique aspect of the control is the direct modulation of the set-
indoor units [11,12]. Individual zone controls, along with seamless point of high pressure to match the heating loads of indoor units.
switchover between heating and cooling operation modes are main The existing studies [26–28] focused on the control of the com-
reasons for the wide use of VRF systems in various buildings such pressor speed to change the heating capacity for a fixed high tem-
as offices, schools, retail stores, and hotels [13]. perature. This study utilizes the high pressure as a control variable,
Prior research on VRF systems has focused on reducing electric- so that the heating capacity can be adjusted with high accuracy to
ity consumption in cooling mode. Simulation and experimental varied heating demands.
results indicated that the water-cooled VRF system considerably The first part of the developed control determines the capacity-
lowered the amount of energy used by the compressor in the cool- weighted part load condition of indoor units in the single refriger-
ing mode [14,15]. Several studies [16–19] have confirmed the ant circuit (Fig. 1). This part to decide the thermal load of indoor
importance of control of electronic expansion value (EEV) opening units is based on the method developed by Yun et al. [24]. The
and compressor control of the part load cooling efficiency of VRF algorithm is applied to a part load condition when the deviation
systems. Tu et al. [16] found that the average suction pressure of (DTrs) of room temperature from the setpoint temperature is less
refrigerant was a reliable variable to control the cooling capacity than 3 K. When the deviation is within 3 K, the algorithm evaluates
of the compressor. Choi and Kim [17] experimentally showed that that the heating load is low enough to apply the developed control.
cooling capacity can be effectively controlled by modulation of EEV The algorithm estimates the two factors of the current indoor unit:
opening to meet the cooling load of two indoor units with a super- Ld1 and Ld2. Ld1 is determined based on the change in the temper-
heat temperature of 4 °C. Control algorithms for multi-evaporator ature of the room over the past five minutes (DTrm) and the crite-
air conditioning systems that regulate the cooling load and EEV rion for the room temperature change (Trc):
openings to determine compressor speed have also been developed
T rc
[18,19]. Ld1 ¼ ð1Þ
There are few experimental studies of the performance of VRF T rm
systems in heating mode. Joo et al. [20] found that the control of Trc can be determined from Table 1 [29]. In the next step, the
the compressor speed and the EEV openings was a key factor in runtime fraction of the indoor unit (Ld2), which is the ratio of
regulating heating capacity. Jeong and Choi [21] investigated the the time that current indoor unit is on over the past thirty minutes
effects of EEV opening and compressor speed on the heating per- is calculated. Then, the smaller value between Ld1 and Ld2
formance of the heat pump and found that heating capacity becomes the heating load coefficient of the current indoor unit
increased with an increase in EEV opening and compressor speed. (Ld(n)). Then, the heating load coefficients for all the indoor units
The setpoint of evaporating temperature remains unchanged in in the refrigerant circuit is calculated. Finally, the capacity
general VRF controls [22], although a rise in evaporating tempera- weighted average heating load coefficient (Ld) is calculated using
ture results in the efficiency of the VRF system [23]. Recent studies Eq. (2), as follows:
[24,25] revealed the large potential of controlling evaporating tem- Pm
perature to improve the cooling efficiency of VRF systems espe- n¼1 ðLdðnÞ  CapacityðnÞ Þ
Ld ¼ Pm ð2Þ
cially for part load conditions because the cooling capacity could n¼1 CapacityðnÞ
be accurately adjusted to cooling load conditions.
Here, m is the total number of indoor units in the refrigerant circuit
The heating capacity of the VRF system needs to be adjusted in
and Capacity(n) is the heating capacity of the indoor unit, n.
order to match the varied heating loads of indoor units in the VRF
The second part of the algorithm decides the dynamic target
system. The common way of modulating the heating capacity of
high pressure as a function of the capacity-weighted average heat-
the VRF system is to change its compressor speed [9]. Awan [26]
ing load coefficient (Ld), as shown in Table 2. A series of experi-
experimentally investigated the change in heating capacity by
ments were carried out to determine the range of target values.
adjusting the compressor speed from 30 Hz to 90 Hz. Kim et al.
Both the stability of the heating capacity with various loads and
[27] proposed the design of the variable speed heat pump that
the pressure range of the compressor with the refrigerant,
could response to direct load control signal. Tu et al. [28] devel-
R410A, were considered to determine the target values. As the
oped a control model for the VRF system, which determined the
heating load decreased, the high pressure of the compressor in
optimal range of compressor frequency. However, the high pres-
heating mode decreased from 30 kgf/cm2 to 25 kgf/cm2. The rela-
sure is fixed in the heating capacity control of the VRF system in
tionship between the discharge refrigerant temperature (Tds in
a normal operation [22], although the change in the high pressures
°C) in Table 1 and high pressure (Ph in kgf/cm2) is given by Eq. (3).
has a direct influence on the heating capacity. Thus, in this study
the high pressure was explored as a control variable in the adjust- T ds ¼ 0:0173ðPh Þ2 þ 2:4443ðPh Þ  7:9824 ð3Þ
ment of the heating capacity to different heating demands.
In order to decrease the energy consumption of the VRF system
in heating mode, this study developed a control strategy, which 3. Methods
regulates high pressure in the VRF system as a function of the heat-
ing load of indoor units. We also investigated the energy implica- 3.1. Experimental methods
tions of the proposed control model when applied to the VRF
system of an office building. After describing the developed control The effects of changes in target high pressure on the heating
in this study, we elucidate the experimental and simulation meth- capacity and power input of the VRF system were measured in a
ods to examine the energy efficiency and annual heating energy multicalorimeter, which consisted of two indoor unit chambers
consumption with the new control developed in this study. Finally, and one outdoor unit camber. Outdoor unit chamber represented
we present results from a series of experiment and simulations and outdoor conditions while indoor conditions were simulated with
discuss main outcomes. indoor unit chambers. The range of the heat dissipation capacity
1388 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395

Fig. 1. The process for determining the setpoint for high pressure.

Table 1
Value for a change in room temperature. the capacity of the outdoor unit. The accuracy error of the mul-
ticalorimeter is within ±2% and enables constant temperature
Difference between air and discharge Criterion for air
refrigerant temperatures (DTre, K) temperature change (Trc, K)
and humidity control. The temperature range of the outdoor unit
chamber was 30 to 60 °C, while the temperature range of 0 to
DTre 6 18 4
18 < DTre 6 23 5
50 °C was simulated in the indoor unit chamber. Thus, the mul-
23 < DTre 6 27 6 ticalorimeter used in this study was adequate for representing var-
27 < DTre 6 32 7 ious indoor and outdoor conditions.
DTre > 32 8 A code tester made in accordance with the calorimeter air
enthalpy method [30] was used to determine the air side heating
capacity of the indoor unit chamber. Air temperature, humidity,
Table 2 and airflow rates of the indoor and outdoor unit chambers were
Target high pressures as a function of capacity-weighted average heating load
recorded by the monitoring system. Dry-bulb and wet-bulb tem-
coefficient.
peratures were recorded at the inlet and outlet of each indoor unit
Level Capacity-weighted average Target high chamber. The monitoring system also measured and saved the
heating load coefficient pressure (kgf/cm2)
temperature and pressure of the refrigerant and the heating capac-
1 Ld = 1.0 30 ity and electricity input. The ratios of EEV openings were also
2 0.9 6 Ld 6 1.0 29
recorded.
3 0.8 6 Ld < 0.9 28
4 0.7 6 Ld < 0.8 27 The outdoor unit chamber was air-conditioned by one 24 HP
5 0.6 6 Ld < 0.7 26 outdoor unit (AM240FXVGHH1) while each indoor unit chamber
6 Ld < 0.6 25 enclosed three 4 HP indoor units (ND11104HXB1). The total piping
length from the outdoor unit to the branch joint was 35 m and
each indoor unit was 15 m from the branch joint. Two compressors
with a power of 12.78 kW each were used in the outdoor unit. The
of the indoor unit chamber was from 5.58 to 55.82 kW, while that power of each fan was 400 W. R410A (14.8 kg) was used as the
of the outdoor unit chamber was from 10.47 to 104.67 kW. The refrigerant. The minimum outdoor temperature that the outdoor
heating capacity of the two indoor unit chambers was equal to unit could operate at was 20 °C and heat dissipation capacity
 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395 1389

was 61 kW at the outdoor temperature of 15 °C. Each indoor unit to test the significance of individual regression coefficients. There
had a heating capacity of 12.8 kW and the power consumption of was a clear difference in the VRF system performance in low and
each unit was 78 W. The specification of the indoor unit is given high outdoor temperature regions, and thus separate curves for
in Table 3. In the experiment, the refrigerant condensing tempera- those two distinct areas were required. The heating capacity ratio
ture was controlled by the target high pressure. The range of the boundary curve was used to separate the low and high tempera-
target high pressure was 25–30 kgf/cm2 at 1 kgf/cm2 intervals. ture areas with different performances. The cubic equation form
The target high pressure was changed by regulating the compres- was used to generate performance curve coefficients of the heating
sor speed. The heating capacity and power consumption of the sys- capacity ratio boundary curve in this study (Eq. (4)):
tem were first measured at the reference high pressure of 27 kgf/
cm2. Indoor dry-bulb temperature ranged from 16 to 24 at 2 °C T OA;WB ¼ a þ bðT I;DB Þ þ cðT I;DB Þ2 þ dðT I;DB Þ3 ð4Þ
intervals while outdoor dry-bulb temperature ranged from Here, TOA,WB is the outdoor wet-bulb temperature, TI,DB is the indoor
20 °C to 15 °C, as shown in Table 4. Heating capacity and electric- dry-bulb temperature, and a, b, c, and are performance curve coef-
ity consumption at other target high pressures were measured ficients. Heating capacity modifier performance curves in the VRF
under standard conditions (indoor dry-bulb temperature of 20 °C, model identified changes in heating capacity as the indoor and out-
outdoor wet-bulb temperature of 6 °C, and outdoor dry-bulb tem- door temperatures were varied. Heating capacity correction factor
perature of 7 °C). (CAPFTheating), which is used to create performance curve coeffi-
cients, is dependent on the dry-bulb temperature of the air entering
the heating coil (Tdb) and the wet-bulb temperature of the air enter-
3.2. Simulation
ing a condenser (Tc). CAPFTheating is the ratio of the actual heating
capacity to the rated heating capacity. The bi-quadratic function
A dynamic building energy simulation program, EnergyPlus
show in Eq. (5) was used.
[31], was used in this study. EnergyPlus calculates thermal loads
of a building and the energy demands of heating, ventilating, and CAPFT heating ¼ a þ bðT db Þ þ cðT db Þ2 þ dðT c Þ þ eðT c Þ2 þ f ðT db ÞðT db Þ
air-conditioning (HVAC) systems using the heat balance method
ð5Þ
[32]. A key feature of EnergyPlus is its rigorous validation, which
guarantees the accuracy and reliability of the simulation outcomes Here, a, b, c, d, e, and f are performance curve coefficients. The heat-
[32]. The VRF heat pump air conditioner model in EnergyPlus [33] ing energy input ratio modifier describes changes in heating power
has been validated against experimental data [34]. input as a function of indoor and outdoor temperatures. The bi-
Fig. 2 shows the conceptual diagram of the VRF system mod- quadratic regression equation for the performance curve of the
elled in this study. A system model consisted of six energy recovery heating energy input ratio modifier consists of the heating energy
ventilators (ERV) linked to a VRF system. A 24 HP outdoor unit input ratio correction factor (EIRFTheating), the dry-bulb temperature
(AM240FXVGHH1) provided heating to a typical office floor. Both of the air entering the heating coil (Tdb), and the wet-bulb temper-
full-load and part-load performance data available from manufac- ature of air entering the condenser (Tc) as shown in Eq. (6).
turers are required as inputs for a VRF model in EnergyPlus. The
full-load performance data identifies variations in power input EIRFT heating ¼ a þ bðT db Þ þ cðT db Þ2 þ dðT c Þ þ eðT c Þ2 þ f ðT db ÞðT db Þ
and heating capacity as a function of outdoor and indoor condi- ð6Þ
tions, while part-load performance data reflects capacity and
power input characteristics as the compressor speed varies. Perfor- To derive EIRFT, the ratio of the actual power input to the rated
mance curves to simulate variations in the power input and capac- power input was first calculated and then divided by CAPFT. Heat-
ity with changes in both indoor and outdoor conditions and ing energy input ratio modifier function of the part-load ratio
compressor speeds are key inputs for VRF models in EnergyPlus. defines the power input of the VRF system as a function of part-
The setpoint of high pressure in the developed control varies load ratio (PLR). In this study, we used the cubic function of the
with the heating load and the performance of the VRF system at part-load ratio (Eq. (7)):
full-load and part-load conditions changes with the varied high EIRFPLROA;WB ¼ a þ bðPLRÞ þ cðPLRÞ2 þ dðPLRÞ3 ð7Þ
pressure. Thus, this study generated a series of performance curves
at the target high pressures from 25 kgf/cm2 to 30 kgf/cm2 to where EIRFPLR is the ratio of operating power to the rated power of
model the VRF system in EnergyPlus. The heating capacity and the VRF system. EIRFPLR was calculated according to the combina-
electricity consumption data measured at the indoor and outdoor tion ratio, which is the ratio of the total capacity of the indoor units
conditions specified in Table 4 were used to create the five sets to the capacity of the outdoor unit.
of performance curves at the setpoints of high pressures from Heating performance changes of the VRF system due to varia-
25 kgf/cm2 to 30 kgf/cm2 at 1 kgf/cm2 intervals. Raustad [35] tion in target high pressure were modelled with a custom com-
developed a method to create the performance curves of the out- puter code to implement the developed control. A programming
door unit for the modelling of the VRF system in EnergyPlus using language called EnergyPlus runtime language (Erl) was used for
experimental data and this study used the method by Raustad [35]. the development of the custom computer code [36]. The custom
Regression analysis was used to generate performance curve computer code first determined the capacity-weighted average
coefficients. The F-test was used to evaluate the statistical signifi- heating load coefficient (Ld) in load responsive control of high
cance of the overall regression model while the T-test was applied pressure at each simulation time step. The custom code obtained

Table 3
Specification of the indoor unit.

Heating dissipation Power Fan Piping connections

capacity (kW)
Voltage Running Type Airflow rate Liquid Suction Drain
(U/V/Hz) current (A) (m3/min) pipe (mm) pipe (mm) pipe (mm)
12.8 1/220/60 0.36 Turbo fan 28.0 9.52 15.88 25
1390 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395

Table 4
Indoor and outdoor conditions for the experiment.

Indoor dry-bulb temperature (°C) Outdoor temperature (°C)

16 Dry-bulb 20 17 15 12 10 7 5 3 0 3 5 7 9 11 13 15
Wet-bulb 21 18 16 13 11 8 6 4 1 2 4 6 8 10 12 14
18 Dry-bulb 20 17 15 12 10 7 5 3 0 3 5 7 9 11 13 15
Wet-bulb 21 18 16 13 11 8 6 4 1 2 4 6 8 10 12 14
20 Dry-bulb 20 17 15 12 10 7 5 3 0 3 5 7 9 11 13 15
Wet-bulb 21 18 16 13 11 8 6 4 1 2 4 6 8 10 12 14
22 Dry-bulb 20 17 15 12 10 7 5 3 0 3 5 7 9 11 13 15
Wet-bulb 21 18 16 13 11 8 6 4 1 2 4 6 8 10 12 14
24 Dry-bulb 20 17 15 12 10 7 5 3 0 3 5 7 9 11 13 15
Wet-bulb 21 18 16 13 11 8 6 4 1 2 4 6 8 10 12 14

Fig. 2. Conceptual diagram of the VRF system in the case study building.

changes in zone air temperatures, setpoint temperature, heating Korea [38]. The thermal transmittances of the external walls and
coil runtime, and operation status of an indoor unit in order to cal- the windows were 0.270 W/(m2 K) and 2.100 W/(m2 K), respec-
culate Ld. The code then evaluated the level of the indoor heating tively. Meteonorm [39] was used to get the weather data of Suwon,
loads. Eventually, the code replaced the heating performance South Korea for EnergyPlus simulations. This study followed the
curves of the VRF system in accordance with the level of the indoor specification of the US DOE medium model for modelling both
heating loads. internal heat gains from people, lighting, equipment, and sched-
We selected the prototype of the medium office model devel- ules, including occupancy and HVAC operations.
oped by the US Department of Energy (US DOE) [37], to evaluate
the energy saving potentials of thermal load based control of the
VRF system. Thee medium office model was divided into six 4. Results
perimeters and two internal zones with a floor area of 1661 m2.
The external walls and the windows was defined according to Fig. 3 illustrates the effect of the target high pressure on the
the specification of the energy saving standard for buildings in electricity consumption and heating capacity of the VRF system
 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395 1391

pressure of 25 kgf/cm2, which was equivalent to an improvement

of 21%. The improvement in the efficiency due to the change in tar-
get pressures was also observed in variations in the power con-
sumption and the heating capacity. The rate of decrease in the
power consumption by lowering the target pressure was greater
than that of reduction in the heating capacity. When the high pres-
sure changed from 30 kgf/cm2 to 25 kgf/cm2, the heating capacity
reduced by 45% and the power consumption was lowered by 54%.
The influence of the high pressure on the VRF system was fur-
ther analyzed in relation to outdoor temperatures, because the
outdoor temperature was one of key determinant of the VRF sys-
tem performance (Fig. 4). The heating capacity started to decrease
at an outdoor wet-bulb temperature of 4 °C and sharply plunged
at an outdoor wet-bulb temperature below 8 °C regardless of the
set points of the high pressure. However, the heating capacity
remained unchanged when the outdoor dry-bulb and wet-bulb
temperatures were over 0 °C at the range of high pressures from
30 kgf/cm2 to 25 kgf/cm2. On the other hand, there was a sharp
drop in power consumption when the outdoor temperature was
over 7 °C. When the outdoor temperature rose from 20 °C to
15 °C, the reduction in the electricity consumption ranged from
9.3 kW at the high pressure of 25 kgf/cm2 to 20.5 kW at the high
pressure of 30 kgf/cm2.
The change in energy efficiency of the VRF system at different
high pressures was analyzed in Fig. 4. In this study, the energy effi-
ciency of the VRF system refers to the ratio of the heating capacity
to the electricity consumption. The general trend was that the
energy efficiency improved with a rise in outdoor temperatures.
This was mainly because the electricity consumption reduced as
the outdoor temperature rose. The positive relationship between
the outdoor temperature and the energy efficiency was found at
the range of high pressures examined in this study.
The experimental data indicated that decreasing the high pres-
sure resulted in the improvement of the energy efficiency. At the
outdoor temperature of 20 °C, the energy efficiency was 1.60 at
the high pressure of 30 kgf/cm2 and this increased to 1.93 at the
high pressure of 25 kgf/cm2. The energy efficiency rose from 3.90
at the high pressure of 30 kgf/cm2 to 4.70 at the high pressure of
25 kgf/cm2, when the outdoor dry-bulb temperature was 15 °C.
The use of dynamic target high pressure control improved the
energy efficiency of the VRF system and the improvement was
more evident as the outdoor temperature rose. Thus, the combina-
tion of the effective operation of the VRF system at part-load con-
ditions and the use of dynamic target high pressure control that
responds to heating demands further improved the energy effi-
ciency of the VRF system.
The monthly heating electricity consumption of a medium
office of the DOE commercial reference building is shown in
Fig. 5. Fig. 5 compares the heating electricity consumption of the
Fig. 3. Effects of high pressure on the performance of the VRF system: (a) heating building with the developed control with that of the same building
capacity, (b) electricity consumption, (c) energy efficiency.
but without the developed control. The same VRF system was used
for the both buildings, but the setpoint of the high pressure
from the multicalorimeter experiment measured at the indoor remained unchanged at 30 kgf/cm2 for the VRF system without
dry-bulb temperature of 20 °C. The outdoor condition was 7 °C the developed control. Thus, a set of the performance curves which
for dry-bulb temperature and 6 °C for wet-bulb temperature. were created from the experimental data measured at the high
Decreasing target high pressure for the refrigerant reduced the pressure of 30 kgf/cm2 were used in the VRF system without the
heating capacity. This diminished from 110.8 kW at the high pres- developed control, while the different sets of the performance
sure of 30 kgf/cm2 to 60.6 kW at the high pressure of 25 kgf/cm2. In curves was applied in the VRF system with the developed control
addition, the power consumption also decreased as the target high depending on the setpoint of high pressures ranging from 30 kgf/
pressure decreased. Power consumption reduced from 33.9 kW at cm2 to 25 kgf/cm2.
30 kgf/cm2 to 15.4 kW at 25 kgf/cm2. Thus, controlling the target The monthly heating energy consumption reached its peak in
high pressure is an effective way to respond to variations in heat- January when the monthly average outdoor temperature was the
ing demand. lowest. The lowest heating energy consumption was observed in
The energy efficiency of the VRF system showed an inverse March. The outdoor temperature and solar radiation during the
relationship with the target high pressure. The efficiency was 3.3 heating season were the greatest in March. The mean monthly out-
at the high pressure of 30 kgf/cm2 and increased to 3.9 at the high door temperature ranged from 3.4 °C (standard deviation,
1392 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395

Fig. 4. Performance of the VRF system as a function of high pressure at different outdoor conditions: (a) heating capacity, (b) electricity consumption, (c) energy efficiency.
 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395 1393

Fig. 5. Heating electricity consumption with and without the target high pressure control.

Fig. 6. Dry bulb temperature of the site used in energy simulation.

SD = 4.9 °C) in January to 6.6 °C (SD = 5.2 °C) in February (Fig. 6). energy delivered to the building on January 3 was decreased by
The mean monthly outdoor temperature during November to 25% when the developed control was used. The heat delivered
February was less than 10 °C. This confirms the crucial impact of was simulated with the code in the simulation model, which chan-
outdoor conditions on the heating energy use of a building. ged the sets of the performance curves according the level of
The use of the heat load based control of the VRF system indoor heating demands. The performance curves included the
reduced the annual heating electricity consumption by 22%. The heating capacity correction factor (CAPFTheating) and the heating
annual heating electricity consumption reduced from 5325 kW h energy input ratio corrector factor (EIRFTheating) as explained in
to 4160 kW h as the developed control was used to operate the Section 3.2. The heating energy saving predicted by the simulation
VRF system. The energy savings by the developed control are sig- model was smaller than a value from the experiment results. This
nificant compared to the advanced control [40] that shows the was because the setpoint of high pressure was not always lower
energy reduction by through the employment of artificial neural than 30 kgf/cm2. There was a variation in the rates of reduction
network models. The heat delivered to the building also decreased in monthly heating electricity consumption when the developed
when the developed control was applied. For example, the heating control was applied. The use of the developed control reduced
1394 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395

the heating electricity consumption in January from 2212 kW h to References

1690 kW h with the developed control. The heating electricity con-
sumption in February reduced by 22 kW h when the developed [1] Intergovernmental Panel on Climate Change, Climate Change 2014: Impacts,
Adaptation, and Vulnerability. Cambridge University Press, New York, USA,
control was applied. Thus, the developed control was useful in 2014.
reducing the heating energy consumption. [2] US Energy Information Administration, Annual Energy Outlook 2014 With
Simulation results are influenced by indoor and outdoor condi- Projections To 2040, 2014.
[3] Intergovernmental Panel on Climate Change, Climate Change 2014: Mitigation
tions, and in particular, internal heating load conditions that deter- of Climate Change, Cambridge University Press, New York, USA, 2014.
mine the heating load levels in the developed control. Further [4] IEA, Transition to Sustainable Buildings: Strategies and Opportunities to 2050,
studies are needed to elucidate how the design and operation of International Energy Agency, 2013.
[5] US Department of State, United States Climate Action Report 2014, US
the buildings influence the performance of the VRF system with
Department of State, 2014.
the developed control. [6] US EIA, International Energy Outlook, US Energy Information Administration,
2015.
[7] M. Sivak, A.C. Will, Put a Chill on the Global Energy Supply?, Am Sci. 100 (2013)
5. Discussion and conclusions 330, http://dx.doi.org/10.1511/2013.104.330.
[8] J. Milnes, Global A/C Market Starting to Warm Up, The Air Conditioning,
Heating and Refrigeration NEWS, <http://www.achrnews.com/articles/
We proposed the heat load based control of the target high 127385-global-ac-market-starting-to-warm-up> (accessed June 18, 2016).
pressure of a VRF system to reduce heating energy consumption. [9] T.N. Aynur, Variable refrigerant flow systems: a review, Energy Build. 42
Experiments demonstrated that the heating capacity of the VRF (2010) 1106–1112, http://dx.doi.org/10.1016/j.enbuild.2010.01.024.
[10] B. Thornton, A. Wagner, Variable Refrigerant Flow Systems, Pacific Northwest
system changed by 45% by modulating the target high pressure National Laboratory, 2012.
(Fig. 3), verifying that controlling the target high pressure is appro- [11] W. Goetzler, Variable refrigerant flow systems, ASHRAE J. (2007) 24–31.
priate for controlling variation in the heating loads of indoor units. [12] ASHRAE, ASHRAE Handbook – HVAC Systems and Equipment, American
Society of Heating, Refrigerating, and Air-conditioning Engineers, 2012.
This is an important outcome because the opening of EEV and the [13] X. Lin, H. Lee, Y. Hwang, R. Radermacher, A review of recent development in
speed of compressor are two mainly control variables in existing variable refrigerant flow systems, Sci. Technol. Built Environ. 21 (7) (2015)
studies [16,18–21]. Only recently, a study [24] developed an algo- 917–933, http://dx.doi.org/10.1080/23744731.2015.1071987.
[14] Y.M. Li, J.Y. Wu, S. Shiochi, Experimental validation of the simulation module
rithm to control target evaporating temperature in cooling mode to
of the water-cooled variable refrigerant flow system under cooling operation,
enhance the VRF energy efficiency at part-load conditions. Our Appl. Energy 87 (5) (2010) 1513–1521, http://dx.doi.org/10.1016/j.
study illustrates that modulating target high pressure of the VRF apenergy.2009.09.018.
[15] Y. Li, J. Wu, S. Shiochi, Modeling and energy simulation of the variable
system in heating mode is an effective mean to control heating
refrigerant flow air conditioning system with water-cooled condenser under
capacity. cooling conditions, Energy Build. 41 (9) (2009) 949–957, http://dx.doi.org/
The experimental results also indicated that the performance of 10.1016/j.enbuild.2009.04.002.
the VRF system enhanced with a reduction in target high pressure. [16] Q. Tu, K. Dong, D. Zou, Y. Lin, Experimental study on multi-split air conditioner
with digital scroll compressor, Appl. Therm. Eng. 31 (14–15) (2011) 2449–
Energy efficiency increased by 21% as the target high pressure 2457, http://dx.doi.org/10.1016/j.applthermaleng.2011.04.010.
changed from 30 kgf/cm2 to 25 kgf/cm2 (Fig. 3). This indicates that [17] J.M. Choi, Y.C. Kim, Capacity modulation of an inverter-driven multi-air
by controlling the target high pressure, heating energy consump- conditioner using electronic expansion valves, Energy 28 (2) (2003) 141–155,
http://dx.doi.org/10.1016/S0360-5442(02)00113-5.
tion can be reduced under part-load conditions. We developed a [18] X. Xiangguo, P. Yan, D. Shiming, X. Liang, C. Mingyin, Experimental study of a
control method that regulates the target high pressure of the VRF novel capacity control algorithm for a multi-evaporator air conditioning
system in response to the heating demands of the indoor units in system, Appl. Therm. Eng. 50 (1) (2013) 975–984, http://dx.doi.org/10.1016/j.
applthermaleng.2012.08.007.
a refrigerant circuit. Thus, the developed algorithm can be used [19] C. Wu, Z. Xingxi, D. Shiming, Development of control method and dynamic
to reduce heating energy consumption by a VRF system. model for multi-evaporator air conditioners (MEAC), Energy Convers. Manage.
This study proposed a novel feedback control that utilized the 46 (3) (2005) 451–465, http://dx.doi.org/10.1016/j.enconman.2004.03.004.
[20] Y. Joo, H. Kang, J.H. Ahn, M. Lee, Y. Kim, Performance characteristics of a
high pressure as a control variable in order to adjust the heat out-
simultaneous cooling and heating multi-heat pump at partial load conditions,
put of the VRF system with high accuracy to varied heating Int. J. Refrig. 34 (4) (2011) 893–901, http://dx.doi.org/10.1016/j.
demands. The electricity consumption and the heat output of the ijrefrig.2010.12.025.
[21] K. Jeong, J.M. Choi, Capacity modulation of a cascade heat pump with the
VRF system reduced when the target high pressure was decreased
variation of compressor speed and electronic expansion valve opening, J.
at the low level of heating demands. The simulation results con- Renew. Sustain. Energy 6 (5) (2014) 53107, http://dx.doi.org/10.1063/
firmed a large potential of modulating the high pressure based 1.4896597.
on the level of heat demands in decreasing the heating energy con- [22] Samsung, Air Conditioner installation manual, Samsung, 2014.
[23] I. Fujimoto, K. Takemura, K. Ueno, K. Saito, K. Ohno, Performance evaluation of
sumption. The annual heating energy consumption of the office VRF systems using refrigerant enthalpy method, in: Proceedings of the 23rd IIR
building was decreased by 22% when load responsive control of International Congress of Refrigeration, Prague, Czech Republic, 2011.
high pressure was implemented (Fig. 5). In particular, monthly [24] G.Y. Yun, J.H. Lee, H.J. Kim, Development and application of the load responsive
control of the evaporating temperature in a VRF system for cooling energy
heating energy consumption decreased by 24% in January, when savings, Energy Build. 116 (2016) 638–645, http://dx.doi.org/10.1016/j.
the outside temperature was the lowest. This implies that the enbuild.2016.02.012.
developed control has high energy saving potentials. [25] D. Zhao, X. Zhang, M. Zhong, Variable evaporating temperature control
strategy for VRV system under part load conditions in cooling mode, Energy
Build. 91 (2015) 180–186, http://dx.doi.org/10.1016/j.enbuild.2015.01.039.
Acknowledgements [26] U.K. Awan, Experimental analysis of variable capacity heat pump system
equipped with vapour injection and permanent magnet motor. [Sweden]:
Royal Institute of Technology; 2012 [cited 2016 Oct 11]. <http://www.diva-
The authors are very grateful to anonymous reviewers for their portal.org/smash/record.jsf?pid=diva2%3A565195&dswid=231>.
constructive comments on our manuscript. This research was sup- [27] Y.J. Kim, J.L. Kirtley, L.K. Norford, Variable speed heat pump design for
frequency regulation through direct load control, in: 2014 IEEE PES T D
ported by the Basic Science Research Program through the National Conference and Exposition, 2014, pp. 1–5.
Research Foundation of Korea (NRF) funded by the Ministry of [28] Q. Tu, D. Zou, C. Deng, J. Zhang, L. Hou, M. Yang, et al., Investigation on output
Science, ICT and Future Planning (2014R1A1A2054494) and was capacity control strategy of variable refrigerant flow air conditioning system
with multi-compressor, Appl. Therm. Eng. 25 (99) (2016) 280–290.
also supported by a grant (15-AUDP-C067817-03) from the Archi-
[29] JSRAE, JSRAE Refrigeration and Air Conditioning Handbook, Japanese Society
tecture & Urban Development Research Program funded by the for Refrigeration and Air Conditioning Engineers, 2006.
Ministry of Land, Infrastructure and Transport of the Korean [30] ISO, ISO 5151 Non-ducted air conditioners and heat pumps – testing and rating
for performance. International Standard Organization, 2010.
government.
 G.Y. Yun et al. / Applied Thermal Engineering 113 (2017) 1386–1395 1395

[31] US Department of Energy, Building Technologies Office: EnergyPlus Energy [36] US DOE, EnergyPlus Documentation: Application Guide for EMS, US
Simulation Software, <http://apps1.eere.energy.gov/buildings/energyplus/> Department of Energy, 2015.
(accessed June 18, 2016). [37] Office of Energy Efficiency & Renewable Energy, Commercial Reference
[32] R.H. Henninger, M.J. Witte, EnergyPlus testing with global energy balance test, Buildings|Department of Energy, <http://energy.gov/eere/buildings/
Gard Analytics, 2013. commercial-reference-buildings> (accessed June 18, 2016).
[33] R. Raustad, Computer modeling VRF heat pumps in commercial buildings [38] Korean Energy Agency, Energy saving standard for buildings, 2015, Ministry of
using EnergyPlus, ASHRAE Trans. 119 (2013) 10. Land, Infrastructure and Transport.
[34] Y.P. Zhou, J.Y. Wu, R.Z. Wang, S. Shiochi, Y.M. Li, Simulation and experimental [39] Meteotest. Meteonorm – Global Meteorological Database, Meteotest, 2015.
validation of the variable-refrigerant-volume (VRV) air-conditioning system in [40] J.W. Moon, S.K. Jung, Development of a thermal control algorithm using
EnergyPlus, Energy Build. 40 (6) (2008) 1041–1047, http://dx.doi.org/10.1016/ artificial neural network models for improved thermal comfort and energy
j.enbuild.2007.04.025. efficiency in accommodation buildings, Appl. Therm. Eng. 103 (2016) 1135–
[35] R. Raustad, Verification of a VRF heat pump computer model in EnergyPlus, 1144, http://dx.doi.org/10.1016/j.applthermaleng.2016.05.002.
ASHRAE Trans. 119 (2013) 101.