Transient and steady-state experimental investigation of flash tank vapor injection heat pump cycle control strategy

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 1 9 2 2e1 9 3 3

www. i ifi i r .org

ava i lab le at www.sc iencedi rec t .com

journa l homepage : www.e lsev ier . com/ loca te / i j re f r ig

Transient and steady-state experimental investigation offlash tank vapor injection heat pump cycle control strategy

Xing Xu, Yunho Hwang*, Reinhard Radermacher

Center for Environmental Energy Engineering, University of Maryland, College Park, 3163 Glenn L. Martin Hall Bldg., MD 20742, USA

a r t i c l e i n f o

Article history:

Received 28 February 2011

Received in revised form

18 July 2011

Accepted 11 August 2011

Available online 17 August 2011

Keywords:

Tank

Injection

Transient state

Steady state

Control

* Corresponding author. Tel.: þ1 301 405 524E-mail address: [email protected] (Y. H

0140-7007/$ e see front matter ª 2011 Elsevdoi:10.1016/j.ijrefrig.2011.08.003

a b s t r a c t

Recent research on vapor injection technique has been mostly focused on performance

improvement using different system configurations. The flash tank cycle typically shows

better performance than the internal heat exchanger cycle. However, the flash tank cycle

control strategy is not yet clearly defined. In this study, a novel cycle control strategy is

proposed for an R-410A vapor injection flash tank heat pump system and its feasibility was

experimentally investigated. The proposed novel cycle control strategy utilized an elec-

tronic expansion valve (EEV) for the upper-stage expansion and a thermostatic expansion

valve for the lower-stage expansion, and applied an electric heater in the vapor injection

line to introduce superheat to the injected vapor by providing a control signal to the upper-

stage EEV. Both transient and steady-state system behaviors were studied. The proposed

cycle control strategy was found to be able to provide reliable control to the system.

ª 2011 Elsevier Ltd and IIR. All rights reserved.

Etude experimentale en regime transitoire et permanent de lastrategie de regulation appliquee a un cycle a pompe a chaleuravec injection de vapeur du reservoir intermediaire

Mots cles : reservoir ; injection ; regime transitoire ; regime permanent ; regulation

1. Introduction

As the scroll compressor has been widely used in the air

conditioning/heat pumping industry, there have been

numerous research projects focused on applying vapor

injection technology in air conditioning/heat pumping

systems. It’s already known that there are two typical cycles

for vapor injection: the internal heat exchanger cycle and the

7.wang).ier Ltd and IIR. All rights

flash tank cycle, as shown in Figs. 1 and 2, respectively. The

wide injection ratio range of the internal heat exchanger cycle

is more favorable, because the injected vapor superheat could

be easily adjusted by the upper-stage expansion valve.

However, the performance of the internal heat exchange cycle

is lower than that of the flash tank cycle (Wang, 2008; Ma and

Zhao, 2008). This is due to the fact that the pressure drop

across the internal heat exchanger yields a reduced injection

reserved.

mailto:[email protected]

http://www.iifiir.org

http://www.sciencedirect.com

http://www.elsevier.com/locate/ijrefrig

http://dx.doi.org/10.1016/j.ijrefrig.2011.08.003



Nomenclature

COP coefficient of performance

EEV electronic expansion valve

h refrigerant enthalpy, kJ kg�1

_m mass flow rate, kg s�1

Ptotal total power consumption

PID proportional integral derivative

Q Cooling and heating capacity

RPM revolution per minute

TXV thermostatic expansion valve

Subscripts

air air side

in inlet condition

out outlet condition

ref refrigerant side

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 4 ( 2 0 1 1 ) 1 9 2 2e1 9 3 3 1923

pressure, which results in a reduced injection ratio. Moreover,

the cost of the flash tank is expected to be less than that of an

internal heat exchanger. Therefore, the flash tank cycle has

received more attention in recent years.

The research of flash tank cycle falls into two categories:

system level research and component level research (Xu et al.,

2011). The system level research is mostly focused on inves-

tigating the potential improvement under extreme weather

conditions by applying vapor injection technology. In low

temperature heating scenarios, Wang (2008) tested the flash

tank cycle at �18 �C and found the maximum capacity and

coefficient of performance (COP) improvements to be 33% and

23%, respectively. Bertsch and Groll (2008) tested a two-stage

heat pump system at very low ambient temperature of

�30 �C. The heating COP was found to be 2.1. For heat pump

water heating, Fan et al. (2008) tested a heat pump prototype

at the ambient temperature as low as �30 �C. The hot water

temperature was found to be between 55 �C and 60 �C. Choet al. (2009) studied a two-stage CO2 vapor injection cycle for

the cooling mode. The cooling COP of such cycle was

improved by 16.5% compared with that of the two-stage non-

injection cycle. In high ambient temperatures, Wang (2008)

also tested the flash tank cycle at 46 �C and found the COP

and capacity improvement to be 2% and 15%, respectively.

Research has also been conducted on components of the flash

tank cycle in order to optimize the overall system perfor-

mance. Examples include research on the different types of

compressors (Park et al., 2002; Liu et al., 2008), injection port

location (Wang et al., 2009a), injection pressure (Wang et al.,

2009a), and injection ratio (Winandy and Lebrun, 2002; Wang

et al., 2009b).

Despite the large amount of research efforts on the flash

tank cycle, few open publications exist regarding flash tank

control strategy. In the internal heat exchanger cycle, the

Condenser

Evaporator

Vapor-injected

CompressorInternal heat

exchanger

Upper-stage

expansion

valve

Lower-stage

expansion

valve

Fig. 1 e Internal heat exchanger vapor injection cycle.

upper-stage expansion valve can be controlled by the degree

of superheat in the internal heat exchanger outlet. However,

in the flash tank cycle, the injected vapor is in a saturated

state, therefore a conventional thermostatic expansion valve

(TXV) would not function properly. Moreover, the control of

the liquid level in the flash tank is critical to the reliability of

the system. The liquid level should be controlled in a manner

such that only vapor refrigerant is injected to the compressor.

Xu et al. (2010) provided detailed analysis of the flash tank

cycle control strategy, yet more experimental data is needed

to validate the proposed approach of using a TXV for the

upper-stage expansion valve control. Jang et al. (2010) pre-

sented the method of using the flash tank inlet and outlet

mass balance in order to estimate the liquid level in the flash

tank. However, it’s not clear whether additional sensors, such

as mass flow meters, are needed to accurately measure the

refrigerant mass flow rate. Consequently, the development of

an effective flash tank cycle control strategy is essential to the

industrial application of the flash tank cycle.

The purpose of the current research is to experimentally

investigate the control strategy of the flash tank cycle for both

transient and steady-state operations. In addition, the liquid

level variations under different operation conditions have

been studied.

2. Experimental setup

2.1. Test facility and instrumentation

Fig. 3 shows the schematic of the test facility of a vapor

injection cycle, with a flash tank using R-410A as the refrig-

erant. It is comprised of a closed air loop and units located in

the environmental chamber. In the closed air loop, the air is

Condenser

Evaporator

Vapor-injected

CompressorFlash

tank

Upper-stage

expansion

valve

Lower-stage

expansion

valve

Fig. 2 e Flash tank vapor injection cycle.



P

P

T

T

P T

R.H.

R.H.

R.H.

Air Handing Unit

Air Flow

Nozzle

HumidifierIndoor Unit

Humidity sensor

9-thermocouple grid

Expansion valve

Check valve

Pressure transducer

Temperature transducer

Three way valve Mass f low rate meter

P T

P T

P T

P T

P TOutdoor coil

Outdoor unit

CLOSED AIR LOOP

ENVIRONMENTAL CHAMBER

Flash tank

Shut-of f valve

P T

1

2

3

4

Four way valve Sight glass

Compressor

Fig. 3 e Schematic of the test facility for a vapor injection flash tank cycle.


driven by the blower of the air handling unit. The air flows

through the nozzle, which measures the air volume flow rate,

and then enters the indoor unit. Within the inlet and outlet of

the indoor unit, two 9-thermocouple grids measure the

temperatures of the inlet and outlet air, respectively. Relative

humidity sensors were installed to measure the relative

humidity of the inlet and outlet air, respectively. An outdoor

unit is located in the environmental chamber. Thermocouples

and dew point sensors were installed to measure the air side

inlet and outlet temperatures and dew points, respectively. In

the cooling mode, the refrigerant leaves the compressor,

entering the outdoor unit for condensing. After the upper-

stage expansion valve (Bertsch and Groll, 2008), the refrig-

erant enters the flash tank; the vapor refrigerant is injected to

the compressor, while the liquid refrigerant enters the lower-

stage expansion valve (Fan et al., 2008), and circulates through

the indoor unit. After evaporating at the indoor unit, the

refrigerant then enters the suction port of the compressor to

complete the cycle. In the heating mode, the refrigerant

leaving the compressor circulates through the indoor unit for

condensing; then it is expanded through the upper-stage

expansion valve (Bertsch and Groll, 2008), and enters the

flash tank. The vapor refrigerant is injected to the compressor;

meanwhile the liquid refrigerant circulates through the lower-

stage expansion valve (Cho et al., 2009), evaporates in the

outdoor coil, and then enters the compressor to complete the

cycle. Pressure transducers and in-stream thermocouples

were installed in the system to measure the refrigerant-side

pressures and temperatures, respectively. Mass flow meters

were installed tomeasure the refrigerantmass flow rate of the

injected vapor and through the condenser. A watt meter was

installed to measure the compressor and outdoor power

consumption.

The compressor used in the experimental study is

a vapor-injected scroll compressor. It has a constant speed of

3500 RPMwith a displacement of 29.5 cm3. The specifications

of the outdoor and indoor heat exchangers are shown in

Table 1. It should be noted that the upper-stage expansion

valve (Bertsch and Groll, 2008) used in the system is an

electronic expansion valve (EEV) that has 500 steps. The

lower-stage expansion valves (Cho et al., 2009) and (Fan et al.,

2008), used in the system, are TXVs with a nominal capacity

of 14e21 kW. The vapor injection control valve (ANSI/

ASHRAE Standard 116-1995) is a manually operated meter-

ing valve. In the vapor injection test, this valve is fully

opened to allow vapor refrigerant to be injected to the

compressor.

Fig. 4 shows the flash tank used in the test. The two-phase

refrigerant enters the flash tank in themiddle part of the tank,

and is then separated into liquid and vapor phases. Liquid

refrigerant exits the flash tank from the port located at the

bottom, and vapor refrigerant leaves the flash tank from the

port at the top. A sight glass was installed in the flash tank to

monitor the liquid level, as well as to visualize the liquid-

evapor separation in the flash tank. A capacitance liquid level

sensor was installed through the top of the flash tank to



Table 2 e Specifications of the flash tank.

Parameter Unit Dimension

Flash tank height m 0.32

Diameter m 0.07

Flash tank volume m3 0.001

Sight glass height m 0.15

Table 1 e Specifications of the outdoor and indoor heatexchangers.

Parameter Unit Outdoor HeatExchanger

Indoor HeatExchanger

Tube length mm 2565 483

Tube outer

diameter

mm 7.9 9.5

Tube wall

thickness

mm 0.8 0.8

Tubes per bank e 32 26

Number of tube

banks

e 2 3

Coil in parallel e 1 2

Tube horizontal

spacing

mm 15.7 25.4

Tube vertical

spacing

mm 24.1 25.4

Fins per inch e 22 12

Fin thickness mm 0.1 0.1

Fin types e Wavy fin Wavy fin


measure the liquid level inside of the flash tank. The capaci-

tance liquid level sensor has a length of 25.4 cm that is

equivalent to 80% of the flash tank height. Therefore the

sensor cannot measure the liquid level that is lower than 20%

of the flash tank height, since the sensor was installed

through the top of the flash tank. The main purpose is to

monitor the liquid level variations during different operating

conditions. Specifications of the flash tank are summarized in

Table 2. The uncertainties of all the sensors used in the

experimental study and performance parameters are shown

in Table 3.

Fig. 4 e Flash tank used in the experiment.

2.2. Control strategy analysis

In order to ensure no liquid refrigerant being provided to the

injection port, an additional device is needed in order to

control the upper-stage EEV. In this experimental study, an

electric heater was attached to the surface of the vapor

injection tube in order to provide superheat to the injected

vapor. The schematic of the flash tank cycle control strategy is

shown in Fig. 5. The heater power was varied to investigate

the performance variations at different degrees of superheat.

The superheatwasmeasured at the location of vapor injection

to the compressor. The EEV utilized the injected vapor

superheat to control its opening. This avoided the additional

cost of installing a liquid level sensor in potential industrial

application.

The EEV control was programmed in data acquisition

software, and the proportional-integral-derivative (PID)

control function built in data acquisition softwarewas utilized

to control the EEV. The user interface of the EEV-PID controller

is shown in Fig. 6. This interface provides the convenience to

control the system either by automatic or manual control. In

the automatic control mode, the PID controller automatically

controls the injected vapor superheat to reach the set point

value by regulating the upper-stage EEV opening. In the

manual mode, the EEV opening can be controlled by the

manual input value of the valve opening. In the experimental

study, the automatic control mode was used to reach the

automatic control of the system.

The injected vapor superheat can be used as the control

signal because it can effectively avoid the liquid refrigerant

injected to the compressor. The control algorithm works as

follows: A target degree of superheat is assigned to the PID

controller to control the EEV opening. If the liquid refrigerant

Table 3 e Uncertainties of sensors and calculatedparameters in the experimental study.

Sensor and Parameter Uncertainty

T type thermocouple (range: -200e350 �C) �0.5 �CPressure transducer (range: 0e3447 kPa) �3.79 kPa

Pressure transducer (range: 0e6895 kPa) �8.62 kPa

Relative humidity sensor (range: 0%e100%) �1.0%

Dew point sensor (range: -80e95 �C) �0.2 �CLiquid level sensor (range: 0e25.4 cm) �0.25 cm

Mass flow meter (range: 0e100 g s�1) �0.2% of flow rate

Watt meter (range: 0e5 kW) �0.5% of full scale

Cooling and heating capacity �3.6% of calculated

value

Cooling and heating COP �5.7% of calculated

value



Condenser

Evaporator

Compressor

Upper-stage

expansion

valve (EEV)

Lower-stage

expansion

valve (TXV)

Electric

heater

Sensing bulb

PID controller

Injected

vapor

superheat

Shutoff

valve

Fig. 5 e Schematic of the control strategy for the flash tank

cycle.


is to be injected with vapor refrigerant to the compressor by

the liquid flooding in the flash tank, the superheat of the

injected vapor would decrease rapidly. In this event, the

upper-stage EEV would reduce its opening to maintain the

target degree of superheat. This reduces the amount of liquid

flowing from the condenser to the flash tank, which reduces

the flash tank liquid level.

3. Test conditions

Both cooling and heating tests were conducted to evaluate the

system performance. The volume flow rate of the air circu-

lating in the closed air loop was set to be 0.58 m3 s�1. The test

conditions followed the ASHRAE Standard (1995), and are

illustrated in Table 4. Moreover, extended conditions of 46.1 �Cfor cooling and �17.8 �C for heating were added to investigate

Fig. 6 e The user interface of the EEV-PID controller.

the system behaviors at severe weather conditions. The

electric heater power input and the injected vapor superheat

were varied to investigate their effect on the system perfor-

mance as well as the liquid level variations in the flash tank.

4. Performance evaluation

In transient conditions, the system cooling and heating

capacities rely on the air side performance, since it’s difficult

to accurately obtain the refrigerant-side enthalpy in the first

few minutes after the system is started. In this scenario, the

capacity is calculated in Equation (1):

Qair ¼ _mair

�hair;out � hair;in

�(1)

Where _mair is the air mass flow rate; hair;out is the outlet air

enthalpy at the indoor heat exchanger, and hair;in is the inlet air

enthalpy at the indoor heat exchanger. The system cooling

and heating COP is defined in Equation (2):

COP ¼ Qair

Ptotal(2)

Where Ptotal is the total power consumption, which includes

the heat pump system power consumption and the electric

heater power consumption.

For steady-state operation, refrigerant-side performance

was used for the evaluation since the measurement in the

refrigerant side is more accurate than the air side. In this

scenario, the capacity is calculated in Equation (3):

Qref ¼ _mref

�href;out � href;in

�(3)

Where _mref is the refrigerant mass flow rate; href;out is the

refrigerant enthalpy at the indoor heat exchanger outlet, and

href;in is the refrigerant enthalpy at the indoor heat exchanger

inlet. The system cooling and heating COP is defined in

Equation (4):

COP ¼ Qref

Ptotal(4)

5. Experimental results

5.1. System startup

Ideally, the injected vapor superheat should always be main-

tained positive if the PID controller functions properly,

regardless of steady-state or transient system operations. The

first step is to examine the superheat variations during the

system startup to investigate the controllability of the PID

controller. Fig. 7 shows the injected vapor superheat varia-

tions during the system startup at different operating condi-

tions and heater power input. It can be seen that the injected

vapor superheat could always be controlled to reach the target

superheat within a relatively short period of time with

different superheat settings. The degree of superheat could

also be maintained to be non-negative to avoid liquid refrig-

erant being injected to the compressor. Moreover, the liquid

level variation is also closely related to the injected vapor

superheat. Fig. 8 shows the flash tank liquid level variations



Table 4 e Test conditions.

Test Indoor Outdoor Operation

DB WB RH DB WB RH DP

Extended condition 26.7 �C 19.4 �C 50.66% 46.1 �C NA NA NA Steady state cooling

A 35.0 �C Steady state cooling

B 27.8 �C steady state cooling

C �13.9 �C �21.41% Steady state cooling, dry coil

D Cyclic cooling, dry coil

High temp2 21.1 �C �15.6 �C �56.42% 8.3 �C 6.1 �C 72.9% 3.7 �C Steady state heating

High temp1 16.7 �C 14.7 �C 81.1% 13.4 �C Steady state heating

Low temp �8.3 �C �9.4 �C 69.8% �12.3 �C Steady state heating

High temp cyclic 8.3 �C 6.1 �C 72.9% 3.7 �C Cyclic heating

Frost acc. 1.7 �C 0.6 �C 82.0% �0.9 �C Steady state defrost

Extended condition �17.8 �C NA NA NA Steady state heating


during the system startup at different operating conditions.

Some fluctuations could be seen in the first 3 min, and this

was due to the opening and closing of the EEV initiated by the

PID controller. It should be noted that the liquid level sensor

can only measure 80% of the flash tank height from the top of

the tank, and therefore as the liquid level is close to or lower

than 20% of the tank height, the liquid level appears to be

perfect steady. The liquid level never exceeds 50% of the entire

tank height, and this ensures reliable system operation. The

injected and total refrigerant mass flow rates in the system

startup are shown in Fig. 9 and Fig. 10, respectively. Large

variations can be observed in the first 5 min, and then the

mass flow rates tend to be steady.

5.2. Cooling cyclic test

In the real application, the air conditioning/heat pump

systems are turned on and off frequently. Therefore, it’s

worthwhile to investigate the system’s cyclic performance.

According to ASHRAE Standard (1995), cyclic cooling and

heating tests are comprised of 6 min compressor “on” time,

followed by 24 min compressor “off” time.

5.2.1. PID tuning of cooling cyclic testPID gains are critical factors that affect the transient behavior

of the PID controller, and therefore it’s necessary to find out

-2

0

2

4

6

8

10

12

14

0 1 2 3 4 5 6 7 8 9 10

Time [min]

Inje

cte

d v

ap

or s

up

erh

ea

t [

K]

Cooling: 46.1ºC

Cooling: 35.0ºC

Heating: 8.3ºC

Fig. 7 e Injected vapor superheat variations during the

system startup at different operating conditions.

the appropriate PID gains in order to reach optimum perfor-

mance for the system control. The output of the PID controller

in this study was normalized to be �100 to 100 to obtain more

general results for PID gains. The proportional gain makes

a change to the output that is proportional to the current error

value. The integral gain is the sum of the instantaneous error

over time and gives the accumulated offset that should have

been corrected previously. The derivative of the process error

is calculated by determining the slope of the error over time

and multiplying this rate of change by the derivative gain.

Different combinations of PID gains were tried, and from the

experiment it was found out that the EEV experienced large

variations when the derivative gain was used, and therefore

the derivative gain was set to be zero. Fig. 11 shows the

injected vapor superheat variations of PID tuning results of

cyclic cooling test. From the results it can be seen that small P

and I gains tend to yield a large variation, andwith larger P and

I gains the control curve looks better. P¼ 5.0, I¼ 0.5, D¼ 0, and

P ¼ 10, I ¼ 0.5, D ¼ 0, together with P ¼ 15, I ¼ 1.0, D ¼ 0 out-

performed other three different combinations of PID gains.

Fig. 12 shows the EEV opening variations during the PID tuning

of cooling cyclic test. The combination of P ¼ 2.0, I ¼ 0.2, D ¼ 0

and P ¼ 5.0, I ¼ 0.5, D ¼ 0 yield the least variations, and other

PID gains result in relative large variation of the EEV opening.

Fig. 13 shows the variations of indoor heat exchanger outlet

air temperature variations. It can be observed that with

0

5

10

15

20

25

30

35

40

45

50

0 1 2 3 4 5 6 7 8 9 10

Time [min]

Liq

uid

le

ve

l p

erc

en

ta

ge

[%

]

Cooling: 46.1ºC

Cooling: 35.0ºC

Heating: 8.3ºC

Fig. 8 e Flash tank liquid level variations during the system

startup at different operating conditions.



-2

0

2

4

6

8

10

12

0 1 2 3 4 5 6

Time [min]

Inje

cte

d v

ap

or s

up

erh

ea

t [

K]

P=1.0; I=0.02; D=0 P=2.0; I=0.2; D=0

P=5.0; I=0.1; D=0 P=5.0; I=0.5; D=0

P=10; I=0.5; D=0 P=15; I=1.0; D=0

Target superheat

Fig. 11 e PID tuning of cooling cyclic test: injected

superheat variations.

0

5

10

15

20

0 1 2 3 4 5 6 7 8 9 10

Time [min]

Inje

cte

d r

efrig

era

nt m

as

s f

low

ra

te

[g

/s]

Cooling: 46.1ºC

Cooling: 35.0ºC

Heating: 8.3ºC

Fig. 9 e Injected refrigerant mass flow rate variations

during the system startup at different operating

conditions.


P¼ 2.0, I¼ 0.2, D¼ 0 and P¼ 5.0, I¼ 0.5, D ¼ 0, the temperature

variation is more smooth, and this corresponds to the varia-

tions of EEV opening, as shown in Fig. 12. From the perfor-

mance point of view, P ¼ 5.0, I ¼ 0.5, D ¼ 0 outperformed

P ¼ 2.0, I ¼ 0.2, D ¼ 0. As a consequence, considering the time

for the injected vapor superheat to reach steady state, the

variation of the EEV opening, and the air side performance,

P ¼ 5.0, I ¼ 0.5, D ¼ 0 was found to be the most appropriate PID

gains for the cooling cyclic control.

5.2.2. Time delay to initiate the PID controllerAfter selecting the optimum PID gains, it’s also interesting to

study whether there is any difference if the PID controller is

initiated with different time delays after the system is started.

Since the cyclic test requires the system to be turned “on” for

6 min, and then turned “off”, therefore the delay time was

selected to be between 1 min and 5 min Fig. 14 shows the

indoor heat exchanger air outlet temperature variations with

different delay time to initiate the PID controller. It can be seen

that only with 1 min delay time, the transition of air side

temperature is not smooth, and there is no difference between

other delay time and the case without delay initiating the PID

0

20

40

60

80

100

0 1 2 3 4 5 6 7 8 9 10

Time [min]

To

ta

l re

frig

era

nt m

as

s f

low

ra

te

[g

/s]

Cooling: 46.1ºC

Cooling: 35.0ºC

Heating: 8.3ºC

Fig. 10 e Total refrigerant mass flow rate variations during

the system startup at different operating conditions.

controller. Therefore it’s recommended to turn on the PID

controller when the system is started, and no delay for the PID

controller is needed.

5.2.3. Cooling cyclic test resultsFig. 15 shows the cooling cyclic test results with vapor injec-

tion “on” and “off”. It can be seen that the indoor heat

exchanger outlet air temperature with vapor injection “on” is

lower than with vapor injection “off”, which means that the

cooling capacity delivered by vapor injection “on” is higher

than with vapor injection “off”. Calculation shows the

improvement is 10.1%. However, the power consumption also

becomes higher as vapor injection is initiated, which also can

be seen from Fig. 15. This results in a degradation of cooling

COP of 2.4%, which is within measurement uncertainty.

Therefore it seems that vapor injection can still be beneficial

when larger cooling capacity is needed in mild temperature

conditions.

5.3. Heating cyclic test

5.3.1. PID tuning of heating cyclic testPID tuning was also conducted for heating cyclic test in

a similar manner as the cooling cyclic test. Fig. 16 shows the

0

10

20

30

40

50

60

70

80

0 1 2 3 4 5 6

Time [min]

EE

V o

pe

nin

g [

%]

P=1.0; I=0.02; D=0 P=2.0; I=0.2; D=0

P=5.0; I=0.1; D=0 P=5.0; I=0.5; D=0

P=10; I=0.5; D=0 P=15; I=1.0; D=0

Fig. 12 e PID tuning of cooling cyclic test: EEV opening

variations.



0

5

10

15

20

25

30

0 1 2 3 4 5 6

Time [min]

Ou

tle

t a

ir t

em

pe

ra

tu

re

[ºC

]

0

0.5

1

1.5

2

2.5

3

Po

we

r c

on

su

mp

tio

n [

kW

]

VI off - inlet air VI on - inlet air

VI off - outlet air VI on - outlet air

VI off - power consumption VI on - power consumption

Fig. 15 e Cooling cyclic test: comparison between VI “on”

and VI “off”.

10

12

14

16

18

20

22

24

26

28

0 1 2 3 4 5 6

Time [min]

Ou

tle

t a

ir t

em

pe

ra

tu

re

[ºC

]

P=1.0; I=0.02; D=0 P=2.0; I=0.2; D=0

P=5.0; I=0.1; D=0 P=5.0; I=0.5; D=0

P=10; I=0.5; D=0 P=15; I=1.0; D=0

Fig. 13 e PID tuning of cooling cyclic test: indoor heat

exchanger outlet air temperature.


injected vapor superheat variations using different combina-

tions of PID gains. Similar trend could be seen as the cooling

cyclic test. Small P and I gains tend to yield a large variation of

the control variable, and P¼ 10, I¼ 0.5,D¼ 0 and P¼ 15, I¼ 1.0,

D ¼ 0 result in the least variations of the injected vapor

superheat. Fig. 17 shows the EEV opening variations with

different PID gains. P ¼ 1.0, I ¼ 0.02, D ¼ 0 and P ¼ 5.0, I ¼ 0.1,

D ¼ 1 result in large variations of the EEV opening, and other

PID gains lead to less variations of the EEV opening. Fig. 18

illustrates the indoor heat exchanger outlet temperature

variations with different PID gains. The optimum perfor-

mance was reached when P ¼ 10, I ¼ 0.5, D ¼ 0. Therefore,

P ¼ 10, I ¼ 0.5 and D ¼ 0 was selected for the PID gains for

heating cyclic test.

5.3.2. Time delay to initiate the PID controllerDifferent time delays for the heating cyclic test were also

performed. Fig. 19 shows the air outlet temperature variations

with different time delays to initiate the PID controller. It was

also found thatmaximumperformance can be achievedwhen

there is no delay in turning on the PID controller, whichmeans

that the PID controller was turned on as soon as the system is

turned on. Therefore, there is no need for time delay in initi-

ating the PID controller.

0

5

10

15

20

25

30

0 1 2 3 4 5 6

Time [min]

Ou

tle

t a

ir t

em

pe

ra

tu

re

[ºC

]

1 min delay

2 min delay

3 min delay

4 min delay

5 min delay

No delay

Fig. 14 e Cooling cyclic test: performance variations with

different time delays for PID controller.

5.3.3. Heating cyclic test resultsFig. 20 shows the heating cyclic test results with vapor injec-

tion “on” and “off”. It can be seen that there is no visible

difference in the air outlet temperature for vapor injection

“on” and “off” modes, which indicates that the heating

capacity with vapor injection “on” and “off” is almost the

same. Calculation shows that the improvement is only 0.6%,

which is within the measurement uncertainty. However, the

power consumption with vapor injection “on” is significantly

higher than that with vapor injection “off”, and this yields

a degradation of heating COP of 13.4%. Therefore it’s recom-

mended to turn off vapor injection during mild temperature

conditions since no benefit could be achieved.

5.4. Cyclic test results and discussions

From the cooling and heating cyclic tests it can be seen that

PID gains are important parameters that significantly affect

the system performance. P ¼ 5.0, I ¼ 0.5, D ¼ 0 was found to

be the most appropriate PID gains for the cooling cyclic

control, and P ¼ 10, I ¼ 0.5 and D ¼ 0 were most suitable PID

gains for heating cyclic test. The difference in P gain value in

cooling and heating tests may be due to different perfor-

mance behavior of lower-stage expansion valve. Although

-4

-2

0

2

4

6

8

10

12

0 1 2 3 4 5 6

Time [min]

Inje

cte

d v

ap

or s

up

erh

ea

t [

K]

P=1.0; I=0.02; D=0 P=2.0; I=0.2; D=0

P=5.0; I=0.1; D=0 P=5.0; I=0.5; D=0

P=10; I=0.5; D=0 P=15; I=1.0; D=0

Target superheat

Fig. 16 e PID tuning of heating cyclic test: injected

superheat variations.



15

20

25

30

35

40

0 1 2 3 4 5 6

Time [min]

Ou

tle

t a

ir t

em

pe

ra

tu

re

[ºC

]

1 min delay

2 min delay

3 min delay

4 min delay

5 min delay

No delay

Fig. 19 e Heating cyclic test: performance variations with

different time delays for PID controller.

0

10

20

30

40

50

60

0 1 2 3 4 5 6

Time [min]

EE

V o

pe

nin

g [

%]

P=1.0; I=0.02; D=0 P=2.0; I=0.2; D=0

P=5.0; I=0.1; D=0 P=5.0; I=0.5; D=0

P=10; I=0.5; D=0 P=15; I=1.0; D=0

Fig. 17 e PID tuning of heating cyclic test: EEV opening

variations.


lower-stage expansion valve for cooling and heating test has

the same nominal capacity, the manufacture imperfection

may introduce difference in its response, which further

affects the PID gains for the upper-stage expansion valve.

Different time delays were tried for cooling and heat cyclic

tests, and it was found that the initiating the PID controller

following the start of system yields the best performance.

For cooling cyclic mode, a capacity improvement of 10.1%

was observed, and a COP degradation of 2.4% was noticed.

For heating mode, a capacity improvement of 0.6% was

observed, yet a COP degradation of 13.4% was seen.

However, it should be noted that the cooling and heating

COP at the cyclic conditions with vapor injection “on” are not

as good as the scenarios with vapor injection “off” are not

simply due to the PID controller, but also due to the fact that

the current system only gives COP improvement at low

ambient temperatures. The current system was evaluated

regarding vapor injection “off” and “on” modes at different

cooling and heating temperatures without any PID

controller, as shown in Fig. 21. It can be seen that the COP

improvement was only found at �8.3 �C and �17.8 �Cconditions, and therefore the PID controller itself should not

account for all the COP degradations.

15

20

25

30

35

40

0 1 2 3 4 5 6

Time [min]

Ou

tle

t a

ir t

em

pe

ra

tu

re

[ºC

]

P=1.0; I=0.02; D=0 P=2.0; I=0.2; D=0

P=5.0; I=0.1; D=0 P=5.0; I=0.5; D=0

P=10; I=0.5; D=0 P=15; I=1.0; D=0

Fig. 18 e PID tuning of heating cyclic test: indoor heat

exchanger outlet air temperature.

5.5. Steady-state behavior of different injectionsuperheats and heater power input

The injected vapor superheat is affected by two factors: the

heater supply power and EEV opening. The EEV opening

affects the injected vapor superheat because it controls the

injection pressure in the flash tank. The system performance

varies depending on the heater power input as well as the

selected superheat setting. Both cooling and heating tests

were conducted.

5.5.1. Steady state cooling testFig. 22 illustrates the steady-state test results of cooling COP

variations with different ambient conditions, heater power

input and degrees of superheat. For 27.8 �C condition, the COP

variationswith 60W and 90Wheater power input is small, yet

the COP tends to decrease with 30 W heater power input. This

is because with the same target superheat setting, decreasing

the heat power input lowers the injection pressure in the flash

tank in order to match up the superheat. Moreover, if the

heater power input is kept constant, increasing the target

superheat value also results in decreased injection pressure in

order tomatch the increasing superheat. This can also be seen

from the liquid level variations, as shown in Fig. 23. The

0

5

10

15

20

25

30

35

40

0 1 2 3 4 5 6

Time [min]

Ou

tle

t a

ir t

em

pe

ra

tu

re

[ºC

]

0

0.5

1

1.5

2

2.5

3

Po

we

r c

on

su

mp

tio

n [

kW

]

VI off - inlet air VI on - inlet air

VI off - outlet air VI on - outlet air

VI off - power consumption VI on - power consumption

Fig. 20 e Heating cyclic test: comparison between VI “on”

and VI “off”.



0

20

40

60

80

100

120

0 2 4 6 8 10

Degree of superheat [K]

Liq

uid

le

ve

l h

eig

ht p

erc

en

ta

ge

[%

]

27.8ºC - 30W heater power 27.8ºC - 60W heater power



Fig. 23 e Flash tank liquid level variations at different

steady-state cooling ambient conditions, heater power

input and degrees of superheat.

-5%

0%

5%

10%

15%

20%

25%

30%

28.2%

[46.1ºC]

15.6%

[35.0ºC]

10.8%

[27.8ºC]

21.0%

[8.3ºC]

25.8%

[-8.3ºC]

28.4%

[-17.8ºC]

Maximum injection ratio

Imp

ro

ve

me

nt

Capacity improvement

COP improvement

Cooling Heating

Fig. 21 e Capacity and COP improvements at different

ambient conditions with the maximum injection ratio

compared to vapor injection “off”, steady-state

performance without any PID controller.


average liquid level for 60 W heater power and 90 W heater

power is higher than that with 30 W heater power at the

temperature of 27.8 �C. Higher liquid level indicates higher

injection pressure, because the upper-stage expansion valve’s

opening is increased in order to increase the injection pres-

sure, and also results in more liquid flowing into the flash

tank. The COP variations for 35.0 �C and 46.1 �C are relatively

small, and this indicates that the performance is not quite

sensitive when the heater power varies from 60 W to 90 W.

The cooling capacity variations show similar trend as the

cooling COP variations, which is shown in Fig. 24.

5.5.2. Steady state heating testThe heating performance variations are different from that of

the cooling results. Fig. 25 shows the steady-state heating COP

variations at different ambient conditions, heater power input

and degrees of superheat. At 8.3 �C condition, the general

trend is that higher heater power input yields higher heating

COP, and lowering the target superheat also results in higher

COP. An explanation for this is that when the target superheat

is set to be a constant value, increasing the heater power input

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

0 2 4 6 8 10


Co

olin

g C

OP




Fig. 22 e Steady-state test of cooling COP variations at

different ambient conditions, heater power input and

degrees of superheat.

induces the EEV to enlarge its opening in order to increase the

injection pressure in the flash tank, thus maintaining the

same degree of superheat. This results in higher injection

ratio and improves heating COP. Moreover, the liquid level in

the flash tank is also increased. Likewise, if the heater power

input is kept constant, decreasing the target superheat value

also results in increased injection pressure in order to match

the decreasing superheat. For �17.8 �C condition with 60 W

heater power input, the trend is similar to that of 8.3 �Ccondition. However, as the heater power input increases to

90 W, the heating COP remains almost constant. This is

because in this condition, the system injection pressure

already reaches its maximum, and therefore lowering the

superheat cannot increase the injection ratio anymore. This

can also be seen in the liquid level variations, as shown in

Fig. 26. At �17.8 �C with 60 W heater power input, decreasing

the superheat shows a trend of increasing liquid level height.

However, for 90 W heater power input, the liquid level in the

flash tank is above 80% of the flash tank height. The upper-

stage expansion valve’s opening cannot be increased

anymore as the superheat decreases because it would result

in liquid flooding the compressor. In this scenario the system

8

9

10

11

12

13

14

0 2 4 6 8 10


Co

olin

g c

ap

ac

ity

[k

W]




Fig. 24 e Steady-state test of cooling capacity variations at





4

6

8

10

12

14

0 2 4 6 8 10


He

atin

g c

ap

ac

ity

[k

W]

-17.8ºC - 60W heater power -17.8ºC - 90W heater power


Fig. 27 e Steady-state test of heating capacity variations at



1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

0 2 4 6 8 10


He

atin

g C

OP



Fig. 25 e Steady-state test of heating COP variations at




stability is also compromised since the liquid level is very

high, which is represented in the error bar in the heating COP.

The heating capacity variation shows the same trend as the

heating COP, as shown in Fig. 27.

5.5.3. Steady-state results and discussionsSystem performance is an important factor in selecting the

appropriate heater power input and superheat settings. For

the cooling mode, there is no big difference with 60W or 90W

heater power input, yet 30 W heater power input seems to be

insufficient in reaching optimum performance. The optimum

performancewas observedwith 4 Ke6 K degrees of superheat.

For the heating mode, 90 W heater power input shows better

performance than 60W heater power input. However, it’s also

very important to consider the flash tank liquid level varia-

tions under different operating conditions. For the cooling

mode, the liquid level with 60 W and 90 W power input

exceeds 60% of the flash tank height, and therefore it’s better

to select power input lower than 60 W. For the heating mode,

90 W heater power input yields a liquid level higher than 80%

at the ambient temperature of �17.8 �C. Therefore, 60 W

heater power input is preferred.

0

20

40

60

80

100

120

0 2 4 6 8 10


Liq

uid

le

ve

l h

eig

ht p

erc

en

ta

ge

[%

]



Fig. 26 e Flash tank liquid level variations at different

steady-state heating ambient conditions, heater power

input and degrees of superheat.

In overall, when the systemperformance and reliability are

both considered, 4 Ke6 K degrees of superheat is preferred. For

heating mode, 60 W heater power input is preferred to 90 W;

for coolingmode, 30Wheater power input is preferred to 60W

and 90 W.

6. Conclusions

This paper investigates the experimental control strategy of

a flash tank vapor injection heat pump cycle. Both transient

and steady-state performances have been studied. In the

experiments, an electric heater was applied to introduce

positive superheat to the injected vapor. An EEV coupled with

a PID controller in data acquisition software was employed to

provide accurate control of the injected vapor superheat. In

the transient study, different PID gains were investigated

regarding their effects on the system performance, and most

suitable PID gains were obtained for cooling and heating cyclic

tests. Different time delays were tried for initiating the PID

controller, and it was found that initiating the PID controller

following the startup of the system yields the best perfor-

mance, and no time delay is needed. Through experiment it

was found out that the PID controller was able to provide

accurate control of the EEV to reach the target superheat. In

addition, the injected vapor superheat was varied to investi-

gate its effect on the system performance and the flash tank

liquid level variations. This also provides useful information

to analyze the feasibility of using the injected vapor superheat

as the control signal. Through steady-state and transient

cooling and heating tests, it was found that the injected vapor

superheat can be effectively used as the control signal of the

upper-stage expansion valve. The effect of different settings

of superheat and heater power input on the system perfor-

mance was also investigated. Considering the system perfor-

mance and reliability of controlling the flash tank liquid level,

4 Ke6 K degrees of superheat is a recommended value. For

heating mode, 60 W heater power input is preferred to 90 W;

for coolingmode, 30Wheater power input is preferred to 60W

and 90 W.




Acknowledgment

We gratefully acknowledge the support of this effort from the

sponsors of the Alternative Cooling Technologies and Appli-

cations Consortium and the Center for Environmental Energy

Engineering (CEEE) at the University of Maryland, and Emer-

son Climate Technologies.

r e f e r e n c e s

ANSI/ASHRAE Standard 116-1995, 1995. Methods of Testing forRating Seasonal Efficiency of Unitary Air Conditioners andHeat pumps. ASHRAE, Atlanta, GA, USA.

Bertsch, S.S., Groll, E.A., 2008. Two-stage air-source heat pump forresidential heating and cooling applications in northern USclimate. Int. J. Refrig. vol. 31, 1282e1292.

Cho, H., Baek, C., Park, C., Kim, Y., 2009. Performance evaluationof a two-stage CO2 cycle with gas injection in the coolingmode operation. Int. J. Refrig. vol. 32, 40e46.

Fan, S., Liu, Q., He, S., Scroll compressor development for air-source heat pump water heater applications, Proceedings ofInternational Refrigeration and Air Conditioning Conferenceat Purdue, 2008.

Jang, Y., Lee, E., Chin, S., Ha, S., 2010. Effects of flash and vaporinjection on the air-to-air heat pump system. InternationalRefrigeration and Air Conditioning Conference at Purdue.

Liu, F., Huang, H., Ma, Y., Zhuang, R., 2008. An experimental studyon the heat pump water heater system with refrigerantinjection. Proceedings of International Refrigeration and AirConditioning Conference at Purdue.

Ma, G.Y., Zhao, H.X., 2008. Experimental study of a heat pumpsystem with flash tank coupled with scroll compressor. Energ.Build. vol. 40, 697e701.

Park, Y.C., Kim, Y., Cho, H., 2002. Thermodynamic analysison the performance of a variable speed scroll compressorwith refrigerant injection. Int. J. Refrig. vol. 25,1072e1082.

Wang, B., Shi, W., Li, X., 2009a. Numerical analysis on the effectsof refrigerant injection on the scroll compressor. Appl. Therm.Eng. vol. 29, 37e46.

Wang, X., Hwang, Y., Radermacher, R., 2009b. Two-stage heatpump system with vapor-injected scroll compressor usingR410A as a refrigerant. Int. J. Refrig. vol. 32, 1442e1451.

Wang, X. Performance investigation of two-stage heat pumpsystem with vapor injected scroll compressor. Ph.D.Dissertation, University of Maryland College Park, 2008.

Winandy, E.L., Lebrun, J., 2002. Scroll compressors using gas andliquid injection: experimental analysis and modeling. Int. J.Refrig. vol. 25, 1143e1156.

Xu, X., Hwang, Y., Radermacher, R., Pham, H.M., 2010. Controlstrategy of vapor injection cycle. International Refrigerationand Air Conditioning Conference at Purdue.

Xu, X., Hwang, Y., Radermacher, R., 2011. Refrigerant injectionfor heat pumping/air conditioning systems: Literaturereview and challenges discussions. Int. J. Refrig. vol. 34,402e415.



Documents

Transient and steady-state experimental investigation of flash tank vapor injection heat pump cycle control strategy