Experimental investigation of an absorption heat pump with organic working pairs

Applied Thermal Engineering 163 (2019) 114311

Contents lists available at ScienceDirect

Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng

Experimental investigation of an absorption heat pump with organic working pairs

T

⁎

P. Chatzitakisa, , J. Safarovb, F. Opferkucha, B. Dawoudc a

Technische Hochschule Nuernberg Georg Simon Ohm, Nuremberg Campus of Technology, Fuerther Str. 246b, 90429 Nuremberg, Germany University of Rostock, Institute of Technical Thermodynamics, Albert-Einstein-Str. 2, 18059 Rostock, Germany c Ostbayerische Technische Hochschule Regensburg, Laboratory of Sorption Processes, Faculty of Mechanical Engineering, Galgenberg Str. 30, 93053 Regensburg, Germany b

H I GH L IG H T S

molar mass and vap. pressure at desorber temperature, absorbents are favorable. • Low point is not a reliable absorbent characteristic. • Boiling vaporization enthalpy refrigerants are favorable. • Higher • Optimized COP and SSC equations show lower deviations from experiment.

A R T I C LE I N FO

A B S T R A C T

Keywords: Absorption heat pump Coefficient of performance Organic working pairs Specific solution circulation

As part of a systematic approach towards the search for alternative absorption heat pump (AHP) working pairs that could potentially provide comparable performance to conventional ones, a previous work performed a detailed theoretical cycle analysis and simulation that revealed concrete correlations between key working fluid thermophysical properties and AHP performance indicators. Following this work, targeted combinations of two organic refrigerants, 2,2,2-trifluoroethanol (TFE) and 2,2,3,3,3-pentafluoropropanol (5FP) and two organic absorbents, 1,3-dimethyl-2-imidazolidinone (DMI) and 2-pyrrolidone (PYR) were tested in a prototype 5 kW AHP, based on a highly compact plate heat exchanger design, which has been previously introduced. The purpose of this effort was to test the findings of the previous work with experimental measurements. The working pair combinations were also subjected to vapor liquid equilibrium (VLE) and viscosity measurements, in order to determine reliable activity coefficient and improve the accuracy of the simulations. The experimental performance data agree well with the COP simulations and show to be consistent with the conclusions derived from the previous theoretical work.

1. Introduction Commercially available vapor absorption heat pump systems have always been dominated by ammonia/water and water/lithium bromide (LiBr) working pairs. Despite their superior thermodynamic performance, critical disadvantages have restricted their adoption and market penetration. More specifically, ammonia/water heat pumps present considerable hazards due to ammonia’s toxicity, corrosiveness and high system pressure whereas water/LiBr systems are considered safer but also handicapped by narrow temperature lifts and even higher corrosion problems. Alternative absorption heat pump working pairs have already been extensively reviewed by many researchers over the years, Macriss et al.

⁎

(1988) [1], Sun et al. (2012) [2], with similar outlooks: The main focus falls on absorbent alternatives for water and ammonia refrigerants, both organic and inorganic. In more recent reviews, Zheng et al. (2014) [3] concentrated exclusively on ionic liquids, whereas Papadopoulos et al. (2019) [4] focused on non-ionic organic working pairs. Iedema (1982) [5] performed a quasi-quantitative analysis of an idealized single and double stage absorption heat pump. The conclusions concerning the thermodynamic performance pointed towards a low ratio of excess mixing enthalpy to vaporization enthalpy, a high vaporization enthalpy, a strong negative deviation from Raoult’s law and low solution specific heat capacities. Hodgett (1982) [6] presented a simplified approach for the coefficient of performance (COP) that depended on three dimensionless parameters, the ratio of excess mixing

Corresponding author. E-mail address: [email protected] (P. Chatzitakis).

https://doi.org/10.1016/j.applthermaleng.2019.114311 Received 2 May 2019; Received in revised form 23 August 2019; Accepted 27 August 2019 Available online 31 August 2019 1359-4311/ © 2019 Elsevier Ltd. All rights reserved.

Applied Thermal Engineering 163 (2019) 114311

P. Chatzitakis, et al.

η ρ

Nomenclature COP CR cp H Mi ṁ j p Q̇ RV SSC T Χ

Coefficient of performance Circulation ratio Average mass specific heat capacity Enthalpy Molecular weight of component i Mass flow of stream j Vapor pressure heat flow Relative volatility Specific solution circulation Temperature Liquid molar fraction

Subscripts 1 2 abs ce cool con des desin eva HP j sol v xsdes

Greek letters γ ζ ξ

Dynamic viscosity Mass density

Activity coefficient Vapor mass fraction Liquid mass fraction

Component 1, refrigerant Component 2, absorbent Absorber conditions Condenser-evaporator average cond. Cooling Condenser conditions Desorber conditions Desorber inlet conditions Evaporator conditions Heat pump Stream number (Fig. 1) Solution Vaporization Excess desorption

enthalpy, a low acentric factor, a low reduced boiling point, a high critical temperature and low critical pressure. On the other hand, the refrigerant should have a high acentric factor, a high reduced boiling point, a low critical temperature and a high critical pressure. Nowaczyk (1991) [10] carried out a literature review summarizing that, the refrigerant should have a high specific vaporization enthalpy, a high critical temperature, a flat vapor pressure curve. In addition, the solution with the absorbent should have a high negative deviation from Raoult’s law, a low excess mixing enthalpy, low specific heat capacity, a low viscosity, a high density and a high difference in boiling points. Alefeld, Radermacher and Hwang (1994) [11] pointed out that the refrigerant should exhibit a low heat capacity to vaporization enthalpy (per unit mass) ratio in order to obtain a high efficiency. Additionally, the resulting pressure ratios between the high pressure and low pressure sides should be low and the refrigerant volumetric heat capacity should be as high as possible. More recent studies, Kühn et al. (2013), have primarily focused on ionic liquids as absorbent replacements [12]. Nevertheless, despite all efforts there is still no recognized alternative working pair with the potential to exceed the impact of the two conventional pairs [13]. The lack of success of these partially qualitative screening processes, as well as the fact that since 2011 the European Chemicals Agency (ECHA) classified 2,2,2-trifluoroethanol (TFE), 1,3-dimethyl-2-imidazolidinone (DMI), N-methyl-2-pyrrolidone and tetraethylene glycol dimethyl ether as toxic to reproduction, and therefore eligible as candidates for the list of substances of very high concern, signified the need for an alternative quantitative approach, towards identification and investigation of alternative substances from the same class to be applied in hardware

enthalpy to vaporization enthalpy, the ratio of pump work to vaporization enthalpy and the generator sensible heat duty to vaporization enthalpy. The final requirements for an efficient working pair included, a high solution density, a low circulation ratio and a low solution specific heat capacity. Perez-Blanco (1984) [7] investigated the influence of non-ideal solutions’ negative deviation from Raoult’s law on the COP and the circulation ratio (CR) of ammonia single stage absorption heat pumps. The conclusions were that there is an optimum temperature and concentration dependence of the activity coefficients and that too strong negative deviations are undesirable. Furthermore, it is stated that the absorbent specific heat capacity must be as low as possible. Eisa & Holland (1987) [8] reviewed previous literature works and collected a list of desirable properties for the working pair. Specifically, the refrigerant should exhibit a high vaporization enthalpy, a low heat capacity per unit mass and low boiling point. The absorbent should have as low a vapor pressure as possible, a low heat capacity per unit mass and create solutions with the refrigerant with high negative deviations from Raoult’s law, low viscosity and density. Furthermore, they also investigated the influence of various ions in water – salt solutions on the heat of solution, vapor pressure lowering and solubility. Narodoslawsky et al. (1988) [9] followed an analytical method based partly on semi-empirical relations and concluded that high performance working pairs should show high refrigerant vaporization enthalpies at normal boiling point and an extremum of excess mixing Gibbs free energy between −1000 and −2000 J/mol preferably located at high refrigerant concentrations. Additionally, it is argued that the absorbent should also have a high boiling point vaporization

Table 1 Molecular structure and key thermophysical properties for the selected refrigerants and absorbents from literature and experimental measurements [17]. TFE

5FP

DMI

PYR

100.04 73.8 2584.0 452 1.9 2.6 1382.9

150.05 81.1 2023.9 311 1.59 3.7 1506

114.15 224.9 31.2 513 1.78 2.0 1052.4

85.1 245 4.5 773 1.95 13.5 1110.2

Molecular structure

Mol. weight Boil. point [°C] p100°C [mbar] Hv5°C [kJ/kg] cp25°C [kJ/kg K] η25°C [mPa s] ρ25°C [kg/m3]

2

Applied Thermal Engineering 163 (2019) 114311

P. Chatzitakis, et al.

can be seen in Fig. 2. The AHP comprises six heat exchangers, the evaporator, absorber, solution heat exchanger, desorber, condenser, exhaust heat exchanger, one rectification column, four pumps (the two not shown are recirculation pumps, for the evaporator and absorber) and two expansion valves. Temperature was measured with PT-100 sensors ( ± 0.02 K), pressure with diaphragm electronic pressure transducers ( ± 0.5 mbar), volume flow with magnetic induction flow meters ( ± 2l/h) and mass flow with a Coriolis flow sensor ( ± 0.2 kg/ h). The rich solution and reflux (streams 1 and 8) mass flows, directly affecting the desorber temperature and condensate purity respectively, were considered as degrees of freedom and were adjusted such, in order to provide optimum COP values. The prototype was connected to a test stand able to satisfy variable heating and cooling loads, for the water and brine circuits respectively, and a broad temperature range. The coefficient of performance (COP) general equation has been analyzed in the previous paper for an ideal system:

already developed for organic working pairs. A previous work [14] focused towards improving the quantitative understanding of the multitude of parameters that influence the performance of absorption heat pumps by developing correlations between the coefficient of performance (COP) and the specific solution circulation (SSC), and readily available basic substance properties, like molecular mass and critical point properties. Based on this methodology, 2,2,3,3,3-pentafluoropropanol (5FP) and 2-pyrrolidone (PYR) were identified as possible absorption heat pump candidates to replace the above-mentioned high toxicity organic working fluids. 5FP has only been mentioned as a potential refrigerant by Ishikawa et al. [15], PYR, due to its higher viscosity, has only been investigated as a heat transformer absorbent, but none has ever been theoretically or experimentally investigated as an absorption heat pump or chiller working fluid. The scope of this paper focuses on the further development of this approach, that could eventually be employed as a screening process for new working pairs, aiming to validate the previous findings with experimental heat pump performance measurements amongst 4 candidate working pairs. More specifically, TFE and 5FP, were paired with DMI and PYR, each with similar chemical structure but significant differences in just a few key properties, like molecular weight, vaporization enthalpy, viscosity and vapor pressure (see Table 1). This effort aims towards a direct comparison of a heat pump’s performance, influenced by these key properties without significant “interference” from the rest of the pure substance thermophysical properties. As working fluids, all are more or less known to the scientific community [15,16].

1− COPcool, ideal = CR

cp,1, ce Hv,1, eva

(Tcon − Teva )

(ξ1,3 cp,1, des + ξ2,3 cp,2, des )(Tdes − Tdesin )

Hv,2, des Hv,1, eva

Hv,1, eva

+

+

ζ1,5 ξ2,4 Hv,1, des ζ1,5 − ξ1,4 Hv,1, eva

+

ζ2,5 ξ2,4 ζ1,5 − ξ1,4

ξ2,4 Hxsdes ζ1,5 − ξ1,4 Hv,1, eva

(1)

COPHP = 1 + COPcool = 1 +

2. Materials and methods

CR = TFE and DMI are well-known and documented and were chosen as a basis for the development effort. All of them are organic compounds and can be considered to represent a middle ground between the traditional working pairs’ pitfalls. Maximum pressures are well within the vacuum region, the operational heat pump temperature lift is comparable to ammonia and corrosion reactions are negligible for stainless steel and copper. A few drawbacks are, TFE’s toxicity and flammability, DMI’s thermal decomposition at 114 °C [15], and that both are classified as damaging fertility or the unborn child [18,19]. On the other hand, 5FP is significantly less toxic and PYR is classified only as an irritant [20,21]. 5FP (> 99 mass %) was supplied by Daikin Industries Ltd., TFE (> 99.5 mass %) by Solvay GmbH, DMI (> 99 mass %) by Mitsui Chemicals Inc. and PYR (> 99.5 mass %) by BASF Corporation. All working fluids were carefully degassed using specific procedures [22,23]. The working pair VLE vapor pressures were measured using two high precision static method apparatuses, whereas the viscosity measurements were performed with a stress-controlled rheometer equipped with a cone-and-plate geometry, both at the University of Rostock. Using the available VLE data, the working pairs’ activity coefficients were regressed and modelled using the NRTL method and were applied within the simulation models developed in the previous work [14]. The VLE, viscosity, as well as the NRTL model data are in the process of publication in a different paper. The simulation calculations were programmed and executed in Microsoft Excel’s® Visual Basic for Applications (VBA) environment. The experimental heat pump performance measurements were based on the German regulatory standard VDI 4650-2 [24], which precisely defines, among others, the brine-water heat pump boundary conditions for each measurement (brine & water return temperatures into the relevant AHP components as well as the heating capacity to be delivered by a gas-driven heating appliance employing an AHP for baseload and a peak load condensing boiler), see Table 2. The hybrid heating appliance comprises the single effect direct-fired AHP-prototype, depicted in Fig. 1, having a rated heating power of 5 kW and based on the highly compact plate heat exchanger design, which has been first introduced by Ishikawa et al. (1995) [15]. A simplified sketch

̇ Qeva ̇ Qdes

(2)

1 − ξ14 1 − X1,4 ⎛ ṁ 1 M = = X1,1 + X2,1 2 ⎞ ṁ 10 ξ1,1 − ξ1,4 X1,1 − X1,4 ⎝ M1 ⎠ ⎜

⎟

(3)

Where

X1,1 =

p1, eva − p2, abs γ2,1 p1, abs γ1,1 − p2, abs γ2,1

(4)

X2,1 = 1 − X1,1

(5)

are the refrigerant and the absorbent rich solution molar fractions, respectively.

X1,4 =

p1, con − p2, gen γ2,4 p1, gen γ1,4 − p2, gen γ2,4

(6)

is the refrigerant poor solution molar fraction. Under equilibrium conditions, refrigerant concentrations are determined partly by the refrigerant vapor pressure ratios between the evaporator and absorber, for the rich solution, and between the desorber and condenser for the poor solution. This signifies that the CR is correlated to temperature Table 2 Heat pump boundary conditions based on the VDI 4650-2 standard. Meas. No.

Therm. Output [kW]

Heating system 65/50 01 3.64 02 4.66 03 2.51 04 4.30 Heating system 55/45 05 3.64 06 4.66 07 2.51 18 4.30 Heating system 40/30 09 3.64 10 4.66 11 2.51 12 4.30

3

Evap. brine inlet temp. °C

Evap. brine vol. flow [l/h]

Abs. water inlet temp. °C

Abs. water inlet vol. flow [l/h]

8.0 7.0 6.0 5.0

500

32.6 35.2 37.7 41.6

700

8.0 7.0 6.0 5.0

500

29.6 31.7 33.9 37.3

980

8.0 7.0 6.0 5.0

500

23.5 24.4 25.3 26.7

980

Applied Thermal Engineering 163 (2019) 114311

P. Chatzitakis, et al.

Fig. 1. Absorption heat pump prototype front view (a) and side view (b).

Fig. 2. Absorption heat pump diagram with main components and sensors. P: pressure, T: temperature, V: volume flow, m: mass flow.

differences between absorber-evaporator and desorber-condenser. Additionally, the activity coefficients, the absorbent vapor pressure and the molar mass ratios influence the CR significantly. The prototype in question is a real system with components of finite size, rendering certain assumptions from the previous work invalid. Eq. (7) below is a slightly modified relation, making accommodations for a realistic rectification column, i.e. rectified vapor concentration ζ1,7 < 1 and rectifier bottom liquid outlet ξ1,6 not in equilibrium with the desorber liquid concentration ξ1,4, and an imperfect solution heat exchanger (η < 1).

1− COPcool, real = CR

ξ2,8

cp,1, ce

ζ1,7 − ξ1,8 Hv,1, eva

(ξ1,3 cp,1, gen + ξ2,3 cp,2, des )(Tdes − Tdesin )

Hv,2, des Hv,1, eva

Hv,1, eva

+

(Tcon − Teva ) +

ζ1,5 ξ2,6 Hv,1, des ζ1,5 − ξ1,6 Hv,1, eva

+

ζ2,5 ξ2,6 ζ1,5 − ξ1,6

ξ2,6 Hxsdes ζ1,5 − ξ1,6 Hv,1, eva

(7) In both cases, a correlation between the COPHP and the heat pump temperature lift, the difference between the average condenser and evaporator temperatures (Tcon-Teva), can be established from the equations above. Heat pump temperature lift is hereby defined as the temperature difference between the condenser and the evaporator. The specific solution circulation (SSC) has also been modeled in the same paper, correlating with the circulation ratio (CR). 4

Applied Thermal Engineering 163 (2019) 114311

P. Chatzitakis, et al.

1,8

1,6

COPHP

5FP-PYR meas. 5FP-DMI meas.

1,4

5FP-PYR Eq.1 5FP-DMI Eq.1 5FP-PYR Eq.7

1,2

5FP-DMI Eq.7

1,0 25

30

35

40

45

50

55

Temperature lift [K] Fig. 3. Comparison between measured and simulation 5FP COP data.

Table 3 Experimental and calculated COP and SSC data for the working pair 5FP + DMI. Meas. No.

Temp. lift [K]

Exp. COP

Sim. COP Eq. (1)

Rel. dev.

Sim. COP Eq. (7)

Rel. dev.

Exp. SSC

Sim. SSC Eq. (8)

Rel. dev.

Sim. SSC Eq. (9)

Rel. dev.

01 02 03 04 05 06 07 08 09 10 11 12

39.2 40.8 42.1 47.3 35.1 38.7 38.7 44.6 24.8 – 27.4 –

1.37 1.26 1.37 1.21 1.46 1.32 1.44 1.26 1.56 – 1.62 –

1.53 1.51 1.50 1.45 1.56 1.53 1.53 1.48 1.66 – 1.63 –

11.4% 20.1% 9.3% 19.9% 7.2% 16.1% 6.1% 17.0% 6.2% – 1.0% –

1.38 1.35 1.33 1.24 1.45 1.39 1.39 1.29 1.63 – 1.58 –

0.7% 7.4% 3.2% 2.3% 0.7% 5.2% 3.8% 1.9% 4.2% – 2.2% –

80.6 118.1 98.6 167.7 63.2 124.7 84.1 126.2 64.4 – 89.9 –

39.1 41.6 43.9 52.9 33.0 38.4 38.3 48.0 19.7 – 22.8 –

51.4% 64.8% 55.5% 68.5% 47.7% 69.3% 54.5% 61.9% 69.5% – 74.6% –

81.9 89.9 97.4 128.1 62.6 79.4 79.2 111.4 25.1 – 33.2 –

1.6% 23.9% 1.3% 23.6% 0.8% 36.4% 5.8% 11.7% 61.1% – 63.1% –

Table 4 Experimental and calculated COP and SSC data for the working pair 5FP + PYR. Meas. No.

Temp. lift [K]

Exp. COP

Sim. COP Eq. (1)

Rel. dev.

Sim. COP Eq. (7)

Rel. dev.

Exp. SSC

Sim. SSC Eq. (8)

Rel. dev.

Sim. SSC Eq. (9)

Rel. dev.

01 02 03 04 05 06 07 08 09 10 11 12

37.4 41.7 40.1 48.6 33.1 37.7 35.9 42.8 26.9 32.2 27.5 34.7

1.54 1.45 1.51 1.34 1.58 1.50 1.57 1.42 1.63 1.58 1.67 1.55

1.60 1.57 1.58 1.51 1.64 1.60 1.61 1.56 1.69 1.64 1.68 1.62

4.1% 8.2% 4.7% 12.7% 3.6% 6.5% 2.9% 10.0% 3.6% 3.8% 0.7% 4.7%

1.52 1.46 1.48 1.36 1.58 1.51 1.54 1.44 1.66 1.59 1.65 1.55

1.4% 0.7% 1.9% 1.7% 0.2% 0.8% 1.9% 1.9% 2.1% 0.3% 1.0% 0.2%

44.2 61.9 60.8 113.1 40.4 64.8 56.2 90.6 24.2 32.3 30.1 33.1

31.5 36.3 34.5 44.6 26.8 31.8 29.8 37.7 20.7 25.9 21.2 28.5

28.9% 41.3% 43.2% 60.6% 33.6% 50.8% 46.9% 58.4% 14.5% 19.7% 29.3% 13.7%

49.0 58.4 54.9 74.7 40.4 49.7 46.0 60.9 29.5 38.9 30.5 43.6

10.8% 5.8% 9.8% 33.9% 0.1% 23.2% 18.2% 32.7% 22.1% 20.3% 1.6% 31.8%

SSCideal =

CR (ρsol, abs (Hv,1, eva − cp1, ce (Tcon − Teva )))

in direct correlation with the heat pump temperature lift. Figs. 3 & 5 and Tables 3 & 4 show experimental COP and SSC data from the combinations 5FP + DMI and 5FP + PYR. Additionally, corresponding COP and SSC data from simulations utilizing Eqs. (1) & (8), a theoretical perfect heat pump, are presented as well as results from simulations taking into account rectification losses, utilizing Eqs. (7) & (9). It should be mentioned that the 5FP + DMI experimental measurements took place while the system degree of freedom optimization algorithm was still under development, therefore the COP and SSC data appear to have larger scattering compared to the rest. On average, the experimental COP measurements were 9.5% higher and the SSC measurements 43.7% lower for PYR.

(8)

CR

SSCreal = ⎛ρ ⎛ sol, abs Hv,1, eva − ⎝ ⎝

⎜

ξ2,8

c (T ζ1,7 − ξ1,8 p1, ce con

− Teva ) ⎞ ⎟⎞ ⎠⎠

(9)

As with the COP, Eq. (9) provides the SSC of a real machine. The SSC is also directly correlated to the heat pump temperature lift. 3. Results All performance data, simulation and experimental, was presented 5

Applied Thermal Engineering 163 (2019) 114311

P. Chatzitakis, et al.

1,8

1,6

COPHP

TFE-PYR meas. TFE-DMI meas.

1,4

TFE-PYR Eq.1 TFE-DMI Eq.1 TFE-PYR Eq.7

1,2

TFE-DMI Eq.7

1,0 25

30

35

40

45

50

55

Temperature lift [K] Fig. 4. Comparison between measured and simulation TFE COP data.

200

SSC [l/h/kW]

150 5FP-PYR meas. 5FP-DMI meas.

100

5FP-PYR Eq.8 5FP-DMI Eq.8 5FP-PYR Eq.9

50

5FP-DMI Eq.9 0 25

30

35

40

45

50

55

Temperature lift [K] Fig. 5. Comparison between measured and simulation 5FP SSC data.

200

SSC [l/h/kW]

150 TFE-PYR meas. 100

TFE-PYR Eq.8 TFE-DMI Eq.8 TFE-PYR Eq.9

50

TFE-DMI Eq.9

0 25

30

35

40

45

50

55

Temperature lift [K] Fig. 6. Comparison between measured and simulation TFE SSC data.

6

Applied Thermal Engineering 163 (2019) 114311

P. Chatzitakis, et al.

Table 5 Experimental and calculated COP and SSC data for the working pair TFE + DMI. Meas. No.

Temp. lift [K]

Exp. COP

Sim. COP Eq. (1)

Rel. dev.

Sim. COP Eq. (7)

Rel. dev.

Exp. SSC

Sim. SSC Eq. (8)

Rel. dev.

Sim. SSC Eq. (9)

Rel. dev.

01 02 03 04 05 06 07 08 09 10 11 12

38.52 36.01 42.42 48.04 35.30 32.38 39.26 44.35

1.46 1.50 1.39 1.32 1.49 1.56 1.45 1.36 – 1.63 1.56 1.46

1.62 1.64 1.58 1.54 1.64 1.67 1.61 1.57 – 1.72 1.67 1.64

10.9% 9.1% 13.9% 16.4% 9.9% 6.5% 10.9% 15.3% – 5.4% 7.2% 11.8%

1.49 1.52 1.43 1.35 1.53 1.58 1.47 1.40 – 1.67 1.59 1.52

2.0% 1.5% 2.8% 1.9% 2.6% 0.8% 1.6% 3.0% – 2.4% 1.9% 4.1%

– – – – – – – – – – – –

37.1 33.0 43.8 54.5 31.9 27.6 38.3 47.4 – 19.3 26.4 33.1

– – – – – – – – – – – –

73.3 63.0 90.7 119.2 60.3 49.6 76.4 100.1 – 30.3 46.7 63.2

– – – – – – – – – – – –

26.12 31.52 36.05

Table 6 Experimental and calculated COP and SSC data for the working pair TFE + PYR. Meas. No.

Temp. lift [K]

Exp. COP

Sim. COP Eq. (1)

Rel. dev.

Sim. COP Eq. (7)

Rel. dev.

Exp. SSC

Sim. SSC Eq. (8)

Rel. dev.

Sim. SSC Eq. (9)

Rel. dev.

01 02 03 04 05 06 07 08 09 10 11 12

38.3 44.0 41.9 51.6 33.3 39.3 37.7 45.0 28.8 32.5 29.0 35.6

1.62 1.51 1.57 1.35 1.68 1.58 1.65 1.45 1.74 1.69 1.76 1.65

1.72 1.68 1.70 1.64 1.75 1.71 1.72 1.68 1.78 1.75 1.78 1.74

6.3% 11.2% 8.1% 21.3% 3.8% 8.1% 4.3% 15.5% 2.4% 4.0% 1.2% 4.9%

1.60 1.51 1.54 1.38 1.68 1.58 1.61 1.49 1.76 1.69 1.75 1.64

1.0% 0.4% 1.8% 2.6% 0.2% 0.1% 2.5% 2.7% 1.1% 0.5% 0.2% 0.6%

43.0 50.9 72.8 87.7 35.2 40.5 58.1 57.5 25.6 28.5 42.9 34.9

31.6 35.2 33.9 39.5 28.2 32.3 31.2 35.8 24.9 27.6 25.1 29.8

26.5% 30.9% 53.4% 54.9% 19.7% 20.3% 46.3% 37.7% 2.8% 3.1% 41.5% 14.6%

48.6 63.4 57.9 85.9 37.6 51.3 47.3 66.2 28.6 35.9 29.1 42.5

13.3% 24.5% 20.5% 2.0% 6.9% 26.7% 18.5% 15.2% 11.7% 25.9% 32.3% 21.8%

Table 7 Key experimental measurements for the working pair 5FP + DMI.

Table 8 Key experimental measurements for the working pair 5FP + PYR.

Meas. No.

Evap. brine outlet temp. [°C]

Cond. water outlet temp. [°C]

Des. outlet solution temp [°C]

Des. inlet solution vol. flow [l/h]

Des./ cond. pressure [mbar]

Abs./ evap. pressure [mbar]

Meas. No.

Evap. brine outlet temp. [°C]

Cond. water outlet temp. [°C]

Des. outlet solution temp [°C]

Des. inlet solution vol. flow [l/h]

Des./ cond. pressure [mbar]

Abs./ evap. pressure [mbar]

01 02 03 04 05 06 07 08 09 10 11 12

6.4 5.4 4.9 3.8 6.3 5.3 4.8 3.5 6.2 – 4.7 –

37.0 40.1 40.7 46.0 32.5 35.4 36.0 40.8 26.4 – 27.4 –

112.5 122.7 113.1 127.5 102.8 113.0 105.0 123.3 95.7 – 85.9 –

60.3 80.0 93.6 106.9 55.3 83.4 58.8 89.6 70.2 – 77.4 –

123.70 144.05 158.35 210.13 97.71 112.44 125.46 155.82 65.94 – 72.55 –

7.36 5.14 9.26 7.49 8.21 3.60 9.36 5.03 3.58 – 7.05 –

01 02 03 04 05 06 07 08 09 10 11 12

6.0 4.8 4.7 3.3 5.9 4.5 4.6 3.0 5.8 4.6 4.6 2.7

36.7 40.3 40.6 46.4 32.5 35.6 36.0 40.9 26.7 27.9 27.4 30.2

110.3 124.3 112.7 136.6 100.0 116.3 103.8 128.1 95.7 104.3 86.5 108.3

49.5 76.5 46.2 109.6 48.9 93.3 46.2 107.0 30.8 45.9 26.6 44.5

124.4 152.1 154.3 215.6 99.4 119.6 119.6 163.2 66.8 72.0 71.1 83.9

8.6 6.7 8.2 6.9 8.5 7.2 7.8 6.7 6.9 5.8 6.5 5.3

extension the SSC had the function of an optimization parameter and therefore a less evident correlation. In the case of one working fluid replacement, the ideal system calculations based on Eqs. (1) & (8) are tendentially correct but show significant deviations, especially at higher temperature lifts, where cycle losses play a dominant role. The modified Eqs. (7) & (9) however, accounting for rectification and solution heat exchanger losses, provide data much closer to the experimental measurements. Tables 7, 8, 9 and 10 contain some of the most important prototype heat pump measurements for the corresponding working pairs, such as the evaporator brine and condenser water outlet temperatures, the desorber solution outlet temperature, the desorber solution inlet volume flow and the system high and low pressures. Fig. 7 makes a direct COP comparison between the toxic TFE + DMI and the safer 5FP + PYR working pairs. The experimental measurements show on average 3% higher COPs for 5FP + PYR, whereas Eq.

Figs. 4 & 6 and Tables 5 & 6 follow the same example, this time for the TFE + DMI and TFE + PYR working pairs. On average, the COP measurements were 8% higher for PYR. Unfortunately, due to a volume flow sensor malfunction, the SSC experimental measurements for TFEDMI were not available. Regarding the validation of the purely theoretical heat pump model, the mean relative deviations for the COP (Eq.1) were 8.6% and 41.5% for the SSC (Eq. (8)). The model accounting for rectification losses shows mean relative deviations of 1.9% for the COP (Eq. (7)) and 19.4% for the SSC (Eq. (9)). It is worth mentioning that the new model simulation data display steeper slopes than the ones based on the previous model, indicating the increasing cycle losses at higher temperatures. The experimental data show a solid relation between the COP and the temperature lift, whereas the desorber temperature and by 7

Applied Thermal Engineering 163 (2019) 114311

P. Chatzitakis, et al.

DMI combinations. It is interesting to note that the boiling point difference between PYR and DMI is only 20 K.

Table 9 Key experimental measurements for the working pair TFE + DMI. Meas. No.

01 02 03 04 05 06 07 08 09 10 11 12

Evap. brine outlet temp. [°C]

Cond. water outlet temp. [°C]

Des. outlet solution temp [°C]

Des. inlet solution vol. flow [l/h]

Des./ cond. pressure [mbar]

Abs./ evap. pressure [mbar]

5.5 5.6 4.2 3.0 5.3 5.5 3.9 2.8 – 5.6 3.7 2.2

37.0 37.4 41.1 45.8 33.0 33.5 36.8 40.9 – 26.1 28.2 30.4

102.5 97.7 109.7 129.8 102.4 93.9 111.8 120.4 – 82.0 96.0 105.1

– – – – – – – – – – – –

201.5 210.5 247.1 299.2 169.2 178.0 199.5 243.0 – 125.7 132.6 151.4

23.2 24.2 21.4 18.0 20.0 23.4 21.2 17.7 – 18.6 20.0 16.0

4. Conclusions This work investigated for the first time, both theoretically and experimentally, an absorption heat pump refrigerant (5FP) and an absorption heat pump absorbent (PYR) and compared combinations of them with a chemically similar refrigerant (TFE) and absorbent (DMI) from the literature. The selection of these new working fluids was based on the further development of a theoretical model from a previous work on the systematic screening of working fluids, with special focus on the molar mass. The previous model, based on an idealized heat pump, is able to perform tendentially correct comparisons between working pairs with one working fluid replacement, but shows significant relative deviations from real system measurements. The model presented in this work performs significantly better in both aspects, COP and SSC, showing good COP accuracy but still high SSC relative deviations from the experimental data. The experimental absorption heat pump measurements are in good agreement with the new theoretical model and show conclusively that PYR is a significant improvement over DMI, with an average of 9% higher COPs and 44% lower SSCs. It is also shown that the absorbent boiling point and viscosity cannot be considered a reliable indicator in the absorbent selection process. An absorbent candidate with lower molar mass and vapor pressure seems to overcompensate for unfavorable flow properties and reduced absorption affinity and results in improved heat pump performance. Overall the results justify the absorbent selection and offer an initial confirmation on the previous work’s theoretical approach. On the refrigerant side, a higher refrigerant molar mass also influences the SSC positively but also results in lower vaporization enthalpies (per mass unit) with a slightly stronger undesired influence on both the SSC, +12%, and the COP, −5%. The COP performance downgrade was anticipated by the new theoretical model, the refrigerant selection however was primarily influenced by the similar chemical structure and the low toxicity of the substance. In conclusion, a replacement of the combination of TFE + DMI, both substances toxic to reproduction, with the non-toxic 5FP + PYR working pair seems viable, introducing a slight improvement, +3%, in the COP values. It should be noted that this is still a work in progress and more working pairs are being currently investigated in order to provide additional solid results for the validation of this approach. Moreover, the theoretical model is being continually developed with the goal of

Table 10 Key experimental measurements for the working pair TFE + PYR. Meas. No.

Evap. brine outlet temp. [°C]

Cond. water outlet temp. [°C]

Des. outlet solution temp [°C]

Des. inlet solution vol. flow [l/h]

Des./ cond. pressure [mbar]

Abs./ evap. pressure [mbar]

01 02 03 04 05 06 07 08 09 10 11 12

5.7 4.3 4.6 3.1 5.7 4.2 4.4 2.8 5.4 3.9 4.5 2.3

37.0 40.7 40.7 46.7 32.4 35.6 36.1 40.9 26.7 28.4 27.3 30.3

100.6 117.7 103.9 130.0 92.2 105.9 93.6 120.1 81.0 91.5 72.0 96.0

54.6 71.8 61.1 88.4 45.7 62.0 55.0 69.5 37.9 49.9 41.4 54.0

183.2 225.8 224.7 308.5 135.5 172.6 175.5 223.6 101.0 105.2 104.9 119.8

20.0 16.9 19.0 15.9 19.5 17.0 18.8 15.5 19.1 16.5 17.8 14.4

(7) based calculations predict 1% higher COPs for 5FP + PYR and Eq. (1) based calculations 2% higher COPs for TFE + DMI. Table 11 compares key thermodynamic and thermophysical solution properties for the four working pairs. PYR solutions show slightly higher activity coefficients, more than twice higher viscosities and almost an order of magnitude higher refrigerant relative volatilities than

1,8

1,6

COPHP

5FP-PYR meas. TFE-DMI meas.

1,4

5FP-PYR Eq.1 TFE-DMI Eq.1 5FP-PYR Eq.7

1,2

TFE-DMI Eq.7

1,0 25

30

35

40

45

50

55

Tlift [K] Fig. 7. Comparison between measured and simulation COP data for TFE + DMI and 5FP + PYR. 8

Applied Thermal Engineering 163 (2019) 114311

P. Chatzitakis, et al.

Table 11 Activity coefficients of simulated rich and poor solutions, as well as poor solution refrigerant relative volatilities in the desorber and viscosities in the absorber. Working pair

γ1 X1 = 40% T = 40 °C

γ2 X1 = 40% T = 40 °C

γ1 X1 = 30% T = 100 °C

γ2 X1 = 30% T = 100 °C

RV1 X1 = 30% T = 100 °C

ηsol [mPa s] X1 = 30% T = 40 °C

5FP-DMI 5FP-PYR TFE-DMI TFE-PYR

0.131 0.237 0.209 0.215

0.608 0.698 0.690 0.445

0.208 0.260 0.226 0.230

0.846 0.886 0.859 0.750

15.9 132.0 21.8 176.1

1.81 4.97 1.77 4.34

providing a reliable tool for the screening of novel absorption heat pump working pairs.

doi.org/10.1016/0890-4332(88)90058-0. [10] U. Nowaczyk, Kriterien zur Auswahl von Arbeitsstoffgemischen für Absorptionsprozesse und Erste Auswahlmessungen, DKV Forschungsberichte 38 (1991). [11] G. Alefeld, R. Radermacher, Y. Hwang, Heat Conversion Systems, CRC Press, 1994. [12] A. Kühn (Ed.), M. Seiler, M. Radspieler, O. Kotenko, H. Moser, R. Rieberer, Ionic liquids as new absorbents for absorption chillers and heat pumps, in: Thermally Driven Heat Pumps for Heating and Cooling, Universitätsverlag der TU Berlin, Berlin, 2013 (Chapter 6). [13] K.E. Herold, R. Radermacher, S.A. Klein, Absorption chillers and heat pumps, second ed., CRC Press, 2016. [14] P. Chatzitakis, B. Dawoud, An alternative approach towards absorption heat pump working pair screening, Renew. Energy 110 (2017) 47–58, https://doi.org/10. 1016/j.renene.2016.08.014. [15] M. Ishikawa, H. Kayanuma, N. Isshiki, Absorption heat pump using new organic working fluids, Proceedings of the International Sorption Heat Pump Conference, (1999). [16] C. Zhuo, Absorption heat transformer with TFE-Pyr as the working pair, Delft University of Technology, Delft, 1995. [17] ASPEN Plus V9.0, Aspen Tecnology Inc., Burlington, 2016. [18] 2,2,2-trifluoroethanol registration dossier, European Chemicals Agency, 2019 February 27, https://echa.europa.eu/registration-dossier/-/registered-dossier/ 13741/2/1. [19] 1,3-dimethyl-2-imidazolidinone registration dossier, European Chemicals Agency, 2019 February 27, https://echa.europa.eu/registration-dossier/-/registereddossier/11777/2/1. [20] 2,2,3,3,3-pentafluoropropanol registration dossier, European Chemicals Agency, 2019 February 27, https://echa.europa.eu/information-on-chemicals/cl-inventorydatabase/-/discli/details/130567. [21] 2-pyrrolidone registration dossier European Chemicals Agency 2019 February 27 https://echa.europa.eu/registration-dossier/-/registered-dossier/14900/2/1. [22] J. Safarov, I. Kul, M.A. Talibov, A. Shahverdiyev, E. Hassel, Vapor pressures and activity coefficients of methanol in binary mixtures with 1-Hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, J. Chem. Eng. Data 60 (6) (2015) 1648–1663, https://doi.org/10.1021/je501033z. [23] J. Safarov, F. Lesch, K. Suleymanli, A. Aliyev, A. Shahverdiyev, E. Hassel, I. Abdulagatov, Viscosity, density, heat capacity, speed of sound and other derived properties of 1-butyl-3methylimidazolidium tris(pentafluoroethyl) trifluorophosphate over a wide range of temperature and at atmospheric pressure, J. Chem. Eng. Data 62 (10) (2017) 3620–3631, https://doi.org/10.1021/acs.jced.7b00618. [24] VDI 4650 Part 2, Simplified method for the calculation of the annual heating energy ratio and the annual gas utilization efficiency of sorption heat pumps. Gas heat pumps for space heating and domestic hot water, Beuth Verlag GmbH, 2013.

Acknowledgements Part of this work was sponsored by the European Union’s “European Regional Development Fund”. Appendix A. Supplementary material Supplementary data to this article can be found online at https:// doi.org/10.1016/j.applthermaleng.2019.114311. References [1] R.A. Macriss, J.M. Gutraj, T.S. Zawacki, Absorption fluids data survey: final report on worldwide data, Institute of Gas Technology, Chicago, USA, 1988, Report No. ORNL/Sub/84-47989/3. [2] J. Sun, L. Fu, S. Zhang, A review of working fluids of absorption cycles, Renew. Sustain. Energy Rev. 16 (4) (2012) 1899–1906, https://doi.org/10.1016/j.rser. 2012.01.011. [3] D. Zheng, L. Dong, W. Huang, X. Wu, N. Nie, A review of imidazolium ionic liquids research and development towards working pair of absorption cycle, Renew. Sustain. Energy Rev. 37 (2014) 47–68, https://doi.org/10.1016/j.rser.2014.04.046. [4] A.I. Papadopoulos, A.S. Kyriakides, P. Seferlis, I. Hassan, Absorption refrigeration processes with organic working fluid mixtures – a review, Renew. Sustain. Energy Rev. 109 (2019) 239–270, https://doi.org/10.1016/j.rser.2019.04.016. [5] P.D. Iedema, Mixtures for the absorption heat pump, Int. J. Refrig. 5 (5) (1982) 262–273, https://doi.org/10.1016/0140-7007(82)90065-2. [6] D.L. Hodgett, Absorption heat pumps and working pair developments in Europe since 1974, Proceedings of New Working Pairs for Absorption Processes Workshop, (1982). [7] H. Perez-Blanco, Absorption heat pump performance for different types of solutions, Int. J. Refrig. 7 (2) (1984) 115–122, https://doi.org/10.1016/0140-7007(84) 90024-0. [8] M.A. Eisa, F.A. Holland, A study of the optimum interaction between the working fluid and the absorbent in absorption heat pump systems, Heat Recov. Syst. 7 (2) (1987) 107–117, https://doi.org/10.1016/0890-4332(87)90075-5. [9] M. Narodoslawsky, G. Otter, F. Moser, Thermodynamic criteria for optimal absorption heat pump media, Heat Recov. Syst. CHP 8 (3) (1988) 221–233, https://

9