COMPACT HEAT AND MASS EXCHANGERS OF THE PLATE FIN TYPE IN THERMAL SORPTION SYSTEMS

Application in an absorption heat pump with the working pair CH30H-LiBr/ZnBr2

H. Becker TR diss 1698

COMPACT HEAT AND MASS EXCHANGERS OF THE PLATE FIN TYPE IN THERMAL SORPTION SYSTEMS

Application in an absorption heat pump with the working pair CH30H-LiBr/ZnBr2

H. Becker

CIP-DATA KONINKLIJKE BIBLIOTHEEK, DEN HAAG Becker, Harry Compact heat and mass exchangers of the plate fin type in thermal sorption systems: appiication in an absorption heat pump with the working pair CH30H-LiBr/ZnBr2 Harry Becker. - Delft: Delft University of Technology, Mechanical Engineering Department Thesis Delft. - With ref. - With summary in Dutch ISBN 90-370-0022-3 SISO 653.3 UDC 621.577(043.3) Subject heading: absorption heat pumps.

Copyright 1989, Faculty of Mechanical Engineering and Marine Engineering Delft University of Technology

All rights reserved. This report, or parts thereof, may not be reproduced in any form without permission of the publisher.

Any use or appiication of data, methods and/or results e t c , occuring in this report will be at user's own risk. The Delft University of Technology, Faculty of Mechanical Engineering and Marine Engineering accepts no liability for damages suffered from the use or appiication.

COMPACT HEAT AND MASS EXCHANGERS OF THE PLATE FIN TYPE IN THERMAL SORPTION SYSTEMS

Application in an absorption heat pump with the working pair CH30H-LiBr/ZnBr2

PROEFSCHRIFT

Ter verkrijging van de graad van doctor aan de Technische Universiteit Delft, op gezag van de Rector Magnificus, Prof.drs. P.A. Schenck in het openbaar te verdedigen ten overstaan van een commissie aangewezen door het College van Dekanen op dinsdag 7 februari 1989 om 14.00 uur.

door

HARRY BECKER
geboren te Amersfoort Werktuigkundig ingenieur

TR diss 1698

Dit proefschrift is goedgekeurd door de promotor prof.ir. A.L. Stolk

CONTENTS SUMMARY SAMENVATTING page CHAPTER 1. THE ABSORPTION HEAT PUMP 1.1 1.2 1.3 1.4 Introduction Working principle The log P - l/T diagram The heat ratio, the cofficint of performance and the circulation ratio 1.5 Research work on absorption heat pumps (AHP) 1.5.1 Working pairs for an AHP 1.5.2 System configurations 1.5.3 Heat and mass transfer 1.6 This research work 1.6.1 Goals 1.6.2 Tools 1.6.3 Set up CHAPTER 2. COMPACT HEAT AND MASS EXCHANGERS 1 Definition 2 Compact and enhanced transfer surface 3 Application 2.3.1 Field of application 2.3.2 Corrugations 2.3.3 Construction materials Characterization 2.4 2.4.1 Identification 2.4.2 Analytical solutions 2.4.3 Experimental results 2.4.3.1 Introduction 2.4.3.2 The work of Kays and London 2.4.3.3 Rectangular offset strip fin surfaces Discussion Application in sorption systems CHAPTER 3. THE ABSORPTION HEAT PUMP (AHP) TEST PLANT 3.1 3.2 3.3 3.4 3.5 Introduction Choice of the working pair Type of compact heat and/or mass transfer surface Overview of the test plant The components 3.5.1 The absorber 3.5.2 The condenser and evaporator 3.5.3 The mixture-mixture heat exchanger 3.5.4 The generator 3.5.5 Distribution of liquid flows 3.5.6 The mixture pump 25 26 26 27 29 30 31 31 31 32 13 13 15 15 16 16 17 18 18 18 21 22 1 1 3 4 6 6 8 10 11 12 12

3.6 The heating and cooling circuits

3.6.1 The cooling system of the absorber and condenser 3 6.2 The heating system of the evaporator 3 6.3 The heating system of the generator 3 6.4 General 3.7 Control 3 7.1 Mass flows 3 7.2 Temperatures 3 7.3 Weight fraction 3 7.4 Pressures 3.8 Measurement 3.8.1 Mass flows 3.8.2 Mixture density 3.8.3 Temperatures 3.8.4 Pressures 3.8.5 Accuracy of measurements 3.8.6 Data registration 3.9 Construction materials and corrosion CHAPTER 4. THE WORKING PAIR : SELECTION AND DATA 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 CHAPTER 5. Introduction Experiments Simulation Literature survey on working pairs Selection of a new working pair Consequences for the test plant Consequences for the simulation program Comparison R123a - DTG with CH30H - LiBr / ZnBr2 Conclusions 39 39 39 39 40 41 41 41 43 32 32 32 32 33 33 33 34 34 34 34 34 35 36 36

COMPUTER SIMULATION MODEL AND PROGRAM 5.1 5.2 5.3 5.4 Introduction Starting point Development of a new simulation model and program Final simulation program 5.4.1 Structure 5.4.2 Heat and/or mass transfer correlations 5.4.3 Computer subroutines 5.5 A simplified simulation program 5.6 Application of the simulation program: Some examples 5.7 Conclusions 45 45 46 47 48 50 51 52 55

CHAPTER 6.

RESULTS OF THE EXPERIMENTS 6.1 Introduction 6.2 Experimental results of the AHP and its components 6.2.1 The absorption heat pump 57 60

6.2.2 The components 6.2.2.1 The generator 60 61 6.2.2.2 The condenser 63 6.2.2.3 The evaporator 65 6.2.2.4 The absorber 6.2.2.5 The mixt.-mixt. heat exchanger 70 6.3 Conclusions from the experiments 71 CHAPTER 7. INTERPRETATION OF AND DISCUSSION ON THE EXPERIMENTAL RESULTS 7.1 Introduction 73 7.2 A qualitative interpretation 73 7.3 A quantitative interpretation 7.3.1 Introduction 74 7.3.2 The mixture-mixture heat exchanger 75 7.3.3 The absorber 77 7.3.4 Discussion and comparison 79 7.4 Detailed absorber simulation model 7.4.1 Introduction 81 7.4.2 Former researchers 82 7.4.3 Results from this research work 83 7.4.4 Discussion on correction factors 84 7.5 Comparison SAT absorber/generator and the CHME absorber 7.5.1 Introduction 85 7.5.2 Compactness (area density) 85 7.5.3 Heat transfer (heat flow density) 86 7.5.4 Discussion and conclusion 88 7.6 Conclusions 88 CHAPTER 8. GENERAL CONCLUSIONS 8.1 8.2 8.3 8.4 Retrospective view Conclusions Considerations Recommendations 91 91 92 93

APPENDIX A - HEAT AND MASS TRANSFER CORRELATIONS OF THE SIMULATION PROGRAM APPENDIX B - CALCULATION SCHEMES OF THE AHP COMPONENTS REFERENCES NOMEMCLATURE CURRICULUM VITAE

97 99 103 109

113

SUMMARY

This dissertation covers a theoretical and experimental study into the possible application of compact heat and mass exchangers (CHME) in a gas fired absorption heat pump (AHP) for domestic heating. The framework of the study is defined by discussing the general principles and the research fields of the AHP. In addition the goals, the means and the set up of this research work are explained. (Chapter 1.) The above-mentioned heat and mass exchangers are of the plate type. The space between the parallel and plain plates is filled up with corrugated plates of a certain height, the corrugation or finned plate. The plain and finned plates are stacked and welded together. This gives a heat and mass exchanger which is very compact expressed by a high area density (m 2 /m 3 ). This leads to heat and mass transfer processes with small temperature and concentration differences. (Chapter 2.) For testing purposes a pilot plant was built using the above type of components in order to test their heat and/or mass transfer performance. Only the generator is of the shell and tube (SAT) type. As the working pair CH 3 0H - LiBr / ZnBr 2 was chosen, with the alcohol as the solvent and the salt mixture as the absorbent. This leads to sub-atmospheric working pressures with only solvent in the vapour phase. (Chapter 3.) A literature survey has been conducted on working pairs for sorption systems in order to update the knowledge in this field and to select a new working pair for the pilot plant. This offers the possibility to verify and validate the simulation and the experimental results. As possible new working pair R123a - DTG was selected and tested in the simulation program. For practical reasons experiments in the pilot plant are outside the scope of this research work. (Chapter 4.) At the same time a computer program has been developed to simulate the test plant, based on heat and mass transfer correlations found in the literature, later to be replaced by correlations based on experimental results. The program consists of three parts. The main program covers the overall calculation and iteration procedures. The component program contains the subroutines of the separate components, while the property program contains the thermodynamic and physical property data of several working pairs and cooling/heating media. (Chapter 5.) Three series of experiments have been carried out, during which the input parameters were varied over a certain range. The heating temperature of the evaporator was between 5 and 10C, of the generator mostly 125C due to the instability of the methanol in the mixture above that temperature). The cooling temperature of the absorber and condenser was varied between 30 - 60"C over a mixture mass flow range of 25 - 125 g/s. (Chapter 6.)

Different approaches have been adopted to interpret and explain the experimental results, the emphasis being on the absorber as the most important and interesting component (simultaneous heat and mass transfer in a film flow). The results have been used to find matching (heat) transfer correlations and to verify the film (heat transfer) and penetration (mass transfer) theory as adopted in the simulation program to match them by means of correction factors. Also the SAT generator/absorber and the CHME absorber have been compared for their compactness (area density) and their heat transfer (heat flow density). (Chapter 7.) Conclusions have been drawn concerning the possible application of the finned plate compact heat and/or mass exchangers in thermal sorption systems, while recommendations have been given for further research work. (Chapter 8.)

SAMENVATTING Dit proefschrift doet verslag van het theoretische en experimentele onderzoek naar de mogelijke toepassing van zgn. " compact heat and mass exchangers (CHME)" in een gas-gestookte absorptie-warmtepomp (AWP) voor individuele woningverwarming. Het kader wordt aangegeven middels een beschrijving van de werkings principes en van het huidige onderzoeksveld betreffende de AWP. Ook worden de doelstelling, de middelen en de aanpak van het onderzoek uiteengezet. (Hoofdstuk 1.) De bovengenoemde warmte- en stofwisselaars zijn van het plaat-type. De ruimte tussen de parallelle en vlakke platen is "gevuld" met een vervormde plaat, de zgn. "corrugation" of gevinde plaat. De vlakke en gevinde platen worden gestapeld en tot een pakket gesoldeerd. Dit geeft een warmte- en stofwisselaar die zeer compact is en heeft een hoge oppervlaktedichtheid (m 2 /m 3 ). Dit geeft overdrachtsprocessen van warmte en stof met kleine temperatuur- en concentratieverschillen. (Hoofdstuk 2.) Voor beproeving hiervan is een testopstelling gebouwd met dit type componenten om de warmte- en stofoverdracht te onderzoeken. Alleen de generator is van het "shell and tube (SAT)" type. Als stofpaar is CH30H - Li Br / ZnBr2 gekozen met de alcohol als het oplosmiddel (solvent) en het zoutmengsel als de absorbent. Dit geeft subatmosferische werkdrukken met alleen solvent in de dampfase. (Hoofdstuk 3.) Er is een literatuurstudie verricht naar stofparen voor sorptiesystemen. Dit om de kennis op dit gebied bij te houden en om een nieuw stofpaar voor de testopstelling te selecteren. Dit laatste biedt de mogelijkheid de simulatie en de experimentele resultaten te verifiren. Als mogelijk nieuw stofpaar is R123a - DTG geselecteerd en beproefd in het simulatieprogramma. Experimenten in de testopstelling vallen helaas om practische redenen buiten dit onderzoek. (Hoofdstuk 4.) Tegelijkertijd is een computerprogramma ontwikkeld dat de testopstelling simuleert, werkend met warmte- en stofoverdrachtsrelaties bekend uit de literatuur. Later zullen deze vervangen worden door relaties komende uit de experimenten. Het programma bestaat uit drie delen. Het hoofd-programma zorgt voor de "overall" berekeningen en iteratieprocedures. Het componenten-programma bevat de subroutines van de onderscheiden componenten, terwijl het stofprogramma de thermodynamische en fysische stofgegevens van meerdere stofparen en verwarmings/koelmedia bevat. (Hoofdstuk 5.) Er hebben drie series van experimenten plaatsgevonden waarbij een aantal invoerparameter over hun werkgebied gevarieerd zijn. De verwarmingstemperatuur van de verdamper lag tussen de 5 en 10C, die van de generator was meestal 125C i.v.m. de instabiliteit van de alcohol in het mengsel boven deze temperatuur. De koelwatertemperatuur (abs./cond.) werd gevarieerd tussen de 30 en 60C, de mengselmassastroom tussen de 25 en 125 g/s. (Hoofdstuk 6.)

Verschillende benaderingen hebben plaatsgevonden om de experimentele resultaten te interpreteren. De nadruk lag daarbij op de absorber als de meest belangrijke en interessante component (simultane warmte- en stofoverdracht in een filmstroming). De resultaten zijn gebruikt om correlaties voor de warmteoverdracht te vinden, om de juistheid van het toepassen van de filmtheorie voor de warmteoverdracht en van de penetratietheorie voor de stofoverdracht te onderzoeken en om correctiefactoren voor deze theorien te vinden. Ook is een vergelijking gemaakt tussen de SAT generator/absorber en de CHME absorber. Vergeleken zijn zowel de compactheid (oppervlaktedichtheid m 2 /m 3 ) als de warmteoverdracht (warmtestroomdichtheid W/m 2 ). (Hoofdstuk 7.) Conclusies zijn getrokken voor de mogelijke toepassing van de gevinde plaat CHME's voor thermische sorptiesystemen. Ook zijn een aantal aanbevelingen en richtingen voor verder onderzoek aangegeven. (Hoofdstuk 8.)

CHAPTER 1.

THE ABSORPTION HEAT PUMP

1.1 Introduction This chapter is meant as an introduction to the absorption heat pump (AHP) in general and focuses on its most important aspects. This subject will be dealt with more specifically in the sections following this introduction. In Section 1.2 an explanation is given of the working principle of the AHP, introducing the different heat and mass flows, the definition of the concentration of a mixture and the need for rectification. In Section 1.3 the log P - l/T diagram, with the concentration as a parameter, of an absorption working pair (mixture / pure substances) is explained from a theoretical basis. The different processes in the AHP are presented in the corresponding diagram. Section 1.4 defines some ratios concerning the performance of the AHP. These are the heat ratio, the cofficint of performance and the circulation ratio. Section 1.5 deals with the research field on AHP's. Apart from a distinction into theoretical and experimental research work, a distinction has been made into three different fields of research. Those fields are research regarding new working pairs, new system configurations and heat and mass transfer improvement in components. The last section, Section 1.6, describes in the framework of the other sections mentioned above the research work reported in this dissertation. It focuses on the goals, the means and the set up of this AHP project.

1.2 Working principle Almost all handbooks on absorption and on heat pumps contain a detailed description of the basic principle of the AHP, its components, the chosen workinq pairs and the processes of flow, heat and mass transfer (Kirn [KI], Bergmahs [BI], Stolk [SI]). The figure on the next page shows a simple scheme of an AHP, with its components, the corresponding heat flows and temperature levels and the different mass flows (Figure 1.1). In an AHP one is deal ing with a working pair, that is a solvent and an absorbent. The first is the volatile component, the second the less (or non-) volatile component. The concentration of a mixture, vapour or liquid, is defined as the weight fraction w, that is w = kg solvent / kg mixture (1)

In the AHP there are two mass flows (liquid/vapour), and both can be a mixture of both components. That is first the fluid that is circulating in the cycle of the absorber, the heat exchanger and the generator, and secondly the fluid/vapour that is flowing through the condenser and the evaporator, and then absorbed in the absorber and generated in the generator (absorbed in and generated from the first fluid). Depending on the boiling point difference AT between the solvent and the absorbent, the vapour generated in the generator can contain a certain amount of the absorbent, the less volatile component.

Qc, Tc

Mv. Wv

COND

RECT

Qr. Tr

GENE
Mp. Wp _

Qg. Tg

H.E.
Mr, Wr

Ph t
PI
Qe. Te

Pp

H_J_
ABSO

EVAP

Qa. Ta

Figure 1.1

Scheme of an absorption heat pump

If so, a rectification column between the generator and the condenser is necessary to ensure a weight fraction w of almost 1.0 in the vapour going to the condenser. The remaining absorbent in the vapour, with or without rectification, will remain in the evaporator and must be transported to the absorber to maintain a steady state in the process. In practice, a boiling point difference AT larger than 200 K will ensure a vapour weight fraction w close to 1.0. So the first fluid is a relatively "rich" or "poor" (solvent) mixture of both components, the second fluid/vapour has a weight fraction w of 1.0 or very close to that, because, if rectification is necessary, it will never be a complete rectification of the vapour. Let us now follow the process cycle, starting in the absorber (see Figure 1.1). A rich mixture, coming from the absorber, is pumped, causing a rise of pressure, through a heat exchanger, where it is raised in temperature, to the generator. By adding a heat flow Q at a "high" temperature level T , vapour is generated from the rich mixture, which, now a poor mixture, returns via the heat exchanger, where it is lowered in temperature, and an expansion valve, where it is lowered in pressure, to the absorber. Now, the vapour enters, if necessary, the rectification column where it is cooled by removing a heat flow Q at a temperature T . The now rectified vapour passes to the condenser and the remaining "very poor" mixture returns to the generator. The vapour, rejecting a heat flow Q at a "medium" temperature level T by condensation in the condenser, returns by another expansion valve, Feducing the pressure, to the evaporator where a heat flow Q is adapted at a "low" temperature level T by evaporation. Then the vapour (pure solvent) flows to the absorber and is absorbed by the poor mixture, rejecting a heat flow Q at a "medium" temperature level T . In the following figure the different temperature and pressure lvels (Figure 1.2) can be found with:

T P w Q

= = = =

temperature pressure weight fraction heat flow

[K] [Pa] [kg/kg] [W]

* Qm
Qcf fQT
Ph
! Qg

P
Figure 1.2 Scheme of the AHP in a P - T diagram PI

Qe

Qal

Te

Tm

Tg

Under most conditions the heat flows in the absorber, in the condenser and in the rectification column are rejected at the same temperature level T_ 'm' so

VVV
Qm = wQ a
H

T m
m
+

(2)

The total heat flow output Q m is

QC

(3)

In the case of our working pair methanol - lithium bromide / zinc bromide (CH30H - LiBr/ZnBr2 (2:1 mol)), the methanol is the solvent. For this working pair AT 200 K, salt has a negligible vapour pressure The same holds for the well-known working pair lithium bromide/water (H20 - LiBr). So in both cases rectification is not necessary. The mixture of our working pair has a weight fraction w between 0.28 and 0.40. In the case of the also well-known working pair NH 3 - H 2 0, the solvent is ammonia, AT = 135 K, so rectification is a necessity.

1.3 The log P - l/T diagram When deal ing with a working pair, one has one extra variable, namely the weight fraction w. In other words, the liquid mixture possesses two degrees of freedom at liquid-vapour equilibrium, so w = w(P,T), P = P(w,T) and T = T(w,P). Derived from the well-known Clausius - Clapeyron law, the following relation between the absolute vapour pressure P and the absolute temperature T, known as the Antoine equation, can be written for the pure solvent and for a mixture log P = a - b r / T log P = a'- b' (r + 1) / T (4) (5)

with

r = enthalpy of evaporation/condensation 1 = enthalpy of absorption/desorption a,a'= constant b,b'= constant

[J/kg] [J/kg]

With this in hand one can draw for this type of mixture a log P - l/T diagram with the weight fraction w as parameter. One should keep in mind that every point in this diagram describes an equilibrium situation, so a process can not be drawn in the diagram, only the two points between which it takes place and the direction. The "processes" in the AHP are drawn in the following log P - l/T diagram (Figure 1.3).

I nP

Figure 1.3 The log P - l/T diagram for an AHP with the weight fraction w as parameter

1 / T
In this diagram two important simplifications have been made, namely that the enthalpy of evaporation/condensation and the enthalpy of absorption/ generation are independent of temperature and weight fraction. Nevertheless this diagram gives a good overview and presentation of the processes taking place in the AHP.

1.4. The heat ratio, the cofficint of performance and the circulation ratio If one considers the AHP as a reversible one, one can, with the ingoing and outgoing heat flows at three different temperature levels (Figure 1.2), define a so called heat ratio , based on the Carnot efficiency:

<YTe I \

T \ I <V e>

(6)

The heat flow Q is adapted in the evaporator at the temperature level T , the ambient temperature. Therefore this heat flow is not to be accountedefor in Equation (6). Furthermore, the process in the generator and in the absorber take place at a temperature range, here a mean temperature is assumed for both. With this in hand, one can also write for the heat ratio, based on the "ins" and "outs" and the enthalpies: in: evaporator r, T, generator r + 1, T out: condenser absorber

r, T
r + 1, T

e - [(r + l) + r] / (r + 1) and with # - 1 / r C - ( 2 + # ) / ( l + # )

(6a) (7) (6b)

Most of the working pairs have a positive 0, so for a (single stage) AHP the heat ratio will never exceed 2. As one can see the heat ratio is a pure theoretical parameter, also some assumptions had to be made, as well as some simplifications. A more practical parameter, mostly based on the output of experiments, is the so-called cofficint of performance, the COP, which is defined as follows: COP = energy output / energy input For the AHP that leads to COP = ( Qa + Qc + Qr) / ( Qg + Pp ) with: Q Q Q Q^ P = = = = = heat flow from absorber heat flow from condenser heat flow to generator heat flow from rectif. column pumping energy to pump [W] [W] [W] [W] [W] (8a) (8)

Since P Q , the pumping energy P is often left out from Equation (8a). This cofficint will be used later on in the experiments with the computer simulation model and program and with the test plant as a main parameter for the performance of the AHP. Another practical parameter is the circulation ratio f, the mass flow ration of the rich mixture and the vapour leaving the rectification column: f = M r / Ms with: f = circulation ratio [kg/kg] M = mass flow [kg/s] r = rich, p = poor, s = solvent (9)

By using two mass balances, of the total mass and of the mass of the solvent, the circulation ratio can be expressed in terms of the weight fraction w: total mass balance mass balance solvent leading to: M w M + M + M = M = w M (10) (11) (9a)

f (1 - W ) / (wr - w )

Together, the total heat production Q , the cofficint of performance COP and the circulation ratio f form thefflainoutput parameters of the AHP.

1.5 Research work on absorption heat pumps The ongoing research activities and those which took place in the past in the field of the AHP can, generally speaking, be divided into three research categories: a. research on (new) working pairs, b. research on new sorption system configurations and c. research on heat and mass transfer (components). Within these fields one can also make a distinction in theoretical and experimental research activities. It is not a surprise that there always is one main goal behind the direct goals of the research work: saving energy. Looking at the last five years, it seems that more and more the research work on sorption system is focussed on new system configurations. Mostly by means of a theoretical approach, that is by using computer simulation techniques. It is in this research field that researchers expect the most promising results concerning energy saving. In the research work on (new) working pairs for sorption systems, only a very few promising ones have been found up until now and they are still in the experimental stage. The well-known working pairs NH 3 -H 2 0 and H20-LiBr are still favourable because of the experience and reliability of these working pairs, also taking their disadvantages into account. Research work on heat transfer, with or without change of phase, takes place extensively. This is not so when also mass transfer takes place. So in components with simultaneous heat and mass transfer, which is the basic principle of sorption, not so much results can be found. Although this research work is mainly focussed on components, on exchangers, for simultaneous heat and mass transfer, for heat transfer with or without a change of phase, all three research fields will be introduced briefly. 1.5.1 Working pairs for an absorption heat pump The well-known working pairs NH 3 - H 2 0 and H 2 0 - LiBr are the most applied working pairs in an AHP, but a world-wide research took place at technical universities and research institutes to find better working pairs, or working pairs that could meet the disadvantages of the above mentioned working pairs. In an AHP for example, the working pair NH 3 - H 2 0 has high working pressures and the necessity of rectification. Limitations in the use of the working pair H 2 0 - LiBr are the solubility (danger of crystallization) and the evaporation temperature (danger of freezing). The following list gives an overview of the most important selection criteria for a working pair for an AHP. Of cause some criteria are more important than others or are in a way depending on one or more other criteria. 1. 2. 3. 4. 5. the the the the the heat of evaporation r [J/kg] of the sol vent, maximum working pressure P, [bar.kPa] of the solvent, circulation ratio f [kg/kg]'oT the mass flows, energy consumption P [W] of the mixture pump, heat capacity c [J/\kg.K)] of the mixture (sensible heat),

6. the critical temperature T [K] of the solvent and the absorbent, 7. the boiling point differente AT [K] between the solvent and the absorbent, 8. the solubility in the desired operation field (no crystallization), 9. the chemical stability at the maximum desired temperature, 10. the corrosion to construction materials, 11. the toxity and 12. the availability and the costs. At the Technical University of Essen, West Germany, a great effort has been made in the measurement of the thermodynamic properties of and the composition of the P - T and h - w diagram for promising working pairs. Based on the list above, they selected the following criteria: 1. the heat of evaporation at 0"C, r 0 [J/kg], 2. the pressure of the solvent at 50C, P [bar], 3. the circulation ratio f [kg/kg], 4. the pumping energy factor N : M p - (f AP) / (rQ . p r ) with: AP = pressure difference over the pump [kPa] p - density [kg/m3] (12),

5. the heat exchange factor N. : N h = ((f-D c p AT) / r Q with: (13),

AT = temperature difference absorber and generator [K] [K], the necessity of rectification and

6. the boiling point difference AT 7. the toxity

In the fundamental experimental research to new working pairs, that is the determination of the thermodynamic and physical properties, a great effort is put over the last twenty years to meet the disadvantages of and replace the well-known working pairs like NH 3 - H 2 0 and H 2 0 - LiBr. Although successes are made in this direction, the disadvantages of the new working pairs and the advantages of the conventional working pairs, leads to a tendency to choose for reliability and experience, so to conventional working pairs. Only a very few new working pairs have been successfully, that is, are applied in experimental test plants of a sorption system. But the conventional working pairs are still favoured in industrial applications, they are the only systems in practical use today. The problem is that most of the new and promising working pairs fail on one or more criteria. In most cases corrosion and/ore chemical stability is the bottleneck, furthermore the toxity seems to play an important role.

Around 1980 methanol (CH30H) seemed to be the solvent of the future, in combination with LiBr or a mixture of LiBr and ZnBr2 as the absorbent. Unfortunately the methanol itself and the methanol/salt mixture showed to be chemically unstable above 110 - 130C, as detected by Koebel [K2]. That meant that only a maximum temperature (in the generator) of about 120"C is allowed and, with an evaporation temperature of about 0C, the useful heat can only be produced at about 40 - 50C. This drastically diminishes the chances for a direct gas-fired domestic AHP heating system. Until now, one of the most promising solvents (working fluids) seems to be the 2,2,2-trifluorethanol (TFE, trifluorethanol, CF 3 CH 2 0H). The thermodynamic properties are extensively investigated by Bokelmann (TUEssen)[B2, B3] and Girsberger (TU-Bern) [Gl, G2]. Working pairs with TFE as the solvent have already been tested in some AHP or AHT pilot plants. Berghmans [B4] tested the working pair TFE - Chinoline (C9H7N) in an absorption heat transformer (AHT) with a heat output of 275 kW. Nakayama [NI] has developed a sorption system (AHT) with the working pair TFE - N-methyl 2-Pyrrolidon (NMP, C 5 H 9 N0). Bokelmann [B3] investigated the possibilities of several working pair with TFE of which the TFE - NMP showed to be the most promising one. Since with NMP there is the need of rectification, he later changed to 2-Pyrrolidon (Pyr, C 4 H 7 N0) [B5]. This last working pair, TFE - 2-Pyrrolidon, will now be tested in an AHT test plant (heat output 10 kW) at the Laboratory of Refrigeration Engineering at the Technical University of Delft, the Netherlands [Wl]. 1.5.2 System confiqurations As the absorption heat pump AHP is in fact developed from the absorption cooling machine ACM, the absorption heat transformer AHT is developed from the AHP. The following figure (Figure 1.4) shows the different temperature levels of these three basic sorption systems. New configuration of sorption systems are more or less based on these three systems.

Figure 1.4 Temperature levels in the ACM, the >

AHP and the AHT

ACM

AHP

AHT

The first to mention is the two or multi-stage configurations to achieve higher temperature differences and/or a better performance (COP) (Hobling [Hl], Alefeld [Al], Ziegler [Zl]. Although some experimental research has taken place with test plants, most is done by means of computer simulation because of the high investment costs of test plants. 8

For the ACM and the AHP a higher temperature lift (double-lift) or a higher COP (doubl effect) can be achieved. For the AHT a higher COP at a smaller temperature lift (doubl effect) or a higher temperature lift at a lower COP (doubl lift) can be achieved [Zl]. Figure 1.5 shows a two-stage AHP system.
dg. Tg

double-lift (temperature)

Ore. Tre

Ppa

Ode. Tde
Qa. Ta *

b. doubl effect (COP)

Qc, Tc

Qg. Tg

Figure 1.5 A two-stage absorption heat pump system

Qe, Te

In practice it seemed that more than two-stage is not yet feasible. For industrial application more and more two-stage configurations are found because of the increasing possibilities to reduce the investment costs. An also possible configuration is the resorption heat pump or heat transformer (RHP / RHT). In the AHP / AHT the evaporator and the condenser circuit is replaced by a solution circuit with a desorber (a generator at low pressure and temperature) and a resorber (an absorber at high pressure and temperature). In fact the complete name should be the absorption / resorption heat pump. So the "evaporator" and the "condenser" are now working with gliding temperature differences. Figure 1.6 is showing a resorption heat pump system. The choice between an absorption or a resorption system is not easy (Westra [W2], v/d Ree [Rl], Baehr [B6]), because the differences in performance are small. The tendency is that the profits from the gliding temperatures are smaller than the losses because of the extra investment costs.

Also in view of the energy efficiency the absorption system seems to be favourable. For the adaptability to changing process temperatures, a resorption system seems to be better.
Ore. Tre

RESO

- -

RFNF

O Tg

H.E.r 1

L KEa -J
Ppr

Figure 1.6 A (absorption/) resorption heat pump system

Qde. Tde

Ph

Ppa

(J

PI

DESO

ABSO

Qa. Ta

An other new configuration is a combination of the RHP / RHT and the compression cooling machine CCM and the compression heat pump CHP. That is the compression heat pump with solution circuit. Instead of the evaporator and the condenser a solution circuit with an absorber and a generator is used. Figure 1.7 shows such a system.
Ore. Tre

RESO

Figure 1.7 A compression heat pump system with a solution circuit

KE. Pp Ph P I
Pcomp

Qde. Tde

DESO

As with the resorption system, there is the possibility to work with gliding temperature differences (Ahlby [A2]) to achieve a higher COP. Another name for such a configuration is a compression/absorption system. 1.5.3 Heat and mass transfer In sorption (absorption, resorption, desorption, etc.) the basic principle is that in the components for these processes, simultaneous heat and mass transfer is taking place, independent of the type of component (bubble, film, spray, e t c ) . On macro-scale the measurement of the transferred amount of heat and mass is not too difficult by means of the different heat and mass balances. 10

But on micro-scale, where one is interested in the temperature and weight fraction distribution (boundary layers), it is very difficult, if not impossible, to measure temperature and weight fraction. To meet this, simulation models are developed based on the different theories on heat and mass transfer (film theory, penetration theory, etc.) like v/d Wekken c.s. [W3] and Grossman [G3] did. But for a better understanding and improving of the sorption processes, this simplification is necessary. In this research work there were used to predict the right surface configuration for the different processes. Key word in this procedure is the area density 0, the transfer surface area [m 2 ] per volume [m 3 ] of the component. 1.6 This research work 1.6.1 Goals The scope of the research reported here is a theoretical and experimental study into the possibilities of the application of so called compact heat and mass exchangers (CHME) in gas fired AHP's for domestic heating. This heat and mass exchanger, the CHME, is of the plate type. The space between the parallel and plain plates are filled with corrugated plates of some height. This is what is called the corrugation. The plain and corrugated plates are stacked and, with headers and pipes, brazed or welded together. This qives a heat and mass exchanger which is very compact and has more than 700 m 2 transfer surface per m 3 . This leads to heat and mass transfer with small temperature and concentration differences. Besides this more qualitative aspects, more quantitative information is needed to say more about the behaviour and performance of this type of heat and mass exchanger. This leads to a theoretical and experimental research into the application of this type as a component of an AHP and well as: - absorber and generator: simultaneous heat and mass transfer - evaporator and condenser: heat transfer with change of phase - mixture-mixture heat exchanger: heat transfer Next to that, the AHP as a whole - its behaviour and its performance at all desired working conditions - will be subject of this research. In that frame work an AHP with heat and/or mass transfer components built as compact heat and/or mass exchangers (except for the generator, which is of the shell and tube type) was tested with the working pair lithium bromide LiBr / zinc bromide ZnBr 2 (2:1) and methanol CH 3 OH, so a salt / alcohol mixture with alcohol as the solvent. The thermodynamic properties of this working pair are well known and it is extensively tested in an earlier AHP test plant by Iedema [II] and Saurwalt [S2]. Therefore this working pair is suited to test the compact heat and/or mass exchanger components in a new AHP pilot plant, although one of the most important and limiting disadvantages of this working pair is the fact that the methanol in the mixture is not stable above temperatures of 120C (393K) and is breaking up, a process which is not reversible.

11

1.6.2 Tools The starting point of this research study was the research work of Iedema, especially his dissertation "The Absorption Heat Pump" [II]. In this one can find the first set up for the development of an AHP for domestic heating. Based on the thermodynamic properties, derived experimentally and theoretically by the fundamental equations of state, of the until that time known working pairs, the most suitable one was chosen, LiBr/ZnBr2 (2:1) CH3OH. With that in hand a computer simulation model and program was developed by Bakker [B7] to simulate an AHP for domestic heating with this working pair. In this program all the components were put in separated subroutines and were of the shell and tube type, with straight or wound tubes. Only the mixture heat exchanger was of the plate type. The complete program was run through until all the components itself and the AHP as a whole were in balance. The main input variables were the ambient temperature and the chosen control strategy (mixture mass flow and additional heating). Furthermore by means of a literature study on flow hydrodynamics, heat and/or mass transfer, all concerning the falling film flow around tubes, a theoretical model was developed for the absorber, consisting of rows of parallel tubes, and for the occurring processes of flow, heat and mass transfer. It should be emphasized that this dissertation was mainly concentrating on the absorber as the most important and limiting component

of the AHP.
Finally an AHP pilot plant was built to test the absorber and the heat pump as a whole. Also here the components were of the shell and tube type, only the mixture heat exchanger was of the plate type. For a better heat and mass transfer and a more compact construction (domestic heating !) the step is made to the above mentioned compact heat and mass exchangers, CHME. So for this research, there are three starting points: a. the dissertation of Iedema [II], mainly the chapters 3 until 7 deal ing with heat and/or mass transfer, the absorber model and the simulation of and the experiments with an AHP, the computer simulation model and program of an AHP for domestic heating with a salt/alcohol mixture [B7], an AHP pilot plant with components built with so called compact heat and/or mass exchangers, only the generator is still of the shell and tube type.

b.

c.

1.6.3 Set up This work, with the above mentioned three starting points, has also three basic elements, namely simulation/modelling, experiments and literature. The next chapters will contain the results of this research work, realized along the above mentioned lines.

12

CHAPTER 2.

COMPACT HEAT AND MASS EXCHANGERS

2.1 Definition Compact heat and/or mass exchangers have a compact transfer surface, with a area density greater than 700 nr/m3, a somewhat arbitrary value. Heat and/or mass exchanger stands for the transfer surface of heat and mass, while compact stands for not only compact itself, but also for enhanced. In short, a compact heat and/or mass exchanger is an exchanger for heat and/or mass of a compact construction with many "extra" transfer surface and a relatively small volume. 2.2 Compact and enhanced transfer surface Concerning the heat transfer, one can define the amount of heat which is transferred from one side of the exchanger to the other side as follows: Q = K A AT with: Q = K = A = AT= heat transfer rate [W] overall heat transfer cofficint [W/m2.K] total heat transfer surface area [m 2 ] mean temperature difference [K] (14)

Furthermore one can define for the heat exchanger 7 = Q / AT with: and 0 = A/V with: f} = area density [m 2 /m 3 ] V = total volume [m3] (16) 7 = specific heat transfer rate [W/K] (15)

For exchangers of the plate type, this leads to

h = A c / V c or *h =A h / Vh
with: c = cold side, h = hot side

(16a>

and for exchangers of the shell and tube type, this leads to 0t = A t / V t with: t = total (16b)

An clear overview of the different types of exchangers concerning their area density is derived from Shah [9] and shown in Figure 2.1. From (14), (15) and (16) one can derive 7 = K fi V (15a)

1 3

-COKPACTNESS?"

* KATTER OF DERf-E

oooooo oooooo

\-"'''"'"

" > " " >>"

o o o o o o o o o o 1 o o o o o o o o o o o o o o o o o o o o o I ^QO_ooooooooy
Gai Turbin. Rotary iHegenccitoi

-O O O-r

KH

Buun

: ...

OIOO^
For X X , 1.B8,

Automotivr fUdiators

Matrix Types, H.r* Screen Sphere Bed, Corrugated Sheets

S t r i p - F i n and Louv^red-Fin H.E.

E j p

Jnd

O O.B!]

SL
P U i n TubuUr, Shell-and-Tub H.E.

COMPACT SURFACES
llydrdullc DiaawtU D L 2

20

10

I I II I
60 100

Tl
200

l l i l il
500

,I

1(

1
2000

I MUI.
5000 10

I
2

I .
1x10

Heat translVi Brfac- *

Figure 2.1 The area density / ? for different types of exchangers from Shah [9]

Figure 2.2 A stack of parallel plain and corrugated plates

plain triangular fin

plain rectangular fin

wavy fin

offset strip fin

round perforated fin

pin fins

Figure 2.3 Different types of corrugated finned plates (corrugations) from Shah [3]

1 4

The goal is to achieve a great specific heat transfer rate in combination with a small mass and volume of the exchanger by means of the use of compact surfaces. This most of the time leads also to a greater overall heat transfer cofficint K which itself also leads to a smaller volume. This compact construction is strong and stable with small wal! thicknesses. This also reduces the volume and to a smaller content the mass of the exchanger. With enhanced surfaces is meant to achieve a greater overall heat transfer cofficint compared to the not enhanced plain surfaces. So it deals with the factor K A (= y ) , achieving an enlargement of K and/or A. This can be done by a. adding extended surface to the prime surface (K and A greater) b. adding turbulators to the surface (K greater) c. reducing the flow passages (hydraulic diam.) (K and A greater) As one can see, a. and c. also increase the transfer surface area. Compact and enhanced transfer surfaces characterizes more the category of compact exchangers than the earlier mentioned area density of 700 m 2 /m 3 . Clearly is shown that in most cases compact and enhanced appear together. This all offers the possibility to apply this type of exchanger with an extensive choice in type, geometry and area density of the surface. So a great flexibility in the choice of surface on both sides and a reduction in the total mass and volume can be obtained. This also offers the possibility of automated production techniques, certainly in the case of the plate fin type exchangers as used in this research work and shown in Figure 2.2. This all can lead to a competitive price of this type of exchanger.

2.3 Application 2.3.1 Field of application The most important application of compact heat and/or mass transfer surfaces concerning the type of mass flow, is on the gas side in heat exchanging processes (in gas/gas, gas/liquid, gas/condensing,evaporating liquids). To a much smaller extend they are used for applications for two phase flow or on the liquid side. It is obvious that for a certain rate of transferred heat a much larger transfer surface is needed on the gas side than on the liquid side because of a normally 10 - 100 times smaller heat transfer cofficint on that gas side. Unfortunately the enhancement of transfer surface by means of corrugations or turbulators gives a much greater pressure drop on the gas side. Therefore the adding of that type of turbulators is limited. Application in the aircraft and automobile industry was the starting point of the development, foliowed by application as heat transfer components in installations [S4]. The application in sorption systems is rather new, up until now hardly no literature is found on that, except for Minkus [Ml]. 2.3.2 Corrugations If one limits oneself to compact heat and/or mass exchangers consisting of a stack of parallel plain and corrugated plates, one can make the following classification for the type of corrugation [S3] (Figure 2.3):

15

2.3.3

plain fins (rectangular, triangular, square) wavy fins offset strip fins louvered fins perforated fins pin fins

Construction materials

Construction materials are mainly aluminium and copper while stainless steel and other corrosion resistant materials are only scarcely used. This is mainly caused by the difficult manufacturing and the brazing of the corrugated plates of this materials (AKG [A3], Trane Company [Tl]). 2.4 Characterization 2.4.1 Identification To identify the performance and quality of each type of surface (here corrugated plate) two characteristic properties are used, one for the heat transfer, j, also called the Colburn factor, and one for the friction, f, also called the Fanning friction factor [K3]. Both are non-dimensional and defined as follows: j = St Pr 2 / 3 f = 2 to / (p v 2 ) with: St Pr Nu Re to p v = = = = = = = Stanton number (St = Nu/(RePr)) Prandtl number Nusselt number Reynolds number surface shear stress [kg/m.s2] mass density [kg/m3] velocity [m/s] (17) (18)

When the thermodynamic properties of the flow medium are known, one can define j and f for a given geometry of the corrugation as a function of the Reynolds number Re. An other form of presentation which is also commonly used in literature is the following one: Nu = a Re b Pr c and with: (here c=l/3) (19) (20) [-] [N/m2]

AP = f d 1/2 p v 2 a, b, c, d = form factors (constant) AP = pressure drop

So in this way the factors j and f, or Nu and AP, are commonly used to express the performance of the type of transfer surface (corrugated plate).

16

2.4.2 Analvtical solutions In Shah and London [S5] one can find a great amount of analytical solutions, that is for simple geometrical configurations like triangular, rectangular and circular channels. It is restricted to fully thermal and hydrodynamically developed laminar flow, that is fully developed temperature and velocity profiles. This gives a constant Nusselt number, independent of the Reynolds and Prandtl number, only depending on the geometry. For three sets of thermal boundary conditions the Nusselt numbers are given: 1. constant wall temperature, 2. constant wall temperature and heat flow in the axial direction and 3. constant heat flow in the axial and peripheral direction. The characteristic length in the Nusselt number and in the earlier mentioned Reynolds number is the so called hydraulic diameter d

h'

dh
with:

4 Ar 1
d, Af A 1 = hydraulic diameter [m] f flow cross sectional area [m 2 ] = total heat transfer area [m 2 ] = flow length [m]

(21)

For most of the geometries one can write: h = -Q-f with: 0 = wetted perimeter [m] For a better understanding of the above mentioned formulas, this means for a rectangular channel (see Figure 2.4) :
d

(21a)

h-i

A r 1 _ 4 (w hl 1 A 2 (w + h) 1

2 w h w +h

(21b)

with: w = flow passage width [m] h = flow passage height [m] 1 = flow passage lenght [m] If w h than d h 2

Figure 2.4 Offset strip fin corrugated plate

flow direction

t = fin thickness [m] x = uninterrupted fin length [m]

1 7

2.4.3 Experimental results

2.4.3.1 Introduction The type of heat and/or mass exchangers that is used in the earlier mentioned absorption heat pump (AHP) pilot plant is of the plate fin type. The transfer surface is corrugated and of the offset strip fin type. Figure 2.4 shows an example of this type of corrugation. In the following the limitation is made to heat exchangers of the plate type and to heat transfer. This because most of the available literature deals with heat transfer, and the exchangers in the pilot plant are of the (corrugated) plate type as shown Figure 2.4. 2.4.3.2 The work of Kavs and London A Standard reference in this field is the book "Compact Heat Exchangers", written by Kays and London [K3], It gives a thorough and detailed survey on compact heat exchangers, not only concerning the geometry, but also a very large amount of experimental results. Not only for transfer surfaces of simple geometry, but also for very complicated geometries experimental results are given for tubes and plates. For all these types data are given for the factors j and f as well as a clear indication of the geometry. Almost all the experiments took place with an air flow through the corrugated passages and a (condensing) steam flow through the other passages. For the overall heat transfer cofficint K in Equation (14) one can write: K = [ 1 /aQ + d y ^ + 1 /as V1 with: (22)

a- heat transfer cofficint on the air side [W/m2.K] d = plate thickness of the wall [m] X thermal conductivity of the wall [W/m.K] a = heat transfer cofficint on the steam side [W/m2.K]

For this kind of heat exchange it is allowed to say: a and X/d L so: K * a (22a) s o ' o o 2.4.3.3 Rectanqular offset strip fin surfaces Since the corrugated surfaces of the components of the pilot plant are of the rectangular offset strip fin type, the focus will be on that type of corrugated surfaces. Wieting [W4] has gathered all the experimental results for rectangular offset strip fin configurations from [K3] and [S6] and correlated the experimental heat transfer and flow friction data over two Reynolds number ranges, that is for Re < 1000 (laminar) and Re > 2000 (turbulent). A clearly defined transitional Re, Re*, was not found. So to minimize the effect of the transitional Re, occurring basically between 1000 < Re* < 2000, on the correlation, this range was excluded. To find these correlation for heat transfer (and for flow friction an analogous one, that is f) the following non-dimensional functional relation was assumed: a

18

j = a (x/d h ) b - (t/d h ) c . (w/h) d - Re e (a, b, c, d, e en d are unknown coefficients.) The Reynolds number is based on the hydraulic indicated variables are as follows: 0.7 0.03 0.162 0.65 100

(23)

diameter d. . The ranges of the < < < < < x/d. t/d" w/hn d. R < < < < < 5.6 0.166 1.196 3.41 10000

Furthermore, the area density p is in the range 1000 < fi < 3000. Leaving out the flow friction factor, Wieting found the following correlations, based on the available heat transfer data of 22 offset strip fin surfaces: j = 0.483 (x/d,)-- 162 . (w/h)" 0 ' 1 8 4 . Re" 0 ' 5 3 5 , for Re < 1000 n j . 0.242 (x/d.)-- 322 . (t/d.) 0 " 0 8 9 . Re" 0 " 3 6 8 , for Re > 2000
n n

(24)

(25)

This with j from Equation (17) and Nu = ( a dh ) / X Re - ( with: r ? = dynamic viscosity cofficint [kg/m.s] The overall discrepancy between the correlations and the experimental results is within 10 %. For the transitional Re Wieting derived from Equation (24) and (25) the following equation for Re*: Re* = 61.9 (x/d.) 0 " 9 5 2 . (t/d.)" 0 - 5 3 . (w/h) - 1 - 1 n for 1000 < Re* <n2000 (29)
fi

(26) (27) (28)

. v d h ) / r\

Pr = ( t) . c ) / X

So if Re < Re*, than use Equation (24) and when Re > Re*, than use Equation (25) in the transitional range. For any surface geometry of the corrugated plate one can with this calculate the Colburn modulus j as a function of the Reynold number Re. In the same way Wieting derived equations for the friction factor f. If one correlates Equation (24) and (25) for all the 22 offset strip fin surfaces, one can find an "average" correlation for this type of surface for all geometries: n ,. j = 0.487 Re" u - D J D for Re < Re* (30) j = 0.149 R e - 0 ' 3 6 8 with Re* = 1185 for Re > Re* (31)

19

For the use of the Equation (24) and (25) and even more of the Equation (30) and (31) one should be careful in extrapolating data for strip fin geometries that have geometrical parameters outside the range of those for the correlations. This one is strictly based on a limited amount of reported test data. Also these correlations may be applicable only for air or gas as the working fluid (Pr 1 ) . Mochizuki and Yagi [M2] have done experimental research with aluminium test cores of plate fin type heat exchangers with (offset) strip fins, also using air as the test fluid and condensing steam as the other medium. Seven types of strip fin surfaces were tested, with a constant height h (10 mm) and fin thickness t (0.2 mm). They were mainly interested in the influence of the fin spacing (width) w and the fin length x. The range of the parameters is here in the same range as those belonging to the data Wieting [W4] used in Equation (23), except for the hydraulic diameter d h , here 3.04 and 4.35 mm. The corresponding area densities , 8 are 700 and 1000 m2/ni3. They also presented their results in the form of the Colburn factor j for the heat transfer and the Fanning factor f for the flow friction as a function of the Reynolds number Re, but in a different way than Wieting did. Here the wall surfaces of the flow passages are interrupted in the flow direction and the process is repeating itself since the air flows along the very short strip surface and then separates at the trailing edge of the strip. Thus the boundary layer is never able to become thick. Therefore, if the pressure loss is considered to consist of the two effects of form drag of the fin and a friction drag of the fin surface owing to the viscosity of the fluid, then the skin friction factor f may be expressed in the following form: Q f = a + b Re 1 , (32) and a similar expression for the Colburn factor j : j = c + d Re"1,0 (33)

For each type of strip fin surface they defined a set of a, b, c and d. For comparison the following averaged correlation for the Colburn factor was derived from their data: j = 4.72 10"3+ 10.06 R e " 1 - 0 for 1000 < Re < 8000 (34)

For comparison with the correlations of Wieting for the Reynolds number range 2000 - 8000, Equation (34) was transformed into the following one: j * 0.135 R e " 0 - 3 5 0 for 2000 < Re < 8000 (35)

Within this range the maximum deviation in j from Equation (31) is less than 6 %.

20

2.5 Discussion Shah and Webb [S3] have discussed both the results of Wieting and Mochizuki and Yagi. They started with a theoretical solution. For the heat transfer they took the Pohlhausen laminar boundary layer solution for a flat plate of "plate length" x [S7]: j = 0.664 R e ^ 0 - 5 (36)

and f o r the f r i c t i o n the modified Blasius laminar boundary layer s o l u t i o n for a f l a t plate using the form drag associated with the leading blunt edge of the s t r i p f i n [S7]: f =(C d to) / ( 2 x) + 1.328 R e x 5 with: C. = form drag cofficint [-] (37)

In order to indicate and compare the performance of offset strip fin surfaces, one can use the factor j/f as an indication. Based on the Reynolds analogy for flow over a flat plate, in the absence of form drag ( C. = 0 ) , the factor j/f should be 0.5 (for Pr*l). Since the contribution of the form drag is of the same order of magnitude as the skin friction drag for such an interrupted surface, the factor j/f will be about 0.25. If one takes the correlations of Wieting of Equation (30) and (31), one will find that for the ranges Re < 1000 and Re > 2000 the factor j/f is smaller than 0.275. Even in the transitional range the factor j/f does not exceed 0.30. Mochizuki and Yagi used the factor j/f to determine the influence of the geometrical factor x/w and the Reynolds number Re. They found, with a negligibly small dependence on the Reynold number Re, for the factor j/f that 0.2 < j/f < 0.6. Shah and Webb have their doubts about the results of Mochizuki and Yagi because the factor j/f > 0.3 and stated that "published data for strip fins are questionable if j/f > 0.3. All the measurements and possible sources of flow leaks and heat losses must be checked thoroughly for all those basic data having j/f > 0.3 for strip fins." Dubrovskii and Fedotova [Dl], [D2] investigated an offset strip fin surface, not a rectangular one but with a slit form, with d, = 2.9 mm and n = 1280 m 2 /m 3 . They found the following expressions for j and f: j = 0.090 R e " 0 - 3 0 f = 1.590 R e " 0 , 2 7 for 800 < Re < 3250 for 1500 < Re < 3250 (38) (39)

As one can see, in this range the factor j/f is almost independent of that Reynolds number and the factor j/f = 0.045. Compared with earlier results this is rather low.

21

Also attempts have been made to predict the factors j and f numerically by a finite difference method for the offset strip fins considering the laminar boilndary layer on each strip fin. In [S3] Sparrow [S8] and Patankar [PI] are mentioned. They compared the numerical results with a strip fin surface from Kays and London [K3] and found a reasonable agreement for the factor f, but the predicted j factors were about 100 % higher. The predicted slopes of both j and f versus Re curves were steeper than the test data. One of the major unpredictable factors, mentioned in both [S3] and [K3], is the existence of small burrs at the leading and trailing edges of the fin during its formation by a shearing operation. Fins of this type are generally constructed by a machine-cutting process that inevitably leaves a slightly bent and grazed fin edge that differs depending upon the fin material and character of the cutting tooi. These burrs increase the effective plate thickness, causing increased form drag. This factor can not be taken into account accurately in the numerical solutions, nor can the influence on the experimental data be estimated accurately. Therefore, the possibie existence of burrs causes uncertainty in the correlations or in the comparison of predicted values with data. Briggs and London [B8] have paid some attention to this point. So far the theoretical, numerical and experimental information found on the performance of rectangular offset strip fin surfaces in literature.

2.6 Application in sorption svstems In the last section of this chapter on compact heat and mass exchangers, a global comparison will be made between the application as described in the sections before and the application in a sorption system, in particular in the AHP test plant. The main important differences are pointed out. First of all should be mentioned that one has to do with simultaneous heat and/or mass transfer and/or with a change of phase, not only with heat transfer. In literature only information is found concerning pure heat transfer. In the cases of the analytical solutions and the experimental results the flow was in the direction of the "most open" side of the corrugated passages. In the exchangers of the AHP the flows on both sides - both sides have corrugated passages - are in the direction of the "most closed" side of the corrugated passages. Because the strip fins are offset, the passaqes are not completely blocked. The working fluid flowing through the corrugated passages was air, with on the other side condensing steam. In the AHP, in the secondary circuits (for cooling or heating) water or liquid methanol were used as media while in the primary circuits lithium bromide/ zinc bromide - methanol (the absorbent or mixture) and methanol (the sol vent), liquid and/or vapour, were the media. To have a rough indication of the thermodynamic properties of these media, in comparison with those of air, one can find in Figure 2.5 the Prandtl number Pr, the kinematic viscosity v and the thermal conductivity \ over a certain temperature range for the different media. By this one can estimate the possibie influences on the heat transfer, keeping in mind the equations for the Nusselt number, the Reynolds number and the Prandtl number (Equation (26), (27) and (28)) and the heat transfer correlation of Equation (19).

11

100 -

50-

20-

10-

0.5PrandN X-10*(W/m-K] V-10* 0.2 -20 20 (*/$)

r60

80

100

120

T (C)

Figure 2.5

The Prandtl number Pr, the kinematic viscosity u and the thermal conductivity A as a function of the temperature T for: air (A), water (B), methanol (C liquid, D vapour) and the working pair lithium bromide zinc bromide - methanol (E)

2 3

In this the kinematic viscosity v is V = v / P [m2/s] (40) The secondary passages of the exchangers in the AHP are completely fil led with the liquid flow media, the primary passages, except for the mixture heat exchanger, are not. In the absorber there is a (liquid) film flow over the fins and a vapour flow in between. In the evaporator and condenser there is a liquid and vapour flow, that is an evaporating and condensing film flow. The generator is left out, this exchanger is not a CHME. For heat exchangers operating with high-density fluids (in this case liquids) the friction power is generally small relative to the heat transfer rate, so has not a great influence. But for low-density fluids like air, as one could see in the sections before, the friction power is of the same order and even greater than the heat transfer rate. As said before, the heat transfer cofficint on the liquid side is 10 - 100 times greater than on the vapour/gas side. So for the exchangers in the test plant, not the factor j/f is of importance but the Colburn factor j. The experimental result came always from a cross-flow configuration to keep the header construction simple, in the exchangers of the AHP always counterflow took place. Only the position of the headers could give very a small cross-flow influence. In short, the thermodynamic properties and the phase of the chosen media as well as the flow and corrugation configuration of the exchangers in the AHP test plant form the major differences with the information found in literature. To find out whether or not the correlations found in literature for heat transfer are in whatever form applicable for the heat transfer processes in the exchangers of the test plant, a computer simulation model is developed for a liquid flow over a offset strip fin surface. In combination with the results of the experiments with the test plant heat and mass transfer correlations will be searched which can be compared with those found in literature. Both the experiments and the simulation model will be discussed later.

24

CHAPTER 3. THE ABSORPTION HEAT PUMP TEST PLANT 3.1 Introduction During the research work of Iedema [II] an experimental absorption heat pump (AHP) test plant was built. Experiments took place mainly to test the absorber. Furthermore to investigate the possibilities of the working pair CH3OH - LiBr/ZnBr 2 . The absorber was a horizontal tube bundie heat and mass exchanger with a liquid film flowing over and dripping between the tubes (Figure 3.1).

Figure 3.1

Film flow over a row of horizontal tubes

This type of film flow gives a good heat and mass transfer in the absorption process because the flow is frequently interrupted. So the film flow, and with that the formation of the thermal and concentration boundary layers will be disturbed. This breaking-up, when falling from one tube and splashing on the next one, gives a good mixing of the liquid film, leading to a more homogeneous temperature and concentration distribution over the film thickness. So in this test plant, the absorber was the main subject of investigation and experiments. The other components in this plant were also conventional heat and/or mass exchangers: a shell and tube bath evaporator, a coaxially wound condenser and a electrically heated bath generator. The mixture heat exchanger was of the plate type, but with a rather bad performance. All these component were only to form an AHP cycle in which this type of absorber could be tested. So the absorber was the determinative and limiting component in the cycle. The test plant was designed to have a total heat output Q (from absorber and condenser) of about 10 kW. The result of the experiments with this absorber and this AHP cycle as a whole, can be found in [II] and [S2]. In the present research work a next step is made to develop an AHP for domestic heating as described in Section 1.6.

25

To this, one must still keep in mind that the goal is to develop such a AHP, in combination with an additional heating system, that can meet the required heat output and heating temperature at a given outdoor temperature in combination with an acceptable cofficint of performance (energetic efficiency) and occupying an acceptable space in the house (volume). To meet these demands to a certain level the idea was generated to investigate the possible application of so called compact heat and/or mass exchangers (CHME) as components in an AHP. So an evaporator, a condenser, an absorber and a mixture heat exchanger of this type were installed while the "old" absorber will now be used as a generator.

3.2 Choice of the working pair Because of the knowledge of and the experience with it, the same working pair as used by Iedema [II] will be used in the new test plant: lithium bromide LiBr / zinc bromide ZnBr 2 (2:1 mol) and methanol CH 3 0H, so an alcohol/salt mixture. In the absorption cycle the alcohol is the solvent and a alcohol/salt mixture the absorbent. With that working pair a good qualitative and quantitative analysis can be made of the performance of this type of exchangers in an AHP by means of experiments. 3.3 Type of compact heat and/or mass transfer surface As earlier mentioned this type consists of a stack of parallel plain and corrugated plates, brazed or welded together. This corrugated plate will be called "the corrugation". So the space between the parallel plain plates is filled with corrugations, in this case even on both sides and of a different type. The type of the corrugation is mainly determined by the kind of medium (mixture, water, methanol, oil), the state of that medium (liquid, liquid film, vapour or a combination) and the type of process taking place that is heat and/or mass transfer, sometimes in combination with a change of state. In all the CHME components the corrugations are of the offset strip fin type. Where there is a liquid (film) flow, sometimes in combination with a vapour flow, through the corrugated passages, the flow is not in the "open" direction of the strip fin but at right angles to it. In that configuration the heat and/or mass transfer will be much better. There will be a more turbulent flow because of the constant change in flow direction. Where a vapour flow is the main flow (condenser and evaporator), the flow is in the "open" direction. This because pressure losses are far more greater in the vapour phase than in the liquid phase. In general three types of corrugated plate surfaces of the offset strip fin type are applied: a. In the primary circuits of the absorber (liquid film flow of mixture) and of the evaporator (vapour and liquid film flow of methanol), a trapezium type (Figure 3.2a):

mun
26

Figure 3.2a

The trapezium type offset strip fin

b. In the secondary circuits of the condenser (water) and the evaporator (methanol), a wavy type (Figure 3.2b):

Figure 3.2b The wavy type offset strip fin c. On both sides in the (mixture) heat exchanger, on the cooling side in the absorber and on the condensation side in the condenser, a rectanguiar type (Figure 3.2c):

-1 1 1 1 ,1
J

1 1
1

1 1
1

Figure 3.2c The rectanguiar type offset strip fin 3.4 Overview of the AHP test plant For a better understanding of the whole AHP test plant, an overview of the test plant is shown by the schematic flow sheet in Figure 3.3. The design quantities of the AHP test plant are given in Table 3.1.
Table 3.1 maximum maximum maximum heating cooling heating Design quantities 10 kW 0.1 kg/s 3.5 g/s 100 - 120 C 30 - 50 C -15 - +15 C

heat production Q rich mixture mass flow M solvent mass flow M temperature generator T temperature abs./cond. f temperature evaporator T

Given the chosen working pair, this means a pressure P. in the generatorcondenser section of 20 - 70 kPa and a pressure P, in the absorberevaporator section of 2 - 7 kPa. Also from this the heat flows can be derived and, with an estimation of the appearing temperature differences and the overall heat transfer cofficint, the required transfer surfaces can be calculated roughly (see Table 3.2).
Table 3.2 component The required heat flows and transfer surfaces required heat flow estimated temp. diff. estimated heat transfer coeff. K [J/kg.K] 600 300 750 750 200 required transfer surf.[m2] 2.0 4.0 1.8 1.8 5.0

Q [kW]
absorber generator condenser evaporator heat exch. 6 6 4 4 10

AT [K]
5 5 3 3 10

27

(^Pump

elec. h r a l i n g

Figure 3.3

An overview of the AHP test plant

In the next section a description of the components in the AHP test plant is given, such as the geometry of the transfer surfaces, the construction materials, the processes of heat and/or mass transfer taking place and the geometry and construction of the component itself. A more detailed description can be found in v/d Welle [W5].

28

3.5 The components 3.5.1 The absorber In the absorber a process of simultaneous heat and mass transfer takes place, the vapour coming from the evaporator is absorbed by the liquid mixture (poor to rich), rejecting the heat of condensation and mixing. That heat is removed by a water flow on the other side. The absorber (Figure 3.4) is a construction of stainless steel, the corrugation are made of Inconel 600 and the rest of the absorber of AISI 316. It consists of 10 primary passages (h = 8 mm) and 11 secondary passages (h = 3 mm), the plate thickness is 1 mm. The plate transfer surface has a length of 700 mm and a width of 240 mm.
poor mixture (in)

cuoling water (out)

cooling water (in)

vapour

rich mixture (out)

Figure 3.4

The absorber

As one can see in Figure 3.4, in the primary passages only 160 mm of that width is wetted by the liquid, over the other 60 mm the corrugations are placed in the opposite direction to get a better distribution of the vapour flow, giving a better contact with the mixture and reducing the pressure losses. Taking this all into account, the absorber has a total (plate) transfer surface of about 2.2 m 2 .

29

3.5.2 The condenser and the evaporator The superheated vapour from the generator is fed to the condenser. It will first cool down till the condensation temperature is reached and then condense, forming a (growing) liquid film flow. The rejected heat is removed by a water flow on the other side of the exchanger. The condenser is a complete aluminium construction. It consists of 10 primary passages (h = 10 mm) and 11 secondary passages (h = 3 mm). The plate transfer surface has a length of 370 mm and a width of 250 mm, giving a total (plate) transfer surface of about 1.8 m 2 . In the evaporator the liquid sol vent (methanol) coming from the condenser via an expansion valve is evaporated by adding a heat flow from the other side of the exchanger. In that secondary circuit, also methanol is circulating to be able to achieve evaporating temperature below 0 C (273K). On the side of the evaporating methanol the exchanger is open, there are no headers. The whole is placed in a glass/stainless steel container, with a distribution tray on top of the exchanger. This component is also made of aluminium and has almost the same geometry as the condenser. The total (plate) transfer surface is about 1.7 m 2 . During the experiments it showed that the evaporator was leaking because the low pressure slowly increased during stable operation. Under water it showed that the exchanger had several holes in the passage. It was replaced by a new one which showed after a short time the same problem. At that time also the condenser showed air leaking in. The main cause of the leakage seemed to be the presence of salt particles in both the components. These must have come from the generator, dragged by the methanol vapour or by a too high liquid level in the generator. To solve this corrosion problem both the components were replaced by stainless steel ones, with a triangular type of corrugation on both sides.The plate spacing is 3 mm on both sides. The corrugated plates have perforations of 0 2 mm every 10 mm as shown in Figure 3.5. Only on the condensation side there are no perforations.

AAAfFigure 3.5

3!

The corrugated plate of the condenser/evaporator

There are 20 passages on the solvent side with the flow direction to the "open" side of the corrugated plate (to avoid vapour friction/pressure drop) and 21 passages on the cooling/heating side with the (liquid) flow to the "closed" side. the width is about 250 mm, the flow length about 270 mm for condensation and 250 mm for evaporation. This gives a total (plate) transfer surface area of about 2.3 m 2 for the evaporator and 2.7 m 2 for the condenser.

30

3.5.3 The mixture-mixture heat exchanger This component exchanges heat between the relatively cold and rich mixture flow coming from the absorber via a pump and the relatively warm and poor mixture flow coming from the generator. A good performance of this component is of great importance for a good performance of the whole AHP. The warmer the rich mixture flow is entering the generator, the less heat is needed to generate a certain amount of methanol and, the colder the poor mixture is entering the absorber, the less methanol will flash after the expansion valve. In general, with a better heat exchanger the COP will increase. The heat exchanger is constructed out of a nickel alloy, Inconel, and has 20 passages on one side and 21 passages on the other side. The passage height is 3 mm and the plate thickness is 0.4 mm. The exchanger has a height of 500 mm and a width of 300 mm, both on the outside, giving a total (plate) transfer surface of about 5.60 m 2 . 3.5.4 The generator In this component the rich mixture flow, coming from the absorber via the pump and the heat exchanger, is partly degassed by adding a heat flow from the other side of the exchanger. In this case electrically heated thermal oil is used. This exchanger is not a heat and mass exchanger of the plate type, nor very compact, but a conventional shell and tube exchanger. The mixture is flowing over the tubes, dripping from one to an other, while the thermal oil is flowing through the tubes and heating the mixture. It is a construction of 10 horizontal and 32 vertical rows (2 x 16 offset) of tubes. The tube length is 0.4 m. This gives a total transfer surface of about 4.80 m 2 (outside tubes). 3.5.5 Distribution of liquid flows (prim.) A good distribution of the liquid mixture flow in the absorber and the generator and of the liquid methanol flow in the evaporator is of great importance for the heat and/or mass transfer in the component. The total transfer surface area of the component is to be used for heat and/or mass transfer. The quality of the distribution should also be independent of the mass flow. To take care of a good distribution of the liquid flow, trays are placed in the top of the component. The trays have a perforated bottom (sieve). To avoid the confluence of the liquid drops into bigger drops or even jets, the perforations in the trays of the absorber and the generator are "filled" with pins. The perforations in the tray of the evaporator are provided with small tubes. More on the distribution trays one can find Table 3.3.
Table 3.3 component generator absorber evaporator Overview of the geometry of the distribution trays frontal area comp. [cm2] 1536 216 277 number of perf. 832 100 88 diam. diam. perf.[mm] pin [mm] 2.2 3.0 2.0 1.5 1.0 remaining flow passage [cm2] 16.9 6.3 2.8

31

3.5.6 The mixture pump Most fluid pumps have dynamic seals which do not meet the vacum requirements. Pumps with a wet rotor and static seals do but have a too high capacity to pressure rise ratio. Here a magnetically driven gear pump was applied with a stainless steel housing and teflon gears, provided with a thyristor-controlled motor. A stainless steel filter was placed between the absorber and the pump to protect the gears against solid particles.

3.6 The heating and coolinq circuits 3.6.1 The coolinq system of the absorber and condenser From both the components the rejected heat is removed by a water flow. This cooling water is coming from a buffer tank (150 litre) where the desired temperature can be obtained (see Section 3.7.2). The water is normally flowing parallel through the absorber and the condenser - each circuit has its own circulation pump -, a connection in series is also possible. It is also possible to reject an amount of heat to the secondary circuit of the evaporator by means of a copper wound doubl tube heat exchanger. 3.6.2 The heating system of the evaporator The evaporator is heated by a methanol flow. The desired temperature can be obtained by exchanging heat with the water circuit of the absorber/condenser and by an additional electric heating element (see Section 3.7.2). This circuit also contains a reservoir (30 litre). 3.6.3 The heating system of the generator The generator is heated by a thermal oil flow. The circuit contains a reservoir (32 litre) where the desired temperature can be obtained (see Section 3.7.2). Furthermore the circuit contains a expansion vessel with a safety valve. 3.6.4 General All the secondary circuits contain a volume flow meter with a scale from 0 until 100 %. With this the mass flow can be calculated with the density (temperature dependent) of the medium and the type of float. With a valve close behind the flow meters the flow rate can easily be adjusted by hand. All the circuits have a centrifugal circulation pump. For Standard operating conditions, the mass flows are listed in Table 3.4. Table 3.4 component absorber condenser generator evaporator The mass flows in the secondary circuits (heating/cooling) medium water water oil methanol temperature [C] 40 40 120 10 mass flow [kg/s] 0.25 0.22 0.69 0.25 ( 80%) (100%) (100%) (100%)

32

3.7 Control 3.7.1 Mass flows The rich mixture mass flow M can roughly be adjusted by hand by means of the thyristor-controlled motor of the mixture pump, with a volume flow meter as an indicator close behind it. When the entering temperatures of the four secondary mass flows are stabie, the mass flow of the poor mixture and the methanol can be stabilized by hand with the corresponding expansion valves and level indication tubes (glass). The position of the valves is changed as long as the levels are unstable. After a while a stabie working condition will be created as far as the mass flows are concerned. One should nevertheless keep in mind that all the other temperatures and also both the pressures must be stabie to have a stabie working condition of the AHP as a whole. If one input parameter is changed, all the output parameters will also change. So a lot of patience is needed to achieve a steady-state equilibrium condition. For the adjustment of the mass flows in the secondary circuits, see Section 3.6.4. 3.7.2 Temperatures This all concerns the adjustment of the entering temperatures of the heating/cooling media in the components. Both the secondary circuits of the absorber/condenser and of the generator have three electrically heated ceramic elements of resp. 2 kW and 3 kW each. The first one is an on/off element, the second one is an adjustable element (between zero and fuil power) and the third one is a PIDregulated element. That third heating element is always in operation, controlling the desired temperature by thermocouples in the exit of the buffer tanks and in melting ice as a reference. Concerning the absorber/condenser, the first and second heating element are only used to start up the heat pump. The adding of cold water in the buffer tank and the heat exchange with the heating circuit of the evaporator, cools down the water and the third element takes care of the temperature by adding a (little) heat. Concerning the generator, the first and/or the second heating element is always in operation, they must take care of the total heat input of the generator. The secondary circuit of the evaporator has one heating element, a PIDregulated one of max. 1 kW. After the heat exchange with the cooling water of the absorber/condenser, the methanol is already heated up a little bit and then the heating element takes care to achieve the adjusted temperature. Also here the controlling takes place by thermocouples in the exit of the buffer tank - here the housing of the heating element - and in melting ice as a reference. 3.7.3 Weight fraction The weight fractions can only be adjusted or controlled roughly. Between the condenser and the expansion valve to the evaporator there is a buffer vessel with a certain amount of methanol. So in combination with the expansion valve, methanol can be added to or removed from the primary circuit. An estimation of the new weight fractions can be made based on the vibration time of the density meter (see Section 3.8.2).

33

3.7.4 Pressures As showed before, when the temperature and the weight fraction are known and there is equilibrium, the pressure is determined as one can see from the log P - l/T diagram. But since in most of the cases there is no equil ibrium because transfer surfaces are always finite, the pressure is an interesting parameter. It is used as an indication whether there is equilibrium between the absorber and the evaporator (low pressure) and between the generator and the condenser (high pressure). For example, an increasing low pressure means that there is more methanol evaporating in the evaporator than the mixture in the absorber is absorbing. Therefore both the output signals of the pressure gauges are continuously visible by means of a two-channel writer. 3.8 Measurement 3.8.1 Mass flows The real measurement of the mass flow takes place between the heat exchanger and the absorber. There a magnetic-inductive flow meter measures the volume flow of the poor mixture V by giving an electric output signal which is scanned by a data-logger. Also the temperature is measured at that place. Between the condenser and the expansion valve before the evaporator a volume flow meter is placed to measure the methanol volume flow V . Also the temperature is measured at that place. With these two the density p and the mass flow mm can be calculated. 3.8.2 Mixture density The density of the rich mixture mass flow p is also measured. A small amount of that mixture is pumped from the small reservoir before the pump through a density meter, working according to the vibrating U-tube principle. The measured vibrating time, given by a digital output signal, and the temperature determine the density and with that the weight fraction of the rich mixture w can be calculated from the thermodynamic properties. 3.8.3 Temperatures The temperatures of all ingoing and outgoing liquid and vapour flows are measured (primary and secondary circuits) with copper-constantan thermocouples, giving output signals in milli-volts, which are scanned and translated to degrees Celsius by a data-logger. 3.8.4 Pressures Both the pressures in the absorber/evaporator and the generator/condenser are measured by a pressure transmitter (strain gauge sensor), giving a signal in the 10 V range. For the low pressure this corresponds with 100 mbar (10 kPa), for the high pressure with 1000 mbar (100 kPa). Both signals are also scanned by a data-logger.

34

3.8.5 Accuracv of measurements Before presenting figures and values on the performance of the AHP and its components, it is correct to give an indication and/or estimation of the accuracy of the results, that is the possible errors in the measured and calculated parameters. First of all the temperature control. The specifications of the manufacturer says that a maximum variation of 0.1 K around the desired temperature will occur. The temperature is measured by means of copper-constantan thermocouples. lts thermo-voltage value has an accuracy of 0.05 K, and, together with the datalogger output, the maximum accuracy in that value will be 0.1 K. The flow meters in the secondary circuits of the absorber, condenser and generator have an accuracy of + 2.0 % at full scale, which is almost always the case. The flow meters in the secondary circuit of the evaporator and in the mixture circuit, close behind the pump, have an accuracy of + 2.5 % at full scale. For the first this is almost always the case, the second is only an indication for adjusting the pump velocity. The flow meter in the primary circuit indicating the methanol mass flow, has also an accuracy of + 2.0 % at full scale. Taking into account the operating range that will be + 4 - 8 %. The flow meter in the primary circuit indicating the mixture mass flow, has an accuracy of + 3.3 %, which is the average over the total range of the scale. With this in hand, an estimation can be made of the accuracy of the calculated heat flows, taking into account the measured temperature increase / decrease in the secondary circuits, as one can find in Table 3.5. Table 3.5 : component absorber generator condenser evaporator heat exch. Indication of the errors in the different (calculated) heat flows error in heat flow primary circuit + + + + + 5% 5% (4 - 8) % (4 - 8) % 3.5 % error in heat flow secondary circuit + + + + + (4 - 12) % (4 - 12) % (4 - 12) % (4 - 9) % 3.5 %

The 12 % in the second column seems somewhat high, but one should keep in mind that this holds for tests with a small mixture mass flow and a cooling water temperature of 50 C, so for very small heat flows. The experimental results show that for the generator the heat flow Q (s=secondary) is always greater than the heat flow Q (p=primary). The 9' difference is never more than 6 %. One may assume thar this is mainly caused by heat losses. For the absorber the heat flow Q . is also always greater than the heat flow Q , with differences up to'i5 %. Taking into account possible heat losses,'this is opposite to what could be expected. Besides the possible errors as stated in Table 3.5, a more constant error must be made by incorrect temperature/mass flow measurement. It was not possible to detect the cause of this problem.

35

The measured heat flows Q and Q in the condenser agreed very well with ,p a maximum difference of 2 %. For the evaporator this was not so, the heat flow Q was always greater than the heat flow Q . This is not surprisingly because the evaporator construction and the low temperatures inside the the housing will cause a heat flow into the housing. Both heat flows Q, in the rich mixture/poor mixture heat exchanger differ no more than 1 %. For calculation of the transferred amount of heat the average of the heat flows on both sides is used. 3.8.6 Data registration All signals from the data-logger can be sent to a micro computer. This computer has a program that asks the volume flow percentage of the secondary flows, the number of scale part of the methanol condensate volume flow and the vibrating time of the density meter. Together with the data from the data-logger all heat flows, mass flows, weight fractions, temperatures, pressures etc. are calculated, including the cofficint of performance COP. This can give a total overview of the performance of the AHP and its components. One calculation should be mentioned, that is the one of the primary mass flows and both the weight fractions. The following equations are available: w r M r = w p Mp + Ms Mr = Mp + M s (11) (10)

K = P n ( w n J J Vn (41) p p p p p With these three equations and the known rich weight fraction w , mass flow solvent M and volume flow of the poor mixture V , the unknown poor weight fraction w and the mass flow of the rich and poBr mixture M and M can be p calculatedpby means of an iteration program. Next to the calculating program, the micro computer has a program with which the data-logger can be programmed to show at any desired frequency all the scanned magnitudes (temperatures, pressures and volume flow).

3.9 Construction materials and corrosion A corrosion investigation program has been performed by FDO Hengelo [Fl] (mentioned in [II]) for the mixture CH30H - LiBr/ZnBr2 with various metallic materials, pure, soldered with several types of solder or welded to other pieces or to the same material. All testings took place at a temperature of 110C (generator level). A general conclusion was that in de presence of oxygen almost every material was severely attacked. It appeared that Inconel, a nickel alloy, performed the best, but this is a rather expensive material. Stainless steel AISI 316, a less expensive material, was also satisfactory as the main construction material.

36

The applied construction materials are the following ones: - stainiess steel type AISI 316: all pipes in the primary circuit and all the components, - stainiess steel type AISI 304: all flanges and valve housings in the primary circuit, - Inconel 600 (a nickel alloy): the corrugations in the absorber and heat exchanger, - aluminium: the evaporator and condenser in the first experiments, - copper: all pipes and valves in the secondary circuits, - viton : synthetic rubber (vacum sealings).

So far the description of the AHP test plant. The simulation model and program of this plant will be discussed in Chapter 5. The experimental results will be presented in Chapter 6. But first in Chapter 4. the selection of the working pair will be shown.

37

CHAPTER 4. THE WORKING PAIR : SELECTION AND DATA 4.1 Introduction Although working pair research and selection were no subjects in this study, sortie neccessary information is already given in the Sections 1.5.1 and 3.1 concerning resp. working pair research and development and the choice of the working pair CH30H - LiBr/ZnBr2 as a test fluid in the test plant. The basic idea behind a limited research/study on working pairs for sorption systems was to have an overview of the available literature and data on this subject. Furthermore the idea ros to test the plant and the exchangers as well as the simulation model/program with a second working pair for validation and verification. 4.2 Experiments For experiments with the test plant (see Chapter 3.) and its components an important limitation in selecting a suitable working pair was the construction of the test plant. The whole plant was designed to work under sub-atmospheric pressure. The vacum connections however, are with a small modification able to resist a maximum pressure of 2.5 bar. So the limitation was to the so called low pressure working pairs. An other restriction is that no rectification may be necessary. So a boiling point difference over about 200 K is desired. Furthermore the resistance of the construction materials, like stainless steel, copper, viton e t c , against the working pair and its components must be assured. Also the acquisition and price can give a limitation. Finally the basic thermodynamic and physical properties must be known to calculate heat and mass flows and weight fractions (enthalpy, density, heat capacity, log P - l/T diagram). 4.3 Simulation For the experiments (see Chapter 6.) only the availability and acquisition of the basic thermodynamic and physical properties data are of importance, but now more than that - as mentioned before - is required like the viscosity and the thermal conductivity. 4.4 Literature survey on working pairs Many theoretical and experimental attempts have been made to find new working pairs for sorption systems. This to replace the well known working pairs NH 3 - H 2 0 and H 2 0 - LiBr and to meet their disadvantages and/or to improve their performance. The first aim was to investigate what is done at other research institutes and technical universities on working pair research and data collection. The results of this survey can be found in [B10].

39

A summary of this survey can be found in Section 1.5.1. The most important conclusion might be that TFE (2,2,2-TriFluorEthanol) seems to be a very promising solvent, in combination with absorbents like (Iso-) Chinoline, NMP and 2-Pyrrolidon.

4.5 Selection of a new working pair Given the limitations described above, and the restriction that the working pair must be applied in both the experimental and theoretical tests, so in the simulation program and in the test plant, the far best working pair seemed to be R123a - DTG: R123a: DTG : 1,2 dichloro 1,1,2 trifluoro ethane (CHC1F-CC1F 2 ) Chemical Abstracts Number: 354-23-4 2,5,8,11,14 pentaoxapentadecane (CH 3 (0C 2 H 4 )40CH 3 ) better knows as Dimethyl ether Tetra ethyl ene Glycol also known as DMETEG, TEGDME and E181 Chemical Abstracts Number: 143-24-8

The choice is mainly based on the research work of Bokelmann [B2,B11]. It meets far best the requirements and the limitations mentioned above. Bokelmann has measured the basic thermodynamic and physical properties and done a limited amount of tests on stability and corrosion [Bil]. The limitation for experimental tests seems anyway the instability and the corrosion problems above 160C. Furthermore the toxity of the R123a is still a point of uncertainty. Information from ICI UK [12] and Allied Chemical USA [A4] led to the conclusion that the toxity of the R123a is acceptable under the normal precautions of handling chemical substances. For the experiments a generation temperature below 150C is no problem but information on stability and corrosion with the R123a and the R123a-DTG in that temperature range are not available. So tests must clear that problem first before applying the R123a-DTG for experiments in the AHP. But in fact the bottle neck in this was the availability and/or the acquisition of the R123a. Intensive contacts with a lot of dealers and producers of chemical substances showed that obtaining the DTG was no problem, but no one could provide the R123a. Finally ICI UK could provide first a simple prescription to produce the R123a out of the R113 and triethanolamine (TEA): R113 : 1,1,2 trichloro 1,2,2 trifluoro ethane ( CC1F 2 -CC1 2 F ) Chemical Abstracts Number: 76-13-1 tri-ethanol amine ( N(CH 2 CH 2 0H) 3 ) Chemical Abstracts Number: 102-71-6

TEA

Latter on they sent a patent of Dow Chemical with a more detailed prescription of the production of the R123a. Because of the estimated time to produce the R123a and to do the necessary stability and corrosion tests, this and the experiments with the working pair R123a-DTG feil out of the scope of this research work.

40

4.6 Consequences for the test plant For the construction of the test plant it means some small modifications. First of all the vacum connections. The 0 - rings must have an outer support to keep the ring fixed under super-atmospheric pressure. The low pressure transmitter can be replaced by the high pressure transmitter. The last must be replaced by a new one in the range of 0 - 4 bar. The pressure levels can be seen from the log P - l/T diagram in Figure 4.1. For comparison the log P - l/T diagram of CH30H - LiBr/ZnBr2 is shown in Figure 4.2. The floating ball in the solvent flow meter must be replaced by a heavier one because of the expected increase of that mass flow. So for the experiments no great and/or costly modifications have to be made. The calculation program must also be modified for the thermodynamic and physical property data of the working pair 4.7 Consequences for the simulation program The property data mentioned above [Bil] meet for a great part the data demands of the simulation program. The missing property data are on the thermal conductivity of the liquid and the vapour, the viscosity and density of the vapour and the mass diffusion cofficint in the liquid. For some data acceptable estimations can be made by using approximation formulas. From Perry [P2] a formula was derived for the mass diffusion cofficint in the liquid, based on a correlation by the Wilke-Chang technique. From Jamieson [Jl] a correlation for the thermal conductivity of the liquid was derived, based on experimental data of R123 ( 1,2 dichloro 1,1,2 trifluoro ethane) and DTrG (Dimethyl ether Triethylene Glycol) and using a mixing formula. The vapour density is based on the law for an ideal gas. The viscosity and the thermal conductivity are less important for the program (only used for the superheated vapour in the condenser) and based on data from comparable gases.

4.8 Comparison R123a - DTG with CHiOH - LiBr/ZnBr? For a quick comparison between both working pairs, some criteria mentioned in Section 1.5.1 are calculated, as wel! as some other parameters, for a theoretical absorption heat pump cycle with T = 0C, T = 50C and T = 125/150C. No temperature differences are assumed in the heat and/or mass exchangers. The generation temperature is chosen at 125C because above that temperature the methanol in the mixture becomes unstable. An other generation temperature is chosen at 150C, probably the stability limit of the R123a-DTG working pair. So this is only a theoretical comparison to have a first impression of the performance to be expected of both the working pairs. This is shown in Table 4.1 At first sight the methanol working pair seems to be better overall. As can be seen the heat of evaporation has a very great influence on the performance. For the R123a this is rather low, also compared to NH 3 , H 2 0 and TFE of resp. 1260, 2400 and 450 kJ/kg.

4]

l'CI 1B0 500 IkPal P

100

120

140

160

R123a-OTG

lUU ~

(kPal
P

^fl -

7/ J>
iv -

f7 7 /

VA
/

IZ / Z i / /, ' / ' /// // / / / I Z '// ZZ / / / / ZZ / / / /y ~z ^ V/ 7J V / < ' / zz / / /1 7^ / / / // ' / 77ZZ / / v Y / / / / A / / , '/ 7/ ' / / / //, % / '// i / > / / / 1

f,
7

7?

$777Mt V V/A
^
te

A 7/

/ J

'7,V

/ y '/ / ./ rj /i / r / /. / y AV/ / // ' y /, A A /

777,

200

VA

> \ /* V/l" 77 / 7 / / i f

.,*

y r-

100

7 72 4 7 V A V
/
/
60

10 -20

7/

VA%
20

10

ao
T CC)

Figure 4.1

The log P - l/T diagram of the working pair R123a - DTG

70 80 90 100

I'CI 120

110

(kPal P

Figure 4.2

The log P - l/T diagram of the working pair CH30H - LiBr/ZnBr2

42

But when higher temperatures are possible with the R123a working pair, this becomes more and more competitive. Information from Allied Chemical USA [A3] mentioned that with aluminium as construction material instead of stainless steel, and the addition of an inhibitor, even higher temperature than the 160C are possible without any corrosion and/or stability problems.
Table 4.1: A theoretical comparison of the working pairs working pair generation temp. T heat evap. at 0C rQ pressure P. at 50C pressure P" at 0C poor weight fract. w rich weight fract. w circulation ratio f density p at 50C pump. fact. N (xlO6) spec. heat c at 50C heat exch. factor N, boiling point diff.AT 0.22 0.28 12.0 1080 102.0 1825 8.2 275 3.8 0.8
co

R123a - DTG

CH30H - LiBr/ZnBr2

unit

125
185 2000 300

150

125
1200 540 40

150

C
kJ/kg mbar mbar

0.13 4.8 40.8

0.31 0.37 10.5 1585 2.8 1360

0.27 6.3 1.7 0.6

kg/kg kg/kg kg/kg kg/m3 m3.mbar/J J/(kg.K)

K *

* : no rectification is needed for both the working pairs. 4.9 Conclusions This survey on working pairs led to the conclusion that for the experiments with the CHME components in the AHP test plant the working pair CH 3 0H LiBr/ZnBr2 will be used because of the experience with this working pair and the availability of almost all data on the thermodynamic and physical properties. This also means that it is suitable for application in the simulation model and program. Furthermore a comparison can be made with the experimental results of the shell and tube absorber from Iedema [II]. Because of the problems with the availability and/or the acquisition of the working pair R123a - DTG, experiments with this working pair to test the CHME components in the AHP test plant feil out of the scope of this research work. The data on the thermodynamic and physical properties are sufficint for the simulation model and program. The missing data will be obtained by approximations by means of empirical relations in literature.

43

CHAPTER 5.

COMPUTER SIMULATION MODEL AND PROGRAM

5.1 Introduction The aim behind the development of a simulation model and program for the absorption heat pump test plant, was to have a basis to develop a more general program that can serve as a design tooi for any sorption system, with a flexible cycle configuration of components, component geometry and working pair. It can also serve as a test program for new types of components and for new working pairs, to indicate whether or not experimental testing might be interesting and worthwhile.

5.2 Startinq point In an earlier stage of the research work a computer simulation model and program has been developed by Bakker [B7] for an AHP system for domestic heating. As the working pair for the AHP lithium bromide / zinc bromide - methanol was chosen. The same model and program was used by Iedema [II]. The model contained a complete domestic heating system, that is an AHP and an additional heating system. At a given outdoor temperature, the required domestic heat flow input and the heat flow output of the AHP at maximum mixture mass flow were calculated. If the required heat flow input was smaller than the produced heat flow output, the mixture mass flow was accordingly reduced. If the required heat flow was greater than the produced heat flow, the additional heating system provided the necessary extra heat flow output. This heating system was a directly gas fired boiler, producing water of 120C at 3 bar, and provided also the heat flow to the generator. Every component was put in a separate subroutine with the corresponding input and output parameters/variables. All the components were linked together by means of a main program. The methanol mass flow was used as the main iteration parameter. The calculation stopped when the equilibrium was reached in and between the components, that is when the different methanol mass flows are with a certain accuracy equal (generated in the generator = absorbed in the absorber = flowing between condenser and evaporator). The whole program consisted of three sub-programs: - the main program, a control program containing the main iteration loops and handling the input/output of information, - a program containing subroutines for the components and for the heat and mass transfer correlations and - a program containing the subroutines of the thermodynamic and physical properties of the different media (water, water/glycol and the alcohol salt working pair). So generally speaking, the only input parameter was the outdoor temperature and most important output parameters were the COP's of the total heating system and of the AHP. The computer simulation output as a whole contained all information on temperature, pressure, heat flow, mass flow, weight fraction, enthalpy, geometry, efficiency etc. of the components and the connecting pipes. All the heat and/or mass exchangers were conventional shell and tube exchangers, except for the mixture heat exchanger which was a plate exchanger.

45

In the program heat and mass transfer correlations were obtained for that type of exchangers from VDI - Wrmeatlas [VI] and Gregoric [G4] for the heat transfer on the cooling/heating side. For the heat and mass transfer on the absorption/generation side the film and penetration theory were used for resp. heat and mass transfer [II], [C2].

5.3 Development of a new simulation model and program The main differences between the AHP which the computer model and program described above is simulating, and the AHP of the test plant are the input parameters and the type of heat and/or mass exchangers. Furthermore the former computer program simulates not only the AHP but a complete domestic heating system. The task is now to transfer this program into a simulation program for the AHP test plant. The first step was changing the condenser and the evaporator from the conventional shell and tube exchangers in plate type exchangers (geometry and transfer correlations). In this the presence and influence of the corrugations was not accounted for because no information was found on this type of heat and/or mass transfer. Also the geometry of the mixture/mixture heat exchanger was changed. A next step was to change also the absorber from the shell and tube type in the plate exchanger type. On the side where the absorption takes place, a new calculation model had to be developed for the process of simultaneous heat and mass transfer. In the former model it was assumed that the absorption takes place during the time the liquid flows around the tubes like a film and that a complete mixing takes place when the liquid drops on the next tube. So the liquid film flow starts on that tube with a homogeneous temperature and weight fraction distribution over the film thickness. This concept is now adapted by dividing the plate in a number of horizontal parts with the absorption in the parts and the mixing process between the parts. A change of the partition from 50 to 200 parts is only causing a negligible improvement of the absorption process. Keeping in mind the presence of corrugations, this is a realistic partition since there are about 110 fins over the total length (700 mm) of the absorber. The next step [V3] was to leave out the domestic heating part. The generator is now heated by thermal oil flow, the evaporator by a methanol flow. The input parameters are now (still) the outdoor temperature and the rich mixture mass flow. As one could deduce from Chapter 3. the input parameters of the AHP test plant are the temperature and mass flow of the cooling and heating flows of the components, the rich mixture mass flow through the pump and the weight fraction of the mixture. As earlier mentioned, in the test plant the last two can approximately be adjusted by means of the pump rotating frequency and the sol vent buffer vessel, in combination with the density meter. So the next important step was to change the simulation program in such a way that the above mentioned parameters became also the input parameters of that simulation program. This was not possible for the weight fraction, this parameter is completely determined by the processes of absorption and generation, so by the equilibrium of state that is reached by means of the iteration program.

46

In the test plant it is also determined that way but can however be influenced by adding or abstracting solvent, so by changing the average mixture weight fraction. The transformation from the simulation of a complete domestic heating system to the simulation of the AHP test plant is now completed as far as the calculation program is concerned. A last modification concerned the program that contained the thermodynamic and physical property dat of the media. It was in such a way extended that the media can be selected, the cooling and/or heating media as well as the working pair. That is, new media can be added. With any set of input parameters as mentioned above the AHP simulation program can calculate all desired output parameters such as the mixture temperatures at any place in the AHP as well as the corresponding pressures, mass flows and weight fractions. Coupled to that calculation program, there is a program that takes care of the output. With this the input and output parameter of all the components and connecting tubes can be listed in a clear overview. 5.4 The final simulation program 5.4.1 Structure
The total program A W P (AHP) consists of a series of related sub-programs as shown in Figure 5.1.

AWP

I
RUN Figure 5.1 Structure of the AHP simulation program OUT MAIN -- C O M P O N SPLIT-^ FYS

The main program MAIN takes care of the input, the starting procedure, the iteration procedure and the output. Furthermore it links the components of the AHP. The program COMPON is cal led upon by the main program and contains the subroutines of the components and its correlations for heat and mass transfer. Every subroutine/component calculates with input from the main program its internal equilibrium of state and send its output to the main program. The program FYS (PHYS) is a property data bank and contains the thermodynamic and physical property data of the applied media.

47

For the working pairs those are CH30H - LiBr/ZnBr2, R123a - DTG and TFE - 2Pyrrolidone, and for the cooling and/or heating media those are water, thermal oil and methanol (from CH30H - LiBr/ZnBr2 for w = 1.0). Concerning the working pairs, the methanol working pair data are complete. The R123a working pair data are complete enough (see Chapter 4.) for calculation and simulation purposes. The TFE working pair data are far from complete and mostly based on the working pair TFE - NMP [Wl]. The program SPLIT links the property data bank with the main program and the components program. The input parameters and the component geometry are provided by the program RUN, called upon by the main program. With the program OUT the main program takes care of the output, sending it to a file, a printer or a screen.

5.4.2 Heat and mass transfer correlations In appendix A one will find a listing of the applied correlations for heat and/or mass transfer in the components of the program C0MP0N. Table 5.1 shows the basic values of the heat transfer coefficients under normal operating conditions. Table 5.1 : Heat transfer cofficint (W/m2.K) (basic values) component working pair side (solvent/mixture) 3000 - 4200 25 / 1070 750 1625 200 / 200 cooling/heating side (water/oil/methanol) 95 225 190 850

evaporator condenser absorber generator heat exch.

The following partly overlapping and connective considerations are of importance for the application of these correlations and the interpretation of the simulation results. First of all the correctness of the applied heat and/or mass transfer correlations for this type of transfer processes can be a point of discussion. Applying the film and penetration theory for the modelling of the simultaneous heat and mass transfer in the absorber and the generator is based on some assumptions. For the heat transfer (see Figure 5.2a) it is assumed that the transfer is a stationairy process with the temperature difference as the driving force and a linear temperature profile. Furthermore that the temperature boundary layer is almost at once formed over the whole film thickness. This leads for the heat transfer cofficint between the wall and resp. the interface and the bulk of the film with Nu = o 5 / X = 1.0 resp. 1.6 to a = (1.0 resp. 1.6) X / 6 with: 6 = film thickness [m] (42a) (42)

This is the so called Nusselt film thickness that can be defined for a film

48

flow over a vertical plane plate as: 6 = (( 3 n M / O ) / ( p2 g)) 1/3 (43)

For the mass transfer (see Figure 5.2b) the assumption is that the transfer is a stationary process with a decreasing driving force (weight fraction difference). The weight fraction boundary layer is formed rather slowly and over a very small part of the film thickness at the interface between liquid and vapour. For the mass transfer cofficint K , that is here the average value over the flow length or contact time, one can define K m = 2 7 (D / ( * t j ) [m/s]

(44)

with:

D = mass diffusion cofficint [m2/s] t = contact time [s]

liquid

vapour

liquid

vapour Wi

Ti

Wf

/Wb film film

interface

wa

wall

interface

Figure 5.2

a) the film theory for the heat transfer and b) the penetration theory for the mass transfer

More detailed information on the assumptions concerning both theories can be found in [W8]. In this simultaneous transfer of heat and mass, the mass transfer is the limiting process. This can be shown by comparing the heat diffusion cofficint a with the mass diffusion cofficint D, or the ratio of both cal led the Lewis number: Le = a / D with: a = \ I {p (45)

(46)

The Lewis number is mostly in the range 10 - 100, for the methanol working pair the range is 40 - 800, for the R123a working pair 8 - 80 ( It should be noticed that the Lewis number is sometimes defined as Le = D / a ! ) . This leads to a another point of consideration, the working pair property data. As mentioned above, the diffusion cofficint for heat and mass, resp. a and D, are of great importance. Property data on the density and the viscosity in the liquid phase are mostly available, but not on the thermal conductivity and the mass diffusion cofficint in the that phase.

49

For the methanol working pair the thermal conductivity is determined by experiments, but the mass diffusion cofficint is based on experiments with the working pair CH30H - Li Br, for only two weight fractions and with doubtful results [B9]. For the R123a working pair both are determined by empirical relations as described in Section 4.7, so also both with an doubtful accuracy. Furthermore the effect and influence on flow, heat and/or mass transfer of the presence of the corrugated plates between the parallel plates. The corrugation causes a repeatedly process of disruption of the formed boundary layers and and mixing of the fluid. It has also influence on the definition of the hydraulic diameter d. in the Reynolds, the Nusselt and the Sherwood number (see Appendix A : flow length, film thickness, plate length, corrugation width, wetted length e t c ) . This holds for all components and on both sides of the transfer surface. With the experimental results in hand an attempt will be made later on to correct and modify the simulation program by means of the heat and mass transfer correlations in such a way that for any set of input parameters the program can simulate the AHP test plant within an acceptable accuracy. 5.4.3 Component subroutines In Appendix B the calculation procedures of the components are shown. Since the components are all counterflow exchangers, an iteration procedure is necessary. The main iteration parameter is the outgoing temperature of the cooling/heating medium. The geometry of the components is described in Chapter 3. and the applied heat and/or mass transfer correlations in Appendix A. The evaporator, the condenser and the mixture-mixture heat exchanger are looked upon as a whole (input and output parameters, equilibrium, total transfer surface), the absorber and the generator are divided in calculation sections (plate section / per tube). Generator. For this shell and tube exchanger the required calculation steps take place per tube in the vertical direction. So the calculation direction is in the direction of the mixture mass flow. Only the first two tubes of a row are divided again in an upper and a lower part. This because the strongest parameter changes take place on those tubes. The major assumption is that the liquid dropped on top of the tube starts to flow around that tube in a homogeneous state (temperature and weight fraction). So during the dripping from a tube, the fall between two tubes and the dropping and splashing on the next tube, complete mixing of the liquid is assumed. A 4th order Runga-Kutta procedure is used to take into account the fact that per tube there is a cross-flow, the component as such is a counterflow exchanger (Figure 5.1). Absorber: The mixture flow in this component is a film flow over/through the finned plate (corrugation). For calculation purposes it is assumed the the film flows over a flat plate. In the flow direction the plate is divided into N sections. Between two sections complete mixing of the liquid is assumed. Taking into account the plate fin geometry (fin pitch) N should be around 100. Since the small output difference for N = 50 and N = 100, N is set on 50 to reduce the calculation time.

50

For both the sorption components the calculation steps can be found in Appendix B.l. Condenser: This component is divided into two sections. One for the cooling down of the superheated vapour and one for the condensation of that vapour. Although the outgoing cooling water temperature is the iteration parameter, the condensation temperature is changed until the transfer surface area of both sections together are equal to the total available transfer surface area (see Appendix B.2). Evaporator: In this subroutine the temperature of the outgoing vapour is the iteration parameter. It is assumed that the main part of the transfer surface area is used for evaporation and a minor part for superheating the vapour (see Appendix B.3). Heat exchanger: One of the outgoing temperatures (here of the poor mixture mass flow) is the iteration parameter (see Appendix B.4). Furthermore the expansion valves and the pump are put in separate subroutines. Pump: The pump is supposed to have an efficiency of 50%. The energy flow P p is assumed to raise the enthalpy of the passing mixture flow. The temperature raise is no more than 0.2 K. Expansion valves: The expansion valve of the solvent takes care of the required pressure fall and checks whether or not some liquid has evaporated to obtain the equilibrium temperature. The heat is attracted from the liquid itself. The expansion valve of the mixture has the same function and checks the amount of flashed solvent from the mixture entering the absorber. 5.5 A simplified simulation program Given the disadvantages of the simulation program as mentioned above and the effort to decrease the calculation time of the program, a simplification was made by deducting an identical program, but now with fixed heat transfer cofficint for all components on both side of the transfer surface. Those fixed coefficients are shown in Table 5.2.

Table 5.2
component evaporator condenser absorber generator heat exch.

Heat transfer cofficint (W/m2.K) in the fixed model/program working pair side (solvent/mixture) 2500 25 / 750 1000 1500 200 / 200 cooling/heating side (water/oil/methanol) 100 250 200 100

These estimations are based on the result of the experiments with the AHP test plant and its components.

51

5.6 Application of the simulation program: Some examples The aim as mentioned in Section 5.1 to develop a simulation model that can serve as a design tooi for any sorption system with a flexible cycle configuration, component geometry and working pair is a goal on the long term. In this research work the practical application is more to analyze the influence of the different input parameters on the AHP. With this in mind three examples will be given. First both the working pairs selected for simulation will be compared in their performance. This is shown in Table 5.3. Table 5.3: Comparison R123a - DTG and CH30H - LiBr/ZnBr2 with the simulation program heating temp.evap. T working pair heating temp.gen. T pressure P . pressure P . rich mass flow M poor weight fract. w rich weight fract. w^ circulation ratio f r total heat output Q pumping energy P m coeff.of performT COP heat flow h.e. Q. = 5C and cooling temp. abs./cond. T R123a - DTG = 45C

CH30H - LiBr/ZnBr2

units

130
2034 284 100 0.212 0.270 13.7 3.97 32.4 1.29 12.7

155
2265 255 100 0.148 0.236 9.7 5.48 37.7 1.32 16.7

130
595 31 25 0.319 0.356 17.7 3.80 1.7 1.66 2.5

155
675 25 25 0.288 0.344 12.8 5.38 1.9 1.66 3.0

C
mbar mbar g/s kg/kg kg/kg kg/kg kW

w
kW

In this comparison two generator heating temperatures are chosen, 130C as the maximum temperature for the methanol working pair and 155C as the assumed maximum temperature for the R123a working pair, both because of instability. So 155C is of course a theoretical temperature for the methanol working pair. The rich mixture mass flow is chosen in such a way that the total heat output Qm is almost the same for both working pairs. Overall the methanol working pair is better in its performance, but if higher temperatures would appear to be possible for the R123a working pair (inhibitors/ aluminium construction), this working pair might be favourable over the methanol working pair. Not only because of the performance itself, but in that case also the cooling water temperature, and with that the heating temperature for a house, could be raised. A second example of the application of the simulation model can be the influence of the thermodynamic and physical properties. In Figure 5.3a. and b. the influence of the mass diffusion cofficint D of the R123a working pair is shown on the performance of the AHP because no experimental data are available on this. The influence on the COP of the AHP shows the importance of an experimental determination of this cofficint. The correlation from Perry [P2] gives D ~ 1.0 - 7.0 x 10-9, which is normally 0.1 x 10-9 < D < 1.0 x 10-9, so somewhat high.

52

In this correlation the mass diffusion cofficint D is a function of the temperature and the viscosity, and by that a function of the weight fraction. The dependence on the temperature is very strong but on the weight fraction it is rather weak. In the absorber at T ~ 50C D * 1.5 X 10- 9 m/s, in the generator at T * 150C D = 6.0 x 10- 9 m/s. By the way the diffusion cofficint D of the methanol working pair is very strongly depending on both the temperature and the weight fraction.
COP OF PERFORMANCE C O P

l 4 0 COEFF.

-13

-tl

-10

-O

EXPONENT MASS CFF.COEFF. D E

h m/s)

MASS FLOW SOUVENT Ms. (o/s)

fiT"' =

40 'C 0 'C M 100 g/s r =

-II -10

= 1?5 'C

-9

EXPONENT MASS DFF.COEFF. D (D h m/s)

Figure 5.3

The influence of the mass diffusion cofficint D on a) the COP and b) the sol vent mass flow M

A third example is to see the influence of a component on the performance of the AHP. For this the influence of the mixture-mixture heat exchanger between the absorber and the generator will be shown. In general its function is to improve the COP of the AHP. On the side of the poor mixture it cools down the mixture before entering the absorber. So it avoids or limits the mixture to enter the absorber in a superheated state and so the flashing of a part of the solvent in the mixture between the expansion valve and the absorber. On the side of the rich mixture the mixture is pre-heated before entering the generator, or even some generation of the solvent takes place.

5 3

So in the absorber and in the generator the amount of sensible heat to be exchanged is reduced, in the absorber the amount of flashed solvent is limited and some solvent is already generated before the generator. This all leads to a reduction of the required transfer surface area of the absorber and the generator and lowers the required energy input of the generator.
COP. OF PERFORMANCE C O P .

lwpOFF.

100

200

JOO

400

500

HEAT TRANSFER COEFIOENT ALFA fW/mZKI

M flash H a

"

M flash aOs.

M cona

| 2 MASS

FLOW

SOLVENT M (p/s)

125C 40C 0C 100 g/s

HEAT TRANSFER COFFICINT ALFA (W/m2JO

Figure 5.4

The influence of the overall heat transfer cofficint of the mixture-mixture heat exchanger on a) the COP and b) the solvent mass flows M

In Figure 5.4 the overall heat transfer cofficint K, is changed between 100 and 500 W/(m2.K) and the influence on the COP and the solvent mass flow is shown. As can be seen a good heat exchange between absorber and generator is essential for an improved performance of the AHP.

54

5.7 Conclusions The now developed simulation program can serve as an important design and testing tooi for any sorption system, with a flexible cycle configuration, choice of components (type and function) and working pair. This because of the flexible structure of the whole program. New theoretically promising working pairs could be tested on their merits for a decision on experimental testing. Property data of that working pair can easily be fit in in the property program FYS. Also the amount of cooling and heating media can be extended. For the components there is the possibility to change the geometry or to add new components to the cycle, like a rectifier between the generator and the condenser, or a heat exchanger between the condenser and the evaporator. Also new/other cycle configurations are possible like the absorption heat transformer (AHT) or the resorption cycle (RHP/RHT). Although it offers great possibilities in testing new cycle configurations, components or working pairs, there are certain limitations in the accuracy and the reliability. The major limitation in this lies in the thermodynamic and physical properties of the working pairs. A certain amount of property data is needed for simulation results with an acceptable accuracy and reliability. This can be increased by experimental research on those properties. Especially the thermal conductivity and the mass diffusion cofficint in the liquid phase are important to determine the heat and mass transfer in the sorption components. In this research the simulation program is first of all meant as a simulation model for the experimental test plant with the CHME components. Those results will be discussed in combination with the simulation results in Chapter 7.

55

CHAPTER 6.

RESULTS OF THE EXPERIMENTS

6.1 Introduction Overall three series of experiments were executed. The first serie took place between April and August 1985, with the main results reported by van der Welle [W5]. The second series took place between February and May 1986. A limited part of the results of that series was presented by van der Velde [V3] because they were only used to have a preliminary comparison with the result of the simulation program as presented in Chapter 5. Both testing series had to be preiiminarily stopped because one of the aluminium compact heat and mass exchangers, the evaporator, showed to have a leakage. This could be observed during the experiments by a slowly increasing low pressure. After the second series it showed that also the aluminium condenser had a leakage. But much smaller and with less influence on the results because of the relatively high pressure. This could not be observed by an increasing high pressure, but by testing it under super-atmospheric pressure (1.1 bar abs.) in a water bath. Also during both these series problems appeared with the distribution trays in the absorber and the generator. Here the liquid flows were hindered and the distribution disturbed by pollution of the liquid caused by corrosion particles that blocked the perforations. Cleaning the trays and reducing the exposure to air solved this problem. A third series took place between March and July 1987 with two new components, a condenser and an evaporator, both out of stainless steel AISI 316. It is planned that this was also the last session of experiments with the working pair CH30H - LiBr / ZnBr 2 , the alcohol - salt mixture. During this series no problems with corrosion, distribution or leakage appeared so it was decided to use mainly the results of this series for further interpretation in this report. The main input parameters are shown in Table 6.1. With the basic values of the second column as a starting point, the parameters were varied over the range mentioned in the first column. Table 6.1 : Working range a nd basic value of the input parameters input parameter generator heating temp. evaporator heating temp. absorber and condenser cooling temperature mixture mass flow weight fraction range 100 - 125 C -10 - +15 C 30 - 60 C 25 - 100 g/s 0.30 - 0.36 basic value 125 C 10 C 40 C 100 g/s 0.33

It should be mentioned that the evaporator heating temperature was raised with 5 K to about 15'C and kept at that level most of the time to avoid incomplete evaporation. The mass flow of the heating medium was kept on its maximum to have the best heat transfer.

57

Because of the maximum heating temperature of 125C and the bad heat transfer in the generator, that temperature is kept on 125C and the mass flow of the heating medium adjusted at its maximum. Later more on this when discussing the results per component. In Figure 6.1 a picture is shown of the whole test plant, the AHP with the CHME components.

Figure 6.1

Picture of the AHP test plant with the CHME components

58

a. total heat production Q

Qm o Tm = 30 C

&

Qm t Tm = 40 C

Qm Tm = 50 C

j j t a l heat production Qm (kW) Te = 5C. Tg = 125C

30

rich mass flow N* (g/s)

b. COP
1.70
140

coeff. of performance COP.

Te = 5 C. Tg = 125 C

LM MO
130
120

1.10

00

70

rich mass flow Kt (g/s)

c. circulation ratio f

circulation ratio f (kg/kg) Te =

80

70

90

rich mass flow f v f r (g/s)

Figure 6.2

The influence of the rich mixture mass flow M on the performance of the AHP for three coolihg water temperatures

59

6.2 Experimental results of the AHP and its components 6.2.1 The absorption heat pump Although in this research the focus is on the components of the AHP, first of all as an introduction some overall results of the AHP test plant. Figure 6.2 a,b and c are giving an overview of the performance of the AHP as a whole, that is the cofficint of performance COP, the total heat production Q and the circulation ratio f. Here the basic values of the input parameiers were used, only the mixture mass flow was varied over its total range and the cooling water temperature was resp. 30, 40 and 50C. The COP seems only slightly depending on the mass flow (Figure 6.2 b ) . For the cooling water temperatures 30, 40 and 50C, the COP is around resp. 1.60, 1.45 and 1.10. As one can see in Figure 6.2 c, the higher the cooling water temperature T , the higher the circulation ratio f. Next to the COP and the total heat production Q of the AHP, this parameter, determined by the process itself, is important for the design of the whole AHP, while the rich mixture mass flow is one of the controlling input parameters and determines the capacity range of the AHP as can be seen in Figure 6.2 a. An overall conclusion from these experiments is that only at low cooling water temperatures, that is below 40-45 C, the AHP gave an acceptable heat production Q , cofficint of performance COP and circulation ratio f. Furthermore it was observed that cooling water temperature below 35"C at a generator heating temperature T of 125C gave a solvent mass flow that was too large for complete evaporatton. Both limited the experiments by adjusting the cooling water temperature T almost always to 40C. 6.2.2 The components 6.2.2.1 The generator For the testing of the components itself, the generator was not that important, especially because the generator itself was not of the CHME type, but a conventional shell and tube exchanger (SAT). Because of the instability of the methanol in the mixture above 120C, the thermal oil entered the generator with a maximum temperature of 125C and was most of the time kept constant at that temperature. The overall heat transfer cofficint K showed to be almost constant and around 45 W/(m 2 .K), with a mean logaritnmic temperature difference AT determined by the cooling water temperature T and the mixture mass ffow. Those were results of the "polluted" generator (distribution tray) in the second session. Some experiments after the cleaning of the distribution tray showed a twice as high overall heat transfer cofficint but an also two times lower temperature difference leading to an almost unchanged heat flow. The same holds for the third session with an overall heat transfer cofficint between 60 and 100 W/(m 2 .K) (Figure 6.3) at a temp. difference around 13K. Only at very low flow rates Kg decreased tremendously, at a higher AT . Assuming that - if the mass flow and the temperature of the oil is kept unchanged - the heat transfer cofficint on the heating side is constant, it means that the heat transfer cofficint on the mixture side is increasing with the mixture mass flow.

60

Kg (W/m2K) Qg (x 10)

Qg kW)

Xg /

K .

Figure 6.3 Influence of the rich mixture mass flow M on the generator performance

rich rrtxoxe maas flow M

Ig/sl

6.2.2.2 The condenser The vapour coming from the generator is entering the condenser in a superheated state (above the condensation temperature) and is leaving the condenser as a liquid at the condensation temperature or in a subcooled state. The condensation takes place at a constant temperature (constant pressure). The evolved heat flow Q is removed by the cooiing water in the secondary circuit. With this one can divide the condenser in three sections as shown in Figure 6.4: for cooiing down the superheated vapour (a), for condensation of the vapour (b) and for subcooling of the liquid (c).
1 1 1
l
b

solvent

1 l

Figure

6.4
__l cooiing water

1 i

Temperatures in the condenser as a function of the length

Lc

To be correct, for each section an overall heat transfer cofficint K should be calculated, but the corresponding transfer surface areas are unknown. As one can see the liquid can even leave the condenser in a sub-cooled state as could be observed during the experiments. It can cool down close to the water entering temperature, but even then the contribution to the heat flow Q is far less than 1 %. So this can be neglected. Iedema [II] has calculated from the experiments with the former test plant that the contribution by the cooiing down of the superheated vapour to the

6 1

k (kW/m2K) / delta T / Maolv

delta T (K)

Msolv (g/s)

3i$

ZJO

uo -

aso -

80

75

100

rich mass flow Mr (g/s)

k (kW/m2K) / delta T / Msolv

delta T < K )

Msolv (O/s)

e
2J0

1.50

^ ~ \ ^ ^

0-30

.....

50

100

25

cooling water mass flow M c w {% of max.)

Figure 6.5

Influence of the rich mixture mass flow M (a) and the cooling water mass flow M (b) r cw on the condenser performance

62

total heat fiow Q was about 7%, with a low heat transfer cofficint of about 25 W/(m 2 .K) c on that vapour side. It was also calculated that for this 10-20 % of the total transfer surface area was used. From the experiments with this type of condenser - the former was a doubl tube type - it was calculated that the contribution to the heat fiow Q was only 3 % and that, assuming a heat transfer cofficint of 25 W/(m 2 .K), 5 10 % of the transfer surface area was used. The simulation program however showed that with the same assumption the cooling down of the superheated vapour uses around 20% of the transfer surface area, contributing only around 10% to the heat fiow Q . If one defines the mean temperature difference as the difference between the condensation temperature and the average cooling water temperature - in order to calculate an overall heat transfer cofficint K of the condensation section - this will be around 3 % too high. To avoid this, the simplification is made that the overall heat transfer cofficint K of the condenser is based on the above mentioned temperature difference, the total heat fiow Q and the total transfer surface area A . This heat transfer cofficint K from the experiments with the aluminium component will first be discussefl shortly. It showed that in the experiments of the second session for cooling water temperatures from 30 to 50C K increased from 500 to 1500 W/(m 2 .K) with an decreasing temperature difference from = 3.0 to 1.0. The influence of the mixture/sol vent mass fiow seemed to be negligible. After the cleaning of the distribution tray of the generator a few more experiments could be done, showing that at a cooling water temperature of 40C K = 200 - 300 W/(m 2 .K) with AT 5 - 7 K. Now with a better heat transfer in the generator the entering vapour was considerably hotter, causing a higher pressure with a higher condensation temperature. The cooling down of the superheated vapour was about the same. The overall heat transfer cofficint of the stainless steel component in the third session showed to be far better as shown in Figure 6.5a and b. Overall it shows that the overall heat transfer cofficint K is strongly influenced by the cooling water mass fiow (Figure 6.5b) and by the sol vent mass fiow (Figure 6.5a). So one may conclude that both the heat transfer cofficint on the cooling water side and on the condensation side are of comparable influence on the overall heat transfer cofficint K . In most condensers it is fully determined by the heat transfer cofficint on the cooling side. 6.2.2.3 The evaporator The liquid solvent coming from the condenser is reduced in pressure by an expansion valve. Almost always a part of the liquid flashes, extracting the heat required from the liquid itself until an equilibrium is reached between the vapour and the liquid at that pressure. Then the liquid starts to fiow over the transfer surface area at almost the evaporation temperature. That temperature is an important output parameter. Under domestic heating conditions this component is supposed to attract heat from the ambient air, so in fact the ambient air temperature determines the evaporation temperature. That heat fiow input Qe is free, and about 30-40 % of the total heat fiow input of the AHP. From the experiments can be deduced what the influence of the different input parameters is on the evaporation temperature. First the experiments with the aluminium evaporator. 63

With the heating/cooling temperatures kept constant, the mixture mass flow did not seem to influence the evaporation temperature T (T-,), it stayed on a level depending on the cooling water temperature T . cnanglng the evaporator heating temperature T also did not influence the evaporation temperature. Also the sol vent mass flow M and the heat flow input 0 , remained constant. This meant that the mean temperature difference A T , while lowering the heating temperature T , was decreasing, leading to an increase of the overall heat transfer cofficint K . That was with the assumption that the whole transfer surface area was usea. During the experiments at T ~ 0 C one could observe that not all the solvent was evaporated and that drops started to fal 1 on the bottom of the evaporator housing. One may conclude that at that point all the available transfer surface area was used, corresponding to an overall heat transfer cofficint K e 180 W/(m2.K) and mean temp. diff. AT 6K. 2 So in this case that T. > 0 C and assuming that K. i * 180 W/(m .K), the total transfer surface area was not fully wetted. Also other experiments confirmed that when all the transfer surface area was used K = 180 - 200 W/(m 2 .K), while AT depended on the other parameters. This was in fact contradictory to the expectation that the evaporation temperature was determined by the heating temperature as in the case of compression heat pumps. Here it seemed that the mixture conditions (T and w) at the end of the absorber determined the evaporation temperature, giving a driving pressure difference AP between the absorber (P(T,w)) and the evaporator (P-,) as shown in Figure 6.6.

w = 1 . 0

w = wr w = wa

Figure 6.6 A log P - l/T diagram with the evaporation pressure and the absorption process

Tev

Tr

Ta

,1/T

The temperature of the vapour leaving the evaporator was measured halfway the connecting tube between the evaporator and the absorber. It could be observed that the vapour at that place was always more then 5K superheated. This superheating can be evolved from the remaining (dry) transfer surface area but can also be caused by a heat flow leaking in through the insulation (T . = 20C) or by a conduction heat flow from the absorber. This is not unrealistic because when the total transfer surface area seemed to be used for evaporation the vapour was still superheated and sometimes the vapour temperature was even higher than the heating temperature T . The heat flow for this superheating can be neglected. With a specific heat c ~ 1450 J/(kg.K) and a solvent mass flow M ~ 2 g/s, the heat flow required to superheat the vapour IK, must be about 3 W.

64

So just a very smal! error was made when assuming that all the heat abstracted from the heating system was used for evaporation. The second session was stopped because of a supposed leakage in the evaporator and a bad performance of the generator. A slow but constant rise of the low pressure was a strong indication for the first. A test in a water bath proved this. The experiments with the stainless steel evaporator in the third session were also not satisfactory. The new type of plate fin was triangular, not offset and with the evaporating film flowing uninterrupted in the open direction. Although the total wetted width was larger, leading to a smaller film thickness and a better heat transfer, the absence of interruptions led to such high flow velocities that most of the time at the end of the evaporator the liquid was not completely evaporated. Precaution had to be taken to meet this problem. First the solvent mass flow had to be limited and/or the heating temperature to be increased and secondly a construction had to be made under the evaporator to transport the not evaporated liquid to the liquid at the bottom of the absorber. This to limit the disturbance of the equilibrium of state in the test plant as less as possible and to obtain it as soon as possible. If there is no such construction the liquid will collect at the bottom of the evaporator housing and must be removed in order to obtain the (thermodynamic) equilibrium. So only a indication is given of the performance of this evaporator. The total transfer surface area seemed to be used at an overall heat transfer cofficint K * 80 W/m2.K and a mean temp. difference AT * 16K. 6.2.2.4 The absorber An important, if not the most important, component of the AHP is the absorber where heat and mass transfer take place simultaneously. The amount of heat and mass transfer can be written as follows: for mass transfer with diffusion and drift (convection):
M

or
M

= K

m *

'

( w

i V

'A

+ w

i * Ms - wi>

(47)

with:

s = K m ' p ' ( wi V i = at interface b = in the bulk

*A I

( 1

(47a>

for heat transfer after Equation (14) of Section 2.2 Q a = K a AT a A a (14a)

So with Equation (14a) the overall heat transfer cofficint K can be calculated assuming that the evolved absorption heat at the interface does not influence the exponential character of the mean (bulk) mixture temperature and so the rich mixture temperature leaving the absorber. Unfortunately the weight fractions w. and w, can not be measured to calculate the mass transfer cofficint K . This is a pure technical problem due to the construction of the absorber and the small film thickness (0.2-0.5 mm). In the following figures on the experimental results of the absorber the heating temperatures T and T were resp. 15C and 125C. The cooling temperature T was 40C. " 65

Ka (kW/m2K)

delta T (K)

20

Ka x 10-1 / delta T

ie -

12

28

60

78

100

rlch mlxtLre mass flow Kt (a/s)

Ms (g/s)

Qa (kW)

Ms / Oa

* 3 -

50

rich mlxtire mass flow M- (g/s)

Figure 6.7

The influence of the rich mixture mass flow M on the performance of the absorber

66

The influence of the rich mixture mass flow M on the heat transfer is shown in Figure 6.7a and b. The overall heat transfer cofficint K and the mean log. temperature difference AT are both increasing with an increasing mass flow M (Figure 6.7a). It coula be observed that there was hardly any influence on this while changing the cooling water mass flow M . That means that the overall heat transfer cofficint K was determined by the heat transfer cofficint on the mixture side. Figure 6.7b shows that, with an increasing mass flow, the heat flow Q and the solvent mass flow M are not increasing in an equal way. The contribution of the sensible heat from the mixture flow to the total heat flow ( 1 is increasing. a
+ Tcwjn A Tcw.out (C) o Tmixth Tmlxtout (C )

(C )
33

(C )

Tcwjn / Tcw.out / Tmixtln / Tmixt.out

33 31 49 47 43 43 41 39 37 35

23

30

75

100

rich mixture mass flow Mr (g/s)

Figure 6.8

Temperatures of the mixture and the cooling water in and out the absorber

This can also be seen in Figure 6.8 were the temperatures of the mixture and the cooling water at the entrance and the exit are shown. The temperature difference at the bottom of the absorber shows to be constant, for a cooling water temperature T that is around 10K. Although the cooling water mass flow does not influence the performance, its temperature does. This already could be observed in Figure 6.2. The following Figure 6.9 and 6.10 show the influence of that temperature and the performance of the absorber, that is here the contribution to the total heat production Q and the amount of absorbed solvent. An increase of the cooling water temperature T and/or of the mixture mass flow (Figure 6.9) give for mixture mass flows over = 40 g/s a growing contribution of Q to Q . Around 40 g/s the contribution seems to independent of the temperature, below 40 g/s the contribution even decreases.

67

Qa/Qm Tm = 3 0 C

Qa/Qm Tm = 4 0 C

Qa/Qm Tm = 5 0 C

70 68 66

Qa/Qm x 100% Te = 15 C.
Tg

= 125 C

ySO

62

60

58

/ S

y^

56

-.^

7^^

34

+
30 70

52 50

*
"O
30

/
|
90

rich mass flow Mr (g/s)

Figure 6.9

The contribution of the absorber heat flow Q. to the total heat production Q for three m cooling water temperatures

+ Ms Tm o 30 C

Ma 0 Tm = 40 C

o Ma e Tm = 50 C

solvent mass flow Ms (g/s)

3.30

Te = 15C. Tg = 125C
t-

2J0

1J0

n
C.30

^___

r\

30

ao

70

90

rich mass flow Mr (g/s)

Figure 6.10

The solvent mass flow for three cooling water temperatures

68

This is caused by the fact that with a decreasing mixture mass flow the mixture entrance temperature also decreases and even the mixture tends to heat up. In that case not all the absorption heat is transferred to the cooling water. Figure 6.10 shows the influence on the solvent mass flow. There seems to be an almost linear dependence between both the mass flows. From Figure 6.2a and 6.10 can be deduced that at cooling water temperature of 50"C or higher the overall performance becomes very poor. The exchange of sensible heat in the absorber and the generator tends to overrule the exchange of absorption heat. So for an acceptable performance and keeping in mind the maximum mixture temperature of 125C, the cooling water temperature should not exceed =45C for the working pair CH 3 0H - LiBr/ZnBr2. In Figure 6.11 the weight fraction of the poor and the rich mixture are
shown.
+ Wp Tm =
A

Wr o 30 C

Wp Tm =

Wr 40 C

Wr Tm

Wr 50C

37

weioht fractiona Wpoor / Wrlch

3B

35

34 39

3 2

1"
30 00 70 90

31 -

30

rich mixture ma3s flow N>r (g/a)

Figure 6.11 The weight fraction of the mixture mass flows M and M The dependence of the mass flows and the weight fraction is already shown before:
w

leading to

p * M p + M s = w r 'M r Mp M s = M M s = M r (wr - w p ) / ( 1 - w p )

(11) (10) (48)

The factor (w - w ) is known as the weight fraction change Aw. From Figure 6.10 and 6.11 one can see that M t : M T, w t and w i, so Aw l and (1 - w ) T. ' r s p r The way bothpweight fractions change over the mixture mass flow range is an indication for the mass of the rich and poor mixture in the AHP.

69

In this range the factor 1/(1 - w ) does not change more than 5% so the solvent mass flow is mainly determined by the adjusted mixture mass flow M and by the weight fraction change Aw. The last is in fact determined by the equilibrium of state in the AHP, that is by the heating and cooling temperatures. With an increasing mixture mass flow the solvent mass flow increases but then the weight fraction change Aw decreases. More discussion on the heat and mass transfer in the absorber will take place in Chapter 7. 6.2.2.5 The mixture/mixture heat exchanger The function of this exchanger is crucial to a good functioning of the AHP as a whole. It task is to exchange heat between the rich mixture flow going to the generator and the poor mixture flow going to the absorber. It is favourable that the rich mixture enters the generator as warm as possible so most of the heat flow Q can be used for the desorption of the solvent from the mixture. Furthermore to take care that the poor mixture enters the absorber as cold as possible to reduce the flashing of solvent when entering the absorber.
+ Khe (WAn2KI

T = 15 C / Tm = O C / Tg = 126 C

a. overall heat transfer cofficint

rich mUxo mna* flow N* tfl/sl

oeita T

Ta = 15 C / Tm 40 C I Tg = 125 C

b. heat flow and temp. difference Figure 6.12

rich mlxOre nrnaa flow fc* Is/s!

The performance of the mixture-mixture heat exchanger

70

In these two way it can limit the exchange of sensible heat in the absorber and the generator, reducing the required transfer surface areas of these components. The heat exchanger in the earlier test plant had a very bad performance. The overall heat transfer cofficint K. was very low and strongly depending on the mixture mass flow (M ~ 30 g/s with K, = 10 W/(m2.K) and M * 100 g/s with K. = 40 W/(m 2 .K))/The experiments WTth the plate fin tpe heat exchanger s"how that the overall heat transfer cofficint K. 140 W/(m2.K) for mass flows over 50 g/s, almost independent of that mass flow and even of any other input parameter (Figure 6.12a). Furthermore Figure 6.12b shows that the mixture mass flow is determining the heat flow Q, . So with a constant factor (K. . A, ) and an almost constant specific heat c of the mixture, the mixture mass flow determines the mean logarithmic temperature difference AT-, over the exchanger and the cooling down and heating up of resp. the poor and rich mixture.

6.3 Conclusions from the experiments To summarize the results, for the absorption heat pump as a whole was found that the cofficint of performance COP was strongly depending on the cooling water temperature T , there was only a minor dependence on the mixture mass flow M . For cooling water temperatures of 30, 40 and 50 C the COP was around resp. 1.60, 1.45 and 1.10. The corresponding total heat production Q was in the range of resp. 1.5 - 2.5 kW, 4.0 - 6.5 kW and 7.0 10.0 kW over a mass flow range of 25 -100 g/s. Concerning the components, for the shell and tube generator the overall heat transfer cofficint K was only between 25 - 100 W/(m2.K) depending on the mixture mass flows. The heating temperature and mass flow were not changed but kept at their maximum. For the stainless steel CHME condenser an overall heat transfer cofficint K was found in the range of 300 - 2500 W/(m2.K) , with a comparable contribution of the heat transfer cofficint on both sides. At the basic input parameters K was around 1500 W/(m 2 .K), reducing the sol vent mass it increased up to 2500 W/(m 2 .K), reducing the cooling water mass flow it decreased strongly down to 300 W/m 2 .K). For the equilibrium of state in the whole AHP it was necessary to have complete evaporation in the evaporator. Due to the not optimal construction of the evaporator, less for the aluminium one, more for the stainless steel one, the heating temperature had to be somewhat high, up to 15"C. In the case of complete evaporation the overall heat transfer cofficint K was around resp. 180 and 80 W/(m 2 .K). The overall heat transfer cofficint K of the absorber was in the rang of 100 - 200 W/(m 2 .K), strongly depending on and determined by the mixture mass flow. Changing the cooling temperature or mass flow had a minor influence on the overall heat transfer cofficint. For the mixture heat exchanger the overall heat transfer cofficint K, was about 140 W/(m 2 .K), independent of any input parameter. So far a summary of the experimental results with the AHP test plant and its components. A further interpretation of and discussion on these results can be found in Chapter 7.

71

11

CHAPTER 7. INTERPRETATION OF AND DISCUSSION ON THE EXPERIMENTAL RESULTS 7.1 Introduction In this chapter a further interpretation of and discussion on the experimental results of the absorption heat pump (AHP) test plant wil 1 take place. This is necessary to be able to give a more explicit judgement on the possibility of the application of this type of compact heat and mass exchanger (CHME), that is the offset plate fin type, in a sorption system. First a more qualitative interpretation will be given based on the observations during and the experience with experimental tests. So a more or less subjective impression. This will be foliowed by a more objective analysis, that is a quantitative interpretation of the experimental results. This will be done by an attempt to translate this to heat and/or mass transfer correlations. Here the focus will be on the absorber, the most important component, and on the mixture mixture heat exchanger, the most simple component (heat exchange with no change of phase). Furthermore a discussion will take place what the consequences will be for the simulation program, given the experimental results. The experimental results will also be used in a more detailed simulation model to verify to what extent the film theory for the heat transfer and the penetration theory for the mass transfer adopted in the simulation program are correct for the simultaneous processes of heat and mass transfer in the CHME offset plate fin absorber. In the next part a comparison will be made between the shell and tube (SAT) generator and the CHME absorber, both from the test plant. Also the results from that generator functioning as the absorber in a former test plant will be incorporated. Also the area density will be discussed again. Finally the most important conclusions will be summarized. For all the above mentioned it holds that the interpretation and discussion are limited to one geometrical configuration per component, so the chosen geometry like length, width, number of passages and type of (offset) plate fin. Because of the costs no component can be exchanged for testing the geometrical influences. This stresses the importance of an accurate and reliable simulation program which does offer these possibilities. 7.2 A qualitative interpretation As mentioned above this interpretation is based on the observations during the experiments and on the thereby acquired knowledge and experience. Important for instance was the liquid and vapour distribution in the components. In the mixture-mixture heat exchanger and on the heating or cooling side in the other components a good distribution was assured by the construction (equal flow friction over the passages) and the fact that there was a plug flow filling the total available space between the transfer surface areas. The same holds in fact for the vapour flow entering the condenser and the absorber. But for the distribution of the liquid flow entering at the top of the evaporator, the absorber and the generator a special tray had to be constructed. This to take care of a homogeneous film flow over/through the plate fins for an optimal resp. evaporation, absorption and generation process.

73

These distribution trays are described in Section 3.5.5. The distribution of the liquid flow in the absorber could be observed by a couple of looking glasses (Figure 3.4) at the entrance and at the exit, but not on the transfer surface in the inside of the component. With mass flows over 30 g/s the distribution at the entrance seemed to be good, although its was not in the form of drops but in jets. The assumption that the plate fin takes care of improving the distribution of the liquid flow over/through itself is realistic. At the exit it could also be observed that the distribution over and in the passages was good. In the evaporator the distribution could be observed very well at the entrance because of the glass housing on the top. The distribution over the passages itself was good, but over the length of the passages themselves rather poor at low mass flow rates. The tray was modified by reducing the pressure drop over the tray what met the problem sufficiently. The stainless steel component even had a distribution system in itself, over the first 50 mm the plate fin was positioned perpendicular so the liquid had to flow through the holes in the fins, giving a good distribution in the passages. The distribution in the generator seemed to be good, based on observation through looking glasses on the level of the tubes. In a quantitative way nothing can be said on this, only that a good distribution of the liquid over the total available wetted length is important for the heat and mass transfer and that such a system is from a constructional point of view difficult to design. Further it showed that where there was a vapour flow in the component, that vapour must have enough space to reach or leave the transfer surface area. Especially in the stainless steel evaporator the plate spacing and the type of plate fin were not favourable. This resulted in incomplete evaporation. The vapour velocity between the plates was too high and pushed the liquid downwards. This effect was intensified because the liquid did not meet any turbulators. So for a good contact between the vapour and the mixture or the transfer surface area the vapour velocity must be low. This and the turbulators enlarge the contact time and the transferred amount of heat and mass. A disadvantage in this was the large volume flow of the solvent in the vapour phase due to the sub-atmospherical pressures. Using the law of an ideal gas to calculate the volume flow and velocity of the vapour this problem will occur most strongly in the evaporator and the absorber. Looking at their construction the evaporator will be (and showed to be) the bottle neck.

7.3 A quantitative interpretation 7.3.1 Introduction An attempt will be made to translate the experimental results to heat transfer correlations. The focus will be on the absorber, the most important component, and on the mixture-mixture heat exchanger, the most simple component looking at both their function in the AHP. Although the limiting process, the mass transfer in the absorber is left out, because the mass transfer cofficint K can not be calculated from the experimental results. This is due to the fact that the driving force (see Equation (44)) can not be measured because of the small film thickness and the compact construction. So for practical reasons this was impossible.

74

7.3.2 The mixture-mixture heat exchanger This exchanger is in fact the most simple component of the AHP, as well as from the point of view of the transfer process (only heat transfer, no change of phase) and of the construction. In the first aspect the possible preliminary desorption of a small amount of the solvent from the rich mixture at the end of the exchanger wi11 be neglected. To come to a more qualitative interpretation for this component a computer program was developed to calculate a heat transfer correlation based on those experimental data. For the overall heat transfer cofficint K one can write
K

he = l / (Vp + V*w + VV

(22)

Given the almost equal temperature change, temperature, mass flow, weight fraction and the plate-fin geometry on both sides, the assumption that the heat transfer cofficint on both side are almost equal is realistic. For the most extreme operating conditions during the experiments a rough calculation leads to the conclusion the deviation will always be under 5%. These conditions are by the way from a practical point of view unrealistic. So this leads to <y= < y a and so a = 2 . (1/(1/K - d ^ A J ) (22a)

Three correlations were applied to fit the experimental data. The first correlation was the well-known Nusselt correlation : Nu = a Re b Pr c The other two were based on the Colburn factor j: j = St Pr 2 / 3 with St = Nu / (Re Pr) or j = Nu / (Re Pr 1 / 3 ) (17) (17a) (19)

In most case this Colburn factor j is presented as a function of the Reynolds number Re: j = F(Re) (49)

So here the Prandtl number has a fixed power. The first correlation of this type will be the one as used by Wieting [W4] to present his experimental data with this plate fin types: j = a Re b or Nu = a . R e b + 1 . P r 1 / 3 (23a)

The second correlation of this type was derived from Mochizuki and Yagi [M2] that is the type by which they presented their experimental data: j = a + b/Re or Nu = (a + b/Re) Re Pr 1 / 3 (32a)

These correlations will be called resp. the Wieting and the Mochizuki correlation and compared to the Nusselt correlation.

75

For the computer program the following simplifications and/or assumptions were made: - the thermodynamic and physical properties were based on the average temperature (of the four temperatures) and the average weight fraction (rich and poor), - the characteristic length, the hydraulic diameter d. was the plate spacing, - the presence of the plate fin between the plates was neglected. The computer program calculates the best fit correlation coefficients (a, b and c) by means of the method of least squares (MLS). To indicate the accuracy (goodness of fit) of the correlation two parameter were used. Those are the average deviation e (AD) and the cofficint of determination R2 (COD). The last U al so known as the regression cofficint. Furthermore the maximum deviation e (MD) is important for the reliability of the derived correlation.
e

1 *c,l V i )'

(50) t51 i r w

av-2/Vi "V i / / N R 2 (e,ri),/ui,

with: y . y ', N ' y,,, = = = =

> ()

measured value of y calculated value of y (from correlation) number of samples (sets of experimental data) average value of the measured values = ( 2 y_ . ) / N

The COD should be at least greater than 0.5 to have much confidence in the correlation. If it is 0.5, than half of the variation of y is not explained by the correlation, since it means in fact the proportion of the variation in y. The limits for the average and the maximum deviation are free to set. Table 7.1 shows the results for the three types of heat transfer correlations. The total number of samples was 45. The samples (data sets) are obtained from experiments with the mixture mass flow (25 - 100 g/s) and the cooling water temperature (30/40/50C) as input parameters. I t showed that for all three correlations the MD was +/- 20%. The AD showed to be very close to one another. It showed to be around 10% and given the MD this must be acceptable and satisfactory enough. The COD for the correlations based on the Colburn factor are very high (COD = 1.0 means a perfect explanation of y) and very close to each other. Over the mentioned Reynolds number range the difference between these two correlations is less than 6%. For the Nusselt correlation the COD is very low, too low to have any confidence in this correlation. It also has a somewhat strange Prandtl power of -0.405. In literature one finds normally a power between 0.2 and 0.6.

76

Table 7.1 type Nusselt Wieting

Calculated heat transfer correlation for the heat exchanger correlation Nu 24.51 Re 0 " 0 9 7 . Pr" 0 " 405 j - 1.30 Re" 0 " 8 5 7 Nu=1.30.Re-143.Pr0-333

C0D
0.270 0.865

AD ( % )

9.2
10.4

Mochizuki

j = (0.150 + 1.142 / Re) Nu = (0.150 + 1.142 / Re) Re P r 0 , 3 3 3

0.904

10.2

N = 45, MD = +/- 20% 0.56 < Re < 3.33, 45 < Pr < 62 and 3.69 < Nu < 6.35. (0.47 < j < 2.65)

The same type of plate fin was applied on the cooling water side of the absorber. Applying these correlations to the heat transfer on that side in the absorber, leads to the following heat transfer coefficients as shown in Table 7.2:
Table 7.2 Cooling water side heat transfer cofficint in the absorber heat transfer cofficint W/(m2.K) 4200 - 5300 900 - 950 8500 - 11000

correlation type Nusselt Wieting Mochizuki

110 < Re < 160, 3.6 < Pr < 5.4

The enormous difference is caused by the much higher Reynolds number range, only less reduced by the lower Prandtl number range. It shows the limited validity of the correlations. To conclude for the mixture-mixture heat exchanger both the Wieting and the Mochizuki correlation seem to match the experimental result adequately. Since the Wieting correlation is of a more common form, this one seems the best to be adopted by the simulation program. 7.3.3 The absorber The same calculations as shown in Section 7.3.2 have been done for the experimental data of the absorber, that is to find the Nusselt and Wieting correlation.The main difference with the heat exchanger is that actually two different heat transfer coefficients must be calculated: One on the mixture film flow side and one on the cooling water side. The experiments showed a minor even negligible influence of the cooling water temperature and/or mass flow on the overall heat transfer cofficint K . So with an overall heat transfer cofficint in the range 170 - 320 W/(ffl2.K) the cooling side heat transfer cofficint was set on 1000 W/(m 2 .K). The hydraulic diameter, that is the characteristic length, on the mixture side is the (Nusselt-) film thickness 6 , between 0.29 and 0.48 mm.

77

Using the same data sets (45) as for the heat exchanger, the COD of both the correlations would be extremely low, resp. 0.17 and 0.42 with an high average deviation of resp. 24% and 29%. A tooi to improve this, and also lower the maximum deviation, is to leave out the "extreme" data sets. The question is to what extent, in order to avoid seeking the best result. Taking into account that for this component there are more parameters that can change or be changed, the criterion was chosen that the MD should not exceed +/- 35%. That meant that only 35 of the 45 data sets could be used to calculate the correlations. The results are shown in Table 7.3. For practical reasons the Mochizuki correlation is left out.
Table 7.3 Calculated heat transfer correlations on mixture side in the absorber with the heat transfer cofficint on the cooling side set on 1000 W/(m 2 .K) correlation Nu = 1.68 10" 4 R e 0 ' 4 5 3 . P r 1 ' 6 5 8 j = 8.10 H f 2 R e ' 0 " 7 0 7 Nu = 8.10 IQ' 2 R e 0 " 2 9 3 . P r 0 " 3 3 3 N = 35, MD = +/- 35% and 1.0 < Re < 6.0, 75 < Pr < 120 and 0.3 < Nu < 1.0

type Nusselt Wieting

COD
0.526 0.750

AD (%) 12.5 17.8

With an increasing heat transfer cofficint on the cooling water side the COD and the AD for both the correlations increase. For the Nusselt correlation the maxima are resp. 0.61 and 10%, for the Wieting correlation resp. 0.81 and 15%. Furthermore the Nusselt correlation has a, although over 0.50, too low COD to be reliable and a strange Prandtl number power over 1.0. Using the correlations derived from the heat exchanger experiments (Table 7.1) on the cooling water side (equal plate fin geometry), the calculated Wieting correlations on the mixture side in the absorber are shown in Table 7.4.
Table 7.4 Wieting correlation mixture side for three cooling side heat transfer correlations (from Table 7.1) Wieting correlation Nu - 8.25 10" 2 R e 0 " 2 9 5 . P r 0 " 3 3 3 Nu = 6.85 lO" 2 R e 0 " 2 9 1 . P r 0 ' 3 3 3 Nu = 6.71 lO' 2 R e 0 ' 2 9 2 . P r 0 ' 3 3 3

type Wieting Nusselt Mochizuki

COD
0.747 0.794 0.801

AD (%) 17.9 15.5 15.1

As can be seen for a cooling side heat transfer cofficint over 1000 W/(m z .K), the correlations are almost identical. That holds for all correlations from Table 7.1 as indicated in Table 7.2

78

So also on the mixture side of this component the Wieting correlation showed to be the best fitting. In combination with the Wieting correlation from the heat exchanger for the heat transfer cofficint on the cooling water side, it also takes into account by approximation the influence of that cooling side heat transfer cofficint. 7.3.4 Discussion and comparison The calculations with the MLS methode of the heat exchanger experimental data gave a heat transfer correlation (Wieting) that represents those data very good. Using all data sets (45) the maximum deviation MD was +/- 20%, the average deviation AD around 10% and the cofficint of determination COD was very close to 1.0, 0.87. The same holds for the Mochizuki correlation but its form is less common. The Nusselt correlation (Prandtl number power not fixed on 0.333) gave a strange Prandtl number power (-0.4) and a very low COD (0.27). The result of the calculation with the MLS methode of the absorber experimental data were not that good. First setting the MD at +/- 35%, only 35 of the 45 data sets can be used. Secondly the cooling water side heat transfer cofficint had to be estimated. Based on the experiments this influence was accounted for by setting this cofficint at the value of 1000 W/(m 2 .K). The resulting Nusselt correlation had an also too low COD of 0.53 and an AD of 12.5%. The Wieting correlation had a higher COD (0.75) and a somewhat higher AD (18%). For both components the Wieting correlations can be adopted in the simulation program having an acceptable accuracy. Using the heat exchanger correlations, based on the working pair, on the cooling side in the absorber (water) given the equal plate fin geometry, this proved the limited validity in the Reynolds number range. Now a comparison between the heat transfer coefficients from the experiments and the ones adopted in the simulation program (see Section 5.4.2 and Appendix A) is possible. This is shown in Table 7.5. Table 7.5 Comparison heat transfer coefficients (W/(mz.K)) for the heat exchanger and the absorber between the experiments and the simulation program a adopted in the simulation program * 240 * 600 - 1000 (Nu m 8.0) (Nu = 1.6) a derived from the experiments = 100 - 170 150 - 250

component heat exchanger absorber

It is not surprising that the theoretical values, based on an ideal process, are much higher than the experimental values, although the presence of the finned plate between the plane plates is not accounted for. For the heat exchanger the value 8.0 is based on the geometrical ratio of the plate spacing and the plate width of the passage (here 0.01) and assumes a fully developed velocity and temperature profile [S5] [K3]. The finned plate however disrupts those profiles continuously and a full development will never take place. Furthermore the used hydraulic diameter d. = 2 d (plate spacing) is not correct in that case, but is depending on the plate fin geometry.

79

Another important influence that is not accounted for is the viscosity of the liquid, especially in the case of the alcohol/salt working pair which has a relatively high viscosity. The heat transfer is hindered by a decreasing velocity. Despite these important difference, the actual values of the heat transfer coefficients are not that much different (factor 1.5 - 2.4). For the absorber that factor between theoretical and experimental value is larger ( ~ 2.5 - 6.0). From the film theory the heat transfer cofficint will decrease when the mass flow M (here per passage) increases, due to an increasing film thickness (see Equation (42a) and (43)). The average film temperature also increases, causing a certain decrease of the viscosity, i.e. ~ 25% for a 10K temperature increase. The density changes only slightly ( = 1%). The compensation is only small because this corresponds with a 400% increase of the mass flow ( 25 to 100 g/s). The experiments however showed an increasing heat transfer cofficint when the mass flow increased. This can be seen from the correlations in Table 7.3. Rewriting Equation (42) for the film theory and the Wieting correlation for the experimental results (from Table 7.3 by using Equation (43) and for the Reynolds number Re = p v 6 / r?) : Wieting correlation film theory: a = 0.081 X (M /(L -r?)) 0 " 2 9 3 . P r 0 " 3 3 3 (53) w a = 1.6 \ ((3 . r ? M)/(/>2 g))"' 333 (54)

The assumed (and observed) mass flow and temperature changes correspond with a decrease of the heat transfer cofficint based on the film theory to 70%. For the Wieting correlation this holds to an increase to 150%. In fact the theory and the experiments are giving a contradictious change of the heat transfer cofficint for the influence of the mass flow. Both the film theory and the correlation based on the experiments do not account for the presence of the finned plate between the plane plates. Also here the liquid (film flow) is continuously disrupted. The film theory is based on a stationary heat transfer process with a constant driving force. Although the temperature profile is formed and stable very quickly, the liquid will be mixed very frequently. It seems that the liquid flowing through and over the finned plate is turbulent. This turbulence is determinative for the heat transfer. Another important phenomena that is not accounted for in both cases is the wetting efficiency of the available transfer surface area. This is determined by the quality of the liquid distribution over and in the passages. Both influences can unfortunately only be indicated in a qualitative way, not quantitatively. But the Wieting correlation seems to be suitable enough to be used in the simulation program. For both components experiments with another working pair must be undertaken to verify the validity and accuracy of the correlations. In the next section another attempt will be made to verify the validity of the film theory, although rejected by the correlation from the experiments, by means of a more detailed simulation program, simulating the simultaneous heat and mass transfer in a liquid film flow over/through a finned plate. 80

7.4 Detailed absorber simulation model 7.4.1 Introduction In his research work deal ing with the fundamental heat and mass transfer in a shell and tube absorber (horizontal type) Wassenaar [W7,8] has developed a simulation program to compare different types of absorbers. It simulates the simultaneous heat and mass transfer from a liquid film flow to the transfer surface area (tube, plate, finned plate, cushion plate etc.) while absorption takes place at its interface. This program was adopted by v/d Wel Ie [W5] for this research work, using the finned plate version. This version was transferred into three models: a. the model of the finned plate version itself, b. a plane plate model leaving out the presence of the finned plate between the plane plates and c. a simplified finned plate model based on the plane plate model. The plane plate model was developed for comparison with the finned plate model to estimate the influence of the finned plate. The simplified finned plate model was developed to reduce the computer calculation time. That time will be important for design purposes when all components of the AHP are incorporated. For that the plane plate model was modified by assuming one film flow per passage (in a. one, in b. two) and a transfer surface area that is equal to the finned plate surface area. Furthermore the film flow is ideally mixed where it is supposed to leave a fin and reach the next as in the finned plate model. The simulation programs of the models consist of two linked parts. The first calculates the heat flows in and through the plane and finned plates, the second calculates the transfer and transport of heat and mass in and through the liquid film. In the case of the plate model, the film theory and the penetration theory for resp. the heat and mass transfer are quite correct but lead to a relatively poor heat and mass transfer cofficint, averaged over the total flow length. Both theories (th) will now be adopted in both the finned plate models, but incorporating the correction factors c and c. for resp. the heat and mass transfer for fitting on the experimental values (exp.):
*exp
K

a'

th
K

=c

1.6.

A/6

(55) (56)

m,exp = c k '

m,th = c k *

2 7 (D / ( , . t c ) )

The basic idea behind this is to attempt to take into account the deviations from the theory like the incomplete wetting of the transfer surface area. This is done by adding correction factors to the theoretical values of the heat and mass transfer coefficients that should incorporate all deviations from the theory on an average basis.

8 1

7.4.2 Former researchers For the shell and tube absorber this methode has already been used by v/d Griend [G5], Tuynman [T2], Bakker [B7] and by Iedema [II]. V/d Griend used factors to correct the theoretical transfer coefficients as described above. For the heat and mass transfer the others adopted a correction factor cal led the wetting efficiency n , a mass flow dependent factor: the higher the mass flow, the better the wetting of the transfer surface area. The correction factors c and c k are linked to the transfer coefficients. The wetting efficiency n howevet* is linked to other parameters. Bakker linked it in his simulation model to the mass flow, leading to an "effective" mass flow M : eff " M / % <57> It decreases the heat transfer by increasing the film thickness and increases the mass transfer by decreasing the contact time due to a higher film velocity at the interface. Tuynman and Iedema linked it to the transfer surface area leading to an "effective" transfer surface area A : Aeff - % (58)
M

It effects both the heat and mass transfer processes in a negative way. For the heat transfer they set the correction factor c on 1.0 (so left out) but for the mass transfer they still used the correction factor c. to give an extra compensation leading to an "overall correction" over 1.0. Latter on Iedema based his correction factors on his own experiments by incorporating the influence of the mass flow in both factors by means of the Reynolds number and of the incomplete wetting in the mass transfer correction factor. This last factor is defined as the effective diffusion ratio ( D e f f / D ) 0 - S : r? = a Re2 + b Re + c
w

(59)

c k = ( D e f f / D ) 0 - 5 = d (Re/r?w(Re)) + e (60) Table 7.6 lists the range of the correction factors as found by the above mentioned authors/researchers. The general tendency is that the heat transfer correction factor and wetting efficiency are lower than 1.0 and that the mass transfer correction factor higher than 1.0. The first because the distribution system can never assure a complete wetting of the transfer surface area so c and r? < 1.0 (100 %). The second because the penetration theory assumes a decreasing driving force for the mass transfer while the film is in reality continuously disrupted by dripping from one tube and falling on the next. All found that adopting the wetting efficiency as a function of the Reynolds number ( or the mass flow or the wetting density) gave a good presentation of the experimental results. Both Tuynman and Iedema (Equation (59)) found an average deviation of 20%. The maximum deviation was resp. -25/+25% and -25/+30% (estimations).

82

Table 7.6

Heat and mass transfer correction factors r ? and c. from different authors/researchers
%
C

author/researcher v/d Griend [G5] Tuynman Bakker ledema [T2] [B7] [II] 0.1 0.4

remarks based on experiments based on experiments based on v/d Griend based on Tuynman based on experiments

- 0.3 (c 0 ) - 0.8

1.0 - 6.0 1.6 - 2.8

0.35 - 0.40 0.3 0.2 - 0.8 - 0.8

3.0
2.5 1.0 - 2.5

Tuynman found no indication of a mass flow dependence for the mass transfer correction factor, the cloud of point could only be expressed by indicating the range of that factor. ledema found an expression (Equation (60)} with an average deviation around 20%, but a maximum deviation of 35%. So far the results of former researchers on this concerning the shell and tube absorber. 7.4.3 Results from this research work For the interpretation of the experiments with the finned plate absorber correction factors according to Equation (55) and (56) will be calculated. So all deviations from the theory are to be explained by one correction factor for each the heat and mass transfer. So both factors are for example taking into account the incomplete wetting of the transfer surface area. For each experimental data set (input and output parameters) both correction factors were calculated by means of an iteration program. During the calculations these factors (matching parameters) were modified until the outlet cooling water temperature and the outlet) weight fraction met with a certain accuracy (resp. 0.1 K and 0.0005 kg/kg) their experimental values. Assuming both factors as constant showed a very strong deviation between the experimental values and the average value of it. The heat transfer correction factor is assumed to be depending on the mass flow and the viscosity. A greater mass flow will raise the wetting efficiency as well as a lower viscosity will improve the spreading of the liquid. The same holds for the mass transfer correction factor. A better spreading and wetting leads to a smaller contact time and a higher diffusion cofficint (viscosity influence). Therefore both will be related to the Reynolds number: c a = a Re b c k = c Re d (61) (62)

Using the experimental results and the Equation (55), (56), (61) and (62), the finned plate and the simplified finned plate model gave the following correlations, shown in Table 7.7.

83

Table 7.7 model finned p l a t e

Calculated correction factors f o r the finned plate models correlation c a 0.17 Re0'80 f o r Re = 1 - 6 0.17 - 0.71 0.70 - 1.16 0.12 - 0.46 AD =17 % =30 % =18 % * 5% M D =25% =40% =25% =20%

c k * 0.70 R e 0 ' 2 8 simplified finned plate c . - 0 . 1 2 Re 0 " 7 5 ck=0.98

For both models the heat transfer correction factors are quite close to one another and also in the range indicated by former researchers (see Table 7.6). Furthermore it shows the expected Reynolds number dependence as the experiments indicated. For the mass transfer the correction factors are less close for both models, but still around the same value of 1.0. The finned plate model shows also the expected Reynolds number dependence, but weaker than for the heat transfer. The simplified model gave a constant value very close to 1.0. So the range of this factor is quite different from the results of the former authors and from what could be expected. 7.4.4 Discussion on correction factors Adding correction factors to the heat and mass transfer coefficients based on resp. the film and penetration theory is a way to match the experimental values with the theoretical expected values. The most important influence is the incomplete wetting of the available transfer surface area. But adopting both theoretical coefficients the correction factors are in fact accounting for all deviations from the real transfer processes. The results from the different authors/researchers with the shell and tube absorber cover each other quite well expressing both corrections by means of the wetting efficiency n linked to the transfer surface area and also for the mass transfer by means of an "extra" correction factor c. directly linked to the theoretical value. Both are a function of the Reynolds number (mass flow, wetting density) and influencing each other. For the finned plate absorber using a very detailed simulation program correction factors were obtained from the experiments that were directly linked to the theoretical values according to Equation (55) and (56). So not by means of the wetting efficiency. The idea is that the corrections are accounting for more than only for the effect of the incomplete wetting. Using the finned plate model and its simplified version and adopting a Reynolds number dependence, the results for the heat transfer were in the same range as reported for the shell and tube absorber. For the mass transfer the correction factor seemed to prove the correctness of the penetration theory, giving values close around 1.0.

84

7.5 Comparison the SAT absorber/qenerator and the CHME absorber 7.5.1 Introduction Here a comparison will be made between the SAT generator and the CHME absorber in the test plant. Also the SAT absorber from the former test plant (the present generator) will be compared with. The comparison will deal with the heat (and mass) transfer and the compactness of the SAT and the CHME component. The heat transfer will be expressed by heat flow density Q/A (kW/m 2 ), the compactness by the area density 0 (m 2 /m 3 ). 7.5.2 Compactness (area density) The area density 0 is defined in Equation (16), (16a) and (16b) for both types of exchangers. The main problem is how to account for the finned plate since it is obvious that the extra surface (the surface in the passage) is not for 100% effective in the transfer process. For the SAT absorber/generator as described in Section 3.5.4 the area density can be written as follows (see Figure 7.1a):

a.
Figure 7.1 SAT:

b.
A cross view of the SAT (a) and the CHME (b) exchanger

fi'*

f0 + d.)/Z/ (2 d 2 )

(63)

For this tube bundie configuration the overall area density 0 ~ 120 m 2 /m 3 (calculated the same way as for the overall heat transfer cofficint). On the cooling water side it is 400 m 2 /m 3 , on the mixture side /} = 215 m 2 /m 3 For the CHME absorber the calculation is for the trapezium form of the fin more complicated (see Figure 7.1b). A first comparison can be for the case when the finned plates are not present. In that case it can be calculated from the plate spacing d (m = mixture side, c = cooling side) and the wall thickness d.

w
CHME (plane plate): $ - 1 / (dc/2 + d w + d n /2) (64) For this case fi = 250 m 2 /m 3 on the mixture side and on the cooling water side even 670 m 2 /m 3 . The overall area density is = 155 m 2 /m 3 .

85

So the area density of the plain plate CHME exchanger is ~ 30 % higher than that of the SAT exchanger. It seems that from this point of view even a plane plate exchanger is favourable over the SAT exchanger, but not for the heat (and mass) transfer coefficients: there is no disruption of or turbulence in the film flow in the absence of the finned plate. Taking the finned plate into account the disruption and the turbulence is accounted for as well as for the extra transfer surface. For the mass transfer it is obvious that only that part of the finned plate is used, that is where the absorption takes place with mass transport only through the film. In that case for the mixture side (see Figure 7.1b): CHME (finned plate/mass transfer): $ * ( (d2 + h | ) 0 * 5 + hj) / (d (h 1+ h 2 )) (65)

In that case the area density / ? = 230 m 2 /m 3 on the mixture side. For the heat transfer the situation is more complicated. In this case the heat is adopted at the finned plate surface and transported by it to the wall, the plane plate. In fact the effective transfer zone is the contact area between finned and plane plate, over the length h,. So: CHME (finned plate/heat transfer): fl = hj / (d (h 1+ h 2 )) (66)

In that case / 3 ~ 40 m 2 /m 3 . That is rather low because except for the contact area the remaining surface area is in no way accounted for. The extreme on the other side is to assume that all surface is for 100% contributing to the heat transfer. That is the way Shah [S3] and Kays and London [K3] are calculating the area density because in their case there is an air flow though the finned plate and no film flow over it. In that case CHME (finned plate/heat transfer): P * ( ( d 2 + h 2 ) 0 " 5 + 2 (h ] + h 2 ) / (d (h 1 + h 2 )) (67)

The area density would be = 450 m 2 /m 3 on the mixture side. By the way, on the cooling water side incorporating the finned plate for 100% the area density is ~ 900 m 2 /m 3 . Both values, 40 and 450 m 2 /m 3 , are extreme. Assuming an average value and taking into account the mass transfer area density, an area density of 250 m 2 /m 3 on the mixture side wil! be a good approximation. In that case the overall area density would be = 200 m 2 /m 3 . Although not exactly to be calculated it may be concluded that the CHME finned plate absorber has an area density that is about 2 times that of the SAT absorber/generator. 7.5.3 Heat transfer (heat flow density) For comparison of the heat (and mass) transfer of the CHME and the SAT exchanger, the heat flow density Q/A (kW/m2) was calculated from the experiments. For the CHME absorber this ratio was based on the plate transfer surface.

86

In Figure 7.2 the overall heat transfer cofficint K is shown as a function of the mean logarithmic temperature difference AT for the CHME absorber and the SAT exchanger as absorber and as generator. Also lines of constant heat flow density Q/A are drawn. For a better interpretation an attempt is made to represent the clouds of points by (interrupted) straight lines. As can be seen the overall heat transfer cofficint of the CHME exchanger (140 - 240 W/m z .K) is 2 - 3 times as large as that of the SAT exchanger (20 - 100 W/m 2 .K) and at a much lower temperature difference (resp. 4 - 12 K and 10 -22 K ) . For comparable experiments the heat flow density was 1 - 2 times greater for the CHME exchanger.
240 -

|W/M 'KI )
K
y 200 1
1 1

\
\ i" \ '' o " ABSORBER ABSORBER SAT CHME /i'* ?/ *

GENERATOR SAT

1
.
160 -

\ /

\
> < ^ /A = 1.0 kW/m
N

\
\

*A \\
'

N N

120 -

X\

\
i

' , Q/A * 0,5 kW/M v ^

^-

60-

^ " ^ : >

40

o -1

1 2

16

20 AT (K)

24

Figure 7.2 Heat transfer comparison SAT and CHME exchanger So a smaller temperature difference and a greater heat transfer cofficint and heat flow density for the CHME exchanger. Since the heat and mass transfer processes are linked to and dependent of each other and since the mass transfer is the limiting process, one may conclude also that the CHME exchanger gives a better mass transfer. This is according to the expectation based on the supposed number of mixing spots of 32 for the SAT exchanger (number of tubes per column) and of 100 for the CHME exchanger (number of fins).

87

7.5.4 Discussion and conclusion It is no surprise that concerning as well the compactness (area density) as the heat (and mass) transfer (heat transfer cofficint, temp. difference and heat flow density) the CHME exchanger is favourable over the SAT exchanger as absorber and/or generator in an AHP. The area density is almost 2 times greater, although one should keep in mind to what extent the contribution of the finned plates between the plane plates is accounted for. The overall heat transfer cofficint is 2 - 3 times greater at 1 - 2 times smaller temperature differences leading to a 1 - 2 times greater heat flow density. Apart from the definition of the area density of the CHME exchanger, the overall conclusion must be that the presence of the finned plates (corrugations) between the plane plates adds such an extra turbulence, disruption and mixing to the film flow, that the improvement of the heat transfer factor (K A) is mostly due to an improved heat transfer cofficint on both side of the transfer surface area than due to the extra transfer surface area itself. 7.6 Conclusions In this chapter it was the aim to give a qualitative and an even more quantitative interpretation to the experimental results in comparison to the results based on a more theoretical approach. A qualitative discussion took place dealing with influences of the construction of the components on the transfer processes. They were certainly present and to be interpreted qualitatively to a certain extent, but certainly not quantitatively as a first indication. The most important one was the liquid distribution in the upper section of the absorber, generator and evaporator. The quality of it could only be observed at the entrance and at the exit of the component by means of looking glasses. The only qualification could be that the distribution was acceptable or good enough. A second problem was the incomplete evaporation due to the construction of the evaporator. This could be observed very well and corrected. In fact due to the compact construction of the components it was impossible to observe what took place between the plane plates and on the finned plates. Then a more quantitative interpretation was given by matching the experimental results of the mixture-mixture heat exchanger and the absorber with different types of heat transfer correlations. The correlations contained a relation between the Reynolds number Re, the Prandtl number Pr and the Nusselt number Nu (non-dimensional output parameters): Nu = a Re b Pr c (19)

With the cofficint a, b and c as variable parameters this was called the Nusselt correlation. Also two other correlations were used, both derived from the Colburn factor j (see Equation (17)) with a fixed Prandtl number power c (c = 0.333).

88

They had a different Reynolds number dependence as indicated by Equation (23a) and (32a) and were used by / adopted from resp. Wieting [W4] and Mochizuki [M2]. By means of the mean least squares method the best set of the cofficint a, b and c was calculated for these three correlations. Also the average and maximum deviation (AD and MD) and the cofficint of determination (COD) were calculated. The latter is in the range 0.0 - 1.0 and should be at least over 0.5. For the heat exchanger the AD and the MD were almost equal for all three correlations (resp. 10 and 20 %) but the COD of the Nusselt correlation was far too low (0.27), certainly compared to the other two (0.90). Since the form of the Wieting correlation is more common, this one is to be adopted for further simulation purposes. By the way, using water as the medium, the correlations gave a very large deviation between themselves. This proved the limited validity (Reynolds and Prandtl number) of the correlations. For the absorber first two problems had to be solved. First the heat transfer cofficint on the cooling water side had to be determined. Based on the observed minor influence of the cooling water temperature and mass flow on the overall heat transfer cofficint, the water side cofficint was fixed on 1000 W/m2.K. Secondly using all experimental data sets (45) the COD as well as the AD of both the Nusselt and the Wieting correlation were very bad resp. 0.17 / 24% and 0.42 / 29%. This was met by leaving out the extreme data sets in such a way that the now extremely high MD should not exceed a certain value and that still an acceptable number of data sets could be used. A compromise was found in MD = 35% and N = 35. The AD was now acceptable for both, resp. 12.5% and 17.8%. The COD of the Nusselt correlation was still low (0.53) but that of the Wieting correlation was good (0.75). It should be noticed that for both the heat exchanger and the absorber the Nusselt correlation gave a very strange Prandtl number power of resp. -0.41 and 1.66. Comparison with the theoretical correlations showed that the heat exchanger results (theory and experiment) are comparable in range. Also the very small Reynolds number influence in the experimental correlation is certainly not contradictory to the constant Nusselt number theory. The absorber results are indeed contradictory given the mass flow and viscosity influence in the film theory and in the experimental (Wieting) correlation. Although quantitatively not to be detected the presence of the finned plates must be the main cause of this deviations. To investigate the validity of the film and penetration theory in the absorber a very detailed simulation model of the simultaneous heat and mass transfer in a liquid film flow over/through a finned plate was developed: the finned plate model. The basic idea was to add correction factors c and c. to resp. the transfer coefficients a and K . Former researchers have done tnis by using a wetting efficiency r ? linked to the transfer surface area for both transfer processes ana by an extra correction factor c. for the mass transfer. The theoretical as well as the experimental approach confirmed the common sense idea that the film theory is too optimistic and the penetration theory too pessimistic. So it is expected that c < 1.0 and c. > 1.0.

89

On theoretical grounds in this research work both correction factors were determined as a function of the Reynolds number. This proved to be satisfactory for the heat transfer giving a correction factor in the range 0.2 - 0.7 (Reynolds number 1 - 6) and an AD and MD of resp. 17 and 25 %. For the mass transfer the correction factor was lower than expected, in the range 0.7 - 1.2 with a rather high AD and Md of resp. 30 and 40 %. Next to this a plane plate and a simplified finned plate model were developed. The first for comparing the finned plate influence, the second to reduce the calculation time. For the latter the result was up until now not satisfactory compared to the extended model. Finally the experimental results were also used for comparison of the SAT absorber/generator and the CHME absorber. Both exchangers were compared for their compactness by means of the area density / ? (m2/ro3) and their heat transfer by means of the heat flow density Q/A (W/m 2 ). Although not exactly to be calculated the CHME finned plate exchanger had an overall area density that was about 2 times that of the SAT exchanger. For comparable experiments it was for the heat flow density 1 - 2 times that of the SAT exchanger expressed by a greater overall heat transfer cofficint (factor 2 - 3) and a smaller mean log. temperature difference (factor 1.0 0.5). From both the energy and the exergy point of view this is favourable. So far the discussion and further interpretation of the experimental results from the AHP test plant. The final conclusion concerning the simulation program is that there are in fact now three different simulation programs. The main difference between them are the heat and mass transfer correlations. In the first program they were based on correlations from literature like the film and penetration theory in the absorber and generator subroutine. To reduce the calculation time of the program a simplified simulation program was developed by using fixed heat transfer coefficients in all component subroutines. They were based on the experimental results. Finally in this chapter correlations were derived for the heat transfer in the absorber and the mixture-mixture heat exchanger. Also correction factors for the heat and mass transfer in th absorber, resp. film and penetration theory, were calculated. Both were obtained from the experimental results. Adding this to the first simulation program a third simulation program appears with heat and mass transfer correlations from literature as well as from the experiments. So a next step is made in the development of a more general simulation program for any thermal sorption system. For now the main limitation lies in the fact that the experimental correlations are based on experiments with only the working pair CH 3 0H - LiBr/ZnBr2. A next important step might be a validation and verification by experiments with another working pair. For the ongoing research an obvious choice might be the working pair R123a - DTG as illustrated in Chapter 4. This would to a large extent improve the accuracy and validity of the simulation program. In Chapter 8. the general conclusions of this research work will be given, going back to the aim of it. The results of this chapter will strongly contribute to the main conclusions.

90

CHAPTER 8.

GENERAL CONCLUSIONS

8.1 Retrospective view The main purpose of this research work was to investigate the possible application of compact heat and mass exchangers (CHME) of the plate fin type in thermal sorption systems, in particular in an absorption heat pump (AHP) in a domestic heating system. In the research the focus was limited to investigate the performance of this type of exchangers in the light of the heat and mass transfer and of the compactness. The framework was set by explaining the main principles of the absorption heat pump and indicating its research field (cycle configurations, working pairs, components/heat and mass transfer). The chosen type of plate fin components was illustrated by an overview of the available forms of the fin and the limited application (air) of the plate type compact heat and mass exchanger [K3], [S9]. An AHP test plant was built with components of this plate fin type (offset), that is an absorber, an evaporator, a condenser and a heat exchanger between the absorber and the generator. The generator was of the shell and tube (SAT) type. Parallel to that a simulation model and program was developed for that test plant, but with a flexibility (cycle configuration, working pair, type and geometry of components) to simulate any thermal sorption system. Also a literature survey have been undertaken on working pairs for the AHP, to update the knowledge in this field and to select a new suitable working pair for experimental testing. The most important part of the research work were the experiments with the AHP test plant. Three sessions took place changing the main input parameters over their ranges. Finally those experimental results were intensively discussed, explained and/or interpreted using computer simulation programs and techniques. This to be able to give a more qualitative indication of the performance of this plate fin type CHME and to verify and/or correct the heat and mass transfer correlations (theory) in the simulation program of the AHP.

8.2 Conclusions Most of the conclusions have already been written down in the discussions in or/and at the end of each chapter. So only a short summary of the most important conclusions will be given here. This means on the possible application of the plate fin type exchangers in thermal sorption systems, here in an AHP. As mentioned before two things are important in that perspective: the heat and mass transfer performance and the compactness of the exchangers. Concerning the heat and mass transfer the (overall) heat transfer cofficint of the CHME components is not extremely high compared to other types but the offset plate fin provides (in the absorber) a good and continuous mixing of the liquid and good contact between liquid and vapour. So this surface is very suitable for sorption processes. Specially concerning the mass transfer since that is the limiting transfer process in the simultaneous process of heat and mass transfer.

91

From the experience with the evaporator one may conclude that the offset type with a plate spacing of 8 - 10 mm is in that perspective favourable. That is on the "film" side. On the cooling and heating side and in the mixture heat exchanger where the space between the plates is completely filled with the liquid medium and only heat transfer takes place, smaller plate spacings are favourable, not only for the heat transfer but also for the compactness. Finally it showed that for the absorber the heat flow density of the CHME component is 1 - 2 times that of the SAT component, this with a higher overall heat transfer cofficint. (2 - 3 times) and at smaller temperature differences (1 - 2 times). This holds for the components in this 10 kW AHP. An optimization must take place in respect to the greater driving and the required transfer surface areas. Also must be emphasized the great importance of a good distribution of the liquid over and in the passages in the film exchangers. Concerning the compactness it showed that also here lies a great advantage of the CHME component over the SAT component, based on the experience with the absorber/generator. Although it is difficult to define the correct area density for the CHME component, staying on the safe side one can conclude that it is at least 2 times that of the SAT component. As far as investigated in this research work, the overall conclusion is that from the point of view of as wel! as the heat and mass transfer performance as the compactness this plate fin CHME's are very suitable for application in an AHP for domestic heating. Compared to the SAT exchanger it is at least 2 times better concerning compactness and transfer performance. An optimization will certainly enlarge that factor.

8.3 Considerations The maximum heat production Q of the AHP test plant was 10 kW. Using it for domestic heating that means tnat until an ambient temperature of about -15C under average weather conditions it can provide the required amount of heat. Given the decrease of the COP when the ambient temperature decreases, the AHP must have in that case large transfer areas, resulting in a great volume and very high investment costs. Next to that such low ambient temperatures appear only a limited number of hours over a year. So an obvious solution might be to design the AHP at a lower maximum heat production Q , say at 4 kW, enough for ambient temperatures over * 5C. That is for more than 60% of the year given the seasonal distribution of the ambient temperature based on hours. For the seasonal distribution of the heat consumption/demand one must in fact multiply the difference between T=18C and the ambient temperature with the number of hours ("degree hours") since the lower the ambient temperature the higher the heat demand [B7], [II]. At lower temperatures the heating system of the generator can function as the additional heating system and even at very low temperature (and a very low COP of the AHP) take care of all the required amount of heat. That temperature will be = -7C. Although the COP of that additional heating system will be lower than 1.0 (= 0.9), the COP of the whole system of the AHP and the additional heating will be over a year higher than 1.0. Bakker [B7] calculated that the seasonal COP will be = 1.2 based on an AHP with shell and tube components.

92

Using the CHME components and given the experimental result with these exchangers, this will be even more favourable over the conventional heating systems. This as an example concerning the design. The final design and control strategy must be based on an optimization to a maximum seasonal COP. Apart from the heat and mass transfer performance and the compactness more aspects have to be included to come to a more complete and final conclusion. Also the aspects of production and construction and of the economy must be incorporated. Although outside the direct scope of this research work these aspects are of great importance. It is already mentioned that this plate type exchangers are very suitable for automated mass production techniques. For application in the aircraft and automobile industry they are already for many years produced in that way. Construction materials are mainly aluminium and copper while stainless steel and other corrosion resistant materials are only scarcely used [K3]. This because on one hand the low thermal conductivity of those materials, but also because of the difficult manufacturing and brazing of the finned plates. As far as the working pair allows it, aluminium is favourable over stainless steel (R123a - DTG versus CH 3 0H - LiBr/ZnBr 2 ). The economy is an important aspect, closely linked with the other aspects and in fact the final decisive factor. The investment costs are far higher those that of a conventional heating system and at an unacceptably high payback time, due to the low energy prices at this moment. For a more detailed overview on this one should see at Meynen [M3]. For this aspect the conclusion must be that up until now from an economie perspective a domestic heating system with an AHP is not favourable over a conventional heating system. The justification of this research work in that way is the supposed rise of the gas prices and the possibility of "upscaling". Concerning heating systems with a much higher heat production (over 100 kW [M3]) it is already favourable, as well as to incorporate an AHP in an existing system as to replace the existing system.

8.4 Recommendations Recommendations for further research are first of all in the field of selecting and experimental testing of new working pairs, specially the working pair R123a - DTG offers possibilities. In this chemical stability, toxity and acquisition are important. Not only to find new working pairs that meet the disadvantages of the present one, but also to verify and validate the experimental result with the CH3OH - LiBr/ZnBr2 and the simulation program in order to give the program a higher accuracy and a greater application. With the last is meant the flexibility in the cycle configuration, the working pair and type and geometry of the components. A second field would be to develop a automatic controlling system for not only the AHP test plant but for a small scale AHP in general. For domestic application this is a necessity.

93

A third field might be the design of a prototype and to investigate the possibilities to integrate all the components in one compact block to realize an advanced cycle (matrix construction). The fourth field is already mentioned and not only a technical one: the economy of a domestic heating system with an AHP. A good and thorough study and evaiuation of the economie validity of such a system is inevitable to come to the final decisive conclusion whether or not to produce such a system on an industrial scale.

94

APPENDIX A : HEAT AND MASS TRANSFER CORRELATIONS USED BY THE SIMULATION PROGRAM component evaporator working pair side (solvent/mixture) Nu Nu Re 6 = = = = 0.876 R e 0 - 1 1 a 6 / X p v 6 / rj film thickness [Cl] cooling/heating side Nu Nu Re L = = = = 0.664 a L p v plate Re 0 " 5 - Pr 0 " 3 3 /X L /7 ? length [VI]

condenser

superheated vapour: Nu Nu Re L = = = = 0.664 R e 0 - 5 . P r 0 , 3 3 o L /X p v L / T) corrugation width w [VI]

see evaporator

condensation: a = 4/3 (A 3 . o1- r q) ' 2 5 (4 r) AT L) L = wetted length [VI],[G4] absorber heat transfer: Nu = a 6 / X = 1.6 S = film thickness film theory [II] mass transfer: Sh = 2 / 7 ( T T Fo) Fo = 1 / (Re Sc 6/x) K = 2 y(D/(7r . t )) D m = diffusion coeff. [m/s] t = contact time at interf. penetration theory [II] generator heat transfer: see absorber mass transfer: see absorber Nu = 3.65 + (0.19-X- 8 )/(l.0+0.117.X 0 - 467 ) X = Re Pr d/L L = pipe length d = pipe diameter [VI] Nu = 8.2353 Nu = a d, / X d. = 2x plSte spacing [S5] see evaporator

heat exchanger

97

APPENDIX B CALCULATION SCHEMES OF THE AHP COMPONENTS B.1. Calculation steps (per sectionl subroutines absorber and generator Calculation of a,,
a c/n

K, K

AQ = K AA ( T - T c / h ) Tw = Tc/h ^
+

AQ ( l/ac/h

d w /A w

) / AA

-Tw+1.6 ( T - T w )

w 1 = f(PJi) dM = K m p . ( w. - w ) AA / ( 1 - w ) dw = (1 - w) dM / M dH = ( H v - H ) dM - AQ ) / M dT = ( dH - (dH/dw) . dw ) / c dT c / h = -AQ / ( M c / h c p > c / h ) Calculation new M, w, T, H, and T , . indices: i = interface f = film c/h = cooling/heating no index = average over the mixture film

99

B.2. Calculation steps condenser subroutine Input: T

r

T 5 , Ms, M c w T, = T c + AT = T,

Estimation condensation temperature Calculation ocon, asup) acw tot = M s * (Hl " H3>
T

7 = Q tot / <Mcw * c p,cw ) + Tt

< W "Ms * <H1 " H2> 'sup

VT7
4T K

"sup / <"cw

p>cw'

log, S up ' T l V

" <T3 " T 6'> / "'S T l " T7>/<T3 " V >

+

sup " > / < '/"sup sup '" sup ' * tot
- A H

V\.

"Ao, ' sup '

log,sup '

A H

con

- A

sup

xon
K

tot

^sup
+

cor, > / <"con

V\

/ia,

log,con,a " xon ' '

4T

con

con '
T

log,con,b T 2 " V

" <T3 '

5 /

lo

9 T 2 " T6>/<T3 " V >

abs ((AT, - ATn . ) / AT, ) < e? vv log,con,a log,con,b' ' log,con,a' Calculation new AT

indices:

con sup cw log tot w a

condensation superheated vapor cooling water logarithmic total wall based on heat transfer correlations based on temperatures

""3

re

100

B.3. Calculation steps evaporator subroutine Input: T r T 5 , *

} v

M g l , Mhf

(M = M $ )

Estimation of the outgoing vapour temperature : T, = T, + AT (Tp = T,) Calculation

Q Q

a. 'ev' sup' tot - Mg,3 sup = Mg,2 '

T n

V g,3 " <M1,1 ' H l , l 9,3 - Hg,2>

0.1

g,i>
(Mg,2=Ml,l)

(H

6 " T5 -

su P / <Mhf ' cp,hf>

T

AT

log )S up = T 5 " T3> - <T6 "

K

2 /
+ 1/tt

l0

9 T 5 " T 3>/( T 6 "

su P = ! / < Wm
A

V\

hf

sup " ^sup ' ^ sup ev

= A

log,sup '

tot " Asup

^ev ~ ^tot " ^sup T

7'V
= Q

Vr / ("hf
+

W
V%r
T

Kev 1 / (/v, V \
4T

log,ev,a

ev '

( K

ev " Aev '

4T

log,ev,b ' V V
4 T log,ev,a "
4T

" <T7 " T l / '9 T6 " T2>/<T7 "

4T

abs

log,ev,b> '

log,ev,a> <

Calculation new AT

ind ev hf log g

ces: : evaporation/evaporator : heating f l u i d : logarithmic : gas (vapour)

w a b

: liquid : wall
: based on heat t r a n s f e r correlations : based on temperatures

101

B.4. Calculation steps heat exchanger subroutine Input: T

p

T 5 , Mr, Mp

(wp = w, = w 3 , w. . f= w, = w 7 )

Estimation of the outgoing temperatures: T-, = Tg + AT, T, = T, AT = T, - T, 3 Q = M AlT w p p p , AT,, = T 7 - T c 'r u c p,p (Qp = Qr)

AT

AT

r = % f <Mr ' 'p.r 1 T 7 = T 5 + AT r

Calculation a ( a * a ) K = 1 / ( l/a + dw/Aw)

AT

log

= (

<T1 " T 7 ) '

(T

3 " Tl} /

l0

9((Tl " V

/<(T3 " V

Q = K A AT l n abs ((Q - Q p )/ Qp) < e ? Calculation new AT

T1

T7

i
poor nnxtire rich mixture

indices

p r w log:

: : : :

poor rich wall logaritmic

T3

TE

102

REFERENCES Al Alefeld G. Research activities on advanced heat pumps at Techn.Univ. Munich Proceedings "International Workshop on Research activities on advanved heat pumps", Technical University Graz, 1986 Ahlby L. and Hodgett D.L. Compression/absorption systems-simulation of two cycles for different application Proceedings of the "17th International Congress of Refrigeration", Vienna, 1987 AKG Autokhler - Gesellschaft m.b.H, Hofgeismar, West Germany Product information bulletin "WaYmeaustauscher" Allied Chemical USA private communication with K.P Murphy Bergmans J.(ed.) Heat Pump Fundamentals NATO Advanced Study Institute Serie E: Applied Science, Sythoff Press, The Hague, 1983 Bokelmann H. und Renz M. Thermophysikal ische Eigenschaften von Trifluorethanol Stoffsystemen fr AbsorptionswaYmepumpen Ki Klima-Kaite-Heizung, 10/1983, Teil 4.3, Seite 499-502 Bokelmann H. Auswahl, Messung thermophysikalischer Eigenschaften und Beurteilung der Eignung von Niederdruck-Stoffsystemen fr AbsorptionswaYmepumpen Forschungsbericht des DKKV, Nr. 12, 1984 Berghmans J. and Vandevijvere D. Development of an AHP for industrial application Proceedings International Seminar "Conservation in industry" Dsseldorf, 1985 Bokelmann H. Industrial heat recovery with heat transformers - Practical application and development of advanced systems Proceedings IEA Heat Pump Conference, Orlando (USA), 1987 Baehr H.D. The COP of absorption and resorption heat pumps with ammonia - water as working fluid International Journal of Refrigeration, Vol. 4, Nr. , 1981 Bakker R.W. Processimulatie van een absorptiewarmtepomp Internal Report ST-199, Laboratory for Refrigeration Engineering, Technical University Delft, 1983

A2

A3

A4

BI

B2

B3

B4

B5

B6

B7

103

B8

Briggs D.C. and London A.L. Heat transfer and flow friction characteristics of five offset rectangular and six plain triangular plate-fin heat transfer surfaces Trans. Intern. Heat Transfer Conf.,ASME, paper 14, pp 122, 1961 Becker H. Elektro-chemische bepaling van de diffusie- en stofoverdrachtscoeff. bij pijpstroming van lithiumbromide - methanol Internal Report S-729, Laboratory for Refrigeration Engineering, Technical University Delft, 1982 Becker H. A literature survey on working pairs for sorption systems Internal Report K-145, Laboratory for Refrigeration Engineering, Technical University Delft, 1988 Bokelmann H. und Ehmke H.J. Arbeitsstoffsysteme fr eine Sorptionswarmepumpe Report of the EEC-sponsered research program EE-A4-031-DCB, 1985 Chun K.R. and Seban R.A. Heat transfer to evaporating liquid films Transactions of ASME, Journal of Heat Transfer, Vol.93, Series C, pp 391-396, 1971 Cussler E.L. Diffussion, mass transfer in fluid systems Cambridge University Press, Cambridge, 1983 Dubrovskii E.V. and Fedotova A.I. Investigation of heat exchanger surfaces with plate fin Heat Transfer - Soviet Research, Vol.4, No.6, pp 75-79, 1972 Dubrovsky E.V. Study on the increase of convective heat exchange in channels of triangular and rectangular section International Journal of Refrigeration, Vol. 7, No. 5, 1984 FDO Hengelo (VMF-Stork) Eindrapport corrosieproeven met het mengsel LiBr/ZnBr2-CH3OH Internal Report nr.A83 - 125, FDO-Hengelo, 1983 Girsberger W. und Trepp Ch. Arbeitsstoffpaare fr Hochtemperatur-Absorptionswarmepumpen Ki Klima-Kaite-Heizung,Ki Extra 14, 1981 Girsberger W. Hochtemperatur-Absorptionswarmepumpen Dissertation, Technische Hochschule Zrich, 1981 Grossman G. Simultanious heat and mass transfer in film absorption under laminar flow International Journal Heat Mass Transfer 26, pp 357-371 (1983)

B9

B10

Bil

Cl

C2

Dl

D2

Fl

Gl

G2

G3

104

G4

Gregoric R. Warmetauscher und Warmeaustausch 2. Auflage, Berlin, 1973 Griend E.J. van de Warmte- en stofoverdrachtscoefficienten in een horizontale pijpbundel filmabsorber met het mengsel LiBr/ZnBr2 - CH 3 0H voor een warmtepomp voor woningverwarming Internal Report S-736, Laboratory for Refrigeration Engineering, Technical University Delft, 1982 Hbling W., Schnitzler H. and Moser F. The COP of two- and three-stage absorption heating systems Proceedings "International Workshop on Research activities on advanved heat pumps", Technical University Graz, 1986 Iedema P.D. The absorption heat pump Dissertation, Laboratory for Refrigeration Engineering, Technical University Delft, 1984 ICI United Kingdom private communication with R.L Powell Jamieson D.T. Liquid thermal conductivity survey Edinburgh, 1975 Kirn H. und Hadenfeldt A. Warmepumpen, Band 1: Einfuhrung und Grundlagen, KWK-Actuel bd 26 3.Auflage, Verlag C.F. Muller, Karlsruhe Koebel M. und Aegerter A. Die thermische Stabilitat von Methanol und des terneren Systems Methanol-Lithiumbromid-Zinkbromid Ki Klima-Ka"lte-Heizung,Ki Extra 14, 1981 Kays W.M. and London A.L. Compact Heat Exchangers 3rd edition, McGraw Hill, New York, 1984 Minkus B.A., Dergatchev A.G., Glinka L.L. and Menshina V.A. Low-temperature absorption refrigeration machines Proceedings of the "14th International Congress of Refrigeration", Moscow, 1975 Mochizuki S. and Yagi Y. Heat transfer and friction characteristics of strip fins Heat Transfer - Japanese Research, Vol.6, No.3, pp 36-59, 1977 Meynen A.J. Absorptiewarmtepompen 48ste vakantie-leergang voor Warmtetechniek, MT-TNO afd. Warmte- en Koudetechniek, Apeldoorn, 1986

G5

Hl

11

12 Jl

KI

K2

K3

Ml

M2

M3

105

NI

Nakayama T., Kanai T. and M. Hashimoto Development of a refrigerant-absorption fluid system for absorption cycles Proceedings of the "17th International Congress of Refrigeration", Vienna, 1987 Patankar S.V. and Prakash C. An analysis of the effect of plate thickness on laminar flow and heat transfer in interrupted-plate passages Int. Journal Heat and Mass Transfer, Vol.24, pp 1801-1810, 1981 Perry's Chemical Engineers Handbook 6th edition, McGraw-Hill, 1984 Ree H. van der en Oostendorp P.A. Resorption heat pumps, in "Heat Pump Fundamentals", edited by J. Berghmans NATO Advanced Study Institute Serie E: Applied Science, Sythoff Press, The Hague, 1983 Stolk A.L. Collegediktaat Koudetechniek Al Technical University Delft, 1984 Saurwalt F.W. Metingen aan een absorptiewarmtepomp Internal Report S-756, Laboratory for Refrigeration Engineering, Technical University Delft, 1984 Shah R.K. and Webb R.L. Compact and enhanced heat exchangers in "Heat Exchangers: Theory and Practice" edited by J. Taborek, G.F. Hewitt and N. Afgan, Hemisphere Pub!. Corp., Washington D.C., pp 425-468, 1983 Shah R.K., McDonald C.F. and Howard C.P. Compact Heat Exchangers History, Technological Advancement and Mechanical Design Problems Heat Transfer Division ASME, HTD - vol. 10, New York, 1980 Shah R.K. and London A.L. Laminar flow forced convection in ducts Suppl. 1 to Advances in Heat Transfer, Academie Press, New York, 1978 Shah R.K. and London A.L. Offset rectangular plate-fin surfaces - heat transfer and flow friction characteristics Transactions of ASME, Journal of Engineering Power, Vol.90, Series A, pp 218-228, 1968 Schlichting H. Boundary Layer Theory 6th edition, Mc Graw-Hill, New York, 1968

PI

P2

Rl

51

52

53

54

55

56

57

106

58

Sparrow E.M., Baliga R.R. and Patankar S.V. Heat transfer and fluid flow analysis of interrupted wall channels, with application to heat exchangers Transactions of ASME, Journal of Heat Transfer, Vol.101, Series C, pp 188-189, 1979 Shah R.K. Classification of heat exchangers in "Heat Exchangers: Thermal-hyfdraulic fundamentals and design" edited by Kakac S., Bergles A.E. and Mayinger F., Hemisphere Publ. Corp., Washington D.C., pp 9-46, 1981 Trane Company, Epinal Cedex, France Product information bulletin "Plate fin heat exchangers" Tuynman K. Warmte en stofoverdracht bij filmabsorptie oop horizontale pijpen Internal Report S-751, Laboratory for Refrigeration Engineering, Technical University Delft, 1983 VDI - Warmeatlas Berechnungblatter fur Warmeaustausch 4. Auflage , VDI - Verlag Dusseldorf, 1983 Velden M. van der Aanzet tot een nieuw mathematisch model voor een absorptie warmtepomp voor het mengsel 1ithiumbromide/zinkbromide - methanol Internal Report S-782, Laboratory for Refrigeration Engineering, Technical University Delft, 1985 Velden M. van der Processimulatie van een absorptiewarmtepomp proefopstelling met het mengsel lithiumbromide / zinkbromide - methanol Internal Report S-800, Laboratory for Refrigeration Engineering, Technical University Delft, 1986 Westra, J.J.W. Een geschikt mengsel voor een absorptiewarmtetransformator Internal Report K-139, Laboratory for Refrigeration Engineering, Technical University Delft, 1987 Westra, J.J.W. About the comparison of the absorption and resorption principle for a heat transformer Internal Report K-122, Laboratory for Refrigeration Engineering, Technical University Delft, 1986 van der Wekke B.J.C., Wassenaar R.H. and Segal A. Finite element method solution of simultanious two-dimensional heat and mass transfer in laminar film flow Warme und Stoffbertragung 22, pp 347-354 (1988)

59

Tl T2

VI

V2

V3

Wl

W2

W3

107

W4

Wieting A.R. Emperical correlations for heat transfer and flow friction characteristics of rectangular offset-fin plate-fin heat exchangers Transactions of ASME, Journal of Heat Transfer, Vol.97, Series C, pp 488-490, 1975 Wel Ie A. van der Metingen aan een warmtepomp met "compact heat and mass exchangers" Internal Report S-788, Laboratory for Refrigeration Engineering, Technical University Delft, 1985 Wel Ie A. van der Theorie en experiment aangaande warmte- en stofoverdracht in een absorber uitgevoerd als 'compact heat and mass exchanger Internal Report S-804, Laboratory for Refrigeration Engineering, Technical University Delft, 1986 Wassenaar R. to be published in 1989 Laboratory for Refrigeration Engineering, Technical University Delft Wekken B.J.C, van der and Wassenaar R.H. Simultaneous heat and mass transfer accompanying absorption in laminair flow over a cooled wall International Journal of Refrigeration, Vol. 11, Nr.2, March 1988 Ziegler F. and Alefeld G. Cofficint of performance of multistage absorption cycles International Journal of Refrigeration, Vol. 10, Nr.5, Sept. 1987

W5

W6

W7

W8

Zl

108

NOMENCLATURE Symbol Name specific heat capacity correction factor plate (wall) thickness, plate spacing Fanning friction factor circulation ratio height Colburn factor length specific cond./evap. heat capacity plate (fin) thickness corrugation time surface shear stress velocity weight fraction width length surface area form cofficint diffusion cofficint function overall heat transfer cofficint mass transfer cofficint length mass flow wetted perimeter pressure pumping energy heat flow gas constant cofficint of determination temperature volume Unit (S.I.) J/(kg.K)
--

d f h J 1 r t to
V

m
---

m
--

m J/kg m s kg/(m.s2) m/s kg/kg m m m2

--

A C D F K L" M 0 P Q R R2 T V (Greek)
a 1 5

m z /s
--

W/(m2.K) m/s
D l

kg/s m kPa, mbar W W 0/(kg-K)

--

K, C m3 W/(m2.K) m 2 /m 3 W/K m
--

n
P
\ v

A i> i

heat transfer cofficint area density specific heat transfer rate film thickness (wetting) efficiency dynamic viscosity density thermal conductivity kinematic viscosity difference enthalpy ratio dilution/evap. heat ratio deviation 109

kg/(m.s) kg/m3 w/(m.K) m2/s

Numbers (heat and mass transfer)

Nu Pr Re Sc Sh St
Indices
Ot

Nusselt number Prandtl number Reynolds number Schmidt number Sherwood number Stanton number

Indication heat transfer cofficint absorber, absorbent, absorption ambient bottom condenser, cold, contact condensation cooling water dilution, drag evaporator, evaporation effective experiments cross-flow generator, generation hot, heat (transfer), hydraulic, high in (side) mass transfer cofficint methanol, medium, mass, mixture out (side) poor, pump, primary plate spacing rich, rectification solvent, secondary, spacing total theory wall, wetting length direction, general index

a amb b c co cw d e eff exp f 9 h

in, i

k m
out, o

P ps r s t . th w
X

Constants (in equations) a, a', b, b' a, b, c, d (4), (5) (19), (20), (32), (33), (59), (60), (61), (62)

110

Abbreviations

ACM AD AHP AHT CCP

CHME

CHP COD COP MD MLS RHP RHT SAT

absorption cooling machine average deviation absorption heat pump absorption heat transformer compression cooling machine compact heat and mass exchanger compression heat pump cofficint of determination cofficint of performance maximum deviation mean least square resorption heat pump resorption heat transformer shell and tube

111

CURRICULIM VITAE The author was born on December 25, 1958 in Amersfoort, the Netherlands. Following his secondary education at the "Johan van Oldenbarnevelt Stedelijk Gymnasium" in Amersfoort. In September 1977 he started his study in the Mechanical Engineering Department at the Delft University of Technology. The degree of "werktuigbouwkundig ingenieur" was granted in June 1983. His graduate work took place at the laboratory of Refrigeration Engineering and Indoor Climate Technology. From August 1983 till December 1984 he fulfilled his alternative military service for the Dutch Organization for International Development Cooperation (NOVIB), a non governmental organization supporting development projects in the Third World. Since December 1984 the author has been a research assistant ("promovendus") at the Laboratory of Refrigeration Engineering and Indoor Climate Technology. He was working on the study presented in this thesis under the supervision and guidance of Prof. Ir. A.L. Stolk and Ing. C.H.M. Machielsen.

113

STELLINGEN

Een belangrijke stap om de investeringskosten van een absorptie-warmtepomp met compacte componenten terug te dringen ligt in een verbetering van de productie-methoden, d.w.z. in het terug brengen van het tot op heden hoge uitvalspercentage bij de productie. De toepassing van "compact heat and mass exchangers" als componenten in een absorptie-warmtepomp lijkt uit economisch oogpunt juist daar te kunnen gaan plaatsvinden waar een belangrijke kwaliteit van haar, de compactheid, van minder groot belang is, namelijk voor gebouwverwarming en industrile toepassingen. Het onderzoek naar en de toepassing van sorptie-systemen zal naast de energetische en economische aspecten meer en meer bepaald gaan worden door de milieuproblematiek. Van de energiekosten krijgt men veelal eerst een al dan niet vrijblijvende offerte, van de milieukosten zou dat wel eens gelijk de definitieve afrekening kunnen zijn. Door de sterke opkomst van de reageerbuisbevruchting, het draagmoederschap, de spermabank en vergelijkbare "zwangerschaps"-methoden lijkt het spreken in termen van "het nemen van kinderen" meer en meer te gaan heersen over dat van "het krijgen van kinderen". Het illegaal copieren van software behoort tot de ernstigste vormen van het zogenaamde proletarisch winkelen. In zijn drang naar zelfbehoud en in zijn zorg voor zijn nalatensschap blijft de mens, ondanks het beperkter worden van de middelen daartoe, hardnekkig vasthouden aan het investeren in het verleden (cultuur) naast dat in de toekomst (natuur): Restaureren we een monument of graven we een gifbelt af ? De explosieve groei van de belangstelling voor de technisch maatschappelijke richting binnen de studie Werktuigbouwkunde vraagt om vervanging van de aanduiding "fietsenmaker" voor de werktuigbouwer door die van "rijwiel manager" of "cycle-yuppie". Een van de hardnekkigste taalkundige fouten in de nieuwsgeving is het gebruik van het woord "gijzelaar" als men bedoelt, iemand die gegijzeld wordt (gegijzelde). Of is een wandelaar iemand die over zich laat lopen of een handelaar handelswaar ? Het alsnog straffen van een speler naar aanleiding van videobeelden (wat de scheidsrechter niet zag), roept sterk de vraag op of ook op dezelfde basis scheidsrechter!ijke dwalingen gecorrigeerd dienen te worden. Ondanks de houding en handelswijze in het verleden van hetgeen zij vertegenwoordigen, dienen de recente initiatieven van Gorbatsjov en Arafat door de westerse wereld aangegrepen te worden als een serieuse stap naar internationale/regionale vrede.