(©1995 by the US. Secretary of Commerce on behalf ofthe United Stats. 3 “This copyright is signed tothe Americas nna of Pysice andthe The Viscosity of Ammonia ‘A. Fenghour and W. A. Wakeham Department of Chemical Engineering and Chemical Technology. Imperial College, London SW? 2BY, United Kingdom V. Vesovie Depuriment uf Minerat Resources Engineering, Imperial College, London SW? 2BP, Untied Kingdom J.T. R. Watson [National Engineering Laboratory Executive Agency East Kilbride, Glasgow G75 0QU, United Kingdom J. Millet ‘Nordum Institut fr Umwelt und Anaytik GribH, Gewerbepark Am Weidenbruch, D-18196 Kessin/Rostock, Germany E. Vogel Fachbereich Chemie, Universitit Rostock, D-18051 Rostock, Germany Received Februry 28, 1995; revised manuscript received June 14, 1995 A new representation of the viscosity of ammonia is presented, The representative equations are based on a set of experimental data selected as a result of a cniucal assessment of the available information. The validity of the representation extends from 196 K to the critical temperature for both liquid and vapor phases. In the supercritical rogion the tomperature range extends to 680 K for pressures at or below ambient and to) {600 K for pressures up to 50 MPa. The accuracy of the representation varies from 0.5% for the viscosity of the dilute gas phase at moderate temperatures to about 5% for the viscosity at high pressures and temperatures. Tables of the viscosity generated by the correlating equation at selected temperatures and pressures and along the saturation line are presented to provide easy reference as well as for the validation of computer codes. © 1995 American Institute of Physics and American Chemical Socier. ‘Key words: ammoni;conelation; cial assessment lig viscosity; representation; vapor vscosly Contents Introduction. 1650 8. Acknowledgments ‘equation of State 16919. Reterences. Experimental Viscosity Data. 1651 10. Appendix . Deviation Pots ofthe Selected 3.1 Soures of Data : 1651 Secondary Experimental Data from the 3.2 Data Selection and Adjust 1651 Couclation Methodology. “165311. Appendix TL. Tables of Caleulated Viscosity 4,1 Viscosity inthe Zero-Density Limit 1653 Values.. 42 Viscosiy in the Ciel Region 165412. Appendix IL. Viscosity of Ammonia alone the 43 Excess Viscosity........ 1655 Saturation Boundary. 44 Titial Density Dependence. ..s-s-v-s ees. 1655 45 Higher Density Coeticiens eee iss List of Tables Results and Discussion, 1656 Tabulations. nen LE, 1659.1 List of experimental soures of data onthe Conclusion. ee 1660 __ ¥scosity of ammonia. 2. List of experimental sources of viscosity date used in the analysis... - Coefficients for the representation of the ‘Amu Chorin Soviet. effective collision cross section of ammonia by ‘Repent avalble foe ACS; see Reprints List back of nse. Eq, (4), 1661 2 1661 1669 1664 1667 1652 1653 16541650 4, Coefficients forthe initial density dependence, Eq, (9). 1656 5. Coefficients for the representation of the excess viscosity of ammonia, Eq. (11). 1656 6. Summary of results for the analyzed data set..... 1660) List of Figures 1. Distibution ofthe available experimental viscosity data of ammonia. 1651 2. Deviations of the extracted 2ero-density and low pressure viscosity data from the present corclation expressed by Lg. (3) t0 (9). 3. Comparison of experimental and calculated reduced second viscosity virial coefficients... 4. Deviations of data of Iwasaki and Takabashil® from the correlation. 5. Deviations of data of Hongo and Iwasaki® from the conelaton.... 6. Deviations of data of Iwasaki era irom the correlation. 7. Deviations of data of Iwasaki and Takahashi! gy snd Takis sd Tash a” from the correlation... 8, Deviations of data of Sun and Storvick'® from the coresation 9, Deviations of data of Carmichael etal from the corelation... 10. Deviations of data of Golubev and Likhachev'* from the conclation. 11, Deviations of data of Sun and Storie. Carmichael et al, and Golubev and Likhachev' from the comelaion 12, Deviations of data of Makhija and Stirs! and Carmichael eta. from the corelation. 13, Devistions of data of Malia and Stair! and Carmichael eta from the correlation. 14, The extent ofthe viscosity representation and its estimated uncertain. AL. Deviations ofa selection of secondary _zascphase viscosity data from the coneation, The deviations are defined as 100(Hay Hee : Deviations ofa selection of secondary ligui viscosity data from the correlation ‘The devntons are deine se 1200p Hea exe Deviations of selection of secondary liquid viscosity data from the correlation. ‘The deviations are defined as: 100(se~ Heal Hex 1655 1656 1657 1657 1687 1657 1658 1658 1688 1639 1659 1659 1661 AL2. 1663 ALB. 1663 1. Introduction Ammonia is an important commercial fluid which ranks among the wp ten bulk chemicals in terms of large-scale production. The nominal world-wide capacity of its manu- facture for the year 1993-1994 is estimated to be in excess ‘of 116 milion moti tons. Owing to its large host of vapor J. Phys. Chom. Ret. Data, ol. 24, No 5, 1965 FENGHOUR ET AL. ization, ammonia originally found large-scale application as a refrigerant, particularly in commercial equipment, but the pervnetinn and conenmion fr hi prpnse has heen ove Shadowed in the reeent past by its use in agricultural frti- inet production, However, the demand for environmentally ftiendy reftigerants seems likely to ensure that ammonia en- joys a resurgence of interest fr use in refigration cycles as ‘pure fui or in mixtures since its nither a greenhouse gas tora harmful agent to the ozone layer, Furthermore, recent developments incite that a maxture ot ammonia and Water ‘an be used asa working Mud in power plants wit resulting improved thermodynamic performance. In any ofits indus- trial applications « knowledge of the thermophysical proper ties forthe fluid over a wide range of temperatures and pres- sures is required, preferably in a machine-usable format. Most of the studies ofthe transport properties of pure sub- stances have been confined to nonpolar materials. Ammonia isa polar substance with a significant dipole moment which confers onthe fui « behavior that difers substantially from that of nonpolar substances. Although several assessments'™* of the viscosity of ammonia have been made in the past, ‘one ofthese treatments has included equation forthe rep- reeualon uf we propeny la die dese QuN! regio, Usig the limited data available to them in 1961, supplemented by ‘estimated values, Groenier and Thodos! developed general- ied graphical comelations for the viscosity of the gas at moderate pressures and forthe uid in the dense phase re gion. The ASHRAE publication” on the thermophysical ‘roperis of refrigerants gives representative equations only forthe temperature dependence ofthe viscosity of the gas at atmosphere pressure and forthe saturated liquid, The most ‘comprehensive assessment to date is that of Golubev et al? in 1978, Using the data avallale o them at at time, ey derived, largely using graphical techniques, tables of the vis- cosity within the temperature range 200 to 750 K and the reese range Ol te £0 Ma. Using, similar graphical smoothing techniques, Stephan and Lucas* produced in 1979. lables ofthe viscosity of ammonia in the temperature range 310 to 600 K and pressure range 0.1 t0 60 Ma This paper contains a correlation of the viscosity data for both gaseous and liquid ammonia based on a critical asess- ‘ment of all published experimental measurements including those which have most recently Become availabe. A core- lating equation is obtained which represents the most reliable sets of data within their estimated uncertainty over a wide tange oF tenjeiatute aid pressure, ad wil sepualinss the ‘known phenomenological behavior of this polar fluid. The representative equation is applicable inthe temperature range {nm the wipe pint ta slmact 700K ant in the presen range from the triple point pressure 10 50 MPa. In order 10 thake use ofthe representation of the viscosity which is wrt ten in terms of the temperature and the density, iis neces- sary to make use of an equation of sate. In this work We have consistently employed the equation of state of Tillnet Roth er al. for all the thermodynamic properties of ammo- “The analysis presented in the following sections conforms ‘closely to that adopted in earlier work®" on the representa- Sion of the taneport properties of ude‘THE VISCOSITY OF AMMONIA 2. Equation of State “The proposed representation of the vscosty-of ammonia expresses the viscosity as a funtion of density and tempera- {ure for reasons associated with the kinetic theory. In prac- tice, the majority of experimental viscosity measurements se reported at specified pressures and temperatures, thus the tse of an accurate equation of state is essential The thermo: ‘dynamite formulation for ammonia used here was the analytic ‘equation developed by Tillner-Roth et al° The equation is applicable inthe range extending from the triple point tem- pstatre to 680 K and for the pressure range extending from the dilute gas to 500 MPa, Temperatures in the Tillner- Roth et al formulation are expressed in terms of the temperature seale TTS-00 The equation of Tillner-Roth ot al 9 ie preferred tothe earlier equation of Haar and Gallagher party because it contains fewer terms and is therefore easier to use, but largely because it provides improved consistency with the ‘equation of state forthe erica region which will be impor tant in the treatment of the thermal conductivity of ammonia subsequently Tu pol of ali he coment uf Ua vieusty lou, te equation of state of Haar and Gallagher could be used to evaluate the density of ammonia and employed in conjune- Sion withthe preent repreenttion af the viscosity sithan incurring an error beyond the uncertainty ascribed tothe or relation, However, it should be emphasized that inthis case temperatures measured on the ITS-90 temperature scale should be converted to the earlier IPTS-68 scale before the ‘equation of ste of Haar and Gallagher is used to evaluate the density 3, Experimental Viscosity Data 3.1, Sources of Data ‘The dynamic viscosity of ammonia in the liquid and gas- ‘cous states have been the subject of study by many investi- gators. A literature survey reveals about forty experimental scterminations dating from the eaily work of Graban ia 1846. The data sources are listed in Table 1 together with details of the method of measurement, the uid phase inves- fignted, and the range af experimental conditions for each study. Though the measurements appear to span a wide range of temperatures ({rom 196 to 989 K) there are large regions of the pressure/temperature plane which are either devoid of data or for which the coverage is sparse as depicted by Fig 1. The liquid region below 300 K and up to the critical pres sure is not adequately covered. No data are available in the supercritical fluid phase above 600 K and at pressures higher than 0.1 MPa. The supercritical region above 523 K and at pressures higher than the critical pressure is not covered by ‘eny data; tho rogion below 200 K hao equally not been in- vestigated. In addition, as already established by the previous assessments of the dat, the agreement between the measure- ments of viscosity reported by the different investigators is ‘not good. 1651 100 ‘RESSURE LOH) 190 390~=~=«S90~S=«C780~=S—«90 TEMPERATURE / (6) Mo. Distapsn ote arated expermestal scot data of amsnoaa ‘The fl ine shows the stration is 3.2. Data Selection and Adjustment ‘Actiical analysis ofthe dat hasbeen carted out in order to determine a primary data set. deally, ora set of measure- rents to qualify for the primary class, it should have been ‘obiained from a high precision instrument for which there is 2 full working equation. Tn practic, these constrains had to ‘be relaxed by inclusion of other data which did not meet the stringent ideal conditions in order to cover as much as pos- sihle ofthe phase space. In sch caee. the rights ascigned to the results were adjusted according to an ascribed uncer tainty. Graphical examination of the data was also used to provide a check on consistency and precision, Major disrep- ancis were oocasionully observed between overlapping sts of data from diferent authors. In each case a careful exami- nation of the experimental method was employed to select the wit sli set of dal. The primary dat st established in this manner! jg employed in the development of the representative equation for the viscosity of ammonia, cubereas the secondary data are ueed for sompasioon por poses only. ‘The majority of the primary set of experimental data was obtained from methods using either an oscillating dise o a ‘capillary flow apparatus. Detailed stdies of the theory and rechanies of operation ofthese instruments" indicate that both are now well understood. It was found necessary to supplement these data by a number of other data sels, A ‘major source of data of ammonia by Carmichael et al.!? was ‘obtained using a rotating cylinder viscometer, for which a trathcrmatiel theory for te conection of end effects hes not yet been developed, The uncertaity introduced by the lack ‘ofa rigorous theory Timits the absolute nature of this method and impairs the accra, Measinements obtained fram sich an apparatus must be considered of lower accuracy thanFENGHOUR ET AL. Yew iter Matha Resor TUK) POMPaD Guhl ® =o gerd 1912 95) Sapitiry 1 “1 Vogel 191409) oiling ie ' 1 Esey 1920 (67) capillary t o1 Rankine and Smith 1921 @7) capillary e 01 Edwards and Worswick 1925 (6) capillary ® 01 Jang and Sick 193078) capiry e 01 Brine and Linke 1930 (18) osillating ise e 01 Fredeabagen capillary 1 01 ‘rau and Hebetng capillary 5 ~01 ‘Moroszon and Pskov eillary 1 ma olos Siateteck 1933 (68) ling eine a 253.3532 Yan Cave and Manss 1835 (28) oeillaing disc gma 01 lank and Hust 193961) capillay 1 78298 05-10 ‘obser and Miler 1941 (69) lin ball © man ~or Pisevic 1948 (70) eapitiary 1 zra3 “o1-20 Shatensein eta 1949 (62) —_capiliay 1 ms208 7-10 Kiyama and Maka 1982 (1) ralng ball 3373-97 Carmichael and Sage 1982172) railing hall 1 Golubev aod Percy 1953 (64) capillary a Carmichael eo 1963 (12) rating eylnder a Shimotake ted Thodos 1963 (73) capillary ' gnici aed Hommel 10K (6M) t Iwata ef ak 1964 (1) eating dis ' CChalrsbori apd Gray 196529) capa ' Burch nd Raw 1967 (17) capiiary : Pat and Rann 167 (5) elaine ie £ Pal and Barus 1967 (16) iain ise . Pal and Barua ‘scilaing ise ® [ata and Takahashi ‘xiang dive ' Bhatacharyya er ' Makin and Stairs 1 Gober and Starpove ' Rakhi and Roy sclning ise 5 Golbey and Likhachey 1974 (18) capillary : Hoag snd ask 1977 10) os se : ‘Sena Store 1979013) espa . ‘Wako and Nagashina 1993 (31) __ oiling cylinder 1 Sip he above "preter he pueous se, those from the more rigorously established techniques. How: ever, the data of Carmichael etal! were compared with those of other authors in regions of overlapping fluid states. Good gereement was found between the data of Carmichee! eral," the later measurements of Golubev and Likhachev"+ and the more recent values reported by Sun and Storvick"* using a capillary flow viscometer along two high temperature isotherms, as wll be shown later in the discussion. Although the data agreed to within the claimed accuracy of the two ‘methods, it was noted that Carmichael er al.” and Golubev and Likluchey"* ubseived au initial deciease i viscosity slong an isotherm with increasing density whereas this was not apparent in the measurements of Sun and Storvick.!? ikewise at low temperatures, the measurements of Car- richael et al. are consistent with the remainder of the pric ‘mary set of data within the estimated uncertainty. The agree- ‘ment thus observed with the results of measurements with more widely used and established techniques suggests that 41. Phws Chem, Ref ate, Vol 24 Nn 5 1095, Tote gud tate, either sobooled or eatarated and “gl” 10 the rotating cylinder measurements of Carmichas o a! ro accurate to within their assessed uncertainties of 2 percent. Al the data of Carmichael er al were therefore included in the primary set except those on the 410.91 K isotherm which were excluded on account of their closeness to the critical temperature. ‘A numberof the published measurements on the viscosity of ammonia have been obtained in a relative manner using the values of the viscosity of other uid as calibration data In each of these cases the source and reliability of the cali- lati valucs used have been checked against current recommendations" ** and adjusted as necessary. Only five sets of data required re-normalizaton in this way. For three ofthese, belonging 10 the secondary st, the magnitude ofthe renormalization involved was small””-® and ofthe order of cone percent; however, for two! the adjustment required ‘was more substantial. The data of Burch and Raw” which span large temperature range (273-673 K) in the dilate gas‘THE VISCOSITY OF AMMONIA region were judged to be important forthe development of the zero-density comeation. However, these measurements were made relative tothe results of Trautz and Hebering™ for nitrogen. It is now firmly established thatthe measure~ inents of Trauz and co-workers inthe 1930's ae in sytem- fide enor and ifr from reve accurate measurements of viscosity in a roughly linear fashion with temperature, the difference rising to some 4% to 54% at 1000 K."***"° The ‘ales of Traut: and Heberling® for ammonia were sale by comparing their reported values for nitrogen and oxygen with the currently accepted values”"* for these fluids. The Comection applied amounted to some +5 percent atthe high- est temperature of measurement. Burch and Raw's measure ments!” were therefore scaled in a similar manner and found then to be consistent with the results of er workers in a region of overep. Above about 723K, the values reports by ‘Trautz and Heberling” are not reliable due to ammonia de- composition into hydrogen and nitrogen. The rate of decom prsten increases wth tomperstre and io eoniderebly af fected by the catalytic nature ofthe materials with Which the 58 comes into contact. For the liquid ohase. there is paucity of reticle da ‘Only the data of Carmichael et al.!? and Makhija and Stairs’ ‘vere judged tobe sufcienty accurate tobe included in the ‘imary set. Near the completion of the present work, new ‘data on the saturated liquid by Wakao and Nagashima” were ‘communicated to us. Because not all of the details of the measurements were avilable, we have been able to include the data only in oor final comparisons rather than in the constroction ofthe representation ‘Additional checks were made on the sources and reliabi- iy of any ancillary data, suchas the density. required inthe analysis ofthe experimental viscosity data. Although dite. ent sources of density data have been used by varius au thors, no. majo differences were found which justified any ‘modification ofthe reported values of viscosity. The vals of temperature recorded in the select works were corrected ‘where necessary to the ITS-90 scale." “Tet sources vf pay espera dat woe the dovelopment of both the zero-desty correlation and the excess viscosity contribution is presented in Table 2 which inclodee also estimated vcersintes am the dain The e- signed accuracy foreach source is based onthe measurement method used, the quoted experimental erors and the extent of consistency with other independent data in overlapping regions. In certain cases, the ascribed accuracies were a> {sed according to temperature range, with higher uncersin- tes given to higher temperature data. 4, Methodology For theoretical as well as practical reasons, the viscosity is usually decomposed into three contributions:°°> Wo.T)= ml T)+An(p.T)+ 8.0.7) (1) We.T)=Ap.T)+dnAp.T), @ 1653 “anus 2 List of experimental sources of viscosity dats used inthe analysis. No-ofpome —— Eadmacd Authors) Ret, meepted—_uncerity Primary date Vors! 19 2 20-30 Bane and Linke is ° 1s ‘raute and Heberlng 2» 4 20-100 (Carmichael et ab 2B 10 20-40 Issa et ot i" 20 os ‘Burch end Raw n in 15 wad and Takaaki 16 106, oss Malia and Stes 1s 20 10 Golaber and Lithschey 1 6 20-30 Hongo and Iwas 0 @ os-a7s ‘Sun an Strick B 35 20 Secondary data Rankine and Smith 2 1 20 Edwands and Worevick 56 3 2s rreseosgen 3 2 be ‘Troe and Heberlng 20 Py 20-100 Moncszon and Plekov 75. 6 50-100 Van Cleave and Mauss 28 5 50 Shaeestein eo ee 3 30 (Golubev and Peto 6 m 109 Carmichael et al 2 37 20-40 ign ang Heol o A ts 2 3 8 5 38 3 6 « 9 4 31 0 where _n(T) is the viscosity in the zero-density limit, An(p.T) an excess viscosity Which represents the increase in the viscosity at elevated density over the dilute gas value at the same temperature, and A7.(p,7) a critical enhance- ‘ment accounting forthe increase in viscosity in the immedi ate vicinity of the critical point. The two first terms in the right-hand side of Eq. (1) are sometimes grouped together ‘aud die scoulting quantity is dea tecued the background con- tribution 39,7). In the development of the viscosity corre- lation, this approach allows an independent treatment of each cconteihution inthe Tight of availahle experimental and then retical information. Indeed, the use of the most recent ad- ‘vances in kinetic theory helps to assess the internal consis- tency of the various sources of experimental data, 4,1. Viscosity in the Zero-Density Limit ‘The viscosity of a fluid in the zero-density limit, 79(7), being an experimentally accessible quantity, is amenable to analysis independently of the influence of the other contribu tions enumerated in Bg. (1). Moreover the existence of a well-developed kinetic theory™ applicable under the dilute gs condition offers some guidance as to the form of the ‘correlation Indeed, in the zero-clensity limit, the viscosity is almost independent of the existence of intemal degrees of1654 freedom, and hence is almost unaffected by inelastic colli- sions. However, despite the availability of a detailed theory ‘of the viscosity in the zero-density limit, it cannot be used for the prediction of the viscosity with an accuracy compa rable with that of experiment owing to our incomplete knowledge of the intermolecular potential for ammonia. ‘Therefore, the zero-density values have to be extracted from the experimental data themselves. ‘The examination of the experimental data reveals a lack Cf useful data in the real limit of zero density. Kor reasons that will become clear, itis important to distinguish the prop- erty in the true limit of zer0 density from use at low pressures (such as 0.1 MPa). The lack of data in this vitally Important region means that they must be first extracted from the available experimental data. The experimental viscosity 1 mengnred along ientherms wore fitted tn palye nomials in density following as far as possible the criteria recommended by Hanley etal.” The fits yielded both the zero-density values as well as the first-density coefficients for the thirty isotherms considered. The isotherms reported by Hongo and Iwasaki,!” Iwasaki et al,!" Carmichael ‘etal. Golubev and Likhachev,!* Sun and Storvick,”® and 1wasdki and Laxanasmi"® which cover the temperature range 293-598 K were used in the analysis. In order to extend the temperature range, the set of zero-density values obtained ‘wae supplemented by the addition of the data of Braune and Linke,!* Burch and Raw,"” and Vogel." The data from these sources have an associated uncertainty somewhat larger than the bulk of the data available at intermediate temperatures. However, since they are the only data sets available at low and high temperatures, they were included in the fitting pro- cedure with an appropriate weight. Moreover, since the sets fof measurements were performed at pressures below 0.1 ‘MPa, no attempt was made to extrapolate the values at par~ ticular temperatures to the Timit of zero density since the dlifferences involved are substantially less than the experi- ‘mental uncertainty, One datum from the set of Vogel"? was ‘measured at 196 K and proved therefore useful for defining the romadonsity| ‘near the fiple point temperate of ammonia (195.48 K). For extrapolation purposes up to 953 K, four data points belonging to the scaled set of Trautz and Heberling?® were included in the fit but with a high uncer- tainty of 10%. Altogether 56 data points were used to de- velop the zero-density correlation. ‘The kinetic theory of dilute gases relates the viscosity of a pure gas to an erfecuve coliision cross section which con- tains all the dynamical and statistical information about the binary collision. This relationship, for practical purposes, can bbe written in the form 0.02135717M}" wlT)= Sg 8 where © is a reduced effective collision eross section, T is the temperature in kelvin, M is the relative molecular mass, «7 isa length scaling parameter in nm, and m9(T) is the viscosity in the limit of zero density expressed in units of Pass, ‘The reduced effective collision cross-section values corre- sponding to the viscosity values in the limit of zero density FENGHOUR ET AL. "ASL 3. Coefficients for the representation of the effete clison eros fection of ammonia with Eq (0, ° a0 1 061123364 2 ° 3 assssi26 4 ~0.1160546 sity eft 386 K, 202057 nem, and M~17.03026 ‘were calculated with the help of Eq. (3). The resulting effec- tive collision cross-section values were then fitted, by means ‘of tatstical weights derived fenm experimental certainties as explained earlier, tothe functional form, waged atte TH! ® where the reduced temperature 7* is given by ATI © and e/k is an appropriate energy scaling parameter in kelvin. It has been found that the primary viscosity data of am- ‘monia in the limit of zero density can be represented, with their estimated uncertainties, by means of Eqs. (3)~(5), the ‘optimum coefficients a,"and the parameters and «. These latter two parameters are, in principle, arbitrary, and the val- ‘ues adopted are discussed in Section 4.4. Table 3 contains the relevant coefficients for the representation of the zero- density viscosity values of ammonia. The resulting correla- tion is valid in the temperature range from 196 to 714 K, and its uncertainty is estimated to be 0.5% in the temperature interval 273-500 K, 1.5% from 00 to 714 K and +2% below 273 K. Figure 2 llustrates the deviations of the 2er0 density and low pressure viscosity values from the calculated ‘ones using the correlation. It shows that all the data are rep- sssented by the zero-density correlation within their aneribod ‘uncertainties. r 4.2. Viscosity in the Critical Region ‘Theoretical and experimental evidence indicates that the viscosity of fluids diverges atthe critical point.** However, unlike the critical enhancement in thermal conductivity, that in viscosity is much less pronounced and is restricted to a ‘much narrower range of conditions around the erteal point. For a number of fluids” it has been found thet the ratio of Arnc(p.T)!nkp.7) is greater than 0.01 only within 1% (~4 K) of the critical temperature. Owing to the complete ab- sence of experimental data for the viscosity of ammonia in the region around the ctitieal point (p,=225.0. kg/m, T.=405.4 K, P= 11,3334 MPa, it has not been possible to establish the form ofthe enhancement term, A 7.(0,7), of Eg, (1), Throughout the remainder of this article, the viscos- ity should therefore be identified with the normal of back: pround viscosity defined by Eq, (1).‘THE VISCOSITY OF AMMONIA 20 1s os ae . % aoa to * ® 1.0 ¥ “ os y -20 180 280 380 480 380 680 780 esmarune a0) 2 enone snes Sry Sow pam vay ‘tom ep rate ere E(w The rer te eed at Qua won a iowa oa 8 ead orto Hea al 7 Se wal Late 4.3. Excess Viscosity ‘The excess viscosity contribution describes how the vi ‘costy of the Nid behaves as a function of density outside of the critical region. The present state of knowledge does not provide a satisfactory theory for the excess viscosity over the whole phase space. However. in practice. a power series ex- pansion in density for the representation of the excess vis- ‘cosity has often been found to be adequate.“ The series adopted is of the following form: Ane.T): T)p+ By nlp.T)=bi(T)p+ & bdTIp', o where b,(7) are empirical functions of temperature. From ‘observations on a number of fluids” it appears that, for su- posciitical teampesatutes, the wueffcients 6((7) ate uf very \weak functions of temperature. However, forthe first coefti- cient b,(7), recent theoretical work®™*** has established a temperatare dependence which ean he quite marked! at sub- critical temperatures, Indeed, the coefficient b,(7) can change sign from positive to negative as the temperature is reduced. Thus, the viscosity in the vapor phase along an isotherm may first decrease owing to a negative value of 4 (7) and subsequently increase owing to the influence of higher order terms as the density increases. This effect has been observed for a number of fuids."” Thus, in order to cobjain an accurate representation of the behavior of the vis- ‘costy in the vapor phase, the temperature dependence of the first density cooffcient of viseosty must be taken into con sideration. 1655 4.4. Initial Density Dependence ‘The viscosity of ammonia vapor at temperatures up to ound the critical temperature has @ negative pressure coef- ficient similar to that found for steam.?*“* This phenomenon was first observed by Twaraki fal! using an eocillatng dise viscometer and subsequently confirmed by other work- es using both oscillating dse and capillary flow viscom- ete, ‘As was remarked earier, a statistical analysis of the pri mary data employed for the evaluation of the zero-density values of viscosity also provided experimental values of the first-density coefficient b,(7). The values of this coefficient were then expressed in tems of the second viscosity viral coefficient B,(T) by mesns ofthe definition, b(T)=B (T) m7) o using experimental values of (7). ‘As a result of both theoretical and experimental studies of the second viscosity virial coefficient, it has been possible to develop & valuable coresponding-states representation of its Yetivior as at funeion of vemperature."*"* This proposed ‘model has been useful for treating noble gases, simple gases such a5 n-alkanes® and aromatic vapors? Vogel et al? have shown that polar as well as nonpolar gases exhibit the same general behavior of B,(7) as a function of reduced temperature. The existence of « model for the second viscosity virial coefficient offers the opportunity of confirming the validity of some of the experimental data as well as of extending its temperature range. For that reason, we have vsed the model of Rainwater and Friend” as modified by Bich and Voge!" to examine the experimental values of B (7) deduced from analysis of the primary experimental data. The model of Rainwater and Friend ie baved upon the behavior of « gus whose molecules interact according to a Lennard-Jones (12-6) potential. It therefore predicts that forall such gases the reduced second viscosity virial coefficient. for practical applications, is defined by* m By 25) TENSE" a where B,, isin units of I-mol ‘The reduced second viscosity vir sal function of 7* and has been represented by Bich and ‘Vogel by the empirical function 1 cocfficient is @ univer Beary 0 Bt ‘with the coefficients ¢; given in Table 4 for completeness. Accordingly we have determined the parameters o and ‘fk which secure the optimum representation of the experi- ‘mental values of B'(7) by means of this function. Figure 3 compares the theoretical and experimental values and serves to confirm thatthe qualitative values predicted by theory are reproduced by the available experimental data, in particular the change of sign of the coefficient.1656 ‘nots 4, Coeficiente fr the inital density dependence Ea, () ° TorTeResoex 0 1 4692601 10? 2 =05460794%10° 4 ~O13019164% 108 3 (03614230%108 6 ~os87i1743x108 7 ‘sees 5 -039834012%10" ° o.36s241xi0* 0 ~0.12027350%108 n (0.243405 10° 2 02080795710 ‘The optimum values are e/k=386 K and o=0.2957 nm, and itis these values that have been used in the development of the correlation of the viscosity in the limit of zero density in order to secure intemal consistency. 1 should be pointed out nere tnat we ascribe no panicutar physical significance beyond reasonableness to the values of and € deduced in this fashion. In terms of the Lennard Jones (12-6) potential modal upon which the analysic of the second viscosity virial coefficient is based, the parameters are the well-depth and the separation at zero potential. This should not be taken to mean that the values have any rela- tionship to the true potential of the nonspherical polar am- ‘monia molecule. Z “° Ba y E g- [REDUCED TEMPERATURE, (T) Fi. 3. Comparison of experiment nd calcuited reduced eeond isco viral coefhiets. = Iwasa and Takahashi!" W Tvasal er al:" C San tnd Storvik:® 0, Hongo and Iwasaki V Golubev and Lichen" @ Cemichael er a;!* —— Eq, FENGHOUR ET AL. ZABLE 5. Cocficons for he representation of he excess visoity of s- moni, Ba. (1) i ay 4, 4) ° ° 017366936% 0 0 1 ° 015425035910" o 2 nasenszasxu0-! ° Lsr668689% 10 3 ° ° ~L49710083%10° 4 ~assesit07 x10"! 0 701207410 4. Higher Density Coefficients In order to evaluate the higher density coefficients [,(7), 2 to n] of Eq. (6) and thus develop the representation of the remaining part of the excess viscosity itis necessary to Perform a stasticat nt to te avaulable pnmary expenmental data. For this purpose, we calculate the higher density excess viscosity, Ayn, Aamlo.T)= ne.) no(7)-b\(Me= Eb (10) ‘Tle 2 lists the primary sources used in the determination of Sun(p,T). We have consistently employed Eqs. (7)-(9) to evaluate the contribution b,(T)p in the determination Whenever possible the values of my(T) reported by each investigator have been employed so as to reduce the effects of systematic errors in individual measurements. Where this, has not been possible, the corelated values given by Es. (3) 10 (5) have been used The coefficients [b,(7),# be of the form 2 to.n] have been assumed to ay ‘where T* is defined by Eq. (5), and the coeficients dy, as wel asthe optimum values of and m, were determined by tho applicatinn af the SEEQ technique 3 Ym this pnnocs each datum was given a weight dependent onthe estimated uncer- tainty, The resulting coefficients are given in Table 5. b= ay, 5. Results and Discussion The extent of the agreement between the representative equations developed in this work and the experimental vis- cosity data is briefly reviewed below. Viscosity values of ammonia calculated by the correlation are compared with experimental data in order to assess the performance of the correlation. The deviations of the experimental data of Iwasaki and Takahashi!® along. six isotherms from the present correlation ate illustrated in Fig. 4. The deviations of the data of Hongo and Iwasaki'® along four isotherms from the correlation are shown in Fig. 5. The deviations of the experimental dita of Iwasaki et al!" along two isotherms from the calculated values are shown in Fig. 6. Comparison fof the measurements of Iwasaki and Takahashi!® with the later measurements of Hongo and Iwasaki, using a similar type of oscillating dise instrument but with additional refine=‘THE VISCOSITY OF AMMONIA 1657 to on Eos E oo -0s coat ae a0 “4 D, * *o 2 s 6 > “to on 02 02 03 DENY tat) ENSEY/ eal tomelation. The deviations are defined ax 100g” Palin 2 298.18 K'isoberm: © 32313 K berm: 34812 K hotherm: ® 373.12 K isotherm; © 398.13 K isotherm; # 408.13 K isotherm. ‘ments to improve accuracy reveals a small but systematic difference between the two sets at elevated temperatires. The data sources agree well at 298.14 and 323.14 K but differ by up to 0.3% and 0.7% on the 348,12 and 373.12 K isotherms, respectively, as shown by Fig. 7. The two isotherms of Iwasaki et al." (293.14 and 303.13 K) are close to the two evistions are defined 86.100, Halty- 4 29814 K ister O 30513 K iroterm, low-temperature isotherms of Hongo and Iwasaki and of Iwasaki and Takahashi!® andi exhibit the same behavior as the latter ones as indicated by Fig. 7. These three sources of data are considered to be the most accurate sets of measurements ‘They are found to be consistent in the regions of overlap and are represented by the euutelation within 1%. os 1s on 10 i os ifs & 7 it z z ‘ | ‘ z 5 él t | £ 0s 4 “10 ous DENSITY mat) Fi. 5. Deviations of data of Hongo ad Iwas” from the colton, The evintions ate defined at 1000 ~ Haig AOR 1K techn: 325.13 K therm; & 48.13 K Woche: @°373.12 K isthe, ‘TEMPERATURE 6) Fo, 7. Devitions of dats of Iasi and Takabshil® A Hoogo and ined 100g Hal Pa ‘J Phus. Chem, Ref. Data, Vol 24, No.5, 19951658 FENGHOUR ET AL. aa 30 : Aa 20 20> as at iat fits 10 eee ‘ a |e a Satta g }Xag > E oo gal eos eA Te -1of co 5 of? : oo, ° . -30 eae nai aan ins DENSITY tol?) DeNSTY mt?) ‘a, 8, Deviations of dats of Sun and Soi" rom the colton. The evinions ae dened 3: 10D(iq9"seag 811 K ober, O 447311 Kistner; 523.11 Kiser; © 373.1 K isthe; © $9811 K hotest. Figures 8 9, and 10 show the deviations ofthe data of Sun and Storvick," Carmichael eral,” and Golubev and Likhachow" from the calculated valves. In the overlapping regions, the magnitudes of the deviations of these three sets are consistent as shown by Fig. 11 and fall within the es rated accuracies The liquid region has also been found to be well repre- sented by the new correlation. Figure 12 displays the devia- tions of the experimental liquid viscosity data from the cal- culated values. Most of the data either on the saturation ‘boundary or in the subcooled liquid region, are reproduced by the sepresentation within 2%. The sets of Makhija and Sian” und Connie et wha Ueeue sepniadunel Uy dhe correlation within the ascribed uncertainties. As anticipated from the eritical assessment of the method of measurement employed hy Carmichael ot af 2 the high temperanine data show the largest deviations as illustrated by Fig. 13, The secondary experimental data have been found to be, in general, represented by the corelation within their esti- mated uncertainties and with no discemible systematic be- havior. The plots containing the deviations ofthe data exam- ined are included in Appendix I. A selection of secondary gas phase data comprising the 410:91 K isotherm of Carmichael etal, the data of Travtz and Heberling” below 800 K and the data sets of Pal and Bhattacharyya,* Chakraborti and Gray Rankine and Smith2” Van Cleave and Maase* Ed- wards and Worswick.* Bhattacharyye etal"? Pal and Barua,® and Rakshit eal have been used to test the per- formance of the representation. Their deviations from the correlation are shown in Fig. AlL1. Most of the data are reproduced within 2% with no apparent systematic trend "Another selection of secondary liquid data made up ofthe 4410.91 K isotherm of Carmichael eral and the datasets of 4. Phs. Chem. Ref, Data. Vol. 28, No. 5.1995 Fa, 9, Devinons of dss of Carmichael eral from he conelton. The ‘uh et. pins See det dente Tard Noget 2 oe aC ° Braue an Linke 9 os, 10 1 (Cermichele o 10 os 10 ° Iwasa er al. aie a ccs) 2 Burch and Raw ale aoa ° anh ae Takahat as oa ° Mabie and Stirs moa 1 Golubev and Likhcher os, 10 ° Hongo and wash 02 0 ° Overall es 06 os 5 Sevonday data Mean mo of poi “autos aes" wanda Rankine and Sith ta ° wards and Worsvick 19 1 ‘rate and Hebring 0s as ° ‘Moneszon and Pskov 4 a ° ‘an Cleve and Mas 16 18 ° Plaka Hn ae eva) ° ‘Sbateashicin etal 19 19 ° Golubev and Peo 40049 4 (Carmichael ea. 12 13 ° CCharabort and Gray 0s as ° Pal and Barua 10 12 ° Pal and Bhatacharya is 16 ° natacarya et on Os 1 CGoiabev ang Sharapova a ° abst er 07 as ° ‘Wakao and Nogshia 20 ° veal ests 27 a7 6 ture at which ammonia starts to decompose, are likely to be affected by the rate of ammonia decomposition which fereavee with temperature. Between 350 K and 580K, tho correlation can be extrapolated up to a pressure of 80 MPa. ‘The values yielded in this case will have an uncertainty of 10% by comparison with the data of Golubev and Petrov and Golubev and Sharapova. 7. Conclusion A critical assessment has been made of the available ex venta date fr dhe dyimunic viscustyy of amon. AM terpolating equation guided by theoretical considerations has been developed to represent the data and the known phe- nomenological behavior of the uid over a wide range of temperatures and pressures. Tables of values of viscosity hhave been prepared together with an assessment of accuracy. ‘The representative equations reproduce the data within their estimated uncertainties, and will be adequate for engineering purposes. The comparison of the viscosity values calculated from tne present representation and the corresponding experimen- 4. Phys. Chem. Ret. Data, Vol. 24, No. 5, 1995 tal data provides the basis for estimating the accuracy of the representation over the validity range of thermodynamic sates, I the tnnyceature semge 290-400 K und fan prcanuses below 6 MPa, the accuracy is +0.5% in the range 400-500 K and for pressures below 10 MPa, the accuracy is +1.0%; in the range $00~600 K and for pressures below 10 MPa, the accuracy is 1.5%; and in the range 600-680 K and for pressures below 15 MPa, the accuracy is +2%, For the lig- hid region, in the range 270 Sioa0 isso moto aoa tatas lear arog os 373533692 BOsT oka) Tas Nar 30 35688 MST 25442165 SOD IRRA 6h Lank m9 S29 sis 899 eost = aanos Teas Nees sak uo 52928 364s «309TH! ORDL KS] MeTD— «15108 120 S126 5843649851105 268.) 23596 dm «RST «16796 SLST 16093523929 6796 MLST— IO 0968 HRSA 130 S522 LDL 36845 3SS ATS DIG =a a_—Ioas 4a 160 $399 a3 mgr aoa inna tsa Bo Saar Maaas maar) ast angk gas ska tna S88, Bo SIS agiT 397 Sg eh nest ions ma) S633, wo sisi2 ara 546 mnt ads ST Esa” mo Snips age ales mes 2uae Goa Tssat T1434 1380 ms ssn02 i536 saoss—mK95 else 2470521909599 18s3——5D 250 5569358130820 96s matas” 98s 7856180 TS soigs 45236 sao alas 2e733 25025 aad) mod 1803S 300 S671 466590178] tom ae 8s aaa] esas ms ss froso— esas. ga ono assent aatss dees Nena) 380 St6as_— iso 02 elas oSn 25995 one aes N87 MeO Ss Sesh Mogi maz ugs) ony deauo cans ames dango rua? wo, Seis Sas9 asad. SersaSansoaeaoe = asao ions sora 2s sos aa7s6 1003350757 ass a2 snos TH so Sess forza tings nay aay atop tomt alt lone nes as @oost ssa? ings st0a woe?) ngs alan lssaa— 77s 500 ssi Sono] ans Men7 as 2aas sane tan m0 pom) mo 0mm a wie 0 oC SCC a5 10s ast) kk 10 sot a2 tm SS 93} 20 130021795 amnostoe tal Stas 30 mae 10am so m3 as HGS Bama 60 Wor ome ge OLD x0 miss ima?) tees. IAG RRS & mss iiiae olarak TSB rss maa nas a a ee rer Mh oe 6m ss) em Se fest bom hase eas? GSS iiss asst as2 a0 2 as TT Gas Sa Wee 626823 sean) 6SS SEIT 19037 iason rise tone oso mak)aTwa 94 ‘wn ian) Hesston? snsh oa} aT G30 a joss asap 1192 dona 9R3D 99921806 SRIT. Las Miss 12911180] N0Rs 98K]. SMM tou teat iiss teem oomet arte eek mR Phys. Chem. Ref, Data, Vol. 24, No.5, 1995‘THE VISCOSITY OF AMMONIA 1665 wo 43033050936 t0836 = tons saat HBO han ooo 37513118 t2gs e096 SOT 9973S a BS MSSt 1325012183286 RD BSD] TRIS 99 3073) T34SR 283021 kI 88S RZ BLS TBD MSs wool 1362424901476 ast 9736 DBS Tea oo 15076 sr? 162910713988} BT RT ms isa? asa 2RUT Nemes kas TBST 330 ISAS) M41 20S = 11925 1m = TOL 9018 OSL Ms 1588214266 = 310527 == tLa2—— RS SA] 220TH wo asta? aaa 130832213128 HOaa 9678S SZ THT 25 se MST ISOS S72 ROLY 430 107447303845 249255007 BS BI BHO BOD 7S 134 TakB 136871252962 K3D— DDB] TAS coo lee isas— seats, BR tos tonsa tea m1) Pow) 400 410 0 60 “0 430 “ 0 40 0 or Ti ase—aTS ase os 1395438782923 16 70a 10 ns 438s 9] 20 Be HTS stat io aT) aSld S895 8S 6TS TSS so 943s ATS 557— tows 50 Mot as dBG S27 sg? t608 gag 72k oy 454. 1555 1s98 M6379 40 Wer 1512S STR 66ST M6176 = T8 ta 00 3163816667037) 110 a2 m9 mass Ro 00 SSS 40 519926 25 HAL 1905.9 90H— 9S 150 S36 425 aS 2308 HMw 160 Sis] 1308S 2686 24k 099 atm 10 5580 San Ss um eos meta 180 56955043 Mal 0 m2 96 mana 10 sm su ms wu nw me mat ao suv S289 yer Skis 2365 28S Bs Gls S846 97a hak RTS SD 20 0 one aT TDS O88 ms #90 ot ast vRS S23 dO 6A HBR 289 sno 576125558 207TH] Ltd ns Bi BOIS SMS STE 380 een 998_ SHS SSL BS T3688, 0 my 66 kM MSL SMM GBT 2s me T6667 aT 5 T as Is 7128 HTT SRR] S54] ROHR a 00 m2 nat RGN S6uS— 3K] = 50g) aTST S59 rx) Pitre) 0 10 20 330 so 30 60 70 sso 0 or vas ~aaS ~~ S~Ctm—~Cd)—~—~Cd@ ~~~ os 1171826! t9ms tats kD 20 v0 1830186919085 98S_ OKO 218 30 9s kia? 1990 0295) oss 50 wor ikaT 18861925208 hal 2D 2136 60 Wis RSS RSS ttm 2160 10 27 ik66 90S toasts aS 2096, ain J. Phys. Chem, Ret. Data, Vol 24, No. 5, 19951666 FENGHOUR ET AL. te 4 rT oe iwss any wa e782 07s ‘na m7 oss B05 mage 2129 hes mt ho 19292797 33 06s aT a 0 wis 1948 BL 20S 050s S622 aT bo war 871 003 03S HM GA 40 9m 97 26 Tasha a8 150 pre fry eee FT ee Pe ee a Se 160 m3 sk tl GD no m7 09) aL2akas aad) as ass 180 uz 0 ks SS 190 im aM she ws aT as 200 2% mo mn 2M mo me mo A 23 266 2s no 3H BAT DMS 250 Be 508 SE 2h? HEATERS ns T= 5HIS STS SS] SHR SSG 8A OS soo amo 8 eS IS aT Sm 8D bas 32 AL O10 350 Bes 3235 op 3S 3S RSID OL as Bas 413 Mss 076m gs RT 400 Me 8 MID tS 303 0581 Bw as sae m3) SD aR SOR 056, so 0m ees] BG BSS as a3 ace ue) S85 SSAA 256, S00 au ase oasis 81S IHG 88 x a) 610 oo 60 Ca «o 60 om 30 on 2s —OOeS*«iSCSSSCSSCSSSSCCSCSSC 05 27000 os DT 20 2% 00 «BM we SHH 30 asl 2 st SBC 28 so 2198 = BDRM BASHING 2893 60 mo 2» nT BM Bm IH 25.00 10 no 80 sts RGSS ST 30 ne 26 BTID RG 2523 Pre a yey | Co no ms m9 B33 MO) aT 2880 Bo nie BI BAT Ish 2280 10 mo ms gos Sas STD 150 mos 3392488 160 na Ss ae SG 902524 2893, 10 nas Tl OS 243515032537 2608 80 nso 387 KID 24S] SIT 2550 26.17 0 BIS OSG SS 2629 2n0 castes ioiashe oe (atssl ae (2amieee | ass | 7sd6) 1 osc 2682 2s we MT 5005285562585 86 267 350 isos sz) St 862m 2896 mis ms Woo 586260668 28SL 8S 77.00 236 200 2638650 8656S aT T2196 ms yu mia ak ra2Sk MG 9G. 28? 2840 380 msi 78 ge RSIS RAT ORGS 286 ms ns = 82ST 8G 81S 833 90 3 3M MAS SS HRT 25 30323016 3005.30] 203, 480 316 3095-5079 30703065 anss, Sg) sO ans iia esicrsys oe stale esviee ocr | iT) 1m size a0 95) ‘J. Phys. Chem. Ref. Data, Vol. 24, No. 5, 1995‘THE VISCOSITY OF AMMONIA 1667 12. Appendix Ill. Viscosity of Ammonia Along the Saturation Boundary 7 ? foe Pe a as) oat (oat) ory we ones 0080 “aoa wa » oost ‘0032 asi sons a oor ia eum 41498 a0 oan ois ‘raat ost aa ‘woe oom aus pao x0 om ‘as foosie 25485 250 169 ‘one a0 mason 260 ass ona 3336 wie a0 13500 ass 269389 1812 20 ora 1356 36.1008 a4 0 10817 asa 3s2298 193 2 isms 8820 333604 rt 330 2aa0s 11086 305 nt xo 5.0803 ass 312681 ws se ios 2398 2a809 ew m0 sam Sams nan so ano 7.4403 3.9584 25059 5831 5s Sasie ste non na 5 99836 7047 nose a5 2 osm 85079 160 20 J Phys. Chem, Ret. Data, Vol. 24, No.5, 1995