Experimental Study on Two-Phase Flow Instability in System Including Downcomers

Transcription

1 Jornal of Nclear Science and Technology SSN: (Print) (Online) Jornal homepage: Experimental Stdy on Two-Phase Flow nstability in System nclding Downcomers Kenji FUKUDA, Masanori TANHRA, Takaaki SAKA, Sh HASEGAWA & Tetsya KONDO To cite this article: Kenji FUKUDA, Masanori TANHRA, Takaaki SAKA, Sh HASEGAWA & Tetsya KONDO (1987) Experimental Stdy on Two-Phase Flow nstability in System nclding Downcomers, Jornal of Nclear Science and Technology, 24:4, , DO: 10.10/ To link to this article: Pblished online: 15 Mar 12. Sbmit yor article to this jornal Article views: 77 Citing articles: 2 View citing articles Fll Terms & Conditions of access and se can be fond at

2 Jornal of NUCLEAR SCENCE and TECHNOLOGY, 24[4], pp (Aprill987). Experimental Stdy on Two-Phase Flow nstability in System nclding Downcomers Kenji FUKUDA*, Masanori TANHRA**, Takaaki SAKAU*, Sh HASEGAWA* and Tetsya KONDO* * Faclty of Engineering, Kysh University ** Mitsbishi Heavy ndstries, Ltd. * nterdisciplinary Gradate School, Kysh University Received Agst 28, 1985 Revised Jly 14, Two-phase flow instability in a system inclding downcomers is stdied experimentally. The system sed for experiments consists of two parallel channels each composed of a "first riser", a downcomer and a "second riser" in order. For series of experiments are carried ot with a first-riser-heated single channel system, a downcomerheated single channel system, a first-riser-heated parallel channel system and a downcomer-heated parallel channel system. A dynamic instability is observed in the experiments with the single channel system, while a kind of static instability, "one-sided flow" as well as a dynamic instability, are observed in the experiments with the parallel channel system. As it is proved that the static instability, in this case, reslts from the negative slope in W-JP characteristics which is mainly de to the term of static pressre loss in the downcomer, it is conclded that this instability is pecliar to systems inclding downcomers. Another instability associated with condensation of vapor trapped in the top of the downcomer is also observed. KEYWORDS: condensation, density wave instability, downcomer, downward flow, flow oscillation, instability, two-phase flow. NTRODUCTON Thogh it is sally ndesirable to inclde downward two-phase flow in a flow system, sometimes we are confronted with this sitation<1)(2). When two-phase flid flows in the downcomer, the slip velocity g- 1 between the gas and the liqid phases becomes negative as the reslt of boyant force exerting on bbbles in the direction opposite to the flow. This is a marked featre of the downward twophase flow and some investigators have reported that the flow easily becomes nstable in this sitationc1)( 6 ). However, it seems that the reported cases of the instabilities are not always consistent with each other. Aritomi( 3 ) performed experiments in a system inclding a downcomer with relation to the steam generator of the fast breeder reactor and fond a "slg excrsion instability". He observed that the vapor slg formed in the heated downcomer reslted in decrease in inlet velocity followed by the slg rise against the flow direction. However, the slg condensed at the top of the downcomer throgh contact with the entering sbcooled liqid, which cased a slg excrsion instability of rather high freqency. n his recent work( 4 ) with a system composed of a single channel with a bypass, he also fond a static excrsion instability indced by vapor obstrction as well as a dynamic one nder a condition i;::;; 0.4 m/s. Nakanishi(s) reported an instability in a. * Hakozaki, Higashi-k, Fkoka 812. ** Bnkyo-machi, Nagasaki 852. * Kasgakoen, Kasga-shi

3 VoL 24, No. 4 (Apr. 1987) conter-crrent air-water flow system bt its cases were not revealed. Related with the stdy on the contercrrent flow limitation in the possible loss-ofcoolant accident (LOCA) in nclear reactors, Wallis( 6 ) and Mrase( 7 ) carried ot experiments sing an air-water parallel channel system and observed a static instability which occrred at the negative slope portion on the W-.:::1 Pcrve. Hayama(s) fond that, even in a heated parallel mlti-channel pward flow system, some channels cold fnction as downcomers where the flid flows downward. (5) Cl H " "" " (mm) Thermocople (1) Lower header (5) Riser (2) First riser (6) Upper header (3) Downcomer (7) Trbine flow meter (4) Second riser (8) Valve Fig. 1 Experimental apparats 267 Since two-phase flow instability in systems inclding downcomers ths has varios featres, we performed an experimental stdy sing test apparats with transparent parallel channels to visalize the flow and observed some of above mentioned types of instability. Then, we tried to explain the mechanisms and to classify the varios instabilities observed. ll. EXPERMENTAL APPARATUS AND PROCEDURE A flow sheet of the experimental apparats is shown in Fig. 1. This is a forced convective flow loop with two heated parallel channels each composed of a first riser (2), a downcomer (3) and a second riser (4) in order. Each heated section is made of transparent 40 mm l.d. glass tbing for the prpose of visal observation, in which 27 mm O.D., 1,400 mm in heatedlength sheathed heater is inserted. The test flid, freon R-113, is circlated by a pmp from a lower header (1) throgh heated sections (2)-(4), nheated risers (5), an pper header (6), a steam drm, where the pressre is at an atmospheric pressre, and a sbcooler to retrn to the pmp. For series of experiments are carried ot; experiments with the first-riserheated or the downcomer-heated system for both the single channel and the parallel two channel system. Keeping inlet flow rate and temperatre constant at 35.0 C, heating power is changed ntil oscillation in flow rate starts. n the parallel channel system, conditions for the two channels are set eqal each other. n spite of this, some differences in flow rate happened to occr de to the slight difference in pressre drop characteristics. The inlet flow rate of each channel and the total inlet flow rate are measred by trbine flow meters (k=.:::1p/(yzptp.d-;;. 370) and the flid temperatre at varios points of the system are measred with C-A thermocoples. The heater srface temperatres at the downcomer are monitored to detect dry-ot initiation. ffi. RESULTS 1. First-riser-heated Single Channel System When heating power is increased gradally over a certain threshold while keeping flow -11-

4 268 J. NcL Sci. TechnoL, rate and inlet temperatre constant we observed an instability, which mechanism is clarified by visal observations as follows. When heating power is increased gradally, bbbles are generated at the top of the heated first riser (Fig. 2(a)). Then the vapor generated is accmlated in the top of the nheated downcomer and forms a liqid level as shown in Fig. 2(b ). Since steam qality at the top of the first riser proved to be nearly zero or slightly negative, the vapor accmlated condenses in contact with the liqid entering the top of the downcomer and the liqid level oscillates cased by periodic condensation and accmlation of vapor. However, the liqid level falls gradally since the vapor is generated continosly and finally reaches the bottom of the downcomer. The vapor, then, is led to the nheated second riser where it flows p growing in size by the decrease in static pressre along the axis (Fig. 2(c)). As the presence of the vapor phase in the riser creates driving force for flow circlation, the inlet flow rate increases sharply and the whole of the vapor accmlated is blown ot from the system (Fig. 2(d), (e)) followed by the initial state (Fig. 2(a)). This is repeated periodically. t shold be noted that hydro-dynamics in a single channel system cople with pmp characteristics sed in the system. However, if a pmp with a bypass is sed, as is the case in this experiment, one is able to make its characteristics close to L1 P = const. Ths one might able to expect the same phenomenon in a single channel system with a bypass. The inlet flow rate corresponding to this oscillatory phenomenon is shown in Fig. 3(a), where flow oscillation with small amplitdes is cased by the periodic vapor condensation mentioned above. On the other hand, the scceeding flow oscillation with bigger amplitdes is cased by the periodic discharge of the accmlated vapor from the bottom of the downcomer to the second riser, ths the presence of the second riser plays an essential role here. t seems that this instability is very similar to the "Type " density wave instability< 9 >, which is originally observed in a vertical system with long riser pipings, in a sense that the behavior T rl.e too '-- wt r v, 40 (O) (d) 50 loosec time (O) heating ;>ower Q 3.12 kw T rl,e (b) (e} :?. 40 loosec (C) (fl Fig. 2 Observation of Type instability for first-riser-heated single channel system time lbl heating :>m er w = 2.55 kw Fig. 3 Typical record of flow oscillations for first-riser-heated single channel system - 12-

5 VoL 24, No.4 (Apr. 1987) 269 of steam bbble in the nheated riser pipe plays the dominant role to case the instability and that it is initiated as soon as the exit steam qality exceeds zero. Ths, the name "Type " instability is given to this instability hereafter. Another example of the Type instability at lower heating power and lower flow rate is given in Fig. 3(b ). Periods of flow oscillations are shown in Fig. 4. As the heating power increases the period becomes shorter and attains its shortest where dry-ot occrs. According to the athors(lo) the period of Type flow oscillation coincides with the transit time of a flid particle throgh the system. However, the transit time evalated by sing average vales of flow rate, pressre etc., proved to be arond one third of the period of oscillation (APPENDX). From the natre of the phenomenon, the transit time or the resident time of a flid particle from its entry to the system to the time it is discharged exactly coincides with the period of oscillation, becase the accmlation of steam void and its discharge constitte a part of one cycle of the oscillation. From this view point, the disagreement between the period and the transit time evalated is incomprehensible. Bt, thinking over the difficlty in evalating the 'real' transit time becase of strong nonlinearity of the flow oscillation, this discrepancy may be probably attribted to the error in evalation of the transit time. 2. Downcomer-heated Single Channel System n this case, boiling occrs first at the bottom of the heated downcomer. When heating power is low, the vapor bbbles generated are condensed and disappear as they flow pward against the direction of the liqid flow. Bt, at high heating powers, all of them do not disappear and the downcomer comes to be filled with the bbbles which eventally enter the nheated second riser. This indces the driving force of liqid flow, which reslts in an increased flow rate. As this manifests itself as the flow oscillation with big amplitdes, its mechanism is clearly the same as the Type instability described in Sec. ill-1. Typical records of flow oscillations and the periods of flow oscillations are shown in Figs. S and 6 respectively. t is fond that the periods of flow oscillation are shorter than those 100. \ '... nea t1 ng a ower 2.25 kw -, kw kw 50 0 tlme tal law flow rate condition 2,98 kw --l kw "' 100 A 23.3 ml/sec e 35.0 ml/sec 0 45,0 ml/sec ml/sec ml/sec Q lkwl Fig. 4 Period of flow oscillation for first-riser-heated single channel system lbl tlme hlgh flow rate condl tlon Fig. 5 Typical record of flow oscillation for downcomerheated single channel system -13-

6 270 J. NcL Sci Technol, given in Sec. ill-1 in the case of the first-riserheated system. Similar to the case of the firstriser-heated system, the transit time evalated reslted in arond one third of the period measred as shown in APPENDX; the explanation abot this might have been given very same to the case above Q (k\1) ml/s 25.0 ml/s e 31.7 ml/s c ml/s /s D 58.3 ml/s ml/s ml/s Fig. 6 Period of flow oscillation for downcomer-heated single channel system A stability map is shown in Fig. 7, where the reslts for experiments described in Sec. ill- 1 is also plotted. As the Type instability for both of these experiments is initiated as soon as the bbbles begin to enter the second riser, it is natral that the stability bondaries for these cases almost coincide with each other and also with a crve Xe=O as shown in the figre. 40 o--tirst-nser-heatea e -downcomer-heated Q lkwl Fig. 7 Stability map for single channel systems 3. First-riser-heated Parallel Channel System Wallis( 6 ) and Mrase< 7 ) fond that the flow rate vs. pressre loss crve at constant heating powers may have negative gradient and that the static instability may occr in the system inclding the downcomer no matter whether it is heated or not. This sitation is briefly described as follows: when flow rate is increased keeping heating power constant, the amont of vapor bbbles in the downcomer decreases. Contrary to the case for the pward two-phase flow, the pressre loss along the flow direction, then, decreases de to the decrease in static pressre ths reslting in the negative. slope on the W-L1Pcrve. The static instability cased by this mechanism is very different to that in the pward flow system, which is cased by the negative slope de to the frictional twophase flow pressre loss characteristics< 9 >. Ths it is conclded that the static instability cased by the static pressre loss characteristics in the downcomer is pecliar to a system inclding downcomers. nstabilities occrred nder this first-riserheated conditions in the parallel channel system were a one-sided flow cased by the static pressre loss effect in the downcomers. A typical record showing a transition to the one-sided flow is given in Fig. 8. The periodic flow oscillation preceding the one-sided flow is the one cased by the periodic condensation of vapor at the top of the downcomer, which was observed also in the single channel system described in Sec. ill-1. That is, the vapor bbbles generated first riser and accmlated in the top of the nheated downcomer forming a liqid level there (the sitation is the same as shown in Fig. 2(b)) condense periodically in contact with the entering slightly sbcooled liqid, which reslts in the oscillation of the liqid level as well as the flow rate. The flow oscillation occrs alternately in both channels and their oscillations of flow rates are ot-of-phase. Bt, after a few sccessive periods, flow rate in Ch. 2, in this case, decreases gradally almost to zero while the flow rate in Ch. 1 increases eventally to a vale of the total flow rate. Ths, this may appropriately be called "one

7 VoL 24, No.4 (Apr. 1987) Unmeasrable time Fig. 8 Typical record of flow oscillation for first-riser-heated parallel channel system showing transition to one-sided flow (Q = 0.99 kw) sided flow". As the flow rate in Ch. 1 increases almost to doble that of the initial vale, the vapor bbbles vanish and the flow inside is the liqid single phase. t is interesting to note that one is able to regard Ch. 1 plays as the bypass and in Ch. 2 blockage occrs, which is very similar to the observation by Aritomi( 4 ) in a system with one downcomer and a bypass. Once one-sided flow occrs, frther increase in heating power sally indces dry-ot. Bt when the heating power is imposed rapidly enogh, from a stable region by an amont large enogh to jmp ot of the one-sided flow region, the vapor bbbles generated are accmlated in. both of the channels, which are discharged from the downcomers to the nheated second riser creating driving forces alternately in these channels to yield flow oscillations as shown in Fig. 9. Ths, this oscillation may be classified into the Type instability cased by the same mechanism as those described in Sees. ill-1 and 2. A typical Type flow instability is shown in Fig. 10 where one may observe very violent flow oscillation with a phase difference of 1 between the two channels. The corresponding stability map is shown in Fig. 11. t is fond that the one-sided flow occrs as soon as sbcooled boiling starts, while the Type instability occrs at heating powers nearly doble those for one-sided flow incipience. t is presmed that this corresponds to the condition of boiling initiation even in the channel with larger flow rate, so that bbbles are generated in both channels to case driving forces when they are discharged to the risers, yielding the alternate flow oscillation in two channels, or the Type instability. (al -! ' "i,,,. ' ;: A ; ':! :-: (b) (C) (d) Channel Channe 1 li r 0: p Fig. 9 Observations of Type instability in first-riser-heated parallel channel system - 15-

8 272 J. NcL Sci. Techno, Cnanne 1 1 ;' llo ' w llo 30 Stable y / ' :z llo lllo llo Wr- WJ- w---- Heating power Q= 2.49 kw. Otpt for reversed flow shows as if it is in the normal direction de to the characteristics of the trbine flow meter. However, the reversed flow is noticed visally and it is marked by slashes. Fig. 10 Typical record of flow oscillation for first-riser-heated parallel channel system Thogh evalation of transit time from averaged qantities throgh the period of oscillation becomes more nreliable in this case, becase of the severe nonlinearity of the oscillation, the transit time evalated proved to have the same trend also as shown in the preceding sections. 4. Downcomer-heated Parallel Channel System The one-sided flow occrred also in this case and a transition to it is shown in Fig. 12. n this present case, another interesting phenomenon, reversed flow, occrred in Ch. 1, "' 10 Fig Q lkwl Type instabi 11 ty Stability map for first-riserheated parallel channel system and conseqently, the flid flows in Ch. 2 with flow rate greater than the total flow rate in liqid single phase. On the other hand, vapor bbbles are generated continosly in Ch. 1 which are accmlated in the top of the nheated first riser forming the liqid level in it. Then the flow rate of the reversed flow decreases gradally, fmally to zero, whereas the flow rate in Ch. 2 decreases to an amont as large as the total flow rate to realize the one-sided flow. n low heating power conditions, there existed a region where the flow was reversed in one channel while the flow in the other channel is in the normal direction bt the flow rate was over the total flow rate. This never trned into another mode, and we called this a "flow reversal". As the present downcomer-heated system is designated as the top-heavy system, where the flid flows with its decreasing specific weight along the axis, it is assmed that the flow easily becomes statically nstable and changes its direction to reslt in the phenomena described in Fig. 12. Then a stability map shown in Fig. 13 is obtained. Once one-sided flow occrred, frther increase in heating power indced dry-ot in the channel with zero flow rate. N. CONCLUSON The flow instabilities in these experiments are classified as shown in Table 1. The onesided flow of a kind of static instability occrs only in the parallel channel system: the insta

9 Vol 24, No. 4 (Apr. 1987) >-- = 40 > ;;:; ' :>: Stab e -r-- eversed flow -r- One-Sided t- 1T:'d'":'e ll flow r- Td 1 1 r rd,e,jr, Td.i J "' 30 ' "' 10 Ch.l Ch,2 Ch.l Ch.2 Ch.l Ch.2 ' :...J L.._...:...,J ' > Heating power Q= 1.36 kw, flow rate marked by slashes denotes reversed flow. Stable One-S!ded reversed flow 1 Q Fig. 12 / / i' / +'"/ / /One-Sided flow (kwl Fig. 13 Stability map for downcomerheated parallel channel system bility is cased by a negative slope on the W-JP crve de to the static pressre loss characteristics which is pecliar to the system with a downcomer, and once it occrs the flow rate in one channel decreases to zero. Typical record of flow instability for downcomer-heated parallel channel system Flow oscillation is cased by periodic vapor condensation at the top of the downcomer associated with periodic liqid level oscillation. n the case of the downcomer-heated single channel system its appearance is not clear be- case the definite liqid level is not formed. However, in other cases, the liqid level is formed in the downcomer as well as in the first riser and it oscillates periodically de to the periodic condensation of the vapor phase. Dynamic oscillation of the Type instability cased by the static pressre loss characteristics in the second riser is also observed. By comparing these reslts with other works as shown in Table 2, it may be conclded that instabilities in systems inclding two-phase downward flow can be classified roghly into three types: static instability, dynamic instability and slg excrsion instability in which condensation plays the important role

10 274 J. Ncl. Sci. Techno/., Table 1 Types of instabilities observed in experiments nstability type Experimental apparats Static instability!l Liqid level oscillation" Dynamic instability" Single channel system First-riser-heated X 0 0 Downcomer-heated X X 0 Parallel channel system First-riser-heated Downcomer-heated 0 0 X t 1 presents itself as one-sided flow or flow reversal. t 2 associated with periodic vapor condensation. t 3 is assmed to be the Type density wave instability. Table 2 Comparison of varios experiments on downward two-phase flow instability Reference Hayama [ Bl Mrase [7] Wallis [6] Experimental apparats D]ll Flid Water Air-l"later Static instability reversed flow in some channles Aritomi [ 3] Aritomi [4] Presdy w Water R-113 R-113 flow blockage one-sided flow reversed flow Dynamic instability Type density wave instability Ch. 1 ' 2 heated system others slg excrsion instability Ch. 1 heated liqid level oscillation [NOMENCLATURE] Ll P: Pressre drop Q: Heating power T: Temperatre : Velocity W: Flow rate x: Steam qality p: Density r: Period of flow oscillation (Sbscripts) d: Downcomer e: Exit, g : Gas phase i: nlet, /: Liqid phase r : First riser, r2: Second riser (MPa/m 2 ) (kw) (oc) (m/s) (ml/s) T: Total : Channel, : Channel 2 ACKNOWLEDGMENT The athors wold like to express their appreciations to Mr. K. Nakagawa for his helps in manfactring the test facility. -REFERENCES- ( 1) NAKAGAWA, H.: J. Jpn. Soc. Mech. Eng., ], ( 2) MSHMA, K., NSHHARA, H.: Ncl Eng. Des., 86, (1985). ( 3) ARTOM, M., et al: Japan-US Seminar on Two

11 Vol 24, No.4 (Apr. 1987) 275 Phase Flow Dynamics, (1984). ( 4) ARTOM, M., et al: Pro c. 22nd Nat/. Heat Transfer Symp. of Japan, C216, (1985). ( 5) NAKANSH, S., et al: Pro c. 1 7th Nat/. Heat Transfer Symp. of Japan, B305, (19). ( 6) WALLS, G. B., et a/.: nt. J. M/tiphase Flow, 7, 1-19 (1981). ( 7) MURASE, M., eta/.: Proc. 21st NatL Heat Transfer Symp. of Japan, 02, (1984), ( 8) HAY AMA, S.: Trans. Jpn. Soc. Mech. Eng., 32 (239] (1966). (9) FUKUDA, K., KOBOR, T.: J. NcL Sci. Techno/., 16(2], (1979). (10) FUKUDA, K., et a/.: ibid., 21[7] (1984). [APPENDX) As shown in e.g. Fig. 5, flow rate oscillation has very strong non-linear characteristics and it ranges between single phase to two phase states. Ths, in order to evalate the 'real' transit time, one shold analyze this record and reqire elaborate works, moreover, take accont of heater dynamics. n Fig. Al, period of flow oscillation is plotted against 'average' transit time which is evalated by average qantities of flow rate etc., becase of this difficlty in evalating the 'real' transit time. "' "0 "" i:; Q.. t 0 A 25.0 " <> (lo.jn(::arer heated 16.7 ml/s Jl ( t)) () H (D) (.A) (t)) '...! 75.0 First-nser-heated "" D 45.0 v ao Transit time T <sec> Symbols in parentheses correspond to data for which exit qalities are slightly below zero. Fig. Al Period of flow oscillation vs. transit time As described in the context, it is predicted that the transit time coincides with the period of oscillation, however, the transit time evalated lies arond one third of the period. The discrepancy may be attribted to the error in evalating the transit time