Latent Heat Recovery from Oxygen-Combustion Boiler

Transcription

1 atent eat Recovery rom Oxygen-ombustion Boiler Masahiro OSAKABE*, Sachiyo ORIKI, ugue IO okyo University o Mercantile Marine, Koutou-ku, okyo 5-85, Japan Kunihiko MOURI Electric Power Development o., td., huo-ku, okyo 4-865, Japan *corresponding author; Phone & AX , osakabe ipc.tosho-u.ac.jp he thermal hydraulic behavior was experimentally studied in a heat exchanger or the latent heat recovery rom a lue gas exhausted rom the oxygen-combustion boiler. he heat exchanger consisted o staggered banks o rows and 6 stages o spirally inned tubes. he parametric study varying the lue gas and eed water low rate was conducted. he temperature distributions o cooling water and lue gas, the pressure loss and the amount o condensate were measured. Based on the previous basic studies, a thermal hydraulic prediction method was also proposed. or the condensation o steam on heat transer tubes, the modiied Sherwood number taking account o the mass absorption eect on the wall was proposed. he experimental results agreed well with the prediction proposed in this study. he total heat o 9k including the latent heat o 7k was successully recovered in the present economizer specially designed or the oxygen-combustion boiler Keyword: atent eat Recovery, Oxygen-ombustion, Boiler, hermal-ydraulic Prediction Introduction he most part o energy losses in a boiler is due to the heat released by the exhaust lue gas to atmosphere. he released heat consists o sensible and latent one. Recently, or a biological and environmental saety, a clean uel such as a natural gas is widely used in the boiler. As the clean uel includes a lot o hydrogen instead o carbon, the exhaust lue gas includes a lot o steam accompanying with the latent heat. As a next generation boiler, the oxygen combustion boiler was planned and developed in Japan. o reduce the total amount o exhaust lue gas and the toxic products generated with the combustion, it is preerable to use oxygen instead o air. Using the oxygen also can burn low-caloric gas. he oxygen-combustion lue gas has the larger amount o steam concentration than that in the air-combustion lue gas including nitrogen. So the latent heat recovery rom the lue gas is very important to improve the boiler eiciency. Shown in ig. is the relation between the boiler eiciency and the exhaust gas temperature in the oxygen and air combustion system. he eiciency is larger than % at the lower exhaust temperature as the boiler eiciency is deined with a lower heating value o A-heavy uel oil. he eiciency in the oxygen combustion is much higher than that in the air combustion due to the lack o heat loss carried out with the nitrogen. he dew point o the oxygen-combustion lue gas including the larger amount o steam instead o nitrogen is higher than that in the air-combustion lue gas. he steep increase o eiciency can be observed at the lower temperature region below the dew points. In this lower temperature region, the latent heat in the lue gas is recovered and the eiciency is increased. So the latent heat recovery rom the lue gas is very important to improve the boiler eiciency in the oxygen combustion system. Boiler eiciency (% A-oil Oxygen combustion µ.5 Air combustion µ. Dew point Dew point 5 Exhaust gas temp. ex ( ig. Boiler eiciency in oxygen and air combustion In the previous basic studies -4, condensation heat transer on horizontal stainless steel tubes has been investigated experimentally by using an actual lue gas rom a natural gas boiler. he experiments were conducted using single and stages o tubes at dierent air ratios and steam mass concentrations o the lue gas in a wide range o tube wall temperature. he condensation heat transer was well predicted with the simple analogy correlation in the high wall temperature region. In the low wall temperature region less than or the high steam mass concentration presuming the oxygen combustion, the total heat transer was higher than that predicted by the simple analogy correlation.

2 he thermal hydraulic behavior was experimentally studied in a heat exchanger or the latent heat recovery rom a lue gas generated with the combustion o oxygen and A-oil. he steam mass concentration o the lue gas was approximately 5% which value was higher than that o an air-combustion lue gas. he parametric study varying the lue gas and eed water low rate was conducted. Based on the previous basic studies, a prediction method or the heat exchanger was proposed 5. In the prediction, the lue gas was treated as a mixture o O, O, O, SO and O, and the one-dimensional heat and mass balance calculation along the low direction o lue gas was conducted. he heat and mass transer on tubes was evaluated with a modiied analogy correlation proposed in the previous basic study 4. he modiication was proposed to take account o the mass absorption eect at wall in the high steam mass concentration. he ollowing points are the diiculty associated with the calculation method and considered to be important in the veriication comparing with the present experimental result.. Availability o the proposed analogy correlation between heat and mass transer.. Proper estimation or physical properties and diusivity o actual lue gas.. Precise calculation or the white uming when the gas temperature merges with the saturation temperature. 4. Eect o condensate on heat transer and pressure loss. 5. Stage by stage calculation method instead o conventional method using a logarithmic dierence. Experimental Apparatus and Method Shown in ig. is a schematic o experimental apparatus. he heat exchanger consisted o staggered banks o rows and 6 stages o spirally inned tubes as shown in ig.. he outer and inner diameters o the base tube installed with the ins are. and 5.4mm, respectively. he thickness and length o the plate in are and 8mm, respectively. o avoid the degradation o in eiciency at the condensing region, the in height was kept as small as possible. he lue gas rom an oxygen combustion boiler is led to the upper plenum and lows downward in the irst heat exchanger and upward in the second one. he lue gas was released to atmosphere rom the outlet plenum. Shown in ig.4 is the arrangement o heat transer tubes. he heat exchanger consists o a staggered bank o row tubes and 7 stages in total. he tubes at each stage are connected with a header to maintain the same low rate o eed water. he eed water is supplied at the downstream o gas low and lows counter-currently to the upstream. he temperature distributions o water and lue gas in the heat exchanger were measured with sheathed thermocouples. he thermocouple signals were transerred to a personal computer with a GPIB line and analyzed. he measurement error o the temperature in this study was within ±. K. he pressure loss and the total amount o condensate generated in the heat exchanger were also measured. he parametric study varying the lue gas low rate, eed water temperature and low rate was conducted. he major test conditions are shown in able.. he oxygen supplied to the boiler was produced wit PSA (pressure swing adsorption method and the purity was approximately 97 % due to the inclusion o Argon gas. he oxygen ratio was maintained at approximately.5 and the lue gas low rate was varied with changing the uel low rate in the boiler. Gas st X 67 nd X ig. Schematic o experimental apparatus in ater in t 94.5 S 7.47 Base tube d i. ig. Schematic o inned tubes rows 67 8 stages 8 stages ater 8 d ig. 4 Arrangement o heat transer tubes 55.5

3 able Major test conditions est no. 4 oad (% 8 5 Oxygen ratio Oxygen purity (% uel low rate (kg/h Gas inlet temp. ( eed water (kg/h ater inlet temp. ( able Properties o A-heavy oil uel raction(kg/kg arbon.86 ydrogen. Sulur S.6 onstitutive Equations or Prediction uel combustion he properties o A-heavy oil uel used in the test boiler are shown in able.. Volumetric concentrations o O, SO, O and O in the dry gas were measured by a gas analyzer. By using the mass low rate o the uel V, the volumetric low rate o the carbon dioxide and monoxide, V OX, is VOX V.4 ( he volumetric low rate o dry gas V d is, V OX V d ( O + O he molar raction o the hydrogen corresponding to O X o mol can be calculated as, R mol By using the measured concentration o O, SO, O and O, the air ratio µ is calculated as O 5. O µ + ( ( + R / 4( O + O + SO he volumetric raction o O in the lue gas can be estimated as ( O + O R / O (4 + ( O + O R / he volumetric low rate o wet gas, V wt, is V wt V d (5 O hen the lue gas temperature is, the steam mass concentration per unit volume o lue gas is, O ( he steam mass concentration per unit mass o lue gas is w O 8 + ( O(O O O 8 + SO 64 + O eat and mass transer in gas side he total heat lux q consists o the convection heat lux q V and the condensation heat lux q as q (q V + q (A B + A η (8 where η is the in eiciency. he convection heat lux is expressed as q V h V ( w (9 he condensation heat lux q can be expressed as, q h ( w ( where w is the mass concentration o saturated steam at the wall temperature w o base tube. he ollowing empirical correlation 6 is available in the range o Re Nu j Re Pr ( where.5 d + j 5 d (.5.5 Re (.5 /S.5.65e + (4 5.7 (5 he in eiciency can be calculated by, (7 d + η Y.45 ln (Y + (6 d where Y X (.7 +.X (7 tanh(mb X mb (8.5 h m λ t (9 b + t / ( he in eiciency strongly depends on the heat transer coeicient, h, in Eq.(9. hese correlations were obtained rom the single-phase experiment where the heat lux is a simple multiple o the constant heat transer coeicient and the temperature dierence between wall and luid. Although the condensation heat lux can not be expressed with the above simple relation, it is considered that the Eq.(9 with the larger heat transer coeicient taking account o the condensation heat transer would give an approximation. As a irst attempt, the ollowing equivalent heat transer coeicient in the condensation region was used.

4 h ( h h V + β ( where β in the present calculation. eat lux Base Dew point in ig. 5 Approximation o heat lux distribution Shown in ig.5 is a schematic image o heat lux distribution on a in. he solid line depicts the heat lux distribution when the condensation takes place at the lower part o in. As the dew point locates at the middle o in, the heat lux sharply increases toward the base o in rom the dew point. he dash line is the calculated result with the equivalent heat transer coeicient o Eq.(. he heat lux at the in base is the same as the actual heat lux shown by solid line and the heat lux gradually decreases toward the in tip without the eect o dew point. Generally the in eiciency evaluated with Eq.( yields slightly the higher value than the actual one due to the higher heat transer coeicient including the condensation eect even in the dry region. owever, at the white uming condition when the saturation temperature merges with the gas temperature, the dew point never locate on the in and the error due to the use o Eq.( becomes smaller. or the condensation o steam on heat transer tubes, the modiied Sherwood number taking account o the mass absorption eect on the wall was proposed 4.. w Sh i jre Sc. ( w w i lue gas was treated as a mixture o O, SO, O, O and O and its property was estimated with special combinations o each gas property proposed by the previous studies 7. or example, the heat conductivity and the viscosity were estimated with the methods by indsay&bromley 8 and ilke 9, respectively. It is considered that a strong correlation exists between the thermal and mass diusivities. As a. irst attempt, the mass diusivity o steam in lue gas was estimated with the well-known mass diusivity o steam in air as κ D D air κ air ( where κ and κ air are the thermal diusivities o lue gas and air, respectively. he diusivity o steam in air can be expressed as, + D air / (. 6. (4 P he one-dimensional heat and mass balance calculation along the low direction o lue gas was conducted. he steam mass concentration and the lue gas temperature at N+th stage can be calculated rom those at Nth stage as shown in ig.6. Nth stage tubes (N (N+ O (N onvection ondensation O (N+ ig. 6 One-dimensional calculation he heat and mass balance equations are; q A.4 O(N V 8 O(N + (5 q A.4 V 8 q A V (N + (N P ρ (7.5 + / 7.5 V (6 where A w A B +A η. It is possible that the gas temperature merges with the dew point which is the saturation temperature corresponding to the partial pressure o steam in the lue gas. hen the gas temperature decreases below the dew point, the condensation o steam in the lue gas takes place and the latent heat increases the gas temperature until the gas temperature coincides with the dew point. In this case, the energy balance gives the relation between the increase o the gas temperature,, and the decrease o steam concentration, O, as; 8.4 P ρ (7.5 + / 7.5 O (7 eat conduction in tube he heat conductivity or the inconel or austenite stainless steel is given with the ollowing approximate 4

5 correlation. λ t. +. t /(m K (8 where t is the average temperature o tube as, + i t (9 where w and wi are the outer and inner wall temperatures, respectively. he heat lux at the outer wall is, t ( w wi q w λ ( d ln(d / d i eat transer in water side eat transer correlation by Dittus-Boelter taking account o the pipe inlet region is used. he coeicient by McAdams is used or the modiication..8.4 d. 7 Nu. Re Pr + ( i where is the heating length o tube. Pressure loss calculation he pressure loss per a stage o tube is, ( P ρ u ( where is the riction actor. or the staggered bank o inned tube, the empirical correlation by ESOA 6 is,.5 6 where 4 d ( d Re (4.7( /S. S..5 (5 d.5n.s /S r. +.8.e e.5n r.6s /S.7.8e e (6 omparison o Experimental Result and Prediction emperature distribution Shown in ig.7 is the comparison o the experimental result and the prediction at % load. he solid lines are the temperatures o gas and water in the bank. he a-dot-dashed line and the two-dots-dashed line are the inner and outer wall temperature o base tubes, respectively. he dashed line is the saturation temperature (dew point corresponding to the partial pressure o steam in the lue gas. As the outer wall temperature is smaller than the dew point, the condensation on the wall takes place throughout the heat exchanger. he dew point decreases with increasing stages as the steam concentration decreases. he key O and are the measured temperatures o gas and water, respectively. he prediction or the water temperature agrees well with the experimental result. he prediction or the gas temperature is slightly higher than the experimental result in the second and ourth positions rom the top. In these experiments o the high uel low rate, the entrained and dispersed condensate in the lue gas tends to wet the thermocouples among the tube bank. he wet thermocouples indicate the lower value than the actual gas temperature. he average thickness o condensate was calculated with the method shown in APPENDIX. It was assumed that all the generated condensate lows on the tubes in the calculation. he maximum thickness was approximately.9mm. he existence o the ilm did not aect the calculated temperature proiles in the bank. he eect o condensate ilm on the tubes was considered to be negligibly small or the heat and mass transer calculation. emperature ater temp. % oad test eed water 98 kg/h A-oil uel 9 kg/h O ratio Stages Sat. temp. Outer wall Inner wall Gas temp. ig. 7 omparison o the experimental result and the prediction at % load Shown in igs.8 to are comparisons o the experimental result and predictions when the uel and eed water low rate were reduced. hen the lue gas low rate and the amount o condensate are comparatively small, the eect o the wet thermocouple is considered to be negligible. he measured gas temperature agrees well with the prediction in the 5 and % load tests. his result also indicates the precise calculation or the white uming condition when the saturation and gas temperature merge. 5

6 emperature ater temp. Sat. temp. Outer wall Inner wall 8% oad test eed water 948 kg/h A-oil uel 8 kg/h O ratio Stages Gas temp. ig. 8 omparison o the experimental result and the prediction at 8% load emperature ater temp. 5% oad test eed water 6 kg/h A-oil uel 54 kg/h O ratio Stages Sat. temp. Outer wall Inner wall Gas temp. ig. 9 omparison o the experimental result and the prediction at 5% load Pressure loss and amount o condensate Shown in ig. is the comparison o experimental result and prediction or the pressure loss throughout the heat exchanger. he previous data 5 using air-combustion lue gas was also compared in the igure. he previous data was obtained in a bank o inned tubes where the in space S was 9mm and the in height was mm. As the empirical correlation obtained in the non-condensing region was used in the present one-dimensional calculation, the experimental results or the banks o inned tubes are slightly larger than the prediction. It is possible that the condensate between the ins increases the pressure loss in the bank. emperature ater temp. Sat. temp. Outer wall Inner wall % oad test eed water 59 kg/h A-oil uel 9 kg/h O ratio Stages Gas temp. ig. omparison o the experimental result and the prediction at % load Predicted p (mmaq 5 4 Oxygen combustion S 7.47, 8mm Air combustion S 9, mm 4 5 Measured p (mmaq ig. Pressure dierence between inlet and outlet o heat exchanger or the accurate prediction o pressure loss, the eect o condensate ilm on the tubes should be considered. Shown in ig. is the predicted and measured amount o condensate at each heat exchanger. he amount o condensate generated in the irst heat exchanger was larger than that in the second one. he prediction slightly overestimates the generated amount at the second exchanger and agrees well with those at the irst one. Shown in ig. is the comparison o experimental result and prediction or the amount o condensate throughout the heat exchanger. he total amount o 6

7 condensate generally agrees well with the one-dimensional mass and heat balance calculation proposed in the present study. he prediction indicates that the amount o condensate generated due to the white uming was approximately % o the total. Predicted m c (kg/h 5 st nd 5 Measured m c (kg/h ig. Predicted and measured amount o condensate at each exchanger..5 Predicted m c (kg/hx...5. Measured m c (kg/h X ig. Predicted and measured total amount o condensate onclusion hermal-hydraulic behavior o heat exchangers or the latent heat recovery was investigated experimentally by using an actual lue gas rom an oxygen-combustion boiler. he parametric study varying the lue gas low rate, eed water temperature and low rate was conducted. Based on the previous basic studies, a prediction method or the heat exchanger was proposed. In the prediction, the lue gas was treated as a mixture o O, O, O, SO and O, and the one-dimensional heat and mass balance calculation along the low direction o lue gas was conducted. he heat and mass transer on tubes was evaluated with the modiied analogy correlation. hen the gas temperature decreased below the dew point, the condensation o steam in the lue gas took place and the released latent heat increased the gas temperature until the gas temperature coincided with the dew point. he eect o condensate ilm on the tubes was considered to be negligibly small or the heat transer and pressure loss calculation. he experimental results or the temperature distributions o water and lue gas in the test heat exchangers with inned tubes agreed well with the prediction. Acknowledgment he authors appreciate the helpul supports by Kawasaki hermal Engineering o. td., he Japan Society o Industrial Machinery Manuacturers and NEDO (New Energy and Industrial echnology Development Organization o Japan. Nomenclature A B : heat transer area o base tube [m ] A : heat transer area o in [m ] : mass concentration per luid o a unit volume [kg/m ] p : speciic heat [J/kg] d: outer diameter o base tube [m] d i : inner diameter o base tube [m] D: mass diusivity [m /s] h V : heat transer coeicient [/(m K] h : mass transer coeicient [m/s] : in height [m] : latent heat [J/kg] Nu: Nusselt number [ h d V / λ ] Nr: total number o stages P: pressure [Pa] q: heat lux [k/m ] Pr: Prandtl number [ ν / κ ] Re: Reynolds number [ ud / ν ] S : spanwise pitch [m] S : low-directional pitch [m] S : in space [m] Sh: Sherwood number [ h c d/ D] Sc: Schmidt number [ ν / D ] : temperature [ ] t: in thickness [m] u: velocity at minimum low area [m/s] V: volumetric low rate [m N /s] w: steam mass concentration [kg/kg] κ: thermal diusivity [ λ /( ρ P ] λ: heat conductivity [/(mk] µ: oxygen ratio ν: kinematic viscosity [m /s] ρ: density [kg/m ] 7

8 subscript a: atmosphere : condensation OX: carbon dioxide and monoxide, d: dry gas : uel or in : lue gas i: condensation surace V: convection : outer wall o base tube N: standard condition at and atmospheric pressure sat: saturated condition o steam sub: subcooling wt: wet gas Reerences Osakabe, M., Ishida, K., Yagi, K., Itoh,. and Ohmasa, M., ondensation heat transer on tubes in actual lue gas (Experiment using lue gas at dierent air ratios, (in Japanese, rans. o JSME, 64-66, B, pp.78-8, 998 Osakabe, M., Yagi, K., Itoh,. and Ohmasa, M., ondensation heat transer on tubes in actual lue gas (Parametric study or condensation behavior, (in Japanese, rans. o JSME, 64-66, B, pp.78-8, 999 Osakabe, M., Itoh,. and Ohmasa, M., ondensation heat transer on spirally inned tubes in actual lue gas, (in Japanese, J. o MESJ, 5-4, pp.6-67, 4 Osakabe, M., Itoh,. and Yagi, K., ondensation heat transer o actual lue gas on horizontal tubes, Proc. o 5th ASME/JSME Joint hermal Eng. on., AJE99-697, Osakabe, M., hermal-hydraulic behavior and prediction o heat exchanger or latent heat recovery o exhaust lue gas, Proc. o ASME, D-Vol.64-, pp.4-5, ESOA INUBE ORPORAION, SOIDIN., JSME, Data Book: eat ranser rd Edition, (in Japanese, 98 8 indsay, A.. and Bromley.A., hermal conductivity o gas mixtures, Indust. Engng. hem., 4, pp.58-5, 95 9 ilke,.r., A viscosity equation or gas mixture, J. hem. Phys., 8, pp.57-59, 95 ujii,., Kato, Y. and Mihara, K., Expressions o transport and thermodynamic properties o air, steam and water, Univ. Kyushu Research Institute o Industrial Science Rep.66, pp.8-95, 977 Osakabe, M., hermal-hydraulic study o integrated steam generator in PR, J. Nucl. Sci. & echnol., 6(, pp.86-94, 989 McAdams,.., eat transmission, McGRA-I, 954 Jakob, M., eat transer and low resistance in cross low o gases over tube banks, rans. ASME, 6, pp.8-9, 98 Appendix hough a part o condensate alls down between the tubes and on the duct wall, it is assumed that all the condensate generated at the upper stage lows on the tubes as a laminar ilm. he momentum balance dominated by viscous and gravity orce gives the velocity distribution at θ rom the tube top: u ( ρ ρg gsinθ y yδ µ (7 Integrating the above velocity proile and using the condensate mass low rate per unit o tube length, m, yields 5. µ m δ ρ( ρ ρg gsinθ he heat conductivity o ilm is / ( λ λ g ρ ρ ρg sinθ K δ 5. µ m he average conductivity rom θ to θπ is ( π λ ρ ρ ρ K Kdθ 7. π µ m G g / / (8 (9 (4 he average heat resistance o ilm is deined as the inverse o the above average conductivity. he average ilm thickness is δ λ K (4 In the calculation, the mass low rate, m, at a certain stage includes the condensate generated at the stage or the conservative estimation. ube ig. 4 eat conductance o condensate ilm g y 8