A parametric study on the factors affecting gas turbine combustion using a CFDbased

Transcription

1 Bulgaran Chemcal Communcatons, Volume 47, Specal Issue D (pp A parametrc study on the factors affectng gas turbne combuston usng a CFDbased approach Farhud Shrnzadeh 1, Reza Mohajer Barough, Al Atashbar Orang 3* 1 Department of Mechancal Engneerng, Scence and Research Branch, Islamc Azad Unversty, Tehran, Iran Faculty of Mechancal Engneerng, Unversty of Tabrz, Tabrz, Iran 3 Young Researchers and Elte Club, Tabrz Branch, Islamc Azad Unversty, Tabrz, Iran Receved June 6, 015; Revsed September 10, 015 Abstract: In ths paper, a numercal smulaton was conducted usng FLUENT CFD package to nvestgate njecton effects of axal and swrl flow nsde the annular combuston chamber wth wall jet. The flow was consdered as threedmensonal, steady, turbulent, ncompressble, vscd, and two-phase and turbulence models ncludng RNG (k-ε and the Reynolds stress model were appled. In order to evaluate the appled numercal method, the results obtaned n the reactng combuston chamber were compared wth expermental ones. The effects of dfferent parameters nsde the combuston chamber ncludng ar-fuel rato, swrl mass flow rate, fuel spray angle and swrl number were studed and optmal values for some of these values were obtaned. The results showed that ncreasng the swrl ar flow rate n the nlet area of the combuston chamber stretches the swrl area toward the fuel njecton nozzles and causes ntaton of combuston near the chamber nlet. The ncrease of the ar-fuel rato results n ncreasng the veloctes and unform temperature profles n the outlet chamber. Increasng swrl number and decreasng fuel spray angle also reduces wall temperature. Optmal values for Sauter mean dameter and fuel spray angle were also determned. Keywords: gas turbne combuston chamber; ar-fuel rato; fuel spray angle; swrl number 1. INTRODUCTION Gas turbne combuston chambers need hgh combuston effcency, low pollutants emssons and hgh performance. Turbulent flow feld nsde the combuston chamber plays a very mportant role n flame formaton. Another mportant factor s the way n whch fuel s njected nto the combuston chamber and ts mxng wth ar. Swrl flow has been mostly used to stablze the flame nsde the combuston chamber of gas turbnes. Combuston modelng has a fundamental role n accurate smulaton of combuston flows. Ths s, n addton to the amount of speces and combuston products, energy release rate resultng from the mxng of the fuel and oxdant s determned by t. An approprate combuston model should take nto account the effect of all these factors. In computatonal flud dynamcs (CFD, the smulaton of processes such as turbulence, combuston and radaton requres mathematcal modelng. Smulaton of these phenomena can be useful and relable f the nfrastructure model or models are accurate enough. The extensve applcablty, hgh accuracy, low computatonal cost and smplcty are four mportant specfcatons of mathematcal models. Cameron and Samuelsen [1] studed the characterstcs of temperature and velocty profles nsde a combuston chamber wth wall jet. The obtaned results ndcated that velocty and temperature profles are at sothermal and reactve condtons, respectvely. They also nvestgated the effect of swrl ar ncreasng and fuel droplet atomzaton. Sue et al. [] nvestgated numercally a gas turbne chamber ntegrated wth the dffuser. They used the KIVA-3V code to analyze the flow nsde the combuston chamber. Flow felds and temperature dstrbuton were studed along the axal and crcumferental drectons and the stable behavor of the flow n the combuston-compressor system was evaluated. Zhang et al. [3] studed the thermodynamc condtons nsde a combuston chamber consdered to be symmetrc and twodmensonal. The effects of equvalence rato, reacton temperature and swrl nsde the cylndrcal combuston chamber were nvestgated. The results obtaned ndcated that the combuston chamber geometry has a trval effect on pressure fluctuaton nsde t. Som et al. [4] studed the effect of nlet ar swrl and fuel njecton angle nsde a gas turbne To whom all correspondence should be sent: E-mal: Al.at.orang@gmal.com 015 Bulgaran Academy of Scences, Unon of Chemsts n Bulgara 31

2 F. Shrnzadeh et al.: A parametrc study on the factors affectng gas turbne combuston usng a CFD-based approach combuston chamber. The results revealed that unlke low swrl numbers, a spral swrl zone could be seen around the ntal central zone at a hgh swrl numbers. Kurosavwa et al. [5] studed the swrl flame structure nsde a gas turbne combuston chamber. They focused on hgh combuston effcency, low emssons nsde the combuston chamber. Olvan et al. [6] nvestgated the structure of reactve and non-reactve swrl flows. They found that although the overall mxng process and the man flame structure were handled by the ar flow swrl, the fuel njecton methods have an mportant effect on them at the ntal zone. Sandararaj et al. [7] studed the effects of jet njecton angle, crossflow Reynolds number and velocty rato on entranment and mxng of jet wth ncompressble cross-flow n ventur-jet mxer. Ths paper deals wth the effects of axal and swrl ar njecton and the determnaton of optmal values for effectve parameters ncludng swrl ar flow rate, expanson angle of nlet ar, dameter of the ntal fuel droplets and fuel spray angle. The results of ths work can be used for the prelmnary desgn of the arcraft gas turbne combuston chambers and lqud fuel mssles. Furthermore, the optmal values obtaned for the average dameter of prmary partcles and fuel spray angle can be used n desgn procedures.. GEOMETRICAL MODEL AND NUMERICAL METHOD The wall jet model acts at atmospherc pressure wth JP4 fuel. Combuston chamber ar s preheated to 600 C and ar flow rate s 163 kg/h. Fuel flow rate and the temperature are 3.7 kg/h and 400 C, respectvely, for an equvalence rato of 0.3. As shown n Fgure 1, ar flow s dvded nto swrl jet, ntal jet and dluted ar jet. The longtudnal four valves are bult around the chamber. Intal jets and dluted ar jets have dameter of 7 mm and 9.5 mm, 3 Fg. 1. Schematc of combuston chamber wall jet model respectvely. A stanless steel channel wth radus of 40 cm and length of 3 cm s consdered whch s blocked from 57 mm onwards. A 60 degree turnng s bult n the nlet chamber. One of the man needs of gas turbne combuston chambers s that the combuston chamber operates at every workng condton. The flow pattern s the frst factor of the chamber that affects flame stablty. The most common method s recrculaton of part of combuston products to the upstream flow and remxng of products wth fuel and fresh ar. One conventonal method nvolves the use of swrl to create a recrculaton. In ths desgn, ar nozzles surround the fuel njectors whch ncrease the shear stress and turbulence ntensty nsde the chamber. The combuston chamber mesh has been bult n GAMBIT as shown n Fgure. It s a mesh along r,,z drectons. The FLUNET 6.3 package s used for the smulatons. A segregated solver s used to solve the governng equatons and an mplct form s used for the lnearzaton of the equatons. Convectve and dffuson terms are dscretzed based on power-law and central schemes, respectvely. Pressure and velocty felds are coupled through the smple algorthm and Eddy dsspaton model s used to model the nteractons among them. Fg.. Combuston chamber mesh

3 F. Shrnzadeh et al.: A parametrc study on the factors affectng gas turbne combuston usng a CFD-based approach 3. GOVERNING EQUATIONS The Naver-Stokes equatons are the fundamental partal-dfferental equatons that descrbe the flow of fluds. Mass conservaton equaton or contnuty equaton, s wrtten as followng, t X (U S Momentum conservaton equatons, p. j ( U ( U U J g S t X X X j Knetc energy of turbulence, m j (1 M ( Dk t k G G Y K b m Dt x (3 x k Dsspaton rate of turbulent knetc energy, D t C ( G C G C 1 k 3 b Dt x (4 x K k Energy equaton, r g g g g g g T q ( c pt ( c put ( c p eff S E t x x x x (5 Dscrete phase model: Fuel spray (contnuous njecton n the combuston chamber, ncludes a lmted number of droplet categores wth a range of specfc szes. Intal droplet sze dstrbuton of lqud fuel spray s assumed to follow Rosn-Rammler dstrbuton functon defned as, exp( bd G ( d n exp( bd n mn exp( bd exp( bd n max n max (6 Speces conservaton equaton, C g g g j ( C j ( U IC j ( Deff SC S C J j t x x x (7 Eddy dsspaton model s, P CEBU YP K (8 The dmensonless number swrl used to determne the swrl appled on the flow, s defned as: 1 where G m, G t and (9 D sw are axal flux of angular momentum, axal flux of axal momentum and swrl outer dameter, respectvely. These parameters are gven n turn by, G t D sw D sw 0 S N Gm D G ru ( U dr rpdr (10 G m D sw 0 0 (11 where W, U and P are axal velocty, tangental velocty and statc pressure, respectvely. There s usually no reverse flow for S N < 0.4. The stream lnes show consderable dvergence for 0.4 < S N < 0.6. Reverse flow was observed for S N > 0.6. The followng relaton s proposed for calculatng the swrl number n one-axal swrl wth flat blades, D 1 ( D Sn 3 D 1 ( D (1 where the model factor s defned as, T Tav P. F max (13 Tav Tn here T max, T av and T n are the maxmum temperature at the outlet secton, weghted average temperature n chamber outlet, and weghted average temperature of all chamber nlets, respectvely. 4. RESULTS AND DISCUSSION Referrng to Fgure 1, swrl ar nlet, ntal jets and dluton jets are consdered as 5, 35 and 40 percent of the total ar flow rate, respectvely [1]. The results obtaned show that the onset of swrl area s n the trap area at X/R=0.75 n reactve condton. Fgures 3 and 4 llustrate the axal velocty and temperature profles n dfferent cross-sectons, respectvely. In order to detect the effects of ntal jets on the swrl area, x R 1 has been used. Regardng the total nlet ar, 50% for swrl ar, 50% for dluted ar and 0% for the prmary ar s utlzed. Accordng to Fgure 5, temperature s ntally hgh across the swrl area except near the wall because the combuston occurs at ths regon. Temperature n the downstream of flow s decreased due to ncreased SW ru ( Wr dr hub sw hub sw 3 t tan( 33

4 F. Shrnzadeh et al.: A parametrc study on the factors affectng gas turbne combuston usng a CFD-based approach Fg. 3. Axal velocty comparson wth reactve and sothermal condtons (Re=19600, S=0.36, A/F=1.5 Fg. 4. Total temperature profles n reactve condton at trap zone (Re=19600, S=0.36, A/F=1.5 penetraton of the jets and extreme enhancement of the axal velocty n the central lne. At outer areas of the combuston chamber due to the penetraton of dluted jets, the temperature profle s relatvely unform. Fgure 6 shows the varaton of radal velocty n the case of swrl ar enhancement. A 45 degree expanson s gradually nfluencng the progress of the swrl flow. The reason for ths s that a mechansm was added to the swrl created n the trap area whch affected the reverse mxture and, as can be seen n Fgure 7, the swrl occurred at x R The effect of ncreasng the fuel-ar rato from 1.5 to 3 on the axal velocty and temperature profles s shown n Fgures 8 and 9, respectvely. Ths effect s mostly evdent n the trap area at x R 0.38, 0.75 and Although, the velocty profles are smlar 34 Fg. 5. Temperature profle at reacton case to those obtaned n the prevous ar-fuel rato (A/F=1.5, veloctes become much larger. The effect of ncreasng ar-fuel rato s gradually reduced at x R.5 and 4.5. Asymmetry n the temperature profles near the nozzle at x R 0.38 s agan evdent. Although the hot strp exsts n these condtons, ts effect s much less. Temperature profles are relatvely symmetrc about the central lne. The cool core resultng from the jets penetraton appears at x R.5 and contnues through the flow up to x R 4.5. The temperature profle s relatvely unform at x R 5. The effects of swrl number on the chamber wall temperature and output temperature are shown n Fgures 10 and 11, respectvely.

5 F. Shrnzadeh et al.: A parametrc study on the factors affectng gas turbne combuston usng a CFD-based approach Fg. 6. Varaton of radal velocty wth swrl ar enhancement. Accordng to Fgure 10, the slope of the curve n trap area s caused by chemcal nteractons. The wall temperature rapdly decreased due to the prmary ar jet enterng to the chamber, and then, because of heat penetraton from the core, the temperature near the wall relatvely ncreases. Then, a large drop n wall temperature occurs due to the enterng of dluted ar followed by an ncrease n wall temperature up to the outlet. By ncreasng the swrl number through the ncrease of the swrl angle, the combuston s prevented from reachng the walls and, as depcted n Fgure 13 the temperature near the wall falls. As can be seen n Fgure 11, by ncreasng the swrl number, the outlet temperature reduces near the wall and ncreases n the central lne. The reason for ths s that by ncreasng the swrl number more combuston occurs on the central lne and t s not drawn near the wall. Accordng to Fgure 1, by ncreasng the fuel spray angle, the wall temperature ncreases due to large radal dsperson of fuel partcles and subsequent combuston of the fuel near the wall. Fg. 7. Velocty profle n the case of 45 0 gradual ncreasng. Outlet temperature varaton based on the fuel spray angle s shown n Fgure 13. As observed n Fgure 13, by ncreasng the angle of fuel spray, the fuel mxes better wth ar and most of the reacton process s occurs n the trap area, consequently, the outlet temperature falls. For ssues related to gas turbne combuston, small Sauter mean dameter of the ntal fuel droplets ncreases effcency and engne performance and reduce pollutants; on the other hand, f the droplet sze s too small, they lose the requred momentum and force to penetrate nto hgh-pressure gases. Fgure 14 depcts the Sauter mean dameter of ntal droplets wth a constant spray angle of Combuston effcency, and P.F. are calculated for several Sauter mean dameters. The mean dameter of ntal droplets of 5 has more sutable combuston effcency and P.F. compared to other dameters. As can be seen n Fgure 15, the spray angle of 80 0 results n the hghest combuston effcency and P.F. Accordng to expermental results avalable n lterature, P.F. s between 0.5 and 0.75 [8]. m Fg. 8. Axal velocty profles at three dfferent ar-fuel ratos. 35

6 F. Shrnzadeh et al.: A parametrc study on the factors affectng gas turbne combuston usng a CFD-based approach Fg. 9. Temperature profles at three dfferent ar-fuel ratos. Fg. 10. Effect of swrl number n dfferent cases of wall temperature. Fg. 1. Effect of fuel spray angle on wall temperature. Fg. 13. Effect of fuel spray angle on outlet temperature. Fg. 11. Effect of swrl number n dfferent cases of outlet temperature. 36

7 F. Shrnzadeh et al.: A parametrc study on the factors affectng gas turbne combuston usng a CFD-based approach Fg. 14. Optmzaton of Sauter mean dameter of ntal droplets. Fg. 15. Optmzaton of fuel spray angle. 5. CONCLUSIONS A numercal smulaton was performed to nvestgate the effects of njecton of axal and swrl flow nsde an annular combuston chamber wth wall jet. Accordng to the results, by ncreasng the flow rate of swrl ar enterng the combuston chamber t s possble to acheve a rapd and nearly complete mxng mode. An excessve ncrease n the swrl flow rate at the chamber nlet allows that the combuston reaches the fuel njecton nozzles. Thus, an optmum value for the swrl ar flow exsts. The effect of swrl number and fuel spray angle on wall and outlet temperatures were also nvestgated. Increasng the swrl ar flow rate n the nlet area of the combuston chamber stretches the rotatonal area toward the fuel njecton nozzles and causes the ntaton of combuston near the chamber nlet. A further ncrease of the swrl flow rate at the chamber entrance makes that the combuston reaches the fuel njecton nozzle; hence, an optmal value for swrl flow rate n the chamber entrance needs to be determned. An ncrease n the ar-fuel rato results n an ncrease n the veloctes and unform temperature profles at the outlet chamber. Moreover, ncreasng swrl number and decreasng fuel spray angle also reduces the wall temperature. Accordng to the results obtaned, by enhancng the nlet swrl ar flow rate and a sudden 45 0 expanson through varaton n the geometry of the combuston chamber, rapd and qute complete mxng can be acheved. However, due to the proxmty of the combuston to the chamber nlet and subsequent damage of the fuel njecton nozzle, optmal values need to be obtaned for the mentoned parameters. S m p j g Ṡ M G K G b Y m k Ṡ E, G( d S Cj YP NOMENCLATURE Mass added to the contnuous phase of the second dffused phase Statc pressure Stress tensor Gravtatonal force along drecton Momentum term n gas phase equaton Turbulent knetc energy Knetc energy of buoyancy Fluctuatng expanson n compressble Turbulence relatve to the overall loss rate Turbulent Prantl number Turbulent Prantl number energy term Mass fracton of spray Source term Varance of product mass fracton REFERENCES 1. C.D. Cameron, G.S. Samuelsen, ASME J. Eng. Gas Turbne Power, 111, 31 ( K. Su, C.Q., Zhou Numercal Modelng of Gas Turbne Combustor Integrated wth Dffuser.34th Natonal Heat Transfer conference Pttsburgh, C. Zhang, T. Zhao, Parametrc Effects on Combuston Instablty n a Lean Premxed Dump Combustor. AIAA ,

8 F. Shrnzadeh et al.: A parametrc study on the factors affectng gas turbne combuston usng a CFD-based approach 4. S.K. Som, A.K. Ghosh, Effects of Inlet Ar Swrl and Spray Cone Angle on Combuston and Emsson Performance of a Lqud Fuel Spray n a Gas Turbne Combustor. Dssertaton, Wrght-Patterson Ar Force Base, Dayton, Oho, July Y. Kurosavwa, S. Yoshda, T. Yamamoto, K. Shodra, M. Gom, K. Suzuk, Structure of Swrler n Gas Turbne Combustor.Techncal Report Natonal Aerospace Laboratory of Japan, A. Olvan, G. Solero, F. Cozz, A. Coghe, Exper. Thermal Flud Sc., 31, 47 ( S. Sandararaj, V. Selladura, Thermal Sc., 16, 07 ( X.R. Duan, W. Merer, P. Wegand, B.Lehmann, Lasers and Optcs, 17, 49 (