You are on page 1of 15

Numerical Methods

AE 305 HW-2

Group 11 28.11.2012 Ali Yldrm-1747146 Durukan Tamkan-1747070 Elif Mehtap PELIT-1747039

INTRODUCTION

The homework problem is about visualizing the temperature distribution over a turbine which is divided into grids as:

Unsteady temperature distribution of the body is described by its heat transfer equation as:

The heat transfer equation is a partial differential equation and to solve this partial expression finite volume method is employed to understand temperature distribution.

It is asked to find unsteady heat distribution of a turbine blade which is proposed to a hot environment due to hot gases in the combustor. Some properties stated in the problem given as:

The convective heat flux on the blade surface and in the cooling hole surfaces are given by:

Where the chord of turbine blade L=10cm, k is the thermal conductivity;

THEORY
This method is used to calculate the values of the conserved variables averaged across the volume. One advantage of the finite volume method over finite difference methods is that it does not require a structured mesh (although a structured mesh can also be used. the values of the conserved variables are located within the volume element, and not at nodes or surfaces. Finite volume methods are especially powerful on coarse nonuniform grids.) Finite difference method which is mainly based on the volume and flux integrals works in such a way and can be formulated as follows: volume integrals in a partial differential are converted to surface integrals by using divergence theorem and the terms obtained from integration become flux which enters to one cell after leaving the cell in its neighbourhood. Since the flux entering a given volume is identical to that leaving the adjacent volume, the method based on the conservation laws. Conservation law( in integral form):

Conservation law( in differential form) is:

Divergence theorem is

for Nth boundary. Where:

gives the fluxes at cell boundaries. It is assumed that constant temperature distribution for each triangular cell. Solution domain and unstructured domain is needed. Sample of the solution domain is as follows: Tbc(1)= 1200 Tbc(2)= 200 Tbc(3)= 200 Tbc(4)= 200

For the solution flux terms are based on cell averaged variables and the FORTRAN program is written as follows:
c..mxc : Max number of cells c..mxn : Max number of nodes parameter (mxc=5001,mxn=3001) common /grid/ ncell,nnode,node(3,mxc),neigh(3,mxc), > xy(2,mxn),area(mxc)

common /var/ time,dt,Tcell(mxc),Tbc(10) common /grad/ dTdx(mxc),dTdy(mxc) data mxstep,iostep/7000,1000/ dt,delTallow/0.1,0.01/ c..Read the input data and initialize the solution call INIT

c..Start the solution loop delTmax = 1. nstep = 0 DO WHILE (nstep .lt. mxstep .and. delTmax .gt. delTallow) nstep = nstep + 1 time = time + dt c..Evaluate temperature gradients for each cell call GRADIENT c..Sweep all the cells and solve for T^n+1 delTmax = 0. do n = 1,ncell c..Evaluate temperature change for the cell delT = dt/area(n) * FLUX(n)

Tcell(n) = Tcell(n) - delT delTmax = max(abs(delT), delTmax) enddo print*, ' Nstep, Time, DelTmax :',nstep,time,delTmax c..Output the intermediate solutions if( mod(nstep,iostep) .eq. 0 .and. nstep .ne. mxstep ) > call TECout(nstep)

ENDDO c..Output the final solution call TECout(nstep) stop 'FINE' end subroutine INIT parameter (mxc=5001,mxn=3001) common /grid/ ncell,nnode,node(3,mxc),neigh(3,mxc), > xy(2,mxn),area(mxc)

common /var/ time,dt,Tcell(mxc),Tbc(10) character fn*16 logical ok

c..Read the grid data fn='tblade.dat' inquire(FILE=fn,EXIST=ok) if( .not. ok ) then print*, ' ', fn, ' does not exist! \n\n' stop endif print*, ' Reading ',fn open(5,file=fn,form='formatted') read(5,*) ncell,nnode read(5,*) (no,(xy(i,n),i=1,2),n=1,nnode) read(5,*) (no,(node(i,n),i=1,3),(neigh(i,n),i=1,3),n=1,ncell) close(5) print*, ' # of cells :',ncell print*, ' # of nodes :',nnode c..Compute cell areas do n = 1,ncell n1 = node(1,n) n2 = node(2,n) n3 = node(3,n) area(n) = 0.5*((xy(1,n2)-xy(1,n1))*(xy(2,n3)-xy(2,n1)) > (xy(2,n2)-xy(2,n1))*(xy(1,n3)-xy(1,n1)) ) enddo c..Set Initial and Boundary Conditions Tbc(1)= 1200. Tbc(2)= 200. Tbc(3)= 200. Tbc(4)= 200. c..Initialize the solution Tic = 25. do n =1,ncell Tcell(n) = Tic

enddo call TECout(0)

return end subroutine GRADIENT parameter (mxc=5001,mxn=3001) common /grid/ ncell,nnode,node(3,mxc),neigh(3,mxc), > xy(2,mxn),area(mxc)

common /var/ time,dt,Tcell(mxc),Tbc(10) common /grad/ dTdx(mxc),dTdy(mxc) DO n = 1,ncell dTdx(n) = 0. dTdy(n) = 0. do nf = 1,3 n1 = node(nf,n) if(nf .lt. 3) then n2=node(nf+1,n) else n2=node(1,n) endif dx = xy(1,n2)-xy(1,n1) dy = xy(2,n2)-xy(2,n1) ne = neigh(nf,n) if(ne .gt. 0) then Tneigh = Tcell(ne) else !..walls !..real neighbor

Tneigh = Tbc(-ne) endif Tface = 0.5*(Tcell(n)+Tneigh) dTdx(n) = dTdx(n) + Tface*dy dTdy(n) = dTdy(n) - Tface*dx enddo dTdx(n) = dTdx(n)/area(n)

dTdy(n) = dTdy(n)/area(n) ENDDO return end function FLUX(n) parameter (mxc=5001,mxn=3001) common /grid/ ncell,nnode,node(3,mxc),neigh(3,mxc), > xy(2,mxn),area(mxc)

common /var/ time,dt,Tcell(mxc),Tbc(10) common /grad/ dTdx(mxc),dTdy(mxc) data alpha /22.5E-6/ FLUX = 0. c..Sum surface fluxes over the cell faces do nf = 1,3 n1 = node(nf,n) if(nf .lt. 3) then n2=node(nf+1,n) else n2=node(1,n) endif dx = xy(1,n2)-xy(1,n1) dy = xy(2,n2)-xy(2,n1) ne = neigh(nf,n) if(ne .gt. 0) then !..real neighbor

flux_x = (dTdx(n)+dTdx(ne))*0.5 flux_y = (dTdy(n)+dTdy(ne))*0.5 else !..walls

flux_x = dTdx(n)*0.5 flux_y = dTdy(n)*0.5 endif FLUX = FLUX + (flux_x*dy - flux_y*dx) enddo FLUX = -alpha*FLUX return

end subroutine TECout(nstep) parameter (mxc=5001,mxn=3001) common /grid/ ncell,nnode,node(3,mxc),neigh(3,mxc), > xy(2,mxn),area(mxc)

common /var/ time,dt,Tcell(mxc),Tbc(10) real Tnode(mxn) character fname*32,string*8,ext*5 c..Evaluate average temperatures at nodes call TEMPNODE(Tnode) c..Set the output file name write(string,'(f8.5)') float(nstep)/100000 read(string,'(3x,a5)') ext fname = 'temp-'//ext//'.tec' c..Output the solution and the grid in TECPLOT format open(5,file=fname, form='formatted') write(5,100) nnode,ncell write(5,101) (xy(1,n),xy(2,n),Tnode(n),n=1,nnode) write(5,102) (node(1,n),node(2,n),node(3,n),n=1,ncell) close(5) 100 format (' VARIABLES= "X", "Y", "TEMPERATURE"'/, > ' ZONE N=', I6,' E=', I6,' F=FEPOINT ',' ET=triangle' )

101 format (3(1x,e12.5)) 102 format (3(1x,i6)) return end subroutine TEMPNODE(Tnode) c..Evaluate node temperatures by averaging the cell temperatures parameter (mxc=5001,mxn=3001) common /grid/ ncell,nnode,node(3,mxc),neigh(3,mxc), > xy(2,mxn),area(mxc)

common /var/ time,dt,Tcell(mxc),Tbc(10) real Tnode(mxn) integer npass(mxn)

do n=1,nnode Tnode(n) = 0. npass(n) = 0 enddo c..Find the contribution of cells to the node temperatures do n=1,ncell do nf=1,3 nn = node(nf,n) Tnode(nn)=Tnode(nn)+Tcell(n) npass(nn)=npass(nn)+1 enddo enddo c..Average the total node temperature with # of contributing cells do n=1,nnode Tnode(n)=Tnode(n)/npass(n) enddo return end

RESULTS & DISCUSSION

With one cooling hole We plotted the graph while our mx step 7000 and iostep 1000. This means we will get 7 temperature distributions with one initial state situation. As we can see blade cooling hole is not enough to provide keep all blade cool. While time is increasing, blade is getting hotter from starting from aft. Nevertheless we can say cooling hole is enough to keep their neighbor environment cool.

It can be observed with considering the temperature scale that due to hot gases turbine blades gets hotter and firstly blade tips are effected.

When a second cooling hole is added temperature distributions are as follows: We changed the given blade.d file to put second cooling hole as follows. First hole has been kept constant to be able to add a second cooling hole onto turbine blade. Secondly we changed the first holes place on coordinate system by changing x and y points of first hole. By adding all these information into blade.d we obtained this illustration as shown in below.

With the second hole temperature distrubutions seems more uniform and the blade affected from hot gases less when the temperature scale is considered. Especially on the middle part of turbine blade temperature seems low if we compare with one cooling hole. Therefore, we can briefly say second cooling hole is usefull to keep temperature less in middle section.

Steady State Temperature distribution

After many step solution steady-state condition has been reached. Then the temperature distrubition keeps its stability over the turbine blade surface. In steady state temperature distrubition turbine blades seem in their neutral situation. In this case we couldnt observe too much difference between one hole and double hole blades.

REFERENCES http://www.mathematik.uni-dortmund.de/~kuzmin/cfdintro/lecture5.pdf Lecture notes

You might also like