Encyclopaedia Index

Inflow boundary, definition of

(also see INLET and VALUE commands)

a) Inflow boundary conditions

An inflow boundary-condition type that is frequently used, and widely represented in the Input Library, is the specification of the mass-inflow rate by means of COVAL(patch,P1,FIXFLU,rate), followed by COVAL's for the other variables of the form: COVAL(patch, variable,ONLYMS, value).

This is often perfectly satisfactory; but it sometimes leads to distributions of inlet-velocity profile that are unexpected, in that the velocities are not exactly the same as the values which have appeared in the fourth argument of COVAL. There is of course no reason why they should be; for the source of momentum brought by the mass flow is only one of the contributors to the cell-wise momentum balance.

It is therefore worth mentioning that inflow boundary conditions can also be specified in other ways. One of these will now be illustrated for single-phase flow through the "west" boundary of the grid, as follows:-

PATCH(INFLOW,EAST,1,1,1,NY,1,NZ,1,LSTEP)
COVAL(INFLOW,P1,FIXVAL,P1IN)
COVAL(INFLOW,U1,FIXVAL,U1IN)
COVAL(INFLOW,V1,FIXVAL,V1IN)
COVAL(INFLOW,W1,FIXVAL,W1IN)
COVAL(INFLOW,H1,FIXVAL,H1IN); etc

Here it should be noted that the mass-flow rate is being specified indirectly, by way of the value of the velocity U1IN which will prevail at the east side of the first column of cells, and by way of the pressure P1IN and enthalpy H1IN which will dictate (it is here supposed, by way of example) the density which is associated with the velocity U1IN.

This boundary-condition prescription takes firmer hold of the values pertaining to the first column of cells than does the one which makes use of ONLYMS; it is therefore sometimes preferable. An example of this is provided in the following item.

(b) Inflow conditions at the circumferential boundary of a cylindrical-polar grid

The specification of the momentum resolutes to be convected into a polar grid (CARTES=F) at its circumferential boundary necessitates the resolution of the external flow vector into the local directions of u and v. These directions change from cell to cell with increasing angle (ie increasing IX).

One approach is to introduce a PATCH for each cell in which the velocity resolutes are set in the 4th argument of COVAL according to the angle between the flow direction and the grid direction. A more economical solution is to provide a GROUND sequence which sets all resolutes for a single PATCH that covers the entire zone of inflow.

Such sequences are provided in subroutine GXPOLR called from Group 13 of GREX3. RSG22 sets the magnitude of the exterior velocity which must be aligned with the radial direction at x = 0. GXPOLR is activated separately for u and v by having two PATCHs named UPOL and VPOL. For example, for inflow over the semi-circle extending from NX/4 to 3*NX/4, the following PIL instructions are required:

PATCH (UPOL,NORTH,NX/4-1,3*NX/4,NY,NY,1,1,1,1)
COVAL (UPOL,U1,FIXVAL,GRND1)
PATCH (VPOL,NORTH,NX/4,3*NX/4,NY,NY,1,1,1,1)
COVAL (VPOL,VI,FIXVAL,GRND1)

The need for two patches of slightly differing extents for u and v is required for symmetry.

Because the cells at NY from NX/4 to 3*NX/4 are isolated from the rest of the domain (the values at all the edges being fixed), it is necessary to fix the pressures in these cells by,

COVAL(VPOL,PI,FIXVAL,0.0)

to prevent the pressures taking arbitrary values.

The values u and v are fixed because this is the only exact representation. It avoids the errors that result from a mass flow into a main cell that brings in an azimuthal momentum which is distributed to the east-face and the west-face velocities that differ in alignment by the angle between the two faces.

Examples of the use of this option are provided in the Input Library in Section 4.6 entitled 'Cross Flow in a Cylindrical- Polar Grid'.


Inflow, setting of

(see Inflow boundary, definition of and INFLO)